The Application of Robotics to Aseptic Environmental Surface Monitoring

Robotic sampling can be applied to the surface sampling of isolators, restricted access barrier systems (RABS), and other critical environments where sampling by human operators results in a potentially risky intervention near sterile surfaces and components or is ergonomically difficult to achieve. Although this robotic environmental monitoring system has previously been reported upon in several public lectures, this article for the first time provides data regarding the performance of this system in surface sampling (1). This automated approach to environmental monitoring—aimed at reducing interventions—was developed in response to the US Food and Drug Administration's 2004 Guideline on Sterile Drug Products Produced by Aseptic Processing as well as the agency's report on Pharmaceutical CGMPs for the 21st Century: A Risk-Based Approach. The ultimate objective is to eliminate human interventions in critical aseptic space and to collect process control data in real time as much as current technology allows (2). This article presents a study of an aseptic environmental monitoring system for surface contamination at critical areas using a robot that can be programmed according to predetermined sampling parameters.

Figure 1

The robotic system described conducts active-air and settle-plate monitoring as well as surface monitoring using swabs. However, this article reports only on studies done on the surface-sampling capabilities of this robotic system. It is anticipated that in production operations, this system would handle surface and air sampling and that nonviable or total particulate sampling would be automated as well, using currently available technology. The authors hypothesized that a robot that can be programmed in terms of pressure applied to a swab, and also can be programmed to sample a given surface in a highly reproducible manner in terms of area sampled, might actually yield better recovery than a human sampler.

In the near future, an automatic environmental monitoring system could also include rapid microbiological methods, which would allow the direct, and nearly real-time enumeration of contamination. The authors understand, however, that because those technologies would enumerate cells rather than colony forming units (CFU), some reconsideration of monitoring levels or contamination incidence rates would be required.

Figure 2

Isolators were introduced for aseptic processing in the 1980s, and for the most part, conducting environmental monitoring has been logistically difficult. In some projects, microbiological monitoring could not be easily accommodated, and modifications to the system's design were required after the fact. At present, the reliability of gloves and half-suits is widely considered a weak point in isolator or RABS operations. The authors believe that the automation of environmental monitoring to avoid direct human intervention will be a requirement in the evolution of gloveless aseptic processing systems. In addition, the automation of environmental monitoring and consideration of locations and sampling points during the construction of the filling system will obviate planning and logistics issues concerning environment monitoring (3,4).

A corollary benefit of an automated approach to microbiological monitoring will be near elimination of false positives (2). The system reported in this article requires no manipulation by human operators either in preparing samples or in handling them. Also, because the sampling materials can be identified automatically by barcode, the control system can correlate the sample identification with location and time that are then stored electronically (1). All sample materials are vapor-phase-hydrogen-peroxide-decontaminated before use.

Methodology

Two species of microorganisms, Micrococcus luteus NBRC13867, a vegetative Gram-positive coccus, and Bacillus atrophaeus ATCC 9372, a spore, were selected as test organisms. These microorganisms were selected because they are representative of common human associated organisms, which are often Gram-positive cocci, and also of spore-bearing organisms that may enter on supplies or apparatus that might not have been properly decontaminated. Small stainless-steel coupons were used as carriers of the test microorganisms. The carrier surface was polished to 320 grit and a predetermined number of CFU was inoculated on the polished surface of the carrier. A metal surface finish of 320 grit was chosen because it is broadly representative of a level of polish applied to critical surfaces of the filing and component feed machinery. In general, the surfaces of isolator enclosures are given a similar surface finish. The inoculated carriers were air-dried under HEPA-filtered uni-directional airflow.

Table I: Swab sampling parameters.

After some preliminary experimentation with commercial swabs (Becton, Dickinson and Company in Franklin Lakes, NJ, and ITW Texwipe in Mahwah, NJ), the authors developed specifications for a swab apparatus that was then custom made. After test operation, all swab rods were incubated in Soya Bean Casein Digest broth at 30° C for two weeks and then inspected for the presence of colonies indicative of microbial growth. The relatively long incubation time was chosen to ensure maximal recovery.

Table II: Dry swab efficacy (robot).

Swab sampling conditions

The swab sampling experiments were conducted using the test parameters listed in Table I. The authors hypothesized that when the wet-swab method was used, the microbial recovery rate might vary as a function of moisture content in the swab tip. Therefore, two moisture levels were evaluated in this study. To ensure a very moist swab in one study, the tip was moistened immediately before use. To achieve a higher moisture level, swabs were soaked one day before use to ensure the full saturation of the swab tip.

Table III: Wet swab efficiency with distilled water as wetting agent (robot).

Results

When the dry-swab method was used, the observed recovery (CFU) correlated well with the inoculation level for B. atrophaeus spores. No M. luteus colonies were observed after incubation. A high swabbing force resulted in better recovery of spores (see Table II). In addition, the wet method, in which distilled water was used as the wetting agent, gave much better recovery efficiency than was observed using dry swabs. Sufficient time for absorption of water into the swab tip resulted in more efficient recovery. In addition, recovery efficiency was significantly affected by the speed at which the swab moved across the surface being sampled (see Table III).

Table IV:Wet swab efficiency with distilled water 10.1%Tween80 as wetting agent (robot) .

When the wetting agent was changed from distilled water to mixture 0.1%Tween 80 in distilled water, the rate of recovery of M. luteus improved, but complete recovery could not be achieved. Increasing the number of back-and-forth swabbing motions actually resulted in a decrease in recovery rates (see Table IV). A comparison of manual sampling to robotic sampling under the same test conditions showed a lower recovery rate associated with manual swabbing. The recovery efficiency observed with manual swabbing was approximately 20%. Additionally, far more variability in recovery was observed with manual swabbing as compared with robotic swabbing (see Table V).

Table V: Wet swab efficiency (manual).

Discussion

Wet-swab sampling often is avoided in aseptic operations because of concerns about residual moisture that might encourage microbial proliferation. However, the experimental data clearly confirm that moistened swabs are much more effective in terms of contamination recovery. We observed that using surfactant Tween 80 resulted in improved recovery as compared with using plain distilled water. It has been reported that dry, agglomerated microorganisms can strongly adhere to surfaces; therefore, it is logical that moistened swabs, particularly with the addition of a surfactant, can recover this contamination from surfaces. Given the benefits of using moistened swabs, surface monitoring could best be executed just after the completion of an aseptic production operation or in advance of breaking the aseptic condition of the isolator or RABS. As demonstrated in Table IV, even when using the robot for sampling, better efficiency of recovery was observed with B. atrophaeus than with M. luteus.

Figure 3

The most significant difference between these two test organisms is that M. luteus is a vegetative cell and B. atrophaeus is a spore. The sizes are also different. The size of the former is ~0.7–0.8 µm, and the latter has a diameter of ~1 × 2–3 µm, although it is unknown to what degree the size of the organism plays a role in recovery efficiency. Consequently, the authors believe the primary considerations in recovery efficiency differences between these two organisms are: spores as compared with vegetative cells, and the agglomeration and adherence to the test coupon. The authors also believe the physical composition of the swab tip may play a role in recovery efficiency. Additional investigation is needed to determine which factors have the greatest influence on recovery rates.

Figure 4

Recovery efficiency was influenced by the time allowed for soaking the swab tip with the wetting agent. Wetting the swab immediately before use does not appear to allow sufficient time for saturation to occur. The authors believe that wetting immediately before use results in a swab that is completely dry before the completion of the swabbing action and therefore less efficient. In this case, the full benefits of wet swabbing are not realized. When the swab tip is allowed to soak overnight before use, however, the swab retains moisture from beginning to end of the swabbing action. The fully saturated swab gave a superior recovery efficiency of microorganisms. Therefore, the degree of wetness of the swabs plays a critical role in recovery efficiency.

Figure 5

Complete recovery of M. luteus could not be achieved in these experiments, which is not surprising. However, the robotic sampling operation could achieve much better recovery of M. luteus than could manual sampling. This better result could be attributed to the robot's more precise control over sampling speed, pressure, and area (5) (see Figures 3–7).

Figure 6

Conclusion

The experimental data suggest that the operating parameters of robotic surface sampling shown in Figure 7 result in the best rates of recovery. However, further investigation will be required to determine whether higher recovery rates can be obtained with other robotic sampling parameters. The authors will investigate both the influence of robotic sampling parameters as well as other variables such as swab-tip composition, flexibility of the swab's rod, and recovery media. It is significant that the recovery rates observed with the robotic swab method were in many cases three to fourfold higher than those that resulted from manual swabbing under otherwise identical experimental conditions. These results indicate that robotic sampling has the potential to improve the limit of detection of surface monitoring and also to reduce variability.

Figure 7

The authors believe the automated approach to monitoring reduces the likelihood of false-positive results, although they did not attempt to quantify this benefit. Of course, because the robotic sampling parameters are variable and user-selectable, it will be possible for users to define sampling conditions that are optimal for their particular application (6). The authors are also aware that the robotic surface-sampling method may afford advantages in terms of chemical recovery. This means that a robot installed in an isolator for microbiological sampling could be used to assess cleaning effectiveness as well. This, of course, would depend upon the location of the robot within an enclosure or RABS environment.

Mayumi Maruyama, Tomoo Matsuoka, and Motonari Deguchi work in the Microbial Control Department at Shibuya Kogyo Co., Ltd. in Kanazawa, Japan. James E. Akers* is the president of Akers Kennedy & Associates, PO Box 22562, Kansas City, MO 64113, akainckc@aol.com.

*To whom all correspondence should be addressed.

References

1. J.E. Akers and Y. Oshima, "PAsepT, Aseptic Vial Filling Processing Based on Principles of PAT," in Presentation at the ISPE Annual Conference (San Antonio, TX, 2004).

2. M. Deguchi, et al, "Development of an Advanced High Speed Aseptic Filling System," in Proceedings of the PDA International Congress, (Kyoto, Japan, 2001).

3. B. Ljungvist, B. Reinmüller, and R. Nydahl, "Microbiological Assessment in Clean Rooms for Aseptic Processing," J. R3 Nordic 23(3), 7-10, 1995.

4. J.E. Akers, "Environmental Monitoring and Control: Proposed Standards, Current Practices and Future Directions," J. Pharm Sci. Techn. 51(1), 36-47, 1996.

5. NASA, NASA Handbook for Biological Engineers (National Aeronautics and Space Administration, Washington, DC, 1971).

6. J. Levchuk, "FDA's Perspectives on Aseptic Process Validation," in Proceedings of the Third Annual GMP by the Sea Conference (University of Rhode Island School of Pharmacy, Kingston, RI, 2003).