Research ArticleTechnology/Application

Use of a Real-Time Microbial Air Sampler for Operational Cleanroom Monitoring

T. Eaton, C. Wardle and W. Whyte
PDA Journal of Pharmaceutical Science and Technology March 2014, 68 (2) 172-184; DOI: https://doi.org/10.5731/pdajpst.2014.00952

Abstract

A sampler that detects and counts viable particles in the air of cleanrooms in real-time was studied. It was found that when the sampler was used to monitor airborne particles dispersed from a number of materials used in cleanrooms, including garments, gloves, and skin, the number of viable particles dispersed from these materials was greater than anticipated. It was concluded that a substantial proportion of these viables were of a non-microbiological origin. When the sampler was used to monitor a non-unidirectional airflow cleanroom occupied by personnel wearing cleanroom garments, it was found that the airborne viable concentrations were unrealistically high and variable in comparison to microbe-carrying particles simultaneously measured with efficient microbial air samplers. These results confirmed previously reported ones obtained from a different real-time sampler. When the real-time sampler was used in a workstation within the same cleanroom, the recorded viables gave results that suggest that the sampler may provide an effective airborne monitoring method, but more investigations are required.

LAY ABSTRACT: The airborne concentrations measured by a real-time microbial air sampler within an operational, non-unidirectional airflow cleanroom were found to be unrealistically high due to a substantial numbers of particles of non-microbiological origin. These particles, which resulted in false-positive microbial counts, were found to be associated with a number of materials used in cleanrooms. When the sampler was used within a cleanroom workstation, the counts appeared to be more realistic and suggest that this type of real-time airborne microbial counter may provide a useful monitoring method in such workstations, but further investigations are required.

Introduction

Real-time monitoring of airborne microbial concentrations within cleanrooms and controlled environments used in healthcare would provide considerable advantages. This is primarily due to the delay in obtaining results caused by the slow growth of microbes measured by conventional sampling methods. The advantages of such systems for monitoring in cleanrooms have been previously discussed (1).

Available real-time monitoring systems for sampling airborne microbes are based upon measuring the quantity of light emitted by particles that fluoresce when excited by laser light at a specified wavelength. The fluorescence produced from certain microbial constituents can be measured by this method, and the detection of a full range of micro-organisms, including viable and non-viable, aerobic and anaerobic, and those that are not cultured by normal methods, is potentially possible. However, the potential problem with real-time microbial air samplers is the tendency to mistake particles that fluoresce for microbe-carrying particles (MCPs) (false positives) and the failure to identify some MCPs. This article is mainly concerned with the problem of false positives.

We have previously investigated an instantaneous microbial air monitoring system and reported that sterile materials in the cleanroom dispersed substantial numbers of non-microbial fluorescing particles that appeared to falsely increase the number of viables. We also found that depending on the clothing worn, the real-time microbial sampler recorded between 7 and 94 times more viable particles than MCPs and that the size distribution of the viable particles more closely reflected the effect of the collection efficiency of the sampler on the particle distribution rather than the expected size distribution of MCPs. Other anomalous results were obtained that suggested that the counts of particles of a non-microbial origin may have considerably influenced the counts (2). Another similar programme of work was completed by Bjerner et al. (3), who reported similar results when comparing the counts from a real-time microbial sampler to those of microbial air sampler. A further investigation of another real-time microbial air sampler has therefore been completed to confirm these results. This present investigation was undertaken in three parts. In the first part, the sampler was used to determine the proportion of viable particles dispersed from materials typically utilised within cleanrooms. In the second part, the sampler was compared with high-efficiency microbial air samplers and a total particle counter within a non-unidirectional airflow operational cleanroom. In the third part, a similar comparison was undertaken during operations carried out within a unidirectional airflow workstation located within the same cleanroom. The programme of work was undertaken at AstraZeneca (Macclesfield, UK).

For clarification, the three types of particles referred to in this article are as follows:

  • Microbe-carrying particles (MCPs): The source of microbes in cleanroom air is personnel who disperse skin particles which carry one, or more, microbes, and are known as MCPs. MCPs are colony-forming units (cfu) and their airborne concentration is determined by conventional (active) microbial air sampling.

  • Total particles: These particles are individually counted and sized by particle air samplers that use light scattering techniques. They include all particles, including any that are also MCPs.

  • Viable particles: These particles are identified and counted by real-time microbial air samplers which analyse the fluorescence emitted by them to identify particles likely to be MCPs.

Materials and Methods

Part 1: Measurement of Viable and Total Particles Dispersed from Cleanroom Materials Using a Real-Time Microbial Air Sampler

Test Equipment

Real-time microbial air sampler:

A BioTrak® 9510-BD real-time microbial air sampler was tested. This samples air at a rate of 28.3 L/min (1 ft3/min). The particles that enter into the unit are firstly counted and sized in the same manner as a standard total particle counter. The air then enters a concentrator that uses the principle of a virtual impactor described by Hinds (4) to concentrate the particles into 1.0 L/min of air, to reduce the air speed though the viability detector to obtain a better response to fluorescence. The remainder of the air is exhausted from the side of the concentrator, at a rate of 27.3 L/min, and passes through a high-efficiency particulate air (HEPA) filter before it is vented from the system. The viability detector measures the intrinsic fluorescence of microbial constituents, such as tryptophan, nictoinamide adenine dinucleotide (NADH), and riboflavin, at a wave length of 405 nm. Any emitted fluorescence light is collimated, separated into two wavelength bands, and collected by two photomultiplier tubes where the optical signals are converted to electrical signals. The two intrinsic fluorescence wavelength bands and the Mie scattering signal are then processed to determine if a particle is viable in nature. Those particles that are considered to be of viable origin are counted and assigned to the following channel sizes:

  • ≥0.5 to <1.0 μm

  • ≥1.0 to <3.0 μm

  • ≥3.0 to <5.0 μm

  • ≥5.0 to <7.0 μm

  • ≥7.0 to <10.0 μm

  • ≥10.0 μm

The operational principles of this sampler are described in detail by Niccum and Hairston (5). When compared directly to a high-efficiency slit-to-agar airborne microbiological sampler, the BioTrak® sampler was reported to have an overall better efficiency of detection than a slit-to-agar method over a range of micro-organisms nebulised into an aerosol test chamber (6). The sampler has two sensitivity settings, “normal” and “high”. The high sensitivity setting increases the likelihood that microbes are detected, and as testing was carried out in EU GGMP Grade A and B areas (7) where the concentration of airborne microbes is low and difficult to detect, all testing results reported in this article was obtained by using the high setting. Additionally, the sampler incorporates a gelatine-based filter (37 mm diameter) that collects all of the particles that have been optically analysed, and it can be removed and transferred to microbiological growth medium and incubated for counting or identification.

Investigation Undertaken

Tests were carried out to determine the number of total and viable particles that were dispersed from a variety of materials used in cleanrooms. The testing was performed in a horizontal unidirectional airflow cabinet. The BioTrak® sampler was positioned outside the cabinet and a sterilised, particle-free tube from the counter was positioned inside the cabinet to provide remote sampling. The materials to be tested were transferred into the cabinet using, where applicable, sealed standard wrappings provided by the suppliers. The materials tested were as follows: sterile and unsterilised cleanroom hood, freshly laundered cleanroom undergarment top, sterile cleanroom wipe, sterile 70% isopropyl alcohol in water for injection (WFI) (70% isopropyl alcohol, IPA), sterile latex gloves, and ungloved hands. The first six items were agitated above the intake pipe, the sterile 70% IPA sprayed above the pipe intake, and either gloved or ungloved hands rubbed together above the pipe intake. These procedures were performed three times for each material tested, and both the resultant total particles at ≥0.5 μm and ≥5 μm and the viable counts were recorded.

Part 2: Comparison of a Real-Time Microbial Air Sampler with High-Efficiency Microbial Air Samplers, and a Total Particle Counter within an Operational Cleanroom

Cleanroom and Test Equipment

Cleanroom

The cleanroom utilised for the investigation was an EU GGMP Grade B (7) pharmaceutical non-unidirectional airflow cleanroom, in which are located two EU GGMP Grade A (7) unidirectional airflow workstations. For routine operation, two personnel may be present, one at each workstation, and the room has a maximum occupancy limit of three people. The room volume is approximately 53 m3 and has in excess of 100 air changes per hour. The cleanroom has been operational for a number of years and has demonstrated a high level of operational environmental cleanliness that easily meets the microbial and particle requirements defined for an EU grade B location (7). The cleanroom is shown in Figure 1.

Figure 1
Figure 1

Non-unidirectional airflow cleanroom.

Test Equipment

Real-time microbial air sampler:

The BioTrak® 9510-BD real-time microbial air sampler, sampling air at a rate of 28.3 L/min (1 ft3/min), was positioned at the back of the cleanroom. The sampler was fitted with the gelatine collection filter which, following testing, was transferred to a trypticase soy agar (TSA) medium plate and incubated under aerobic conditions at 32.5 °C for 5 days, and the number of colonies were counted and recorded.

To obtain a size distribution of the airborne viable particles, the mid-point of each channel size was plotted against the frequency of occurrence of that size of particle. The median particle size of the distribution was obtained by cumulating the counts and calculating the cumulated counts at each size as a percentage of the total count. The cumulative percentages were then plotted against the mid-point of the channel size to obtain the 50% cumulative count; this is the median size.

Andersen air samplers:

Two identical Andersen air samplers were used to sample MCPs in the experimental cleanroom. Further details regarding this type of sampler are described by Andersen (8). These samplers are cascade microbial air samplers that have six stages, each stage having 400 holes through which the sampled air passes, these holes decreasing in diameter down through the stages. Each stage has an agar plate below it and the impaction velocity onto each agar surface increases down through the stages; as the air passes down through the stages, the size of particle that is efficiently deposited onto the agar plate becomes smaller. This allows the median diameter and size distribution of the MCPs to be calculated by the method given by Kethley et al. (9). The size distribution is typically log-normal and hence a logarithm of the D50%, of the stage above, was plotted against the percentage cumulated counts calculated on a “less than stated size” for each stage. A regression equation was obtained of the plot, and the equivalent particle diameter at the 50% cumulative count point was determined. The values for the 50% cumulative particle size impacted on each stage of the Andersen samplers, in terms of equivalent particle diameter, were obtained from published results (9), and are as follows:

  • Stage 1, no stage above

  • Stage 2, 9.8 μm

  • Stage 3, 6.2 μm

  • Stage 4, 3.8 μm

  • Stage 5, 2.2 μm

  • Stage 6, 0.9 μm

When tested in comparison to other air samplers, the Andersen sampler is one of the most efficient available (10), especially when the intake cone is removed (11). Consequently, both samplers were utilised without the intake cones. One sampler was used with TSA medium plates and incubated aerobically, and the other sampler was used with Columbia horse blood agar plates and incubated anaerobically. Following testing, all plates were incubated at 32.5 °C for 5 days under aerobic and anaerobic conditions, respectively. After incubation, the number of colonies were counted and recorded. The Andersen units sampled at a rate of 28.3 L/min (1 ft3/min) and were connected in parallel to a vacuum pump, which was located in one of the EU GGMP Grade A (7) cabinets to ensure the emitted particles were not released into the cleanroom. The Anderson samplers were placed adjacent to the BioTrak® sampler.

AirTrace® Microbial Air Sampler:

To provide additional information regarding the airborne microbial concentrations, an AirTrace® slit-to-agar sampler that is routinely used to monitor the cleanroom air during manufacturing was used. It sampled at a rate of 28.3 L/min (1 ft3/min) and was placed in close proximity to the BioTrak® and Andersen samplers. The evaluation of the performance of this instrument was completed by the Health Protection Agency, UK, who reported it to have high collection efficiency (12) when tested in the manner suggested by ISO 14698 (13). Whyte et al. (14) have reported the D50% value for this sampler using a simple analytical approach was 0.25 μm, and using computational fluid dynamics was 0.23 μm. Standard 140 mm diameter plates filled with TSA were used which, following testing, were incubated under aerobic conditions at 32.5 °C for 5 days.

Lasair II® Total Particle Counter:

In addition to the total particle counts measured by the BioTrak® sampler, a Lasair II® total particle counter was used. This unit continuously samples air at a rate of 28.3 L/min (1 ft3/min) and, after each 1 min sample, the concentration of total particles per 28.3 L (1 ft3) at ≥0.5μm and ≥5.0μm are reported.

Investigation Undertaken

Test with Cleanroom Undergarments

For this test, seven people wearing only polyester/cotton cleanroom undergarments of the type shown in Figure 2 were utilised. The undergarments were laundered at a specialist garment company and a fresh set used every day and worn under full cleanroom attire. The 7 seven maximised their airborne contamination dispersal rates by marching on the spot, while swinging both arms, at a rate of approximately 1 beat per second for 4 min until the airborne particle contamination concentrations, reported by the Lasair II® total particle counter, indicated that a steady state condition had been attained. The people continued to march on the spot for an additional 5 min as the BioTrak®, AirTrace®, and Andersen samplers were all activated, and the Lasair II® total particle counter continued to sample. After 5 min, the BioTrak®, AirTrace®, and Andersen samplers were turned off, each having sampled 141.5 L (5 ft3) of air. All the microbiological media plates were removed and incubated, and the colonies were counted.

Figure 2
Figure 2

Testing in cleanroom undergarments.

Test with Full Cleanroom Attire

For this test, the same seven people were used. They were placed in the same locations, and moved in the same manner to that described for the first test, but dressed in standard full cleanroom attire. The cleanroom garment fabric is polyester with a fabric pore diameter of 13 μm. A one-piece coverall with hood and overboots were complemented by goggles, facemask, and double latex rubber gloves to provide full skin coverage. With the exception of the goggles, which were disinfected, all items are sterilised by Gamma radiation. This attire, which is shown in Figure 3, provides a very high level of personnel contamination containment; operational airborne microbial monitoring of the test cleanroom, with two personnel present and wearing this attire, gave an average count of 0.1 cfu/m3. Hence, seven people (the maximum number that could safely occupy the cleanroom) were utilised to obtain a higher concentration of airborne contamination that was within the accurate range of measurement of the microbial air samplers. When the airborne particle contamination concentrations, reported by the Lasair II® total particle counter, indicated that a steady state condition had been attained, the people continued to march on the spot for a further 15 min as the BioTrak®, AirTrace®, and Andersen samplers were all activated, and the Lasair II® total particle counter continued to sample. After 15 min, the BioTrak®, AirTrace®, and Andersen samplers were turned off, each having sampled 424.5 L (15 ft3) of air. All microbiological media plates were then removed and incubated, and the colonies were counted.

Figure 3
Figure 3

Testing in full cleanroom attire.

Part 3: Comparison of a Real-Time Microbial Air Sampler with a High-Efficiency Microbial Air sampler and a Total Particle Counter within a Cleanroom Workstation

Cleanroom Workstation and Test Equipment

Cleanroom Workstation

In these experiments, one of the EU GGMP Grade A (7) unidirectional airflow workstations in the cleanroom that is shown in Figure 1 was utilised. The workstation is supplied with HEPA-filtered air in a vertical unidirectional manner at a velocity of 0.46 m/s, and the operational environmental cleanliness easily meets the microbial and particle requirements defined for an EU grade A location (7).

Test Equipment

The samplers used in the previous experiments were used again. The BioTrak® and the AirTrace® samplers and the Lasair II® total particle counter were all positioned adjacent to the workstation, and the intake of each unit was connected to sterilised, short lengths of particle-free tubing fitted with isokinetic sampling heads. The three isokinetic sampling heads were all positioned together and located in the critical workstations adjacent to the weighing unit.

Investigation Undertaken

The workstation is routinely used to aseptically weigh individual, solid doses of product. This activity and other incidents were performed and monitored. An incident is considered to be a finite particle–generating activity in which the number of contaminants are counted as a single event and reported independently of the air sample volume. Six tests were sequentially completed using similar procedures. These were as follows:

  1. At rest: There was no activity within the workstation and the production equipment was not functioning. Sampling was carried out for 5 min.

  2. Set-up: The operator transferred and assembled the sterilised parts onto the weighing unit ready for the production activities. Sampling was carried out for 5 min.

  3. Production: The operator continuously added product onto the weighing machine. Sampling was carried out for 5 min.

  4. Incident 1: The production equipment within the cabinet was turned off and the operator repeatedly opened and closed long-length sterilised forceps, of the type routinely utilised within the workstation, with the pointed ends of the forceps positioned above the various sampler air intakes, for a period of 15 s.

  5. Incident 2: The operator rubbed his gloved (latex, sterilised) hands above the sampler intakes for a period of 15 s.

  6. Incident 3: The operator removed the gloves and rubbed his bare hands above the sampler air intakes for 15 s.

RESULTS

Part 1: Measurement of Viable and Total Particles Dispersed from Cleanroom Materials Using a Real-Time Microbial Air Sampler

The number of total particles, at ≥ 0.5μm and ≥ 5μm, and the associated viable particle counts that were measured by the BioTrak® sampler when dispersed from each of the tested materials are shown in Table I.

View this table:
Table I

Average Total and Viable Particle Counts from Typical Cleanroom Materials Recorded by the BioTrak® Sampler. Shown in Parenthesis are the Viable Particle Counts as a Percentage of the Total Particle Counts

Part 2: Comparison of a Real-Time Microbial Air Sampler with High-Efficiency Microbial Air Samplers, and a Standard (Total) Particle Counter within an Operational Cleanroom

Airborne sampling was carried out in an operational cleanroom using either cleanroom undergarments or full cleanroom attire. The air sampling methods were as described in the Test Equipment section. To enable direct comparisons of the cleanroom undergarments with the full cleanroom attire, the airborne concentrations from the cleanroom undergarments testing are reported throughout as counts per 424.5 L (15 ft3).

Total Particle Counts

The counts of total particles that were simultaneously recorded by the BioTrak® sampler and the Lasair II® total particle counter were found to be in good agreement and are shown in Table II. The total particles counted by the BioTrak® sampler are also given as a percentage of those counted by the Lasair II® total particle counter and show that the BioTrak® unit counted between 3% and 18% more total particles than that counted by the Lasair II® sampler. Also shown in Table II are the reductions in total particles, resulting from wearing the full cleanroom attire compared to those counted when wearing only the cleanroom undergarments. It can be seen that the reductions recorded when wearing the full cleanroom attire compared to the undergarments are similar for both instruments.

View this table:
Table II

Total Particle Counts (≥0.5 μm and ≥5 μm) per 424.5 L (15 ft3). Shown in Parenthesis are the Total Particles Counted by the BioTrak® Sampler Given as a Percentage of those Counted by the Lasair II® Counter

Viables and MCP Counts

The number of MCPs recovered by the Andersen and the AirTrace® samplers were compared to the number of viable particles recorded by the BioTrak® unit and are shown in Table III. For testing performed with personnel wearing full cleanroom attire, the number of aerobically incubated MCPs recovered by the AirTrace® air sampler was in reasonable agreement with the number of aerobically incubated MCPs recovered by the Andersen sampler.

View this table:
Table III

Airborne Viable and Microbiological Counts per 424.5 L (15ft3) When Cleanroom Undergarments and Full Cleanroom Attire were Worn

Although it is usual practice to incubate cleanroom microbial samples aerobically, there are a number of microbes that grow only anaerobically and may be part of the explanation as to why the count of viable particles was higher than MCPs. Therefore, all samples from one of the Andersen samplers were incubated anaerobically. There is also a group of micro-organisms that will grow in both anaerobic and aerobic conditions and known as facultative microbes. Although it is therefore not scientifically correct to add both types of micro-organisms to obtain a combined number of MCPs, doing so provides a best case comparison with the viables counted by the BioTrak® sampler, and therefore that approach has been used.

Viable Particles and MCPs as a Proportion of the Total Particle Counts

Shown in Table IV is a further analysis of the viable and total particle counts obtained in the cleanroom from the BioTrak® sampler, and the MCPs obtained from the Andersen sampler. Also shown in parenthesis is the percentage of viable particles and MCPs as a percentage of the ≥0.5 μm and ≥5 μm total particles counted by the BioTrak® sampler.

View this table:
Table IV

Viable Particles, Total Particles and MCPs per 424.5 L (15ft3) in Cleanroom Air. Shown in Parenthesis are both the Viables and MCPs Counts as a Percentage of Total Particle Counts Recorded by the BioTrak® Sampler

Sizes Distributions of Viable Particles and MCPs

The sizes distributions of viable particles recorded by the BioTrak® sampler, and the MCPs recovered by the Andersen samplers, are shown in Figure 4. The two methods by which the size distributions of the viable particles and the MCPs can be obtained is explained in Part 2. These methods are different, and to make a direct comparison between the two size distributions, while acknowledging that the size distribution of the MCPs is not log-normal, the method described for use with the viable particles was also used for the MCPs.

Figure 4
Figure 4

Size distributions of viable particles (BioTrak® sampler) and MCPs (Andersen samplers).

The median particle sizes of the size distributions were calculated using the methods given in Part 2 (without modification) and are shown in Table V along with the number of viable particles and MCPs recorded.

View this table:
Table V

Median Diameter of Viable Particles and MCPs

Part 3: Comparison of Real-Time Microbial Air Sampler with a High-Efficiency Microbial Air Sampler and a Standard (Total) Particle Counter within a Cleanroom Workstation

Airborne sampling of the total particles, viable particles and MCPs was carried out during the six tests in the unidirectional airflow workstation. The results are given in Table VI. It should be noted that the results of the first three tests are given as counts per 141.5 L (5ft3) and the rest of the tests as the number of particles counted over 15 seconds.

View this table:
Table VI

Total Particles, Viable Particles, and MCPs per 141.5 L (5 ft3) or per 15 s in the Unidirectional Airflow Workstation Air

DISCUSSION

A previous investigation of an instantaneous microbial air sampler showed that the sampler appeared to count many non-microbiological fluorescing particles in the air of a cleanroom as MCPs (2). Because of this, another sampler (BioTrak® 9510-BD real-time microbial air sampler) was studied.

It can be seen in Table I that most of the materials used in cleanrooms disperse particles that are counted by the BioTrak® sampler as viable particles. The percentage of viable particles ranged from 0.9% to 9.6 % of the ≥0.5 μm total particles although this was considerably less than the corresponding percentages (14% to 74%) recorded during the previous investigation of the same cleanroom materials with an alternative instantaneous airborne microbial detection system (2). The materials investigated were mainly sterile and should not disperse any MCPs. In addition, they were prepared and secured in highly controlled environments to ensure a very low bioburden challenge and, following sterilisation, would give a very low concentration of residual dead microbes that may be counted as viable particles. Nevertheless, the sterile cleanroom wipes, which were considered to be the most securely prepared cleanroom items, still recorded 3% of the ≥0.5 μm total particles as viable particles. Other items such as the filtered and sterilised 70% IPA and the sterile latex gloves recorded viable particles that were 6% and 9.6%, respectively, of the ≥ 0.5 μm total particles. These results suggest that particles dispersed from commonly used items in a cleanroom may fluoresce and increase the concentration of viable particles above the actual concentration of MCPs in cleanroom air. Morgan et al. (15) showed that fibres from materials such as nylon, acrylic, cotton, and polyester produced fluorescence during excitation at the wavelength used by the real-time microbial sampler (405 nm). In addition, the use of a simple laser pointer pen that produces a beam of light at 405 nm also reveals strong florescence from certain materials used in cleanrooms. However, there is no direct evidence that real-time microbial samplers would mistake these fibres for viable particles. However, it is interesting to note that a change in these experiments from dark-blue polycotton undergarments to full attire made from light-blue polyester effected a large drop in the airborne concentration of MCPs but a relatively small drop in the viable particles. This is discussed in further detail later in this discussion, and it is our opinion that this indicates that the type of clothing fabric has an influence on the number of false positives. The possibility should also be considered that materials that fluoresce may be introduced during the processing of garments in cleanroom laundries. This may occur when fluorescent additives are used in the wash, or by transfer from packaging materials. It would be useful if further research was carried out into the questions raised in this and the previous paragraphs.

A comparison was made in an EU GGMP Grade B (7) operational pharmaceutical cleanroom of the airborne concentration of MCPs, viable particles, and total particles. Shown in Table II is a comparison of the airborne concentrations of total particles measured by the BioTrak® sampler and Lasair II® total particle counter, when wearing either cleanroom undergarments, or full cleanroom attire. These concentrations are in good agreement with each other. In addition, both instruments recorded similar reductions of total particles when full cleanroom attire was worn in place of cleanroom undergarments.

The results in Table III show that the MCPs recorded by the Andersen air samplers and the AirTrace® sampler were in reasonable agreement with each other. When personnel wore undergarments, the BioTrak® sampler recorded 2.3 times more viable particles than the combined MCPs sampled by the Andersen samplers. The BioTrak® gelatine filter, which collects all of the particles that have been optically analysed by the viable detector, recovered only 225 MCPs (22.7%) of the 990 viable particles. When full cleanroom attire was worn, no MCPs were recovered by the Andersen and AirTrace® samplers, but 188 viables were counted by the BioTrak® sampler and only 1 MCP was recovered from the gelatine filter, which was only 0.5% of the 188 viable particles.

Further analysis of the airborne sampling results obtained in the cleanroom is given in Table IV and shows that the number of MCPs recovered from the BioTrak® gelatine filter when personnel wore full cleanroom attire was 225 times less than when cleanroom undergarments were worn. Similarly, the total number of MCPs recovered from the air by the Andersen samplers showed a greater than 441 times reduction. Unpublished work carried out at AstraZeneca Macclesfield, using a personnel dispersal chamber of the type described by Whyte et al. (16), gave similar differences. It might therefore be expected that the BioTrak® sampler would give a similar reduction of viable particles, but only a 5.3 times decrease was measured. This result is also contrary to what might be expected from the results obtained from the BioTrak® unit and Lasair II® samplers shown in Table I, which give relatively uniform reductions of total particles when full cleanroom attire was worn instead of cleanroom undergarments. The only reason for the difference was the clothing worn, and the anomalous result for the viable counts is considered to be caused by a difference in the rate of dispersion, and perhaps the type of fluorescent, of particles shed by cleanroom undergarments and full cleanroom attire. This is further highlighted in Table IV, where the viable particles and MCPs are shown as a percentage of the associated total particle counts measured by the BioTrak® sampler. The viable particles were found to account for over 47% of the total ≥5 μm particles present in the room air when the full attire was worn, although the MCP contamination was zero. These results show that when real-time air sampling is carried out in an EU GGMP Grade B cleanroom (7), considerably more viable particles will be counted than MCPs obtained by traditional methods. If the viable counts also show variation between different types of real-time samplers, this could lead to difficulties in setting airborne microbial limits that would conform to those limits set by the regulatory authorities.

It should be noted that each test was only completed once. This was partly due to limited availability of the cleanroom, and further studies would be ideally required to demonstrate the repeatability of the results. However, the results obtained were consistent with those obtained from a similar but more extensive set of experiments (2) performed under identical conditions and in the same cleanroom but with a different real-time microbial sampler. They also confirm the results of the comparison of the counts from a real-time sampler and a microbial air sampler carried out by Bjerner et al. (3), providing some confidence that the results of these experiments are reasonably representative of what would be found when sampling in a cleanroom.

Microbes in the air of occupied rooms are derived from personnel who disperse skin cells that carry micro-organisms (17, 18), and many such MCPs pass through cleanroom clothing (16) and into the cleanroom air. It is expected that the size distribution profile of MCPs sampled by the Andersen sampler should be very similar to the viable particles. Figure 4 shows that this was the case with viables recovered from the cleanroom undergarments, but in the case of full cleanroom attire, the MCPs were much larger and more evenly distributed than the viables. The viable particles had a definite peak and a median diameter of 2 μm, and this diameter is very much smaller than the MCPs found in the experimental cleanroom and reported in other occupied rooms (16, 19, 20). Information about the concentrator in the BioTrak® unit (6) shows that it has a 50% recovery at a particle cut-off size of 2μm and, below that size, the recovery efficiency drops. The efficiency remains at around 55% in the range between 2 and 6 μm, but above 6 μm the recovery efficiency again drops. This is a likely explanation for the shape of the distribution of the sizes of the viable particles given in Figure 4. These results suggest that the proportion of viable particles sampled is likely to below half of the total number in the air. However, this should not necessarily be considered a problem, as the collection efficiency of routinely-used microbial air samplers is often similar or poorer that this (10, 14). The median sizes of the combined aerobically and anaerobically incubated MCPs measured in these experiments was 6.8 μm, and within the expected size range.

In this investigation, the microbial samplers were selected because of their high collection efficiencies, and the combined counts of MCPs included microbes that were incubated anaerobically. However, these additional sampling measures failed to sufficiently increase the number of MCPs to account for the much larger number of viable particles detected by the BioTrak® sampler. The BioTrak® sampler would be expected to count a larger number of micro-organisms than traditional methods, as viable particles may include non-viable microbial cells and microbes that do not grow using normal methods. However, evidence of the much greater number of viable particles than MCPs found in clean areas, and the substantial dispersion of viables from sterile materials used in the cleanroom, suggests that the BioTrak® sampler was unlikely to be counting only MCPs. Improvements to obtain viable counts that are closer to the concentration of MCPs measured by traditional means are required. This is likely to be difficult, as the instrument should recognise and count MCPs without additionally counting non-microbial particles. To provide confidence that the instrument works well in this role, validation is required. However, it seems sensible to validate the instrument using MCPs of the type found in cleanrooms, which are mostly skin cells that carry microbes and are dispersed by people, rather than particles that can be estimated from information in the pre-validation testing report (6) to contain about 20 microbes per particle. These artificially generated particles may register a larger and different response than actual MCPs in cleanroom air, and this possibility should be investigated.

The usefulness of a real-time viable particle air sampler was also investigated in a unidirectional air flow workstation. Unfortunately, because of the very low concentration of airborne contamination in the workstation it was not possible to collect sufficient samples to determine the actual concentrations. However, the results shown in Table VI shows that during the “at rest”, “equipment set-up”, and “production” testing, no viable particles were recovered by the BioTrak® sampler, and no MCPs were recovered by either the AirTrace® sampler or gelatine membranes. Further tests that generated greater amounts of airborne contamination than normal were carried out using forceps and both gloved and ungloved hands. When sterilised forceps were used to generate a contamination incident, a few total particles were collected, but no viable particles were recovered by the BioTrak® sampler or MCPs recovered by the AirTrace® sampler or gelatine filter. This is expected, as any particles associated with the sterilised forceps are inanimate. Similarly, the gloved hand generated only a single total particle that was counted by the BioTrak® sampler. However, ungloved hands would be expected to disperse both total particles and MCPs. Total particles were counted by both the BioTrak® and Lasair II® samplers, and the BioTrak® sampler counted viable particles, but no MCPs were recovered by the AirTrace® sampler or gelatine filter. Although this investigation within a cleanroom workstation is limited, the results suggest that where the airborne contamination of the air is very low, false positives might be expected but are unlikely to be present in sufficient numbers to cause the regulatory limit to be exceeded. Where this does occur, there is likely to still be an advantage to more efficient production because product that is filled a few minutes before the viable particle is registered can be rejected and normal filling can be carried on. However, further studies of this topic are required.

CONCLUSIONS

Sampling airborne MCPs in cleanrooms in real time would be of considerable benefit in pharmaceutical production, and samplers that measure and analyse the fluorescence from airborne particles are being developed for such a task. However, these samplers should not erroneously count large numbers of MCPs or particles that are not MCPs (false positives). This paper is mainly concerned with false positives.

It was shown in the experiments described in this article that sterile materials commonly found in cleanrooms, such as wipes, IPA disinfectant, garments, and so on, disperse large quantities of viable particles that are measured by real-time microbial air samplers.

The BioTrak® sampler was studied in an operational non-unidirectional airflow cleanroom, and the airborne concentrations were found to be unrealistically high and variable in comparison to MCPs simultaneously measured by efficient microbial air samplers. The viable particles also had a different size distribution and a smaller median size than the MCPs. It was concluded that the real-time system was measuring substantial numbers of viable particles of non-microbiological origin (false positives). A direct comparison of the traditional and real-time microbial air sampling methods is not easy. This is because the two methods are altogether different and the new technology has the potential to detect a fuller range of microbes, including those described as “viable but non-culturable”. However, traditional air sampling methods based on growth based recovery of airborne contaminants on solid media are required to demonstrate compliance with regulatory standards. If future regulations are to be adapted for use with real-time microbial sampling, appropriate limits will need to be defined. Such limits must take into account that the real-time samplers give considerably different counts from the traditional methods, and that fluorescing materials used in the cleanroom have a strong influence on these counts.

Of most concern to the pharmaceutical industry are airborne particles carrying viable micro-organisms that may deposit onto, or into, products and then proliferate to cause contamination, spoilage, or infection. Other types of particles, such as dead microbes, or inanimate particles that fluoresce when exposed to laser light, are of much less a concern. Consequently, unless the real-time sampler could more accurately differentiate between micro-organisms and non- microbiological fluorescent particles, this type of real-time microbial detection system is not considered to be suitable for monitoring microbiological contamination within EU grade B, C, and D (7) cleanrooms. Also, as skin particles dispersed from personnel are the source of MCPs in cleanrooms, it would be useful to carry out correlations of counts from real-time microbial air samplers by using skin particles rather than artificially-generated particles containing microbes. By such means, a better correlation to the type and size of MCPs actually found in cleanrooms might be achieved.

When the BioTrak® air sampler was used within an operational cleanroom workstation where the airborne concentration of MCPs is low, the sampler appeared to have the potential to provide real-time airborne microbial monitoring. Although only a small number of results were collected, the results obtained suggest that this type of airborne monitoring in workstations may provide a desirable and effective method, and further investigations in such areas would be useful.

Conflict of Interest Declaration

The authors declare that they have no financial or non-financial conflicts of interest regarding this article.

Acknowledgements

The authors are thankful to TSI Incorporated for the loan of the BioTrak® 9510-BD real-time microbial air sampler.

References

  1. 1.↵
    1. Sandle T.
    Real-time counting of airborne particles and microorganisms: A new technological wave? Clean Air and Containment Review 2012, 9, 4–6.
    OpenUrl
  2. 2.↵
    1. Eaton T.,
    2. Davenport C.,
    3. Whyte W.
    Airborne microbial monitoring in an operational cleanroom using an instantaneous detection system and high efficiency microbiological samplers. Eur. J. Parenter. Pharm. Sci. 2012, 17(2), 61–69.
    OpenUrl
  3. 3.↵
    1. Bjerner G.,
    2. Kene V.,
    3. Åkerlund E,
    4. Andersson K.
    Real-time microbiological air monitoring. Eur. J. Parenter. Pharm. Sci. 2012, 17(3): 52–55.
    OpenUrl
  4. 4.↵
    1. Hinds W. C.
    Chapter 5. Virtual Sampling. In Aerosol Technology, 2nd ed.; John Wiley and Sons: New York, 1999.
  5. 5.↵
    1. Molendhauer J.
    1. Niccum D.,
    2. Hairston P.
    Chapter 13. Real Time Clean Room Monitoring for Total and Viable Particles: BioTrak® Viable Particle Detector. In Environmental Monitoring: A Comprehensive Handbook, Molendhauer J., Ed.; Vol. 6, pp 219–240. PDA DHI Technical books, Bethesda, MD, USA (ISBN: 1-933722-63-0).
  6. 6.↵
    BioTrak® real-time viable particle counter prevalidation test report. January 2013. TSI reference 5001493A.
  7. 7.↵
    European Commission Guide to Good Manufacturing Practice, Annex 1: 2008; European Commission, Enterprise Directorate-General: Brussels, 2008.
  8. 8.↵
    1. Andersen A. A.
    New sampler for the collection, sizing and enumeration of viable airborne particles. J. Bacteriol. 1958, 76(5), 471–484.
    OpenUrlFREE Full Text
  9. 9.↵
    1. Kethley T. W.,
    2. Cown W. B.,
    3. Fincher E. L.
    Bacteria Load in Simulated Operating Rooms. Progress report OH-19. Engineering Experiment Station, Georgia Institute of Technology: Atlanta, GA, 1962.
  10. 10.↵
    1. Ljungqvist B.,
    2. Reinmüller B.
    Active sampling of airborne viable particles in controlled environments: a comparative study of common instruments. Eur. J. Parenter. Sci.1998, 3(3), 59–62.
    OpenUrl
  11. 11.↵
    1. May K. R.
    Calibration of a modified Andersen bacterial aerosol sampler. Appl. Microbiol. 1964, 12(1), 37–43.
    OpenUrlPubMed
  12. 12.↵
    Health Protection Agency. Physical Efficiency Testing of the Air Trace Environmental Slit to Agar Impaction Sampler. Health Protection Agency (Porton Down) Report 843/0, 10.03.04; Porton Down: Wiltshire, UK.
  13. 13.↵
    ISO/FDIS 14698-1(2003). Cleanroom and associated controlled environments – Biocontamination control. Part 1: General principles and methods Annex B (informative) Guidance on validating air samplers, 2003.
  14. 14.↵
    1. Whyte W.,
    2. Green G.,
    3. Albisu A.
    Collection efficiency and design of microbial air samplers. J. Aerosol Sci., 2007, 38(1), 101–114.
    OpenUrl
  15. 15.↵
    1. Morgan S. L.,
    2. Hall S. H.,
    3. Hendrix J. E.,
    4. Bartick E. G.
    Pattern recognition methods for the classification of trace evidence textile fibres from UV/visible and fluorescence spectra. Proceedings of the FBI Trace Evidence Symposium: Clearwater, FL, 2007. Available at http://projects.nfstc.org/trace.
  16. 16.↵
    1. Whyte W.,
    2. Hejab M.
    Dispersion of particles and microorganisms from people. Eur. J. Parenter. Pharm. Sci. 2007, 12(2), 39–46.
    OpenUrl
  17. 17.↵
    1. Davies R. R.,
    2. Noble W. C.
    Dispersal of bacteria in desquamated skin. Lancet 1962, 280(7269), 1295–1297.
    OpenUrlCrossRef
  18. 18.↵
    1. McIntosh C.,
    2. Lidwell O. M.,
    3. Towers A. G.,
    4. Marples R.
    The dimensions of skin fragments dispersed into the air during activity. J. Hyg. (London) 1978, 81(3), 471–479.
    OpenUrlPubMed
  19. 19.↵
    1. Lundholm I. M.
    Comparison of methods for quantitative determinations of airborne bacteria and evaluation of total viable counts. Appl. Environ. Microbiol. 1982, 44(1), 179–183.
    OpenUrlAbstract/FREE Full Text
  20. 20.↵
    1. Noble W. C.,
    2. Lidwell O. M.,
    3. Kingston D.
    The size distribution of airborne particles carrying micro-organisms. J. Hyg. (London) 1963, 61(4), 385–391.
    OpenUrlPubMed
Back to top

In This Issue

Print
Download PDF
Article Alerts
Email Article
Citation Tools
Use of a Real-Time Microbial Air Sampler for Operational Cleanroom Monitoring
T. Eaton, C. Wardle, W. Whyte
PDA Journal of Pharmaceutical Science and Technology Mar 2014, 68 (2) 172-184; DOI: 10.5731/pdajpst.2014.00952
Twitter logo Facebook logo Mendeley logo

Jump to section

Keywords