Annex 1 Misses the Mark–Expanded Version

A previously published article presented difficulties with the revised European guidelines on sterile manufacturing (1).  The authors included a brief summary of the comments developed on the draft document.  This article expands upon that summary, outlines the authors’ rationale, and highlights the most difficult aspects of the revision draft.

The December 2017 revision of EudraLex, Volume 4, EU Guidelines to Good Manufacturing Practice Medicinal Products for Human and Veterinary Use, Annex 1, Manufacture of Sterile Medicinal Products (2), when finalized, will be legally binding. Non-compliance could result in regulatory action, refusal, or revocation of marketing authorizations, and civil and criminal penalties. Given the revised Annex 1’s legal standing, it is vital that the document delineates appropriate regulatory expectations and accepts contemporary manufacturing and control procedures and practices.

The European Medicines Agency (EMA) has put forward Annex 1 for comment, but the document is so badly written that understanding it, much less commenting on it, is far more difficult than it should be. As it stands, this version of Annex 1 will inevitably generate thousands of comments that will be difficult for EMA to reconcile within the framework of the current document. Preliminary review of the document indicates that it is poorly worded, ambiguous, and technically incorrect in many areas. The following comments represent a fraction of those identified during the authors’ review. They are categorized by general areas of concern.

General Comments

1. The document is inconsistent in its use of references; they are present in some sections and absent in others. The document should include a comprehensive set of reference documents to support the points made in the text. The inclusion of appropriate references is especially important for those aspects of the draft that are most contentious.

2. The document presents a view of sterile product manufacture inconsistent with that developed elsewhere, as codified in regulations, international standards, and pharmaceutical compendia, for example:

a. US and Japanese guidance on sterile product manufacturing differ markedly from what is presented.

b. The cleanroom content in the draft does not conform to International Organization for Standardization (ISO) 14644 and perpetuates the myths that cleanrooms can be classified microbiologically and that microbiological testing can enhance sterility assurance (3).

c. United States Pharmacopeia (USP) chapters <1211>, <1228>, and <1229> provide more contemporary and appropriate guidance for sterile product preparation (4–6).

d. There are also conflicts with current EMA guidance (e.g., the water for injection [WFI] Q&A paper and Guidelines on Good Manufacturing Practice specific to Advanced Therapy Medicinal Products) (7-8).

3. The revised Annex 1 does not provide separate consideration of conventional cleanrooms, restricted access barrier systems (RABS), and isolators; they are demonstratively different in many ways, and considering them together denigrates isolator performance, elevates RABS capabilities, and does not consider the limitations of the far less capable barrier-equipped conventional cleanroom.

4. EMA’s illogical and arbitrary insistence upon maintaining its Grade A/B/C/D system should be abandoned and the ISO 5 classification system used throughout the document. We have addressed these in some instances, but other citations to Grades A/B/C/D may remain as well as individual citations to Grade A, Grade B, and the unclear Grade A/B. Grades A, B, C, and D should be replaced with the corresponding classes of ISO 5, 6, 7, and 8.  All references to the arbitrary and confusing Grade B, and the totally undefined Grade A/B, should be removed.

5. Classification of controlled environments is limited to non-viable particle monitoring as described in ISO 14644. There are no means to classify cleanrooms based upon the microbial enumeration. Any suggestion that classification includes a microbial population requirement should be removed from the document.

6. The document has a perspective on microbial monitoring that is inconsistent with scientific reality. The limit of detection for microbial testing is substantially higher than one colony forming unit (CFU), a level which is used throughout the document. This erroneous belief leads to numerous misconceptions and overstatements regarding what value environmental monitoring has in the preparation of sterile medicinal products.

7. The document frequently asks for testing of materials, containers, and surfaces, with the expectation that the testing can somehow assure quality. The founding principle of validation is that it can assure confidence in the reliability and appropriateness of the process in ways that testing can never provide.

8. The document needs better organization to place related content together (e.g., environmental monitoring, terminal sterilization, barrier system, and lyophilization content can be found in multiple locations). In addition, there is non-exact redundancy of the content (i.e., environmental monitoring, terminal sterilization), which only causes additional confusion.

Incorrect application of technology

The use and application of barrier systems, isolation technology, and environmental and personnel monitoring are incorrectly described or applied, as follows:

2. Scope–m), Lines 99–101

m) Monitoring systems-including an assessment of the feasibility of the introduction of scientifically sound, modern methods that optimize the detection of environmental contamination.

Comment: This bullet item suggests that there are available means to demonstrate ‘sterility’ or ‘asepsis’ through the use of expanded monitoring. While these methods can be of value, expanding the application of these methods to enable detection of contaminants not currently found is scientifically incorrect because of the analytical and statistical limitations inherent in them. The environments used for sterile (aseptic) operations need not be, and by in large cannot be, ‘sterile’. The conflation of the condition of asepsis with the concept of sterility is both semantically and scientifically erroneous. It is impossible to test sterility in a product environment using any analytical technology. In addition, monitoring must never be invasive to the production processes in a manner that increases risk by adding activity in the critical zone. The increased interventional frequency that would result from the undefined practice of continuous monitoring (entails excessive risk for no additional useful information gained.

Proposed change: Monitoring activities must not increase the risk of contamination ingress due to their proximity to critical activities.

4. Personnel–4.4, Lines 205–206

This monitoring should take place immediately after completion of a critical intervention and upon each exit from the cleanroom.

Comment: Monitoring of personnel after each “critical intervention” increases risk of contamination by allowing personnel to reenter the critical zone after monitoring. Furthermore, there is no description provided in the document as to what constitutes a “critical intervention,” which would be required to interpret this statement. Monitoring after interventions attempts to do the impossible, which is to link contamination to specific interventions, and it implies that detection of contamination means sterile products will be contaminated. Neither of these objectives is analytically possible. Monitoring is an intervention subject to microbial contamination independent of any process-related interventional activity. Environmental monitoring in aseptic processing is already done far more intensely than is analytically or statistically warranted. Additional monitoring is not justifiable, as the actual limit of detection of the method is in the range of 10–100 CFU. Industry is already operating below the limit of detection of the method, and additional environmental monitoring is unwarranted as it has no quantitative or qualitative meaning.

Proposed change:  Grade A personnel monitoring should be performed in Grade B adjacent to the Grade A environment. For personnel working exclusively in Grade B, monitoring should be performed upon exit from the aseptic core.

5. Premises–5.3, Lines 319–321

Normally, such conditions are provided by a localized air flow protection, such as laminar air flow work stations or isolators.

Comment: The use of unidirectional air flow in isolators provides no operational advantage over turbulent air. Years of experience in sterility test isolators where extensive manual activity is required have not demonstrated any contamination problem that could be resolved by the use of unidirectional air flow. Laminar air flow is a misnomer, as truly laminar air flow does not exist in any cleanroom or separative technology environment. The authors are establishing expectations that cannot be objectively evaluated or established.

Proposed change: Normally, such conditions are provided by a localized air flow protection, such as provided in unidirectional air flow work stations.

5. Premises–5.3, Line 321–323

Unidirectional air flow systems should provide a homogeneous air speed in a range of 0.36–0.54 m/s (guidance value), the point at which the air speed measurement is taken should be clearly justified in the protocol. During initial qualification and requalification air speeds may be measured either close to the terminal air filter face or at the working height.

Comment: There are no reliable and reproducible means to measure air velocity at work height. The most appropriate means is to measure the air velocity proximate to the filter face. Airflow near the work surface is typically impeded by the presence of equipment or machines oriented perpendicular to the general direction of air movement. These conditions render air velocity measurements close to the work surface non-reproducible and thus meaningless. See Mason, W., et al, “Working Height Velocity Measurement in Conventional Cleanrooms, Pharmaceutical Engineering, July/August 2009, online at www.ispe.org/PE. Isolators because of the completeness of their separation do not require the same air velocities as conventional cleanrooms or RABS designs. Air velocities in isolators can be substantially less and non-unidirectional.

Proposed change: Unidirectional air flow systems should provide a homogeneous air flow over exposed sterile materials. Air velocity measurements should be taken to define the operating condition, however in isolator enclosures or other low volume work areas air velocities can be considerably lower than is customary for manned clean rooms. Unidirectional air is not required in isolators.

5. Premises–Barrier Technologies, Line 433 (and elsewhere)

Comment: The document uses the term barrier in a manner that fails to distinguish between highly capable technologies such as isolators, which are among the most capable of current aseptic processing technologies, and less capable designs such as open RABS and simple curtain/partial barriers designs. Open RABS systems allow similar intervention strategies to those found in cleanrooms and are substantially less effective in achieving full personnel separation as compared to isolators or closed RABS. To identify these varying capability systems together as is done throughout the document is to equate their performance, which is substantially different. The text could be substantially improved by separate treatment of isolation technology and RABS (open and closed) technologies and by making a distinction between those designs and ordinary manned cleaned rooms, the majority of which rely on some form of minimally effective barrier.

Proposed change: Provide separate consideration of conventional cleanrooms, RABS, and isolators because they are demonstratively different in many ways and considering them together denigrates isolator performance, elevates RABS capabilities (especially for open RABS designs), and implies that the far less capable barrier designs used in conventional cleanrooms are their equal.

5. Premises–Clean Room and Clean Air Device Classification–Note, Lines 493-495

Note: Classification is a method of assessing the level of air cleanliness against a specification for a cleanroom or clean area device by measuring the airborne particle concentration. The classification is part of the qualification of a clean area.

Comment: The content is correct; however, its placement immediately after content that discusses ‘environmental cleanliness level in the operational state’ suggests that classification can be accomplished in the operational state. Classification of environments is restricted to ‘at rest’ or static conditions as stated in ISO 14644. It is not scientifically possible to classify an environment microbiologically; the growth-based methods used for such purposes have a limit of detection between 10–100 CFU and are statistically and analytically incapable of classifying an environment.

Proposed change: Relocate this content to follow lines 497-499.

Note: Classification is a method of assessing the level of air cleanliness against a specification for a cleanroom by measuring the airborne particle concentration. The classification is part of the qualification of a clean area and is performed under ‘at rest’ or static conditions.

Section 9–Viable and non-viable environment & process monitoring–Section 9.13–Note 2, Line 1659–1662

Note 2: With regards to the monitoring of 5.0 μm, the limit of 20 is selected due to the limitations of monitoring equipment. It should be noted that alert limits should also be set based on historical and qualification data, such that frequent sustained recoveries below the action limit should also trigger an investigation.

Comment: The inability to accurately count low numbers of non-viable particles with current technology precludes the imposition of limits for 5 μm particles in ISO 5. This objection has been raised innumerable times by numerous individuals and organizations since the first issue of Annex 1 in 1999. Maintaining the façade that this is a statistically meaningful measurement is inappropriate as it perpetuates a technical fallacy. There is no scientific justification for retaining a monitoring requirement for 5 μm particles in this guidance.

Proposed change: Delete the columns relating to 5 μm particles in Table 5 and Note 2.

Delete content relating to 5 μm particles from the document.

Technical process errors

Technical errors dot the pages of the document. The following are some of the most egregious:

5. Premises–Section 5.8, Line 357–359

Floor drains in lower grade rooms should be fitted with traps or water seals to prevent back flow and should be regularly cleaned and disinfected.

Comment: “Traps”, which are intended to allow air and condensate exit from a system while retaining steam, cannot prevent back flow.

Proposed change: Floor drains in lower-grade rooms should be designed to prevent back flow and should be regularly cleaned and disinfected.

5. Premises–Section 5.6, Line 351

Materials liable to generate fibers should not be permitted in clean areas.

Comment: This sentence would essentially eliminate the presence of gowned personnel, who, studies have shown, continuously shed particles and fibers. Also, there are almost no aseptic gowns that are not made of fibers.

Proposed change:  Delete this content.

8. Production and specific technologies–Section 8.26, Lines 931–932

Critical defects should not be identified during any subsequent sampling of acceptable containers as it indicates a failure of the original inspection process.

Comment: Inspection of containers for defects is not an absolute process even if automated. The potential for a defective unit to be accepted is always present.

Proposed change: Critical defects should not be identified during any subsequent sampling of acceptable containers as it indicates a possible failure of the original inspection process.

8. Production and specific technologies–Section 8.30, Lines 957–960

Where it is not possible for a product to undergo a sterilization, consideration should be given to using terminal bioburden reduction steps, such as heat treatments (pasteurization), combined with aseptic processing to give improved sterility assurance.

Comment: This sentence adds no value to the document. The included and correct definition of terminal sterilization embraces a wide range of processes when it is understood that the target of the sterilization process is the bioburden present and not the biological indicator (see USP <1229>). Sterilization processes are intended to destroy the bioburden present (as opposed to reducing its population) and thus, the conditions necessary can be less rigorous than those previously utilized, which are fixated on killing high populations of a resistant biological indicator. The false expectation that mandates biological indicator destruction reduces the use of terminal sterilization rather than increasing it. Reduction of biological indicator population actually supports extremely safe terminal processes for the substantially less resistant bioburden.

Proposed change:  Delete this content from the document.

8. Production and specific technologies–Section 8.51, Lines 1086–1087

Chemical or biological indicators may also be used, but should not take the place of physical measurements.

Comment: Physical measurements rely upon mathematical models from microbiological data, which estimate the lethal effect a process asserts on microorganisms. As such, physical measurements are substantially less reliable (though easier to acquire) than the biological results they attempt to mimic. They should never be given preference over the results of a microbiological challenge study. Additionally, there are locations within items requiring sterilization where physical measurements are taken so remotely that their ability to “demonstrate” lethality is either extremely limited or non-existent.

Proposed change: The results from biological indicators and physical measurements must be considered in validation activities.

8. Production and specific technologies–Section 8.78, Lines 1251–1253

If a liquid product cannot be terminally sterilized by a microbiocidal process, it should be sterilized by filtration through a sterile, sterilizing grade filter …

Comment: The draft expects that a liquid product is to be “terminally sterilized by a microbiocidal process” an undefined process instead of simply stating that “a terminal sterilization process” be used.

Proposed change: If a liquid product cannot be terminally sterilized, it should be sterilized by filtration through a sterile, sterilizing grade filter …

8. Production and specific technologies–Section 8.78, Lines 1287–1289

f) Permit in-place integrity testing, preferably as a closed system, prior to filtration as necessary. In-place integrity testing methods should be selected to avoid any adverse impact on the quality of the product.

Comment: Pre-use post-sterilization integrity tests (PUPSIT) of sterilizing filters should only be performed where a closed system is present throughout. Requiring PUPSIT where the system is not closed creates a greater risk than any value the integrity test can provide. This same expectation is repeated at lines 1331–1340. At the very least, content on the same subject should be together in the document.

Proposed change: Delete this content from the document.

9. Viable and non-viable environment & process monitoring–Section 9.22, Lines 1703–1705

Although monitoring of ≧ 5.0 μm particles are not required for room qualification and classification purposes, it is required for routine monitoring purposes as they are an important diagnostic tool for early detection of machine, equipment, and HVAC failure.

Comment: See the earlier comment on lines 1659-1662. Further, the suggestion that larger particles would serve as a superior “important diagnostic tool” than smaller particles, which are more accurately counted, is not supported by documented evidence. Persistence in including this “requirement” in the face of objective scientific evidence to the contrary is completely without justification. 

Proposed change:  Delete all content relating to 5 μm particles from the document.

Application of risk analysis

There is a misdirected focus toward factors and suggested activities that do not increase patient safety or relate to the safety, identity, strength, quality, or purity of a drug product beyond the official or other established requirement.  In some cases, the recommendations add to patient risk.

4. Personnel–Section 4.4, Lines 205–206

This monitoring should take place immediately after completion of a critical intervention and upon each exit from the cleanroom.

Comment: The location of the monitoring is not specified at this point in the document. Table 6 includes expected levels for Grades A/B with different values for each. This implies that Grade A monitoring of personnel take place in Grade A. This is both impractical for the sites identified (see lines 199–200) and excessively risky. Monitoring of personnel in Grade A increases the potential for contamination ingress into that environment because a) personnel (the major source of contamination) are present in Grade A for longer periods, b) monitoring in Grade A increases activity in close proximity to sterile materials, and c) added activity in Grade A increases contamination risk. Monitoring cannot confirm “sterility” or “asepsis” in any event, but there is no benefit to having this sampling being performed in Grade A. Insistence on the Grade A personnel monitoring and consideration of risk should allow for that sampling to occur in Grade B adjacent to the Grade A environment with the Table 6 limits. For personnel working exclusively in Grade B, that monitoring can be performed upon exit from the aseptic core. Personnel monitoring should never be performed in close proximity to sterilized materials.

Proposed change:  Grade A personnel monitoring should be performed in Grade B adjacent to the Grade A environment. For personnel working exclusively in Grade B, monitoring should be performed upon exit from the aseptic core.

8. Production and specific Technologies–Section 8.18, Lines 883–886

Containers closed by fusion (e.g., form-fill-seal small volume parenteral (SVP) and large volume parenteral (LVP) bags, glass or plastic ampoules) should be subject to 100% integrity testing.

Comment: This is nothing more than testing quality into the product as opposed to validating the sealing process and running confirmatory process checks.  Annex 1 references quality by design, statistical process control, and process validation approaches and then undercuts the asserted value of these well-recognized quality management tools with a poorly considered focus on end-product testing.

Proposed change:  Delete this content.

8. Production and specific Technologies–Section 8.18, Lines 1331–1334

The integrity of the sterilized filter assembly should be verified by testing before use, in case of damage and loss of integrity caused by processing, and should be verified by on-line testing immediately after use by an appropriate method such as a bubble point, diffusive flow, water intrusion, or pressure hold test.

Comment: The document incorrectly applies risk analysis (e.g., integrity testing of a sterilized filter assembly before use [PUPSIT] and on-line integrity testing immediately after use).  PUPSIT adds risk by potentially requiring manipulations downstream of sterilized filters.

Proposed change: The integrity of the filter assembly should be verified, in case of damage and loss of integrity caused by processing, immediately after use by an appropriate method such as a bubble point, diffusive flow, water intrusion, or pressure hold test.

Unusual concept of sterile product manufacture

2. Principle, Lines 79, 81, 83

Comment: In many instances, it is essential that the suppliers of API, raw materials, containers, closures, and other critical items (sterilizing filters, single use disposables, gowning materials, etc.) be subject to more than mere evaluation upon receipt. Quality agreements, periodic audits, and other measures should be required where appropriate. The absence of such requirements is a serious shortcoming in the document. Proposed change: Add content that addresses the importance of and specific measures to control the supply of critical items used in the manufacture of sterile medicinal products.

2. Principle, Line 114

Comment: The document is inexplicit regarding the references its authors consider appropriate. In particular, this document clearly does not conform to the ISO 14644–Controlled Environments standard (which it should), yet specific mention is made of ISO. It is noteworthy that many FDA and USP documents are in conflict with this draft. It is inappropriate to have it both ways. By what virtue does EMA decide which standards are to be followed and which should be ignored as well as where harmonization should exist and where it should not? There is singular set of scientific principles used to develop standards; EMA should not ignore standards that are inconvenient, especially when those standards are objectively correct.

Proposed change: The document should explicitly state the reference standards that it follows. It should be extremely careful in proposing any practice, measurement, or value inconsistent with accepted international standards

Conclusion

This document includes only the more “exemplary” of the authors’ comments. Our submitted comments total more than 170 (8 general comments, 52 major comments, and 127 minor comments) and force us to conclude that EMA should withdraw this draft of Annex 1. We strongly recommend it be significantly revised to include scientifically founded expectations, incorporate contemporary technologies, conform to existing standards, and fully consider the pre-existing global regulatory and compendial guidance on sterile product manufacture and control and then reissued for comment once its many faults have been corrected.

NOTE 1: The comments included in this article provide additional clarifying text that was not a part of the submitted comments.  This content was added for improved readability. 

NOTE 2: Where indicated, recommendations for deletion would be for the entirety of the cited text

References

1. R. Madsen, J. Agalloco, and J. Akers, Pharmaceutical Technology Europe30 (2) (2018).
2. European Commission, EudraLex, Volume 4, EU Guidelines to Good Manufacturing Practice Medicinal Products for Human and Veterinary Use, Annex 1, Manufacture of Sterile Medicinal Products, December 2017.
3. International Standards Organization, ISO 14644–Cleanrooms and Associated Controlled Environments - Part 1: Classification of Air Cleanliness by Particle Concentration, 2015.
4. USP, <1211> Sterilization and Sterility Assurance of Compendial Articles, USP 40–NF 35, 2017.
5. USP, <1228> Depyrogenation, USP 40–NF 35, 2017.
6. USP, <1229> Sterilization of Compendial Articles, USP 40–NF 35, 2017.
7. European Medicines Agency, EMA/INS/GMP/443117/2017, Questions and answers on production of water for injections by non-distillation methods–reverse osmosis and biofilms and control strategies, August 2017.
8. European Commission, EudraLex, Volume 4, EU Guidelineson Good Manufacturing Practice specific to Advanced Therapy Medicinal Products, November 2017.