The particle question
Allen Humphries, General Manager of PMT (GB) discusses particle monitoring and the implications of the GMP Revision to Annex 1
The EC guide to Good Manufacturing Practice (GMP), covering the manufacture of sterile medicinal products, revision to Annex 1 became operational on 1 September.
The proposed changes have focused attention on particle counting and raised questions throughout the industry with regard to monitoring techniques and procedures.
This paper seeks to highlight some of the issues and provide information that may be of assistance to those persons tasked with ensuring compliance.
Key issues
While ISO 14644-1 cleanroom standards are now widely recognised and understood it may not be immediately apparent that there are major differences between this standard and the EC GMP guideline figures for acceptable particle population levels, especially for the cleanest grades A and B. Grades A and B 'at rest' and also Grade A 'in operation' allow no more than 3,500 particles per m3 greater than 0.5 micron and 1 (previously 0) particles per m3 greater than 5.0 microns. This compares to ISO 14644-1 class 5 limits of 3520 p/m3 > 0.5 micron and 29 p/m3 > 5 microns.
The GMP guideline is, as can be seen, intent in maintaining exceptionally low particle populations in the larger size range, which in itself poses fundamental questions regarding monitoring procedures and equipment.
The outgoing guide which required 0 particles/m3 at 5 micron for these classes was statistically impossible to achieve without sampling for infinity. 1 particle/m3 is achievable, however if the general rule (as applied in ISO14644-1 and Fed Std 209E) is used whereby a volume of air is sampled such that if at the maximum particle concentration allowable then a minimum of 20 particle events should be measured, using a conventional 1ft3/min sample flow rate particle counter then sample times would need to exceed 700 minutes, bearing in mind that there are over 35ft3 in a cubic metre.
It is therefore not surprising that the guideline states that a continuous measurement system should be used for monitoring the concentration of particles in the grade A zone and is recommended for the surrounding grade B areas. Continuous is a term which itself has in the past caused concern with regard to definition, as it has been used to describe both sequencing manifold systems, which automatically and regularly revisit one of a number of sample locations, and the more recently developed miniaturised particle sensors, which will monitor and generate data truly continuously.
Portable instruments
Most commonly used cleanroom particle counters operate at 1ft3/min (28.3 litres/min). For routine testing GMP requires that at least 1m3 (35.3ft3) of sample is taken for each class zone, the outcome of this is that additional time is probably going to need to be spent conducting routine testing than is currently done. From an instrument design stand point, the move to providing ultra high flow rate instruments is very unlikely as sample flow rate very closely defines both the size and dimensions of the optical system and internal air pump. In fact, industry needs have driven manufacturers to produce lightweight, highly portable instruments, based upon the flow rate norm of 1ft3/min, to fulfill ever more pressing Health and Safety issues relating to lifting and carrying.
For GMP compliance in monitoring (as opposed to routine testing), especially in Grade A and B areas, portable instruments could be used, however in most manufacturing environments the number of locations and availability of instruments would preclude this as a workable option.
Continuously monitoring
These small discrete instruments are now widely available and with their increased use across a wide range of industries, are a feasible, cost effective option. For pharmaceutical applications they are designed to measure continuously at 0.5 and 5 microns at an airflow rate of 1ft3/min, reporting data directly to a PC or to an in-house data management system. By running constantly during the defined periods (i.e. at rest or in-operation) air volume/flow rate is no longer an issue and, if relatively short data collection periods are set, then episodic events associated with individual particle bursts may be identified and distinguished from general overall background levels.
As with all data, whether collected via discrete sensors or portable instruments, the loss of larger (i.e. 5.0 micron plus) particles in sample tubing must be considered. Therefore the measuring device needs to be as close as practically possible to the sample location with the minimum length of tubing and bends.
In the absence of recommendations either in the current standards or guidelines, it is not unreasonable to refer to the old FED-STD-209E which states that for particles in the range 2-10 microns the transit tube should be no longer than 3m (at an airflow rate of 1.0ft3/min).
Sequencing manifold sampling
This type of system was, and in fact still is, widely used for regular automated monitoring. A single particle counter is linked to a sequencing scanning manifold, which automatically samples up to 32 individual ports. Radiating from the manifold are lengths of special low particle shedding, low attraction tubing, which may be up to 125ft (38m) in length.
For critical class A/B regions the main concern with this type of monitoring lies in the fact that large particles will be 'lost' within tubing runs even when the manifold system is linked to a high flow rate auxiliary vacuum pump, which creates turbulent flow within the sample lines.
Essentially the larger the particle, the longer the run, and the greater the number of bends: the greater the particle loss. Small particles (< 1.0 micron) remain airborne and transport very efficiently. The diagram indicates four particle sizes and their transportation efficiency in tubes of up to 125ft (38m) at an (assisted) airflow velocity of 3ft3/min. For general cleanroom monitoring where confidence in room stability is important and where a rapid indication of out of specification particle events is required, the data generated by a manifold system for 0.5 micron and above particles is very sound.
If we assume that the ISO 14644-1 (or old Fed Std 209E) standard approximates to a typical particle distribution curve likely to be found in a clean environment, then the ratio of 0.5 to 5.0 micron particles is in the order of 120:1. Therefore if all larger particles were to be lost then the accumulative total above 0.5 would be only marginally affected.
The system becomes questionable however, if absolute numbers are required to be measured at the 5 micron and above size, as tube losses will reduce the total count above this threshold. The actual magnitude of loss at each point is extremely difficult to determine as there are numerous factors creating the loss, and in-situ comparison testing is fraught with problems and open to experimental error.
While the above comments are focused on large particle losses in manifold tubing it is worth noting that losses will occur in the tubing of the industry standard 1ft3/min portable particle counters, and therefore if precise counts are required, at larger sizes then the tubing length should be minimised, or removed entirely. In many cases of course this is simply not practical, however extending the tubing length past manufacturers recommendation should be avoided otherwise the user will simply create a filter device.
While changes to guidelines in themselves always create concern and anxiety, the rationale behind the changes should not be forgotten, nor should the realisation that confidence in the operation and control within critical manufacturing environments can only be addressed by a rigorous risk assessment from which will naturally develop a fuller understanding as to the implications of the changes and whether or not alterations to current testing and monitoring procedures are required.
