Abstract
Marketed drugs and devices possess specifications including critical microbiological quality attributes purposed to assure efficacy and patient safety. These attributes are legislated requirements intended to protect the recipient patient. Sampling, microbiological testing, interpretation of data for final products, raw materials, and intermediates all contribute to a cohesive assessment in the assurance of finished product quality. Traditional culture-based microbiological methods possess inherent and unavoidable variability, recognized by the compendia and which might lead to erroneous conclusion pertaining to product quality. Such variability has been associated and intrinsically linked with data integrity issues; manufacturers have subsequently been encouraged by regulatory authorities to introduce multiple microbiologists or checks to prevent such issues. Understanding microbiological variability is essential such that genuine data integrity issues are identified. Furthermore, a range of meaningful preventative strategies are feasible beyond increasing the capacity of the quality control microbiological laboratory. This short review describes the legislative requirements, inherent microbiological variability, and realistic actions and activities that genuinely assure patient safety.
LAY ABSTRACT: Marketed drugs and devices possess specifications including critical microbiological quality attributes purposed to assure efficacy and patient safety. These attributes are legislated requirements intended to protect the recipient patient. Sampling, microbiological testing, interpretation of data for final products, raw materials, and intermediates all contribute to a cohesive assessment in the assurance of finished product quality. Traditional culture-based microbiological methods possess inherent and unavoidable variability, recognized by the compendia and which might lead to erroneous conclusion pertaining to product quality. Such variability has been associated and intrinsically linked with data integrity issues; manufacturers have subsequently been encouraged by regulatory authorities to introduce multiple microbiologists or checks to prevent such issues. Understanding microbiological variability is essential such that genuine data integrity issues are identified. Furthermore, a range of meaningful preventative strategies are feasible beyond increasing the capacity of the quality control microbiological laboratory. This short review describes the legislative requirements, inherent microbiological variability, and realistic actions and activities that genuinely assure patient safety.
Introduction
The assessment of the microbiological quality of pharmaceutical finished drug products and devices is mandated through legislated requirements (21 Code of Federal Regulations [CFR] Parts 211 and 800 in the US and Volume 4 of Eudralex in the EU). These laws are purposed to help assure the safety of the recipient patient population from potential microbiological adulteration of drugs and devices. Such adulteration may be regarded as adventitious contamination manifested as the non-sterility of items purported to be sterile or the possession of an item with a certain quantity of microorganisms (bioburden) or type of microorganism (objectionable) occurring within items marketed as non-sterile. The actual technical, microbiological criteria for sterile and non-sterile products are regarded as microbiological quality attributes and defined in the pharmaceutical compendia recognized by the relevant health authority. It is against these standards by which legislated requirements governing a finished product or device are assessed and determined to be fit for market. Similarly, the pharmaceutical compendia detail the reference or referee method or analytical test by which a drug or device is sampled, tested, and assessed against the microbiological criteria (or offer routes through which alternative methods can be adopted). Assurance of the microbiological quality of an item is not solely based upon end product sampling, testing, and assessment; rather it is based on a complete, cohesive, and coordinated plan of process inputs (raw materials, additives, processing aids), process intermediates (including process steps such as intermediate hold) and environmental sampling, testing, and assessment. It is the complete, integrated, and scientifically principled management of sampling, testing, and assessment that provides the entire microbiological data that permits the disposition of product meeting its licensed label claim. This complete end-to-end cohort of data for each batch of finished drug product or device is also legislated through the current good manufacturing practices (cGMPs), for example, 21 CFR Parts 211.110 and 211.113.
It is the combination of legislated requirements, established microbiological attributes, sampling, testing, and analysis that is purposed to help assure patient safety. Patient safety is the fundamentally crucial guiding principle and ethic in the provision of pharmaceutical finished drug products and devices. Industry and regulators are entirely aligned in this, and what could be more unethical and paradoxical than the provision of harmful therapies supplied to treat patients? Equally the provision of safe therapies and assurance of patient safety in terms of microbiological quality is benefited with recognition of the following:
Microbiological data does not per se materially control or modulate the bioburden or the sterility of non-sterile products or products purported to be sterile. Historically, legislation has primarily relied upon data acquired from an analytical test as the proof of safety for the patient. This is understandable given the evidence-based foundation of judicial systems supporting legislation. In actuality, the assurance of the microbiological quality of products is better served with an emphasis upon design and control (as set out in International Conference on Harmonization guidance Q8 Pharmaceutical Development, Q9 Quality Risk Management, Q10 Pharmaceutical Quality System, Q11 Development and Manufacture of Drug Substances). Microbiological data has value; however, this is only one part of the overall design, controls, and checks and balances necessary for assuring microbiological product quality.
Acquisition of microbiological data is reliant upon a sample upon which a particular test is performed. Meaningful microbiological data providing supporting evidence that a well-designed process is in control and that the end product meets the required microbiological attribute can only be gained if a representative sample is obtained. Samples must not only be authentic of the material tested, that is, free from contamination during the activity of sampling, but removed at a location and time that serves the purpose of the test (process control or material meeting a microbiological attribute). Typically a worst-case time for sampling is at the end of any process hold period. Additionally the sample must be representative of material and account for any lack of homogeneity. An end-to-end integrated sampling plan based upon a genuine understanding of microbiological risk and control is the means to achieve this (1).
Each sample is subject to a particular test methodology that is reliant upon a suitable method development, qualification, and accompanying controls. Execution of the test is highly dependent upon the adequacy of the associated standard operating procedures (SOPs) and the acumen of the microbiologist performing the method.
Once a sample has been tested, the test is inspected and microbiological data is interpreted in order to make the microbiological assessment. Traditional microbiological test methods invariably rely upon the visual inspection of a liquid or solid medium and either the enumeration of colony-forming units (CFUs), an assessment of the presence/absence of CFUs, or the presence of turbidity in liquid media. The CFU enumeration is an arbitrary estimate at best, as the only cells able to form colonies are those that can grow under the conditions of the test. These test conditions include incubation temperature, incubation time, type of culture media, oxygen conditions, and so on.
A subsequent recorded number of CFUs are not raw data, but rather the CFU count is the microbiologist's interpretation of the number of colonies on the plate (2). This interpretation is affected by the number of colonies present; where there is too high a number leads to counting errors as a result of confluence or overcrowding across the surface of the plate; and where the number is too low error arises because the CFUs follow Poisson distribution.
Of late several competent regulatory authorities have made inspectional observations of firms that are applying the microbiological test in alignment to the legislated cGMPs. Those observations note that a single microbiologist is visually inspecting the test and making an interpretation of the test. For example, there have been regulatory comments where CFUs have been miscounted. This can arise due to inappropriate light or magnification, a failure to use a suitable counting device, the incorrect multiplication of a dilution, and so on. These observations have been categorized as data integrity issues. It has thus become popular to suggest that deployment of multiple microbiologists is an effective and expedient preventative action of microbiological data integrity issues. At first glance this may appear the most appropriate rectification; however, this in itself is not a ubiquitous solution, may not be necessary, and may in actuality prevent realization of our common goals of reduced patient risk and maximized patient safety.
Our necessary history of evidence-based judicial systems and legislation founded upon indisputable and uncontestable data (laws of evidence or the rule of evidence) has helped to serve our industry well; however, it does not match well with microbiological data with its well-recognized inherent variability (3). Variability can be divided in to two categories (4):
Avoidable variability—variability due to poor practice (5)
Inherently unavoidable variability—variability due to limitations of the methods and the vagaries of dealing with biological samples (6)
Assurance of microbiological quality is not primarily achieved through test data. It frequently appears confoundlingly difficult to reconcile the constraints of microbiological tests, interpretative assessment of those tests with the precise details of design control, and sampling. There is a need to ensure the data collected and purposed to demonstrate process control and product achieving microbiological quality attributes is as accurate as feasible; however, there are many alternative means to achieve this other than multiple microbiologists performing the inspection and assessment. This short review is purposed to illustrate the inherent variability of culture-based microbiological test methods (which cannot be reduced with any significance) and provide the range of options to ensuring the avoidance of potential data integrity issues.
Microbial Quality Attributes
All finished drug products and medical devices must be fit for use and granted marketing licenses based upon appropriate specifications, in combination with well-designed and controlled manufacturing processes conforming to the legislated cGMPs. For finished drug products this means a therapy must have appropriate characteristics of safety, identity, strength, purity, and quality. Microorganisms (viable and non-viable), their products, or their inherent biochemical and physiological activity on or within therapies may adversely affect safety, strength, purity, and quality. Unlike an inanimate chemical or physical impurities or contaminants, the animate and dynamic nature of microorganisms poses a far greater risk to fitness for use and ultimately patient safety. In a certain quantity certain species of microorganisms may cause infection when introduced through the vehicle of a therapy; metabolically active microorganisms may transform a drug product's formulation resulting in chemical changes or changes in efficacy. Lastly, the vestiges of microorganisms or sloughed off cell well and cell membrane may cause pyrogenic or toxic effects. The microbial requirements of finished product specifications are therefore arguably the most important and are often underappreciated in their potential to adversely affect the recipient patient population. Concomitantly pharmacopoeial compendia are very clear in stating the necessary microbial specifications of finished drug products. Sterility required for all parenteral products are stipulated in by the United States Pharmacopeia (USP), (<71> Sterility tests), Pharmacopoeia Europa (EP), (2.6.1 Sterility), and Japanese Pharmacopoeia (JP), (4.06 Sterility test). Non-sterile finished drug product are permitted a level of bioburden that is based upon the dosage form and route of administration and stipulated in the USP (<1111> Microbiological examination of non-sterile products: acceptance criteria for pharmaceutical preparations and substances), EP (2.6.12 Microbiological examination of non-sterile products: total; viable aerobic count), and JP (4.05 Microbial limit test). The potential impact of microorganisms on or within therapies is undisputable and clearly represents a source of risk to patient safety; however, it is imperative that we recognize the precise specifications for bioburden as described by the industry-standard pharmacopeia. USP <1111>, EP 2.6.12, and JP 4.05 list out the maximum limits of bioburden defined as total aerobic microbial count (TAMC) and total yeast and mold count (TYMC) within different dosage forms of drug products, categorized as 10 CFU, 100 CFU, and 1000 CFU. Each pharmaceutical compendia subsequently qualifies the interpretation of these maximum bioburden limits, with the three harmonized compendia (USP, EP, JP) all stating: “When an acceptance criterion for microbiological quality is prescribed, it is interpreted as follows: 101 cfu: maximum acceptable count = 20; 102 cfu: maximum acceptable count = 200; 103 cfu: maximum acceptable count = 2000; and so forth”.
All compendia provide due and appropriate recognition to the inherent variability of microbiology and microorganisms; the expectation is that the expert microbiologist appropriately interprets the data. A 200% leeway is permissible; this is essentially the acceptance of a two-fold difference or variability in the enumeration of CFUs. This variability is necessary because of the inherent dynamic nature of microorganisms, microbial populations, and the means by which samples are handled and tested by culture-based methods. The same is true for process inputs (raw materials, additives, processing aids), process intermediates (including process steps such as intermediate hold), and environmental testing. While enumeration to established microbial quality attributes is an important component of quality assurance, there is recognized variability. In addition to microbial quality attributes, the pharmacopoeial compendia detail the standardized methodology of testing a sample purposed to enumerate bioburden or establish a presence or absence. These methods are recognized as the referee or reference method that a firm needs to optimize and qualify for each of their sample formulations, assuring the recovery of any resident culturable (using the media and conditions stated) microorganisms: the microbiology quality control (QC) laboratory tests.
QC Microbiological Tests
Generally, the harmonized pharmacopoeial reference and referee microbiological tests methods are purposed for the assessment of a material in achieving the necessary microbiological quality attributes. Test methods are growth-based culture methods that either determine the presence or absence of microorganisms or enumerate bioburden. Presence/absence tests include the test for specified microorganisms per USP (<62> Microbiological examination of non-sterile products: tests for specified microorganisms), EP (2.6.13 Microbiological examination of non-sterile products: test for specified micro-organisms), and JP (4.05 Microbial limit test), and USP <71>, EP 2.6.1, and JP 4.06; these are binary in nature, relying upon the inspection of media for the physically visible signature of microbial presence. In contrast, microbial enumeration tests per USP (<61> Microbiological examination of non-sterile products: microbial enumeration tests), EP (2.6.12 Microbiological examination of non-sterile products: microbial enumeration tests), and JP (4.05 Microbial limit test) rely upon inspection of the test with an enumeration of discrete recovered colonies to make an assessment of the material. Similar non-compendial bioburden test methods may be used for the enumeration of microorganisms within process inputs (raw materials, additives, processing aids), process intermediates (including process steps such as intermediate hold) and environmental monitoring samples. The following two sections briefly detail the pharmaceutical compendial sterility test, and the microbial enumeration test and non-compendial bioburden test methods, discussing their constraints and the fundamental microbiological facts related to potential data integrity issues.
Sterility Test
In the harmonized compendial sterility test, an observation is made by a trained professional microbiologist rather than a measurement made and recorded by an instrument. This evaluation may be prone to misinterpretation and to inconsistency of discrimination, and it may result in the erroneous conclusion of an article's sterility. Errors can arise based on the need for the microbiologist to visually assess the presence or absence of turbidity. This requires, in addition to prerequisite of good eyesight and a suitable light source, an understanding of the variations of microbial growth. Issues can arise where product residues react with the culture medium causing turbidity (a problem that affects the direct inoculation method moreso than the membrane filtration method). The definition of turbidity is left up to the practitioner. A potential issue with discerning turbidity and microbial growth is the presence of debris; this can originate from the elastomeric closures of product, the media, and rinse solution bottles.
Sterility is a microbial quality attribute required of certain therapies (drugs, biological, and devices) determined by both the product and its route of administration. All injectable products (cutaneous, sub-cutaneous, and parenteral) and ophthalmic and aqueous oral inhalation products are required to be sterile in the US market as per legislated 21 CFR 200.50 and 21 CFR 200.51, respectively. Sterility of a therapy is defined as “Within the strictest definition of sterility, a specimen would be deemed sterile only when there is complete absence of viable microorganisms from it” per USP (<1211> Sterilization and sterility assurance of compendial articles). This definition and similar definitions have long endured, to date adequately serving the patient population, regulatory authorities, and manufacturing firms for the primary purpose of patient safety. This is regardless of significant scientific flaws (7, 8), misalignment with contemporary microbiology, and the undisputable fact that it is not statistically or technically provable through product testing (sterility testing). The trinity of EP, JP, and USP provide the precise harmonized test methodology for the referee sterility test per USP <71>, EP 2.6.1, and JP 4.06. One of the critical elements of the definition of sterility is the term complete absence, which infers that there is a complete and absolute empirical absence of microorganisms from any item which is purported to be sterile. Sutton (3) illustrates the inability of a test that proves a statistical assurance of sterility of every item labeled as “sterile” within a manufactured batch. Despite the technical flaws of a test for sterility (7), and the statistical constraints, the definition and sterility test has utility in assuring patient safety and remains a legislated requirement.
cGMP regulations are promulgated in CFR Part 211 Good Manufacturing Practices for Finished Pharmaceuticals. Legislative requirements governing sterility and sterility testing in 21 CFR Part 211 are supplemented in 21 CR Part 600–680. The requirement for the microbial quality attribute of sterility is referenced in 21 CFR Part 211.165(b), which states: “There shall be appropriate laboratory testing, as necessary, of each batch of drug product required to be free of objectionable microorganisms”. Here, “objectionable” is any viable microorganism (within the current definition of sterility) within a product labeled sterile. More specifically, 21 CFR Part 211.167(a) states that “for each batch of drug product purporting to be sterile and/or pyrogen-free, there shall be appropriate laboratory testing to determine conformance to such requirements. The test procedures shall be in writing and shall be followed.”
Notwithstanding an approved filing of parametric release (see USP <1222> Parametric release), a test for sterility is always necessary. This applies even under exceptional circumstances, for example, 21 CFR Part 211.165(a): “Where sterility and/or pyrogen testing are conducted on specific batches of shortlived radiopharmaceuticals, such batches may be released prior to completion of sterility and/or pyrogen testing, provided such testing is completed as soon as possible”.
In addition to legislating the application of a test of sterility for articles purporting to be sterile, the CFRs require any test applied to be fit for purpose. 21 CFR Part 211.165(d): “Acceptance criteria for the sampling and testing conducted by the quality control unit shall be to assure that batches of drug products meet each appropriate specification and appropriate statistical quality control criteria as a condition for their approval and release. The statistical quality control criteria shall include appropriate acceptance levels and/or appropriate rejection levels.”
In terms of the harmonized referee sterility test, this highlights that a firm's procedures must clearly distinguish what constitutes a passing from a failing test. The sterility test is reliant upon a professional microbiologist's acuity of vision coupled with an informed decision on whether growth in the sterility test canister has occurred. Those criteria are fundamentally dependent upon an individual's capability and competency (experience and expertise) in microbiology. If we literally apply 21 CFR Part 211.165(e), then it reinforces that a firm must describe and establish the discriminatory capability of the sterility test and therefore the microbiologist: “The accuracy, sensitivity, specificity, and reproducibility of test methods employed by the firm shall be established and documented. Such validation and documentation may be accomplished in accordance with 211.194(a).”
Furthermore, 21 CFR Part 211.160(b) is clear in its intent that for any test that the associated instrumentation is calibrated, maintained, and in a condition that will assure the consistency in the determination of whether an article meets the requisite specification: “(4) The calibration of instruments, apparatus, gauges, and recording devices at suitable intervals in accordance with an established written program containing specific directions, schedules, limits for accuracy and precision, and provisions for remedial action in the event accuracy and/or precision limits are not met. Instruments, apparatus, gauges, and recording devices not meeting established specifications shall not be used.”
Interpreting this literally in terms of the sterility test, assuring a consistent deterministic application of the referee method would require the calibration and maintenance of the microbiologist's visual acuity and professional competency.
The sterility test is widely recognized by industry and regulators as possessing two fundamental constraints. Firstly, the limit of detection of the sterility test is greater than a single viable cell (9). Physiological, genotypic, and phenotypic variability of microorganisms (10) essentially ensure that the sterility test will only ever recover (culture) a fraction of potential contaminants; some microorganisms will be unculturable due to the limitations of the medium (11), others unculturable due to their physiological prerogative (12⇓⇓–15). Secondly, the sterility tests cannot be applied on a meaningful statistical basis. Odlaug (16) used eq 1 (below) to demonstrating the probability of the test succumbing to a Type II error—a false negative, that is, failing to identify non-sterility (17).
where
P is the probability of failing the sterility test (a positive result)
e = 2.7182818
λ = likelihood of a contaminated unit
Using eq 1, and assuming at least one microorganism is required to cause non-sterility, sampling 20 units (n = 20) within a batch of items, the probability of detection of a non-sterile unit are summarized in Table 1. Obviously, where all units are contaminated (100% per the first column in Table 1), the probability of a positive sterility test, that is, identifying non-sterility, is 100%. Concomitantly per Table 1, when the frequency of contaminated units in a batch is 1% there is only an 18% probability of detecting non-sterility with a positive sterility test.
Probabilities of a Positive and Negative Sterility Test with Varying Frequencies of Contaminated Units in a Batch (16)
The insufficiency of the sterility test in assuring sterility is remarkable and recognized by regulators and industry alike as only capable of detecting gross contamination events (18⇓⇓–21). The contributory value of the sterility test to assure microbial quality, reduce patient risk, and maximize patient safety is likely insignificant to the sum aggregate of design and controls of manufacture. Indeed, deploying additional microbiologists to visually check for growth in a sterility test does not guarantee microbial growth in the test will be observed and does little to improve our odds in assuring sterility. There are a number of means of aiding the retirement of data integrity issues associated with sterility tests (see later).
Microbial Enumeration and Bioburden Tests
Assessment of microbial levels within samples per the compendial microbial enumeration test and on materials per non-compendial biobiorden tests is an important part of pharmaceutical process control. With most pharmaceutical processes, samples are drawn from intermediate product at defined stages (with stages based on risk assessment, designed to inform about process risks and hold times). Bioburden testing allows for the microbial levels to be tracked from upstream processing to downstream processing, with an expectation that the microbial levels decrease, or at least remain unchanged, provided they are below an acceptable action level. For sterile products, there is an expectation that the sample prior to final filtration is not of a level that would pose a significant challenge to the sterilizing capabilities of the filter.
Bioburden refers to the microbial content of a material (or on the surface) at a given point in time. This could be prior to sterilization or in relation to a process hold time (22). As well as providing data essential to final product release and material release (microbial enumeration testing), bioburden enumeration (e.g., environmental monitoring and in-process testing) supports long-term trending for identifying:
Samples that exceed an alert or action level
Changes in the total or mean count
Changes in speciation profiles of recovered microorganisms
Both microbial enumeration and bioburden testing is typically performed by a plate count method; either a pour plate agar method or, more preferably (due to the larger sample size), membrane filtration. In all circumstances recovered discrete colonies are quantified as CFUs. These methods are inherently variable.
With the plate count methods, care needs to be taken in relation to selecting an appropriate agar and with incubation time and temperature, in order to ensure optimal recovery (23). This can be assessed by microbial challenge recovery assessments. Collected samples should be assigned an expiry time (which will need qualifying), and storage of samples prior to testing would ordinarily be at 2–8 °C in order to slow down or suspend the rate of microbial growth.
With the above factors, which introduce an element of variability, there remain many variations with the plate counting methods used to assess bioburden. The first of these is with the limitations of the culture media. No single culture medium or the conditions to which it is subjected to can detect all potential microorganisms in the sample that are culturable. A bias can be built in towards microorganisms associated with the human skin microbiota (if that population is desired) by selecting a general purpose medium (like soybean casein digest medium) and incubating it at 30–35 °C for 72 h or longer. Here the microbiologist will need to consider the populations of concern. It remains that in many cases test regimes are biased towards aerobic, mesophilic microorganisms. This is because such organisms are common to the environment; they will often be a problem should they contaminate the product because they are the most likely to grow, and because most human pathogens fall within this grouping (24).
Secondly, as discussed above, the CFU, the end result of the test, is most appropriately regarded as an estimate of the number of viable bacteria or fungal cells in a sample. The visual appearance of a colony requires significant microbial replication; furthermore it is unknown if the progenitor of the colony was a single microorganism or several microorganisms in close proximity. Hence, when counting colonies it is uncertain if the colony arose from one cell or 1000 cells, and importantly CFU is not a direct accurate measure of microbial numbers.
Thirdly, inaccuracies can also occur with the act of plate counting. Here a distinction needs to be drawn between the limit of detection for microbiological agar plates (assumed to be 1 CFU, but in reality untestable with conventional methods) and the limit of quantification (which is based on the countable range, which is partly a function of the size of the test plate).
Due to the size of the agar plate there will be an optimal counting range; errors will occur where microbial numbers are above an upper countable limit (due to confluence or overcrowding) or below a lower limit (due to statistical error in relation to accuracy of the count, particularly where dilutions have been performed) (25). The USP guidance chapter (<1227> Validation of microbial recovery from pharmacopoeial articles) recommends 25–250 CFU for bacteria and yeast, such as Candida albicans; and 8–80 CFU for filamentous fungi. These are in relation to 90–100 mm agar plates. For 55 mm plates (as might be used with membrane filtration methods), countable ranges reduce to 20–80 CFU.
According to Sutton (2), the upper range is due to a combination of colony size, colony behavior (including the effect of swarming or spreading), the surface area of the agar plate, together with the general effects of confluence or overcrowding where large numbers of colonies are concerned. Combined these factors lead to an assumption of error whereby at the upper limit of the range the observed numbers of CFUs will fall off relative to the expected numbers and the divergence becomes significant enough to render the observed count as too inaccurate to be of scientific value.
Similarly, with the lower range, error arises as the pattern of CFUs recovered become subject to what Jarvis (6) terms total error. Total error associated with colony counts from spread plates, pour plates, and membrane filtration total counts can be expressed as:
where
A = % sampling error
B = % distribution error
C = % dilution error
The issue is compounded by the three different sources of error, which may act singularly or in combination. Examples of each error are
Sampling error: weighing and mixing
Distribution error: counting errors and recording errors
Dilution error: pipetting volumes and diluent volumes
As a result of these errors, low counts follow Poisson distribution.
Fourth, a further source of error can arise with rounding up or down or through averaging. Therefore, an estimation of microbial numbers by CFUs will, in most cases, undercount the number of living cells present in a sample (26). Related counting inaccuracies and error can also occur with the act of colony counting. The counting of colonies manually is normally carried out using an artificial light source, such as a colony counter.
A fifth area of concern is with spreading microorganisms, which are problematic for colony counting. Colony counting error arises due to indistinguishable colony overlap (i.e., masking). For spreading colonies, there are usually of three distinct types:
A chain of colonies, not too distinctly separated, that appears to be caused by disintegration of a bacterial clump
One that develops in a film of water between the agar and the bottom of the dish
One that forms in a film of water at the edge or on the surface of agar
Thus methods for enumeration of microorganisms (compendial microbial enumeration and non-compendial bioburden tests) can only have the objective of providing the best indicator possible of the microbial bioburden but not the absolute bioburden.
Variations are also assumed with the assays to which the pour plate methods are applied. These methods are assessed using microbial challenges for which an acceptable recovery is expressed as 50–200%. Challenges are undertaken during method qualification to show that the method is suitable and that any antimicrobial substances have been neutralized or adequately diluted out. Even where this occurs, it is acknowledged that recoveries of microorganisms under the conditions of the test are variable.
In-built within this range are the variabilities described above. To add to this are the variations associated with sample handling, especially when it is subject to procedures like centrifugation, voretxing, and hand-mixing. Microbial recovery can also be affected by resuspension into a different liquid (as with an enrichment or a dilution step), especially when the liquid is of a different composition, osmolarity, pH, or temperature to the original growth conditions.
Specific to the recovery ranges, the upper range of the 50–200% is more likely to be a product of pipetting or dilution error, as well as reflecting the biological complexities when microorganisms grow under one set of conditions and not under another.
The lower range is a reflection of occasional cell death or damage to challenged microorganisms that might lead to suppressed recovery (this can increase in relation to any hold time). It is also important to keep in mind that no ingredient, even in a chemically defined culture medium, is absolutely pure, which can add a further variation.
Strategies for Assuring Microbiological Data Integrity
To error-proof the sterility and other microbiological tests is impossible; it is fundamentally flawed. However, potential data integrity issues associated with the test can be addressed. In terms of microbiological quality control tests, an appropriately qualified microbiologist performs a visual inspection or enumeration of a test result, uses good judgment to interpret or assess, and accurately documents the conclusion. The discriminatory capacity of the test is therefore dependent upon visual observation and interpretation. Both visual observation and inspection are areas in which the sources of data integrity—system error or an individual's honest mistake—can be addressed by many individually appropriate strategies.
Designing Out the Compendial Method
Deming famously stated, “Inspection with the aim of finding the bad ones and throwing them out is too late, ineffective and costly. Quality comes not from inspection but improvement of the process” (27). The same thinking that fundamentally recognizes testing as an inefficient means of assuring quality is articulated by regulatory agencies (28). In the case of the sterility test options for parametric release exist (21), totally removing the potential data integrity risk associated with this microbiology test.
Data integrity is fundamentally a design issue that permits or offers an opportunity for a human to make an error. Due to the inherent flaws of microbiological quality control tests and especially the sterility test, removing or obviating the need for the flawed test is the most effective means of removing sources of data integrity, supporting cGMPs, and assuring the highest level of patient safety. Designing out the test can be achieved in several ways. First and foremost, a parametric program of sterile product release offers the greatest assurance of sterility through an appropriately designed and managed program. Parametric release is described in standards USP <1222> and US Food and Drug Administration (FDA) regulatory guidance (29); it has also been covered in PDA Technical Report 30—Parametric Release of Pharmaceutical and Medical Device Products Terminally Sterilized by Moist Heat (30). For an organization to adopt a parametric program of release in place of sterility testing it must be filed with and approved by a competent regulatory authority, thereby meeting the established standards and guidance. Parametric release is a far superior method of product release (31). Currently parametric programs of sterile product release apply to those items terminally sterilized and to specific short-lived radiopharmaceuticals per 21 CFR Part 211.165(a) and 21 CFR Part 212. In the latter circumstance the other contributing elements of manufacturing design, management, and control are entirely and collectively leveraged to provide an approved assurance of sterility via a non-terminal sterilization process. Hitherto, other non-terminal sterilization process of manufacture producing sterile drug products may also be featured in forthcoming standards and represent candidates for parametric release.
Alternative to a parametric program of release is the adoption of modern test methods that are equivalent per USP (<1223> Validation of alternative microbiological methods) to the compendial method and that remove the elements of visual acuity and judgment-based interpretation. There are several systems that are marketed and available that provide for this opportunity; furthermore, several of these systems have been successfully implemented.
Improve the Test Method
Process improvement of the compendial test method is feasible to assist or aid the visualization of microbial growth: turbidity in the case of the sterility test and CFUs for agar plate-based methods. Machine, laser, and optical technology can assist or entirely remove the need for a microbiologist to visually discriminate and assess microbial growth. Assessment using technology resulting in digitized images provides the means not only to make a conclusion on the test but also to record, archive, reference, and retrieve data, fulfilling the complete elements of data integrity. Visual discrimination by a microbiologist can also be significantly improved, removing human error by the provision of lighting systems and light boxes (where ambient white light provides better visual acuity than yellow light) and simple enhancements to minimize erroneously and accidently failing to identify or accurately enumerate growth. Throughout the test method life cycle it is prudent to establish and maintain a design history file or equivalent that records the methods development, including archiving of images and supporting information.
Enhance Personnel Knowledge and Know-how
Irrespective of the sophistication of a quality system, or test method, pervading culture and individual knowledge and know-how provides the greatest means of assuring cGMPs. Training, knowledge management, and education are pivotal to provide the necessary experience and expertise in discriminating the visual signature of microbial growth within microbiology quality control tests, irrespective of whether it is within a sterility test canister or upon an agar plate. Equipping microbiologists with the fundamental knowledge and principles of the test and moreover the information (especially archived images) within a test method design history file is one of the most effective, expedient, and efficient means of addressing data integrity issues. It is not uncommon for certain product formulations to possess inherent physical or chemical characteristics (e.g., alumina adjuvants in vaccines) that may obscure a visual identification of growth or that could be mistakenly interpreted as growth. A program of interactive knowledge management and a library of images is exceptionally effective in preventing data integrity issues.
Ongoing Expert Qualification
It is also necessary to accompany any expert training program with an initial and ongoing qualification program that not only establishes a microbiologist's credential to make the assessment, regularly affirms that competency, and ensures new information (as part of the test method life cycle and design history file) cogent to the test is provided to the microbiologist. Such training could include assessing the ability of the microbiologist to spot turbidity using standards of turbidity (such as solutions prepared along the McFarland scale) or broth cultures challenged with a range of microorganisms designed to show different growth patterns in different media (Streptococcus bacteria grow very differently in tryptone soya broth compared with fluid thioglycollate medium, for example). As eyesight inevitably declines with ageing, periodic reassessment is recommended. A different example is with tests for color blindness, which are important for differentiating some microbial identification tests and the red-to-blue spectrum that applies to the Gram stain. As with any test, if there is any dubiety in the interpretation of the data, then this should automatically trigger a proceduralized escalation to a further form of evaluation, either by additional qualified microbiologists or other technology.
Trending and Analysis
Thorough, routine trending and evaluation of microbiological data, although primarily purposed as a monitor of manufacturing performance control, also permits the evaluation and identification of potential data integrity issues. Evaluation of microbiological data is a suitable diagnostic to identify any inconsistencies in microbiological data that might originate from a data integrity issue. This is an especially powerful means of addressing potential data integrity risks when multiple microbiologists are routinely scheduled to perform tests, inspect the test, and interpret results.
Second Person Verification
Including a second person verification (an additional microbiologist of equal or better expertise and training to visually examine and interpret) doubles the number of personnel making the visual discrimination and conclusion. It can be argued that this provides an additional level of assurance of achieving the correct interpretation; however, this in itself is a flawed argument because any assessment is fundamentally based upon the competency of the person making the assessment. If a training and knowledge management program is insufficient then it may not matter how many microbiologists verify the test method data. In other words, this solution will only be effective if the visual acuity, experience, and knowledge of both are adequate. This solution does not provide a means of addressing the maleficent intent of an individual or an organization; if there is a determination to manipulate data, an individual will always find a means.
Discussion
Data generated in accordance with cGMPs must be attributable, legible, contemporaneous, original, and accurate. These five adjectives describe the integrity of data as described by guidance from the Medicines and Healthcare Products Regulatory Agency (32), FDA guidance (33), Who Annex 5 (34), and Pharmaceutical Inspection Convention and Pharmaceutical Inspection Co-operation Scheme (PIC/S) guidance (35) and must be operational throughout the entirety of the data life cycle. Although these critical elements of data integrity are tangible, and are easily measured and assessed, data integrity itself is reflective of an organization's culture. Moreover, irrespective the adequacy of an organization's quality management system, adequacy of process, and system designs, an organization's culture will always counterbalance or confound those processes and systems. Historically many microbiological quality control test methods have been reliant upon an experienced microbiologist executing a visual inspection or enumeration of a test result, using good judgment and accurately documenting the conclusion of the test. Historically there have been and there still remain today opportunities or steps which are prone to what we now term data integrity issues within these tests. Their susceptibility to data integrity issues is, doubtless, inexorably linked to an organization's culture; however the actual impact, significance within the entirety of manufacture and assurance of microbiological quality attributes remains subject to assessment.
Sources of data integrity issues originate from a spectrum that spans system errors, individual genuine mistakes, individual malfeasance, and institutional malfeasance (36). The nature of the compendial referee sterility test and its reliance upon executing a visual inspection or enumeration of a test result, using good judgment and accurately documenting the conclusion of the test result, is opportune across the entire data integrity spectrum (system errors, individual genuine mistakes, individual malfeasance, and institutional malfeasance). Microbiological test are highly prone to organizations or individuals who have a motivation to deliberately fabricate, manipulate, or falsify data, and they are becoming a focus as the likes of the FDA increase enforcement activity on data integrity (37). We have discussed potential data integrity issues in the context of microbiological tests associated with system error or an individual's honest mistake. These unintended errors are attributable to human error (e.g., visual acuity), ignorance, or carelessness.
Undeniably cGMP are minimum requirements administered through several legislations and guidance which include Food Drug and Cosmetic Act; 21 CFRs Part 210, 211, 600; and ICH Q7. Data integrity is the bedrock of cGMP, for without data that can be relied upon there can be no confidence or assurance in the manufacturing process or product quality. Issues in data integrity (whether known, unknown from intent, or unintentional) may themselves obscure other issues; a useful diagnostic of an organization's culture represents a tangible means by which a regulatory authority can exemplify fundamental issues within an organization. The sterility test perhaps represents one of those opportunities, although successfully performed for decades with an absence of examples supporting a broad detrimental impact to patient safety due to characteristics prone to data integrity issues.
In general all microbiological quality control test methods are multi-dimensional in nature. Those dimensions include but are not limited to data criticality and data risk, speed versus accuracy and precision, quality management system, knowledge management, and technology. Figure 1 diagrammatically illustrates this dimensionality within which microbiology quality control test methods occupy different locations. By plotting in between these two dimensions of each test, an assessment can be made regarding the need to address potential data integrity issues associated with certain microbiology test methods. With such an understanding a manufacturing organization may then determine the appropriate course of action (if necessary) to address potential data integrity issues by test redesign, improvement, personnel enhancement, or second person verification. Irrespective of microbiological quality control test method, the activity addressing data integrity must be commensurate with the overall contribution to patient safety.
Dimensionality of microbiological quality control test methods (adapted from 32). CQA: critical quality attribute, CPC: critical process control, IPC/Other: in-process control, QMS: quality management system.
Conflict of Interest Declaration
The authors declare that they have no competing interests.
- © PDA, Inc. 2018
