Introduction
This paper discusses microbiology in the pharmaceutical quality control environment and the opportunities for development and application of new microbiology methods. Many new methods use technologies developed for space research [1], clinical studies [2], and the food industry [3]. While it may seem odd that the pharmaceutical industry lags behind in implementing new microbiological technologies, it can be readily explained as a resistance to change spawned, in part, by assay complexity and regulatory pressures.
In a regulated environment, once a method is accepted, there is significant corporate pressure to maintain procedures the same, thereby avoiding delays associated with regulatory scrutiny. As such, “accepted methods” are used repetitively, and often without question. Therefore, unfortunately, without critical evaluation, pharmaceutical microbiology and process understanding fail to advance and serve only to satisfy regulatory requirements.
Quality Control and Quality Assurance
Established standards for “quality control” are required by the Food, Drug and Cosmetic Act (FD&C) and the Code of Federal Regulations (CFR). For new drug products, FD&C section 505(b)(1)(D) requires full descriptions of manufacturing controls in a drug application. Regulations [21 CFR 314.50(d)(1)] require that in-process control tests and specifications be described in the chemistry, manufacturing, and controls (CMC) technical section of a new drug application submitted for review. Analogous requirements exist for generic drugs, animal drugs, biologics, and devices. Furthermore, current good manufacturing practice (CGMP) regulations (21 CFR 211) establish minimum practices to assure that products meet quality and purity expectations. The QA program has the responsibility of assuring the product meets these requirements, and the QC program assures the reliability of data used in that determination [4].
There is sufficient overlap in the roles of QA and QC that the responsibilities are often blended. In fact, there is no specified distinction between QA and QC within the CFR. The regulations (21 CFR 211.22) stipulate the requirement and responsibilities of a “quality control unit.” In general, it is up to individual companies to determine how the functions of the “quality control unit” will be apportioned between QA and QC. However it is done, the roles and responsibilities of QA and QC units should be clearly described in the company’s SOPs. The roles of the pharmaceutical engineers, chemists, and microbiologists should provide seamless support to the overall function of the quality unit, and that can only be done when each technical unit is led by technically independent managers with appropriate decision-making authority.
FDA representatives view quality assurance as a system of activities to assure a product meets its defined standards with a level of assurance [5]. For example, in producing a parenteral product, the defined standard of “sterile” requires a greater level of assurance than that afforded by a compendial sterility test. For that reason, the FDA’s 1994 guidance for submitting documentation for sterilization process validation stated, “The efficacy of a given sterilization process for a specific drug product is evaluated on the basis of a series of protocols and scientific experiments designed to demonstrate that the sterilization process and associated control procedures can reproducibly deliver a sterile product” [6]. This approach overcomes the inherent variability in sterility testing with an overall quality control program that uses process design and process control tests to assure product sterility.
Pharmaceutical products and processes use fewer microbiology tests than chemistry tests, so microbiology resources (personnel, funding, and space) are proportionally less. As awareness of microbiology principles becomes proportionally less, the managers of quality units are less likely to understand microbiology data and conclusions. This is a historical trend and one not limited to the pharmaceutical industry. For example, environmental microbiology suffered similarly during the era of EPA’s “superfund” as Quality Assurance (QA) programs focused predominantly on chemical assays were managed by chemists [4]. Under these conditions, pharmaceutical microbiologists build quality measures based on chemistry measurement principles, which is not a good fit.
However, microbiology plays a critical role in pharmaceutical quality control, specifically evaluating raw materials, process controls, product release tests and product stability tests. The quality and interpretation of these tests’ data critically impacts product safety. It is the Quality Control (QC) unit which assures that data from these tests are meaningful (reliable and precise) and have a minimum of error. QA units must evaluate the suitability for use of microbiology tests, the limitations of their applicability and measurements, and whether acceptance criteria were met. Therefore, to understand a process and product’s microbiological attributes, the overall quality unit (QC and QA) must understand both the nature of the tests and the data derived from them.
Product Quality
In the larger scheme, quality management is an overall corporate “mindset.” No individual or corporate unit “owns” product quality even though selected organizational units are responsible for its design, performance, and verification. However, within the operations of a robust QA unit, there should be clearly defined areas of responsibility for management, supervisors, technical staff, and contract staff [4]. Furthermore, the quality unit must be responsive. Regulation (21 CFR 211.22(d)) states, “The responsibilities and procedures applicable to the quality control unit shall be in writing; such written procedures shall be followed.” Yet, these written procedures must not be so inflexible as to present barriers to improvement. FDA’s initiative to improve product quality, “Pharmaceutical cGMPs for the 21st Century: A Risk-Based Approach” [7], paved the way for process analytical technology (PAT). As stated in the PAT guidance [8], “In a PAT framework, validation can be demonstrated through continuous quality assurance where a process is continually monitored, evaluated, and adjusted using validated in-process measurements, tests, controls, and process end points.” Further, as reported in the 2003 progress report of the Pharmaceutical GMPs [9], the initiative included a focus on the science of manufacturing in order to identify “...efficient approaches for characterizing and controlling critical manufacturing process parameters and quality assurance.” Implementation of new process monitoring and control methods will bring the industry closer to this quality goal. Whereas the latest PAT technologies encompass “real time” and “at-, on- or in-line” process assessments, most rapid microbiology tests are, at present, “off-line” laboratory tests. However, FDAcontinues to support the development of new microbiology methods and recognizes that appropriate new microbiology technologies will be part of up-to-date quality systems. The microbiological aspects of product safety must be considered early in manufacturing process design. Regulators and pharmaceutical industry scientists similarly view the goals of drug product quality microbiology in the following simplified terms:
1. limit adventitious microbial contamination and prevent the presence of harmful organisms (and their byproducts) in non-sterile products (e.g., component and process control tests)
2. minimize the chance of microbial contamination in products purporting to be sterile (e.g., component, process controls, and validation of the sterilization process)
3. maximize product safety and quality by controlling the presence of microbes that gain entry inadvertently (i.e., preservatives effectiveness)
Therefore, QA units must evaluate what the microbiology tests are designed to assess, and whether those data actually do, or do not, demonstrate process control and product safety. To accomplish this, test methods must measure “meaningful” attributes in distinguishable data ranges.
Evolution of Microbial Counting Methods
The history of microbiology is filled with timely observations and refinements of prior advances [10]. Microbial count methods have gone through notable changes in that way. From its beginning, microbiologists have struggled to gain understanding of what a “count” means. The “colony count” is an indirect count with variable insensitivity [11], and is imprecise when very few colonies are counted [12]. Recognizing that the count was imperfect, researchers focused on applying standard methods that allowed data comparisons.
The observations of J.R. Postgate [11] accurately describe the dilemmas faced when using cultivationbased methods for microbial counting:
“An organism is viable if it is capable of multiplying to form two or more progeny in conditions that are “optimal” for the species and strain of microbe concerned: the question of what is optimal is often a matter of informed opinion rather than of scientific fact.”
It is the subjective issue of “informed opinion” that makes absolute standards for microbial counting something of a dilemma. Thus, the established acceptance criteria for any sample must be linked to a particu- lar test method shown to be most suitable for determining the microbial population of that sample.
The term most commonly used for microbial population counting is “Colony Forming Unit” (CFU). This is an artifact-based count relying on cellular replication to produce a visible speck of cells (the “colony”) on the growth medium. If the medium or physical conditions are not adequate, no colony appears. Also, if a clump of many cells lands in one place and only a single colony forms, then the count of “one” underestimates the total. Therefore, plate counts are not always precise or accurate.
Another counting term is “Most Probable Number” (MPN), which uses visible growth (turbidity) among a series of broth-filled containers inoculated with portions of the sample. The principle behind the MPN is “dilution to extinction,” and the MPN uses multiple containers and dilutions of sample in a given test. The MPN is also dependent on a cell’s ability to multiply in growth medium under the physical incubation conditions. CFU and MPN are not the same, but a great deal of effort has gone into correlating their results.
More significant changes in data come from wholly different counting strategies. During the 1970’s, studies using direct counts showed that cultivation detected only a portion of the living microorganisms in water samples. Direct counting of microorganisms became feasible using microscopy aided by vital dyes [13,14]. Additionally, direct microscopic counts were found to reflect public health risks from certain “viable but non-culturable” pathogens [15]. Bypassing the need for incubation, these methods returned data in less time, thus advocating for rapid microbiology counting methods.
Other research exploited the physiological activity of microorganisms. Research at NASA used ATP bioluminescence for detection and counting, especially in the area of remote sensing [1]. This work was applied to clinical microbiology [2] and food and water analysis [3]. However, whether the methods use dye and microscopes, cultivation in growth medium or physiological activity, they are all population measures subject to unique, and sometimes substantial, variabilities or artifacts.
Microbiological Taxonomy Methods
“Rapid Microbiological Methods” are not new ideas. For example, in 1957 the Manual of Microbiological Methods discussed “fast biochemical assays” for characterizing microorganisms. These taxonomic methods were miniaturized assay methods and provided early strategies for developing identification systems that have served well since the early 1970s.
Historically, identification of microorganisms was based on phenotypic characteristics. For many organisms, sets of characteristics were established that remain quite acceptable today. For example, USP <61> (Microbial Limit Tests) provides a set of basic characteristics to identify four types of indicator organisms that are often used as part of acceptance criteria for pharmaceuticals. However, phylogenetic classification based on 16S rRNA sequence data [16] has now become a definitive method of taxonomy. The utility of this method offers great precision advantages. For investigative purposes, this tool is very precise for identification to the species level. However, even genetic sequence systems have vulnerabilities [17]. Also, genetic systems may not be suitable for analysis above or below the species level (for example, when differential characteristics are at the infrasubspecific rank, as in biovar, serovar, or pathovar).
Microbiology Tests and Time
At many manufacturing sites, the slowness of conventional micro- biological data gathering is a chief complaint. The greatest benefits of rapid microbiology, then, are derived from its use in release testing during actual production. However, a sole emphasis on product release testing will diminish its value in process control assays. As mentioned, rapid microbiology methods for process control have not yet included “on-line” assays (although candidate technologies exist), but existing rapid methods may allow process adjustments for faster corrective actions and downstream adjustments.
Often overlooked, “product development” testing presents multiple opportunities for accelerated microbiological data collection to improve pharmaceutical quality or manufacturing efficiencies. While useful in the research and development (R&D) phase, rapid microbiology testing increases its benefits as manufacturing process development moves, for example, into sterile process qualification and validation studies.
Rapid detection of microbiological problems in the process stream allows for investigation and correction before the batch processing (e.g., raw materials and bulk formulation testing) and packaging. These offer enormous benefit especially earlier in the process and go to the heart of the FDA’s product quality initiative [18]. Technical discussions of alternative and rapid microbiology methods may be found elsewhere [19, 20].
Microbiology Tests: Sensitivity and Accuracy
In an effort to improve microbiological quality of drugs, acceptance criteria for pharmaceutical component testing or environmental monitoring have moved to ranges so tight that significant variability exists and data are meaningless [21]. Additionally, microbiological data are exquisitely dependant on the methods used to produce those results. Method-dependant variability was observed by Jones [22] in population studies that showed different counts could be obtained when growth media were adjusted to reflect the source environment. Reasoner and Geldreich [23] advanced this thinking and produced R2A medium, which was shown to increase colony counts for samples of potable water. R2A was the most significant improvement in cultivation-based counting since the introduction of “Plate Count Agar.” Although these methods may have contributed “data” variability, they greatly improved the sensitivity of test methods.
Sample collection and preparation changes may also be a source of variability that can bias test conclusions. Lemmen et al. [24] showed that contact plates were more capable of detecting Gram-positive cocci than swab samples. Differences can also be expected when changing between air sampling systems [25]. Since several environmental sampling methods exist, the “total count” method used should be the most sensitive, which means it would yield the greatest counts. Increased detection sensitivity will certainly improve microbial count data.
This presents us with a dilemma. If data are to be compared over time, then test methods must remain the same, which is fundamental to trend analysis. However, to accelerate data collection, methods must change. Some changes will be insignificant (and test method validation may show no difference), and some will change data greatly. Often, new methods rely on a completely different body of information, some may be direct measurements, some indirect. In either event, previous acceptance criteria may not be applicable. Therefore, implementation of newly developed, or more rapid, microbiology methods may also require establishment of new acceptance criteria. Ultimately, trending of data may be lost in order to bridge the gap between “old” and “new” data analysis.
Developmental Data and Decision Trees
In an ideal world, when a method is translated to the field, all experimental variables are understood and laboratory data are faithfully produced. In microbiology, this ideal is not always achieved, especially in environmental studies where the microbial population changes regularly. This does not mean that a method is inappropriate for a certain set of samples. It may simply mean that the method’s assumptions were not adequately considered during data evaluation. Samples evaluated by different methods often demonstrate this as well, and reinforce John Postgate’s observations about the optimal method that, “...optimal is often a matter of informed opinion rather than of scientific fact.”
In this sense it is important to re-emphasize that microbiological methods are based on secondary or artifact systems: dye uptake, ATP formation, or cell division in specific conditions. Every method, new or old, will have situations that work well and situations that do not. The user needs to be familiar enough with the weakness of the test to understand its limits of use.
Too many laboratories use conventional methods and base important decisions on insignificant data. This is particularly evident when methods such as plate counts, producing less than 10 colonies, are used as acceptance or rejection criteria. Even worse are the methods for monitoring cleanrooms that employ acceptance criteria that are below plausible quantitative ranges. Prince and Prince [21] point out that data (from air, water, surfaces, and personnel) must show changes that are measurable and reproducible if process control improvement is the goal. If the change cannot be measured, then change-related improvement cannot be demonstrated. This is inconsistent with recently established expectations that “critical-area” environmental samples yield no counts. New methods are needed to provide meaningful measures of process control improvement in this situation [26].
Development of any new method should generate sufficient data in support of its suitability for use. During development of rapid sterility testing methods for cell-based products, comparative studies have shown that some methods detect different species with different frequencies [27]. This is expected, just as the compendial sterility tests do not detect all species. The development and implementation of any new method requires an assessment of the incidence of false positives and false negatives. When implementing new methods, procedures must be established to recognize the impact such methods have on the data, and to provide actions for investigation and correction of deviations. For counting methods, the correct action may be to change the acceptance criteria based on risk analysis.
FDA is currently training its regulators to understand the limits of tests and the value of their data, using a risk-based approach. FDA regulators will work with manufacturers as data reveal hitherto unforeseen challenges to the analysis and interpretation of existing and new quality control tests.
In summary, although microbiology tests represent only a small portion of a pharmaceutical quality testing program, their importance is critical to product safety. It is therefore incumbent upon the quality unit, both QA and QC, to understand the strengths and weaknesses of conventional, recent, and emerging microbiology technologies. New microbiology methods can offer advantages of speed and precision for solving microbiological problems associated with materials or environmental influences. Neither Corporate economics nor regulatory attitudes should be a barrier to the use of new testing technologies or different measurement parameters. In fact, if we are to increase our understanding of quality-based products and processes, then quality by design principles and risk analysis methods must be extended to the development of new microbiological technologies. This approach will drive process engineering to yield real, measurable gains in microbiological product quality assurance.
References
1. MacLeod, N.H., E.W. Chappelle, and A.M. Crawford. 1969. “ATP Assay of Terrestrial Soils: A Test of an Exobiological Experiment.” Nature 223 (203):267-268.
2. Conn, R.B., P. Charache, and E.W. Chappelle. 1975. “Limits of Applicability of the Firefly Luminescence ATP Assay for the Detection of Bacteria in Clinical Specimens.” Am. J. Clin. Pathol. 63(4):493-501.
3. Ukuku, D.O., V. Pilizota, and G.M. Sapers. 2001. “Bioluminescence ATP Assay for Estimating Total Plate Counts of Surface Microflora of Whole Cantaloupe and Determining Efficacy of Washing Treatments.” J. Food Protect. 64(6):813-819.
4. Cross-Smiecinski, A.J. 2002. “Quality Assurance,” pp. 158-165. In: Manual of Environmental Microbiology, (ed., C.J. Hurst). ASM, Washington, DC.
5. Taylor, J.K. 1987. “Quality Assurance of Chemical Measurements.” Lewis Publishers, Inc., Chelsea, MI.
6. Food and Drug Administration. 1994. “Guidance for Industry for the Submission Documentation for Sterilization Process Validation in Applications for Human and Veterinary Drug Products.” Office of Training and Communication, Division of Drug Information, CDER, FDA. Available on-line at <http://www.fda.gov/cder/guidance/cmc2.pdf>.
7. Food and Drug Administration. 2002. “Pharmaceutical cGMPs for the 21st Century: A Risk-Based Approach.” FDA Concept Paper, August 21, 2002. Available on-line at <http://www.fda.gov/oc/guidance/gmp.html.>.
8. Food and Drug Administration. 2004. “Guidance for Industry PAT - A Framework for Innovative Pharmaceutical Development, Manufacturing, and Quality Assurance.” Office of Training and Communication, Division of Drug Information, CDER, FDA. Available on-line at: <http://www.fda.gov/cder/guidance/6419fnl.pdf>.
9. Food and Drug Administration. 2003. “Pharmaceutical cGMPs for the 21st Century: Second Progress Report and Implementation Plan.” September 3, 2003. Available on-line at < http://www.fda.gov/cder/gmp/ 2ndProgressRept_Plan.htm>.
10. Cundell, A.M. 2003. “Historical Perspective on Microbial Method Development,” pp 9-17. In: Application and Acceptance of Rapid Microbiological Methods in the Pharmaceutical Industry, (ed., Martin C. Easter). CRC/Interpharm Press, Boca Raton, FL.
11. Postgate, J.R. 1969. “Viable counts and viability,” pp. 611-628. In: Methods in Microbiology, Vol. 1, (eds., J.R. Norris, and D.W. Ribbons). Academic Press, New York.
12. Pelczar, M.R. et al. (eds.). 1957. “Manual of Microbiological Methods,” Society of American Bacteriologists (now ASM). McGraw Hill, New York.
13. Daley, R.J., and J.E. Hobbie. 1975. “Direct counts of aquatic bacteria by a modified epifluorescence technique.” Limnol. & Oceanog. 20: 875-882.
14. Hobbie, J.E., R.J. Daley, and S. Jasper. 1977. “Nuclepore filters for counting bacteria by fluorescence.” Appl. Environ. Microbiol. 33(5): 1225-1228.
15. Hussong, D., R.R. Colwell, M. O’Brien, A.D. Pearson, R.M. Weiner, and W.D. Burge. 1987. “Viable Legionella pneumophila not detectable by culture on agar media.” Bio/Technology 5(9): 947-950.
16. Woese, C.R., O. Kandler, and M.L. Wheels. 1990. “Towards a natural system of organisms: proposal for the domains of Archaea, Bacteria, and Eucarya.” Proc. Natl. Acad. Sci. 87: 4576-4579.
17. Dewhirst, F.E., Z. Shen, M. S. Scimeca, L. N. Stokes, T. Boumenna, T. Chen, B. J. Paster, and J. G. Fox. 2005. “Discordant 16S and 23S rRNA gene phylogenies for the genus Helicobacter: Implications for phylogenetic inference and systematics.” J. Bacteriol. 2005: 6106-6118.
18. Food and Drug Administration. 2002. “FDA Unveils New Initiative To Enhance Pharmaceutical Good Manufacturing Practices.” Press release P02-28, August 21, 2002. Available on-line at <http:// www.fda.gov/bbs/topics/NEWS/2002/NEW00829.html>
19. Parenteral Drug Association. 2000. “Technical Report No. 33 - Evaluation, Validation and Implementation of New Microbiological Testing Methods.” PDA Journal of Pharmaceutical Science & Technology. 54(3) supplement TR 33: 1-45.
20. M. J. Miller (ed.). 2006. “Encyclopedia of Rapid Microbiological Methods, Vol. 1.” Davis Healthcare International Publishing. River Grove, IL.
21. Prince, H.L., and D.L. Prince. 2005. “Contamination control - The life and death of bacteria and other germs.” Contract Pharma. 2005 (June): 1-7.
22. Jones, J.G. 1970. “Studies on freshwater bacteria: Effect of medium composition and method on estimates of bacterial pollution.” J. Appl. Bacteriol. 33: 679-686.
23. Reasoner, D.J., and E.E. Geldreich. 1985. “A Mew Medium for Enumeration and Subculture of Bacteria from Potable Water.” Appl. Env. Microbiol. 49(1): 1-7.
24. Lemmen, S.W., H. Häfner, D. Zolldann, G. Amedick, R. Lütticken. 2001. “Comparison of Two Sampling Methods for the Detection of Gram- Positive and Gram-Negative Bacteria in the Environment: Moistened Swabs Versus Rodac Plates.” Inter. Journ. Hyg. and Environ. Health. 203: 245-248.
25. Jensen, P.A., W.F. Todd, G. N. Davis, and P.V. Scarpino. 1992. “Evaluation of Eight Bioaerosol Samplers Challenged with Aerosols of Free Bacteria.” Amer. Indust. Hyg. Assoc. J. 53(10): 660-667.
26. Hussong, D., and R.E. Madsen. 2004. “Analysis of Environmental Microbiology Data from Cleanroom Samples.” Pharm. Technol. - Aseptic Processing Supplement (May): s12 -s15.
27. Khuu, H.M, F. Stock, M. McGann, C.S. Carter, J.W. Atkins, P.R. Murray, and E.J. Read. 2004. “Comparison of Automated Culture Systems with a CFR/USP-Compliant Method for Sterility Testing of Cell- Therapy Products.” Cytotherapy 6(3): 183-195.
Robert Mello, Ph.D., is a review microbiologist with the New Drug Microbiology Staff, Office of Pharmaceutical Science, CDER, FDA. He received his Ph.D. in biochemistry at Johns Hopkins University School of Medicine and did his post-doctoral research at University of Texas Southwestern Medical School. As an Assistant Professor in the Department of Ophthalmology (Wilmer Institute), Johns Hopkins University School of Medicine, he conducted basic research on anti-angiogenesis before joining a Baltimore-based pharmaceutical R & D company. He has over twenty years of pharmaceutical industry experience in production, clinical supply, aseptic processing, sterilization, validation, facility design, QA/QC, and regulatory affairs. He most recently served as the PDA VP of Education and Director of the PDA Training and Research Institute where he directed training programs for both industry and government health authorities.
David Hussong earned his M.S. and Ph.D. in microbiology from the University of Maryland under the direction of Drs. Ronald Weiner and Rita Colwell. He is a commissioned officer in the U.S. Public Health Service and currently serves as the Associate Director for New Drug Microbiology in the Office of Pharmaceutical Science at FDA's Center for Drug Evaluation and Research (CDER). He has served over 21 years at the FDA. He is also a member of USP’s Microbiology and Sterility Assurance Expert Committee. He previously was a research microbiologist at the USDA’s Agriculture Research Service and at the US Navy’s Infectious Disease Program Command. His research has included microbial taxonomy, detection methods development, ecology, infectivity, and death kinetics.