Article | Open Access | Published 10 Jan 2020
Pharmaceutical Cleanroom Classification using ISO 14644-1 and the EU GGMP Annex 1
Part 2: Practical Application
Classification of cleanrooms and clean zones associated with the manufacture of medicinal products has been assessed in two articles. The first article discussed the classification requirements and principles associated with ISO 14644-1 and Annex 1 of the EU GGMP, and a suitable classification test method for aseptic manufacturing was derived. This second article considers the practical application of the method for the classification of a pharmaceutical cleanroom, and isolator located within it.
Key words: Cleanroom classification, ISO 14644-1, EU GGMP Annex 1
Annex 1 of European Union Guidance to Good Manufacturing Practice (EU GGMP) (1) specifies the environmental conditions that must be provided for the manufacture of sterile medicinal products and requires the classification of different grades of cleanrooms and clean zones to be carried out in accordance with ISO 14644-1 (2). With the correct interpretation and application of the information given in ISO 14644-1 and Annex 1 of the EU GGMP, as well as more current expectations of the regulatory authorities, an appropriate testing regime has been established in our first article that is suitable for the classification of a pharmaceutical cleanroom and clean zone utilised for aseptic manufacture (3). The testing regime method is applied to the classification of a pharmaceutical cleanroom and isolator but the same approach can be used, with some modifications, for most types of healthcare cleanrooms.
2. Description of cleanroom and isolator
The classification method described in the first article is applied to a cleanroom and isolator used for the aseptic filling of a liquid formulation into vials, which are then sealed with sterile closures. Shown in Figure 1 is a schematic diagram of the cleanroom and isolator.
The cleanroom (EU GGMP Grade C) that contains the isolator is a non-unidirectional airflow (non-UDAF) cleanroom with ceiling-mounted HEPA filters, that supply air via swirl diffusers, and low-level air extracts that return the cleanroom air to a central air conditioning system. The cleanroom is operated at positive pressure with respect to a surrounding EU GGMP Grade C corridor. Typically, 4 people work in the cleanroom but a maximum of 6 people may be present, wearing a polyester smock with full hood, overboots, mask, and sterile, double sets of latex gloves.
The isolator (EU GGMP Grade A) is of rigid construction and supplied with UDAF through terminal HEPA filters that cover the entire isolator ceiling area and is maintained at a positive differential pressure with respect to the cleanroom. It interfaces with a vial cooling zone linked to a washing, depyrogenation and cooling tunnel located within an adjacent preparation cleanroom that provides prepared vials for the filling operation. This preparation cleanroom is also used for the formulation of the product in a stainless-steel vessel that then supplies the liquid product directly into the isolator through pre-sterilised single use, sterilising grade product filters and tubing, connected to 4 filling needles. The tubing is transferred into the isolator via a secure liquid transfer port following decontamination of the closed isolator with vaporised hydrogen peroxide. A peristaltic pump, located next to the preparation vessel, is used to transfer the liquid from the vessel to the needles for the subsequent filling operation. The product outfeed (exit mousehole) is used for the transfer of the filled vials secured with rubber stoppers directly into a UDAF workstation in an adjacent cleanroom, where capping and crimping of the closures is completed. Sterilised rubber closures are transferred into the isolator from the cleanroom using a secure split butterfly transfer device that is connected to the isolator. It is changed four times during the course of the filling in order to supply the required number of closures. Vials with an internal neck area of 2 cm² are aseptically filled with 2 ml of product solution in batches of 4 000, which takes approximately 4 hours.
3. Classification of the EU GGMP (2008) cleanroom and isolator in accordance with ISO 14644-1: 2015
A review of the EU GGMP (2008) and ISO 14644-1: 2015 classification requirements and principles has been undertaken in the first article (3) and used to provide a classification testing rationale for pharmaceutical cleanrooms used for aseptic manufacture. This is applied to the example described above for the classification of the cleanroom that contains the isolator, and then to the isolator.
The EU GGMP requires classification to be carried out ‘at rest’ and ‘in operation’. The information given in this article is for the ‘in operation’ classification, and it is assumed that a successful ‘at rest’ classification has already been completed. The ‘at rest’ classification is very similar to the ‘in operation’ classification, but no production is carried out during the testing, and has been discussed in the first article.
3.1 Cleanroom classification
The testing rationale and the associated considerations for the ‘in operation’ classification of the isolator cleanrooms is shown in Table 1. This is an abbreviated version of Table 1 given in the first article (3), and if more comprehensive information is required, that table should be consulted.
3.2 Determination of cleanroom sampling sections and sampling positions
Is necessary to divide the cleanroom into the number of sampling sections given in Table A1 of ISO 14644-1. As the cleanroom contains an isolator, the floor area is not a simple square or rectangle but is asymmetrical, and the method discussed in Appendix D of the first article (3) should be used to divide the floor area of the cleanroom. The application of the method is explained in Appendix A of this article, where it was found that the cleanroom needed to be divided into 15 sections, and these are shown in Figure 2.
As there is no product exposure in the cleanroom, ‘hot spots’ of operator activity were identified to determine the position in each of these sections where air sampling should be carried out. These are lettered and shown in Figure 2 at the identified sampling locations. Sections A4, B2, B6 and D2 had no ‘hot spots’ and sampling was carried out in the centre of the sections as these were considered to be representative of the characteristics of the sections.
3.3 Results of airborne sampling in cleanroom
Using the information given in the previous section, sampling was carried out at each of the locations, and the results shown in Table 2. It can be seen that, with one exception at location A2, all the samples had air concentrations below the EU GGMP Grade C limits set for particles ≥0.5 µm and ≥5 µm.
3.3 Investigation and rectification of the high airborne particle concentration
As shown in Table 2, the concentration of particles ≥5 µm at location A2 exceeds the classification limit of an EU GGMP Grade C cleanroom. ISO 14644-1 allows out-of-specification counts to be retested but on retesting, similar high counts were obtained. The reason for high particle counts in a cleanroom is caused by either an unusually high dispersion rate of particles, or insufficient ventilation. An investigation was carried out to determine the cause of the high count, and to rectify the problem.
The high concentration of particles appeared to be connected with the ventilation effectiveness of the air supply system. The overall average concentrations in the cleanroom of particles ≥0.5µm and ≥5µm were 101 533 /m³ and 6 632 /m³ respectively, which are well below the airborne concentrations specified for an EU GGMP Grade C cleanroom. Had the average count been above the Grade limit, it would indicate that there was insufficient air for the given amount of particle dispersion, and that the air supply to the cleanroom needed to be increased, or the dispersion of particles from machinery or personnel reduced. These particle concentration results suggest that the problem did not lie with an insufficient supply of filtered air.
Table 2 shows that the airborne concentrations at locations in the cleanroom were very variable and the ventilation effectiveness was investigated. The Performance Indexes (PI) was calculated for each sampling location using the following equation (5).
The PI results included in Table 2 confirm that the airborne concentrations to be very variable. If the supply air was consistently well mixed with cleanroom air, all the PI values would be close to 1, and this would have been the ideal situation. However, it can be seen that at location A2 the PI was 0.2, which suggests that the location had 5 times less clean air than the average in the room, and further ventilation effectiveness tests were carried out.
The recovery rate was measured at location A2 according to the method given in ISO 14644-3 (6) and explained elsewhere (5). Test particles were released and then discontinued, and the rate of decay of the particles measured and the recovery rate calculated. Using hours as the units of measurement, the recovery rate was found to be 5 /hour. The recovery rate at a location is known to be the same as the air change rate at that location (6), and if this recovery rate (local air change rate) is compared to the overall air change rate in the room, the Air Change Effectiveness (ACE) index is obtained. This index shows how much clean air a specified location receives, compared to the average in a cleanroom. The overall (average) air change rate in the cleanroom was 30 per hour and therefore the ACE index at the location A2 was 0.17 (5/30). As expected, the PI and ACE indexes were similar. This showed that location A2 receives about 5 times less air that the average in the room, and this was most likely to be the reason for the high concentration of particles at that location.
The problem of uneven distribution of clean air was likely to be solved by improving the air distribution by either installing another ceiling air supply inlet, or air return/extract, close to location A2. It was decided to install an additional exhaust outlet but not increase the overall extract air rate. The air extract system was then rebalanced. The recovery rate was then re-measured at location A2 and found to have increased to 15 /hour, and the ACE index to 0.5. This was considered acceptable but, if it had not been so, an additional air supply could have been installed close to location A2 and, with the same overall room rate of supply volume retained, the air supply system rebalanced. Had neither of these remedial actions been adequate, an increase in the air supply rate and rebalancing to give more air supply around location A2 would have been the next consideration. Reducing the dispersion from machinery, or improvement of the cleanroom garments to reduce the dispersion from personnel, would have been another possibility. However, a new measurement of the airborne particle concentration at location A2 gave a concentration below that required.
ISO 14644-1 allows remedial action to be taken in a cleanroom where an out-of-specification count found at a location is attributed to a technical failure of the cleanroom or equipment. The cause should be identified, remedial action taken, and retesting performed of the failed sampling location, the immediate surrounding locations, and any other locations affected. However, in this case, it was decided that the remedial action was substantial, and required a full reclassification of the cleanroom. This was carried out and the cleanroom passed the EU GGMP Grade C classification.
3.5 Isolator classification
The testing rationale and associated considerations for the ‘in operation’ classification of the isolator is shown in Table 3.
3.6 Determination of cleanroom sampling sections and sampling positions
The 13 sampling locations are shown in Figure 3 and the associated 15 sampling activities are detailed in Table 4.
3.7 Results of airborne sampling in isolator
Sampling was carried out at each of the identified sampling locations, and the results shown in Table 4. It can be seen that all the samples had air concentrations below the Grade A limits set for particles ≥0.5 µm and ≥5 µm in the EU GGMP.
4. Discussion and conclusions
A method for carrying out the operational classification of cleanrooms and clean zones used in aseptic pharmaceutical manufacturing has been discussed in the first part (3) of a two-part article. The method was derived from the classification requirements and principles given in ISO 14644-1:2015 and Annex 1 of the EU GGMP but included more current expectations of the regulatory authorities. To demonstrate how this method can be used, an example is given in this second article, of a cleanroom and isolator used for aseptic filling of a liquid formulation into vials.
For the classification of the EU GGMP Grade C cleanroom, a method explained in the first article (3) to divide up the cleanroom into sampling sections, as required by ISO 14644-1, is explained. Also described is the method to identify the locations within each sampling section where the airborne concentration has to be measured. This is carried out using a formal risk assessment process that locates where the highest concentrations of particles caused by personnel activity are likely to be. The particle concentrations measured at these sampling locations are given and it was found that one result at the ≥5 µm particle size exceeded the limit. This result was used to illustrate two problems that may be encountered during classification.
Firstly, although the air sampling identified a location that failed to meet the airborne concentration limit for particles ≥5 μm, the corresponding concentrations at the ≥0.5 μm particle size passed. The distribution of particle sizes within pharmaceutical cleanrooms has previously been investigated (9) and it was shown that the limit allocated to particles ≥5 μm of 29 000 per m³ in the EU GGMP is too stringent when compared to the corresponding concentrations of ≥0.5 μm particles and airborne microbial contamination. For a Grade C area, with a stated concentration limit for particles ≥5 μm of 29 000 per m³, a more appropriate limit would be 88 000 per m³. The origin of the particle concentration limits in ISO 14644-1, and hence Annex 1 of the EU GGMP, and the reasons for these discrepancies are discussed in the first article (3). It is more likely that the classification (and the subsequent monitoring) would fail at the ≥5 μm size than at the ≥0.5 μm size. It is therefore important that classification is undertaken at the ≥5 μm size as well as the ≥0.5 μm size in order to avoid failures during monitoring.
Secondly, although the cleanroom is non-UDAF, and the airborne particle concentrations are expected to be reasonably even throughout the cleanroom, this was not the case in the cleanroom investigated, and the airborne concentration at the location that failed was much higher than the average in the cleanroom. An investigation was carried out into the ventilation effectiveness of the location that failed. This was firstly undertaken by calculating the Performance Index (PI) using the airborne concentrations obtained during the classification. Further experimental work was also carried out with test particles to determine the Air Change Effectiveness (ACE) index. With both of these indexes providing similar values, the high concentration of particles at the failed location was considered to be caused by the location receiving significantly less clean air than the average in the rest of the room. This was successfully addressed by the installation of an additional air extract at this location.
The classification method derived in the first article was also illustrated by its application to an isolator. The same method used for the cleanroom was used to divide up the isolator base area into the sampling sections, as required by ISO 14644-1. Also, a formal risk assessment method is the expectation of the regulatory authorities to identify the sampling positions to be used during classification. A risk assessment method described in the first article (3) considers risk factors that relate to product or critical surfaces exposure area, time of exposure, type of ventilation, and associated operator activities, to identify locations where the risks of product contamination might occur. This approach was used and illustrated for the isolator. It should be noted that the number of sampling locations identified by risk assessment was greater than calculated by the ISO 14644-1: 2015 method, but in line with regulatory authority expectations.
It should be noted that the risk assessments methods employed in this article for non-UDAF cleanrooms and UDAF clean zones can also be used as the bases of determining the locations for environmental monitoring (microbial and non-viable) of both cleanrooms and clean zones during manufacturing.
Appendix A: Determination of sampling sections and sampling locations in non-UDAF cleanroom
1. The cleanroom floor area (not including the isolator) should be divided into suitable sizes of rectangular sub-areas, starting with the largest area and working towards the smallest. These divisions are shown in Figure A1 by the black full lines, and the sub areas lettered A to E.
2. The floor surface areas (m²) of sub-areas A to E are added together to obtain the total area of the cleanroom.
Total floor area of cleanroom = A(2.8 x 6) + B(4.2 x 5) + C(0.8 x 3.2) + D(1.7 x 4.2) + E(1.7 x 5.1)
= A(16.8) + B(21.0) + C(2.6) +D(7.1) + E(8.7)
= 56.2 m²
3. Table A1 of ISO 14644-1 gives 12 as the minimum number of sampling locations for a cleanroom with a floor area of 56.2 m². If all sections are equal, this requires a minimum area of 4.7 m² per sub-area, and it is therefore assumed that any additional sub-areas should not be larger than 4.7 m².
4. The required number of sections in each rectangular sub-area of the floor is calculated using the following equation;
Number of sections = floor area of sub-areas x minimum no. sampling locations
total floor area
Where, ‘minimum no. of sampling locations’ is given in Table A1 of ISO 14644-1 and is 12.
Taking the sub-area B in the cleanroom, with the largest floor area of 21.0 m², as an example;
Number of sections in sub-area B = 21.0/56.2 x 12 = 4.5
This calculation should be repeated for all the 5 sub-areas A to E and the numbers rounded up to whole numbers (any number under 1 is assumed to be 1). These sampling sections total 15, and as this number exceeds the minimum number of 12 required by ISO 14644-1, it is acceptable. The dimensions of these divisions are shown in Figure A1 by use of red dash lines for the 15 sampling sections.
5. Starting with the largest rectangle sub-area B (4.2 m x 5 m), this floor area (21 m²) is divided by the rounded number of sections (5) to obtain the area of each section. However, with consideration for the rectangular shape of sub-area B, it would be most appropriate to utilise 6 sections which would readily fit the sub-area, each with an area of 3.5 m². This can be achieved by two options, namely, sections of either 1.67 m x 2.1 m, or 2.5 m x 1.4 m. With knowledge of the cleanroom activities associated with the generation of contamination, the most appropriate dimensions can be chosen that will place the activity locations closest to the centre.
The locations where personnel activities are carried out are shown as red spots in Figure 2. Also assessed is the amount of activity, and this is shown by the size of the red spot. Using this information, the most appropriate dimensions for each sub-area is considered to be the 1.67 m x 2.1 m configuration. This process is repeated for the other sub-areas, with the exception of sub area A where there is only one possible configuration, and for sub-area C where there is only a single section. The resultant sampling sections within each rectangle sub-areas are shown by dashed red lines in Figure A1 as A1, A2, etc. and detailed in Table A1.
Appendix B: Determination of sampling sections and sampling locations in UDAF isolator
Division of isolator base area into sampling sections
The process described in Appendix A to determine the number of sampling sections for the cleanroom is similarly applied to the isolator. The isolator base area (15.83 m²) is divided into 3 rectangular sub-areas, A (13.5 m²), B (1.53 m²) and C (0.8 m2) as shown by black solid lines in Figure B1. The resultant dimensions of each section associated with each of the 3 rectangle sub-areas are shown in Table B1 and also included in Figure B1 by the dashed red lines. These total 8, and as this number exceeds the minimum number of 6 determined by ISO 14644-1, it is acceptable.
The sampling locations within each of the 8 sections should be derived by further considerations of the activities within the isolators and use of a risk assessment. The risk assessment method is discussed in the next section and the resultant locations shown in Figure B1 as A1.1, etc. Also included are the other sampling locations discussed in section 3.5.
Selection of sampling locations in sections by risk assessment
The current expectation of the regulatory authorities is for sampling to be carried out in the isolator where the risks from airborne contamination are highest. The chosen locations should be in proximity to critical surfaces, such as where product, components or product contacting surfaces, are exposed to airborne contamination. An appropriate risk assessment method is described in the first article (3), which uses the risk factors and scoring system shown in Table B2.
The level of risk at each location can then be obtained by the following equation;
Risk = Severity x Occurrence
= (Personnel activity score x Ventilation type score x Surface exposed) x Time exposed
For the isolator manufacturing activities, the risk scores are calculated using this equation and are shown in Table B3.
a. It is assumed that all contamination on the internal vial surface is subsequently transferred to the product solution.
b. Each of the 4 needles have a diameter of 0.3 cm and length 7 cm and so a worst-case horizontal area that is exposed is assumed to be 2.1 cm².
c. All contamination on the external needles surface is assumed to be subsequently transferred to the product solution during the filling when the needles external surface contacts the solution and is distributed equally into the 1000 vials associated with each needle.
d. It is assumed that the stopper internal area is 2 cm² but only 25% are actually exposed on the upper most layer of the hopper and approximately 50% of these will be exposed with the internal surface facing upwards. Therefore, an average surface area of 0.5 cm² is assumed.
e. All manipulations from cleanroom without intrusion into the isolator.
f. All contamination on the internal stopper surface is assumed to be subsequently transferred to the product solution.