EU GMP Annex 1 clears the air for sterile manufacturing

Revisions to EU GMP Annex 1 clearly outline appropriate air cleanliness measures to be taken

By Mark Hallworth, Particle Measuring Systems

The Good Manufacturing Practice (GMP) guidance for sterile manufacture was revised in 2003 to accommodate changes from various cleanroom standards to create a single unified cleanroom standard, ISO 14644-1. The introduction to ISO 14644-1 states this as:

Annex 1 of the EC Guide to Good Manufacturing Practice (GMP) provides supplementary guidance on the application of the principles and guidelines of GMP to sterile medicinal products. The guidance includes recommendations on standards of environmental cleanliness for clean rooms. The guidance has been reviewed in the light of the international standard EN/ISO 14644-1 and amended in the interests of harmonisation but taking into account specific concerns unique to the production of sterile medicinal products.1

To summarize, the method to certify a cleanroom needed to comply with the rules and format of ISO 14644-1 guidance; this European Commission annex includes a modified ISO standard that addresses sterile medicinal products. To support this, a table of cleanroom certification values that roughly translated to ISO 14644-1 was defined.

For clarity, a series of notes appended the table. Unfortunately, the first, “Note a,” caused some confusion.

This confusion has been remedied in the 2008 release of the EU GMP Annex 1, which clearly outlines three phases that need to be performed:

  1. Certification: Each cleanroom and clean air device should first be classified.
  2. Monitoring: The cleanroom should then be monitored to verify that conditions are being maintained relative to product quality.
  3. Data review: The data accrued from the monitoring must be reviewed in light of the risk to finished product quality.


To perform the required certification it is important to understand ISO 14644-1 and how to certify a cleanroom in accordance with that standard, rules on number of sample points, sample point location, and volume of sample to be taken at each location, along with the rules on statistical analysis of cleanroom data that need to be followed. However, rather than use the table for classification limits prescribed in ISO 14644-1, technicians should be using the table shown here, as printed in the revised guidance document.

Click here to enlarge image

Other expectations are also defined by the GMP, such as the sample volume for Grade A cleanliness, which should be 1 m3 per sample location, and that a minimum length of sample tubing should be used due to the high precipitation of 5.0-μm particles in transport tubing. Ideally, no sample tubing should be used. Also, recertification of the cleanroom should follow the guidance given in ISO 14644-2–that is, once per year for ISO Class 6 and greater and once every six months for ISO Class 5 and cleaner; concessions are made for extending the ISO Class 5 areas if a monitoring system has been implemented. Suitable times to perform certification are media fills or simulated filling runs.


After the cleanroom or clean air device has been certified, the room must be monitored, relative to risk, to prove that the aseptic manufacturing environment can be maintained and proven to be maintained.

The Grade A zone, which is the environment of greatest potential risk to the finished product, should be monitored for the full cycle of production, including setup. The frequency of monitoring should ensure that any interventions, short-duration events, or general deterioration in conditions will be measured and alarms triggered if alert/action limits are exceeded. This requirement of all events essentially precludes the use of manifolds in these areas due to the sequential nature of the sampling being performed; concessions are made for the use of manifolds if they have been sufficiently validated as suitable for the relevant manufacturing type.

Grade B areas follow the same rules as Grade A. However, the frequency of sampling can be reduced. Grade A is maintained under unidirectional airflow, and so short-burst events may be localized and of a very short duration, excluding some catastrophic failures. However, Grade B is turbulent mixed airflow and reflective of the general environment in which the operators occupy. A low level of continuous particulate activity in this area is normal; and the system’s response is to alarm when general control of this area is out of tolerance. Therefore, an immediate spike in contamination is less likely to have a significant impact on product quality. This becomes more pronounced when looking at background support areas beyond the zone in immediate proximity to the filling line or other Grade A areas.

In the 2003 GMP, there was confusion over the sample required for monitoring the Grade A and Grade B areas due to the phraseology used. The 1-m3 sample was to meet the calculation required by ISO 14644-1 and not a risk-based monitoring value. However, clarity is improved in the revised guidance:

The sample sizes taken for monitoring purposes using automated systems will usually be a function of the sampling rate of the system used. It is not necessary for the sample volume to be the same as that used for formal classification of clean rooms and clean air devices.2

Therefore, a system using a 28.3 L/min particle counter would ideally sample continuously, from setup through the entire filling period and slightly beyond, taking minute-by-minute samples, normalizing data to counts/m3, and setting appropriate alarm and alert limits on the normalized values. The key to monitoring is to be able to respond in a timely manner to events that would show the area is no longer in environmental control.

Data review

There is a relationship between non-viable particles in a cleanroom and the viable contaminants (see USP Chapter <1116>, “Microbiological Evaluation of Clean Rooms and Other Controlled Environments”). There are also studies that show the size of viable particulates free-floating in a cleanroom. When combining these two independent studies together, it is apparent that if the operator can control the large particles in a cleanroom, control over the viable risk in a cleanroom can also be demonstrated. Empirically this is difficult to show due to the statistics of the small numbers generated–that is, <1 particle and <1 CFU. However, the 5.0-μm particle size is of particular importance when reviewing environmental data within the cleanroom.

Figure 1. Lasair III particle counter. Photo courtesy of
Particle Measuring Systems.
Click here to enlarge image

Occasional high counts may be due to interference with the particle counter electronics, although some particle counters have components that reduce these effects as well as the effect of random particles within the cleanroom. Given the fact that random events cannot be interpreted in small numbers with statistical reviews and have very little correlation to the general production activities, they can be reviewed at a later stage when doing longer-term analysis of cleanroom performance. What is key is the consecutive or regular counting of low levels of particulate that may give clues to a possible contamination issue that should be investigated.

Click here to enlarge image

Figure 2 shows three conditions:

  • Continuous: If continuous levels of 5.0-μm particles are seen in a cleanroom, an investigation should be undertaken as it is unlikely that large particles would penetrate a filter. Therefore, the contamination is arising from a source that can be contained.
  • Frequent: When large particles occur with a frequency that is not random, then a source of these particles should be determined and, where possible, rectified. The effect of the particles can be correlated against finished product testing to define what level of particle can be deemed a nuisance.
  • Random: When particles show little or no pattern of occurrence, then a frequency of N of M should be determined–i.e., no more than three particles in any 12 minutes or similar. Again, correlation back to finished product testing should validate the data used in routine monitoring.

The definition of the alert and alarm set points is also examined in the current GMP guidance:

Appropriate alert and action limits should be set for the results of particulate and microbiological monitoring. If these limits are exceeded operating procedures should prescribe corrective action.2

The determination of appropriate alerts is that proof of control over the environment, relative to product quality, should be maintained. Therefore, using the limits assigned by the certification data alone may not always be prudent. Rather, limits that better reflect the production environment of each particular facility, filling line, or similar ought to be determined.

The guide is for continuous particle monitoring in Grade A and immediate Grade B areas using an in situ particle counter. This is because the risk of contamination in finished product is very high and the greatest risk factor, the operator, is in close proximity. The operator is not only the greatest risk posed to product but also a random generator of particles. These are not all inert particles; some will be viable, which poses an even greater risk to the finished product. Since we cannot control the risk, we must measure it. If it is in excess of proven acceptable limits, then the system must alert the users. How quickly the facility monitoring system alarm should alert the users is dependent upon the risk (see “Recommended System Setup for a Grade A Zone”).

Analysis of risk can be considered as how resilient the filling operation is to potential contamination events while still able to protect product. If the system is very robust (isolator, restricted access barrier system), then an event has a relatively low risk of contamination. If it is an ampoule line with curtain protection then small deviations could have a greater impact. No answer will fit all applications because all risk is variable. Considerations such as what gowns are used, what undergarments are supplied, air changes per hour rate, number of personnel in room, etc. are all important factors.

Lyophilized product

Product that has been filled aseptically and is to be freeze dried should be maintained within a Grade A environment, from the point of stopper insertion to the freeze dryer. If this is done via a mobile cart, then this mobile environment must be shown to maintain a Grade A environment. When a stopper is not fully inserted, the vial is deemed to be open, and any aseptic vial open to the environment must be maintained within a controlled environment.

Once freeze drying is completed, the stopper is pulled down into the vial or a mechanical pressure is applied to ensure closure, and the stopper is proven to be fully seated via a validated protocol, the vials should be maintained within a Grade A air supply until the cap is in place and crimped. Recall the table from certification: A Grade A environment is essentially an ISO Class 5 environment. Therefore, the quality of air being supplied to the crimping process is better described as being ISO Class 5 quality, from a particle perspective. If the capping activity is performed as an aseptic process, then a Grade A environment must be proven.

Mark Hallworth is pharmaceutical business manager at Particle Measuring Systems in Boulder, CO (


  1. ISO 14644-1, Cleanrooms and associated controlled environments–Part 1: Classification of air cleanliness, 2003.
  2. European Union Good Manufacturing Practice Annex 1, 2008.


Lasair® is a registered trademark of Particle Measuring Systems, Inc.

Recommended monitoring/alarming system setup for a Grade A zone

Step 1: Set all values in the facility monitoring system to m3

Step 2: Set the 0.5-μm alarm channels (1 = alert, 3 = alarm) to 1,625 and 3,250 n/m3. (These values are temporary until the real values are discovered from the process.)

Step 3: Set the alarm level 1 (alert 0.5 μm) to react on a frequency of 2:2 events. So two consecutive events will trigger an alert = orange light.

Click here to enlarge image

Step 4: Set the alarm level 3 (alarm 0.5 μm) to react on a frequency of 3:3 events. So three consecutive events will trigger an alarm = red light.

Click here to enlarge image

Step 5: Set the 5.0-μm alarm channels (0 = alert, 2 = alarm) to 71 and 35 n/m3. (These are identical, but we will use a different frequency to determine risk.)

Step 6: Set the alarm level 0 (alert 0.5 μm) to react on a frequency of 2:2 events. So two consecutive events will trigger an alert = orange light.

Step 7: Set the alarm level 2 (alarm 5.0 μm) to react on a frequency of 3:10 events. So three consecutive events will trigger an alarm = red light.

Click here to enlarge image

This rationale allows technicians to respond quickly to 0.5-μm events but not be alerted for nuisances


Easily post a comment below using your Linkedin, Twitter, Google or Facebook account. Comments won't automatically be posted to your social media accounts unless you select to share.