Effects of the 2006 Proposed Revisions to Annex 1 of the European Union Good Manufacturing Practices

May 1, 2006
Pharmaceutical Technology, Pharmaceutical Technology-05-01-2006, Volume 2006 Supplement, Issue 2

The author explores the importance of the proposals to revise Annex 1 of the EU GMPs in the context of the desire for science-based, internationally respected GMPs. Commentary also is provided about the relationship between this annex and CEN–ISO cleanroom standards.

Particularly in the fields of cleanroom technology and sterile products manufacturing, good manufacturing practices (GMPs) regulations and guidances and technical cleanroom standards are mutually dependent on each other. The classification of air cleanliness specified in the International Organization for Standardization (ISO) 14644-1:1999 (1, 2) is quoted as a tool in Annex 1 of the European Union (EU) GMPs (3) and in the US Food and Drug Administration's 2004 aseptic processing guidance (4). The mutual dependence of GMPs and cleanroom standards is clear in that the ISO cleanroom standards are not application specific. The standards don't advise which levels of cleanliness are applicable to specific processes. Similarly, the EU GMPs do not explain how to measure airborne contamination or how to classify a controlled environment.

In May 2003, the European Medicines Agency (EMEA) published a revision to Annex 1 of the EU GMPs that was implemented in September 2003 (5). The 2003 version of the sterile products requirements was created in response to the publication of ISO 14644-1:1999, the internationally accepted standard for airborne-contamination control. In particular, this version of the EU guidance tried to clarify a perceived requirement to control ≥5.0-μm airborne particles at very low levels with the ability to measure such low concentrations reliably. The net effect of the September 2003 changes was to frustrate and confuse industry with requirements that were written poorly in some areas or were plain bad science in others. In response to industry's negative reaction, the EMEA initiated another revision to the Annex 1 document. Proposals for the further changes were published on the Eudralex Web site in November 2004, with a closing date of April 30, 2006 for public comments (6).

The EU GMPs have a much greater impact than on just EU nations and those supplying them with medicines. The complete EU GMPs are adopted by the Pharmaceutical Inspection Convention and the Pharmaceutical Inspection Co-operation Scheme and are referred to as the PIC/S GMPs (7). In turn, many nations adopt this document as their GMPs. Thus, the effect of poorly written documents with elements of poor science could reverberate around the world and lead to confusion and unnecessary environmental control, monitoring, and deviation investigation work.

At present, industry has the opportunity to correct these problems and move forward in a positive way by ensuring that it makes significant improvements to Annex 1 of the EU GMPs. Failure to improve this document could endorse bad science.

Table I: Proposed number of airborne particles in environmental grades A–D (per the November 2005 draft proposals).

Grade limits for airborne particles: classification

The following list includes the effects of the November 2005 draft proposals for airborne particles (see Table I) and associated requirements for classification and monitoring:

  • Industry wants and expects to have excellent control of ≥0.5-μm particles and expects negligible concentrations of ≥5.0-μm particles in Grades A (at rest and operational) and B (at rest).

  • Industry wants to avoid extremely large sample sizes for formal classification that would be required by specifying a very low limit.

  • 1-m3 samples must be taken at each location for a formal classification of Grade A zones at rest and in operation, which is much more demanding than the current requirement to collect a total of 1 m3 for the zone under consideration.

  • The sample size for Grade B at rest is not provided in the guidance. It might be thought that the 1-m3 sample size applies, however. If this is the case, then industry will bear a huge burden of airborne particle counting with limited value. It is for this reason that a formal classification of ≥5.0-μm particles in Grade B at rest should be avoided.

  • Industry must avoid triggering classification or monitoring failures because airborne particles measurements could be in a range for which real and false counts might be confused. In formal classification, the proposed acceptance of a limit of 20 particles/m3 will be very helpful.

  • A full classification of "in operation" conditions is required. This process would require an invasive testing regime not demanded previously. It always has been accepted that adequate confidence can be obtained from the "at rest" classification combined with operational monitoring. The guidance proposals suggest this "operational" classification could be carried out during media fills. Many parties will find it is an inappropriate time to add challenge to the process environment.

Additional support notes. Airborne particle concentration is perhaps the best known and the quickest way of determining the cleanliness of a working environment and detecting deviations from target values in real time. The EU GMP Annex 1 defines airborne particle control requirements for the working environment. Tables I–V show the evolution of airborne particle limits related to the environmental Grades A–D. Note that both the "at rest"and "in operation" conditions are defined in the EU GMPs. This is an important difference from the requirements in the equivalent US guidance.

The "0" requirements in Table II were both unrealistic and inappropriate to use as a limit for airborne particle counting for classification or monitoring.

Table II: Number of airborne particles in environmental grades A–D (before the May 2003 guidelines).

Both regulators and users of clean-air devices complying with the EU GMPs guidance accept that information relating to "at rest" and "in operation" conditions are appropriate. From the viewpoint of cleanroom or clean-air device procurement, it is important to have a performance benchmark to differentiate the basic technical performance of the equipment from the operating conditions that might prevail.

Table III: Number of airborne particles in environmental grades A–D (per the May 2003 guidelines, if they had been fully aligned with EN ISO 14644-1:1999).

The requirement to consider ≥5.0-μm airborne particles continues to be somewhat more contentious because the current limit for Grades A and B "at rest," set at 1 particle/m3 (see Table III), can be less than the false count rate of typical particle counters. Although arguments for and against the consideration of ≥5.0-μm particles are likely to continue, it is more important to set meaningful and measurable levels.

Table IV: Number of airborne particles related in environmental grades A–D (per the published May 2003 guidelines, when submitted to industry).

Close analysis of the air sample size calculation in EN–ISO 14644-1:1999 shows that the sample volume required at each location for classification is defined as the air volume required to obtain a count of a minimum of 20 particles at the class limit for the largest particle size considered. The equation is

Vs = 1000 × (20/[Ncl])

in which Vs is the sample size in liters and Ncl is the number of particles at the class limit. If this algorithm is applied to the various class limits previously mentioned, we would find the following sample sizes. (It must be remembered that the minimum sample size is 2 L and the minimum sample time is 1 min.)

  • A class limit of 29 particles/m3 indicates a sample size of 690 L. The collection time would be nearly 25 min at a sample rate of 28.3 L/min.

  • A class limit of 20 particles/m3 indicates a sample size of 1000 L. The collection time would be just more than 35 min at a sample rate of 28.3 L/min.

  • A class limit of 1 particle/m3 indicates a sample size of 20 m3. Collection would take nearly 12 h at a sample rate of 28.3 L/min. Because this sample is impractically large, the Annex 1 proposals recommend a sample size of 1000 L for Grade A and B areas. This testing might be practical and acceptable in Grade A critical core areas but would present a huge, unnecessary burden in Grade B areas. This sample size would be adequate for a class limit of 20 particles/m3, which is from where the suggested practical limit of 20 particles/m3 is derived. Therefore, if one were to rationalize this approach, the table might be adjusted as shown in Table V.

Table V: Suggested number of airborne particles in environmental grades A–D to resolve the problem of 5.0-μm particles more practically.

Grade limits for airborne particles: monitoring

Real-time particle monitoring also is important, and the acceptance criteria will be based on Table I. Once again, the challenge is to accomplish real-time monitoring with ≥5.0-μm particles because there is the potential that false counts may be greater than real counts at such low concentrations. When looking at frequent zeros and occasional ones or twos it is better to look at the frequency of events rather than extrapolating short-term data. Remember, 1 particle/ft3 will be expressed as 35 particles/m3.

The following requirements are found in the draft guidance:

  • Automated systems are expected for Grade A and are desirable for Grade B zones. Such systems are clearly intended to be used to monitor for excursions and events. The guidance offers no information about their use in Grade B zones.

  • Grade A zones must be monitored for ≥0.5- and ≥5.0-μm particles. The guidance goes on to suggest that both manifold and local counters can be used. This is erroneous because manifold and tube multipoint systems will always suffer from an unacceptable loss (dropout) of particles.

  • The guidance accepts the possibility of occasional counts that exceed the published limits, but it doesn't effectively explain how to deal with them. What is needed is clear guidance about how, in principle, to examine the average performance and identify unsatisfactory events. Many applications of real-time systems evaluate small samples taken frequently and consider cumulative count over a greater time. When considering ≥5.0-μm particle counts, three sequential samples of 28.3 L with counts of two or three particles could be considered an event, whereas a single isolated sample with one or two counts could be ignored as being insignificant.

Unidirectional airflow

Unidirectional airflow (UDAF), formerly referred to as laminar flow, is a key method of protecting the critical aseptic core and other clean zones through the delivery of a controlled airflow of defined particulate quality.

In recognition of the vast range of various clean-air devices, EN–ISO 14644-4 simply specifies that the velocity should be >0.2 m/s for unidirectional airflow systems (8). The standard emphasizes that demonstrating the effectiveness of the airflow by determining the airflow uniformity is more critical than choosing only a specific velocity. This focus on the demonstrable effectiveness of the clean-air system also is reflected by the current text in Annex 1. There is confusion about the range of velocities given and where to measure it. EN–ISO 14644-3 clearly differentiates between the filter-face velocity and the system's working level velocity (9).

For most UDAF systems, the well-established value of 0.45 m/s (±20%) will be appropriate. Higher values of approximately 0.6 m/s and lower values of approximately 0.3 m/s may be needed to protect hot bodies and isolator internals, respectively. The current Annex 1 also is not clear about the acceptability of using non-unidirectional airflow in devices such as isolators. The text should be revised to endorse the use of such systems where appropriate.

New requirements for handling stoppered vials

The text proposals recognize the advantage of segregating particulate-generating processes such as vial capping from other more-sensitive activities. Progressively, industry has developed a preferred practice to over-seal vials outside the aseptic area. This practice is now threatened by the proposed text that seems to define the over-sealing as an aseptic process. Industry believes that the following are reasonable expectations:

  • Over-seals can be applied outside the aseptic processing area.

  • Over-seals need not be sterilized.

  • The transport route (conveyor) from the exit of the filling machine's stopper insertion zone to the boundary of the Grade B aseptic processing area should be within a Grade A environment.

  • The transport route to and the capping station should be protected by UDAF ISO Class 5 air.

  • The protected capping station can be located in a Grade D environment.

  • Additional security can be gained by introducing stopper position detection with automated rejection of unsecured units.

Summary and conclusion

At a time of significant change to the European Union good manufacturing practices (EU GMPs) requirements for sterile products, there is still a need for a symbiotic working relationship between GMPs and the technical cleanroom standards. The standards, for example, can be considered as the basic tools and the GMPs' specifications as the requirements and principles.

The current proposals for a revision of Annex 1 of the EU GMPs are an opportunity to get it right in the world of risk-based decision making. The authors of the EU GMPs have a great responsibility because the guidance will be incorporated as part of the Pharmaceutical Inspection Convention and the Pharmaceutical Inspection Co-operation Scheme (PIC/S) GMPs.

Gordon Farquharson is principal consultant at Bovis Lend Lease Technology, Tanshire House, Shackleford Road, Elstead, Surrey, GU8 6LB, United Kingdom, tel. +44 1252 703663, fax +44 1252 703684, gordon.farquharson@eu.bovislendlease.com


1. International Organization for Standardization (ISO) 14644-1, Cleanrooms and Associated Controlled Environments, Part 1: Classification of Air Cleanliness (ISO, Geneva, Switzerland, 1999).

2. ISO 14644-2, Cleanrooms and Associated Controlled Environments, Part 2: Specifications for Testing and Monitoring to Prove Continued Compliance with ISO 14644-1 (ISO, Geneva, Switzerland, 2000).

3. Eudralex, Volume 4: Medicinal Products for Human and Veterinary Use: Good Manufacturing Practice, Annex 1: Manufacture of Sterile Medicinal Products (Eudralex, May 2003 revision).

4. US Food and Drug Administration, Guidance for Industry. Sterile Drug Products Produced by Aseptic Processing—Current Good Manufacturing Practice (FDA, Rockville, MD, 2004).

5. European Medicines Agency (EMEA, London, UK, 2003).

6. Eudralex, http://pharmacos.eudra.org/F2/home.html, accessed April 13, 2006.

7. Pharmaceutical Inspection Co-operation Scheme (PIC/S, Geneva Switzerland), http://www.picscheme.org, accessed April 13, 2006.

8. ISO 14644-4, Cleanrooms and Associated Controlled Environments, Part 4: Design, Construction and Start-Up (ISO, Geneva, Switzerland, 2000).

9. ISO 14644-3, Cleanrooms and Associated Controlled Environments, Part 3: Metrology and Test Methods (ISO, Geneva, Switzerland, 2002).