Sterile medicines: Annex 1 arrives

The long awaited revision of EU GMP Annex 1:20081 has finally been published and is a significant improvement over the 2003 version.2 In particular, the requirements for monitoring particulate concentrations of ?0.5µm and ?5µm particles are much clearer and practical monitoring strategies are now feasible where they weren’t before.

This is because the maximum permitted number of ?5µm particles in Grade A areas “at rest” and “in operation” has been raised from 1 to 20, and for Grade B “at rest” from 1 to 29. Harmonisation with ISO 14644-1:19993 is also claimed, but on closer examination there remain some discrepancies.

The biggest disappointment is that the requirement to measure the concentration of ?5µm particles remains. This is not a requirement of the US FDA cGMP4 and there is a well-argued view that counting of ?5µm particles is of limited usefulness. Papers5,6 have been written that demon-strate a correlation between the concen-trations of ?5µm particles, ?0.5µm particles and microbe carrying particles (MCPs) and therefore, if a certain number of ?0.5µm particles are present, then the number of ?5µm and to an extent, the number of MCPs are approximately predictable.

This has become such an issue that the author would like to hear from anyone who has documented evidence of instances where ?5µm particle concentrations have been in excess of the maximum permitted while those for ?0.5µm particles have remained within limits.

Annex 1:2008 states that “the monitoring of the ?5µm particle concentration takes on a particular significance as it is an important diagnostic tool for early detection of failure.” A useful method for monitoring is to take a continuous sequence of 1 ft3 samples. This method has been termed the “rolling window” system. Using this system, it is possible to get the same response time for excessive particle counts at both ?0.5µm and ?5µm. All but the most recent particle counters draw air in at 28.3 litres/minute, which is 1 ft3/minute, so the good old cubic foot still has its uses!

For classification, ISO 144644-1 specifies that the single sample volume is that volume of air that would contain 20 particles at the class limit for the largest particle size being considered at the designated ISO class. For the Annex 1:2008 at grade A, this is 20 ?5µm particles/m3 and therefore the single sample size is 1 m3 (1,000 litres), which is what Annex 1:2008 specifies.

However, if only the ?0.5µm particle concentration were specified, then the single sample size would be 20/3520 m3, i.e. 5.68 litres. Thus the time to take a single sample is 35.3 minutes if the maximum considered particle size is ?5µm and just 12 seconds if it is ?0.5µm.

So here is a compromise suggestion from the author. For classification (and reclassific-ation) the particle concentration should be checked at ?0.5µm only. This will reduce the overall time for testing and yet allow a larger number of sampling locations within that time. For monitoring, the rolling window system should be used, at both ?0.5µm and ?5µm, with appropriate alert and alarm levels, so that any occurrence where there is an unusually high proportion of ?5µm particles is quickly picked up.

The Pharmaceutical and Healthcare Sciences Society (PHSS), formerly the Parenteral Society, has set up a Special Interest Group to produce a Monograph to be called Best Practice for Particle Monitoring in Pharmaceutical Facilities. The first issue of this should be published in the second quarter of this year.

Annex 1:2008 has moved a long way towards harmonisation with ISO 14644-1:1999. However it still specifies an intermediate ISO class of 4.8 to satisfy the EU requirement for a concentration of 20 ?5µm particles per m3 for Grade A at rest and in operation – not so different from ISO 5 and 29 particles in the FDA guidelines. There is a further discrepancy where Annex 1:2008 gives concentrations of 2,900 and 29,000 at ?5µm instead of 2,930 and 29,300 for ISO Grades 7 and 8 respectively. This is not a material difference, but there is no reason for it.

For classification purposes, Annex 1:2008 refers to the methodology of ISO 14644-1:1999 with respect to the minimum number of sample locations and the sample size. The sample size has already been discussed, but in a pharmaceutical context, it is likely that some form of risk assessment will determine the number and location of the sample points and the number is unlikely to be the minimum specified in the ISO.

It is very disappointing that Annex 1:2008 still uses the term laminar air flow. Worse still, it uses laminar air flow in the context of Grade A work stations and unidirectional air flow in the context of closed isolators as if the terms were intended to have different meanings.

In fact, “laminar flow” means something quite different in other branches of engineering and technology such as aerodynamics, aeronautics and fluid flow and has not been used in clean air standards for many years (see box below).

The misuse started at the very beginning of our technology and a more accurate term would be unidirectional low perturbation airflow. For those who are interested, laminar flow is pure streamline flow with no turbulence. In true laminar flow, each notional layer or lamina of air or fluid has its own velocity and this increases from zero where the flow is along a surface to the maximum some distance from the surface.

In the case of laminar flow in a pipe, if a graph is plotted of the fluid velocity across the diameter, the shape of the graph will be a parabola. This is a good example of how laminar flow does not necessarily have the same velocity across an entire cross section.

Reynolds number is a dimensionless scaling factor that is used, for example, to calculate the wind tunnel velocity that produces the same flow characteristics over, say, an aircraft wing of a given scale as over the full scale wing in a real life flying situation. As length is part of the equation for Reynolds number, it cannot be an absolute number, but rather a number that is specific to a given situation. It is, as has been said a scaling factor, and therefore not relevant to clean air technology.

Another unnecessary invented term that is used in Annex 1:2008, as it was in previous issues, is laminarity. ISO 14644 uses the perfectly good term uniformity of airflow, and ISO 14644-3:2005 - Part 37 gives infor-mation on the testing of uniformity of airflow.

Although Annex1:2008 gives quantitative guidance on airflow velocities for Grade A areas (a precedent for recommending values), it does not give any guidance on HEPA filter specifications and maximum permissible penetrations for the in situ aerosol leak test of filter systems. This would have been particularly welcome given the apparent relaxation of the maximum penetration in ISO 14644-3:2005 Part 3.

The author understands such guidance will be available for NHS installations when the NHS QA committee publishes a test protocol in the near future.