Abstract
For several years, automated colony counting systems have been available with varying degrees of automation. Ever more sophisticated instruments are now increasingly used in microbiological laboratories of pharmaceutical quality control. In addition to the colony counting device, the instruments are now also equipped with robotic systems performing the entire handling of the petri dishes, e.g., automated internal transportation of petri dishes from the incubator chamber to the instrument’s enumeration device and back. Moreover, the subjective evaluation of microbial enumeration tests by analysts is replaced with a more accurate and precise process. This leads to significant improvements to data integrity compliance. Automated colony counting systems also often enable cost reduction in the microbiological laboratory, e.g., by not requiring a contemporaneous verification by a second analyst. They also enable direct integration of count data into an existing laboratory information management system, reducing the hands-on time, costs per test and also preventing human errors caused by manual transcription. Altogether, these instruments will lead to improved monitoring and assurance of control of biopharmaceutical processes and manufacturing environments, as well as shortened cycle times in the supply chain. Regulators are encouraging the biopharmaceutical industry to adopt these innovative systems. For example, this year a BioPhorum member company received the first health authority approvals from EU, US, CH, Canada, Australia, and China for the use of automated colony counting systems for in-process bioburden testing and the release of drug substance lots, with an incubation time reduced by about 50%. Although these approvals are for release testing of drug substance lots, the instruments can also be used for environmental monitoring, testing of water samples, etc. This article describes a systematic 9-step approach to the evaluation, equipment qualification, and deployment of automated colony counting systems, which can be applied by biopharmaceutical companies wanting to take advantage of their numerous benefits.
- Alternative and rapid microbiological method
- Automated colony counting
- Data integrity improvement
- Environmental monitoring bioburden testing
- Equipment qualification microbiological method validation
- Product bioburden testing
- Water bioburden testing
Introduction
Traditionally, microbial colonies have been generated and observed on agar media, sometimes aided with various forms of magnification, often with the use of a pen or an electronic colony counting pen to mark colonies as they are counted on the cover of the plate. The reliability of the result has always been limited by the abilities of the human observer and data recorder. For example, humans can only see colonies once they are approximately 105 cells (approx. 100–200 µm in size) (1), and often merged colonies are not easily differentiated by analysts. Visual acuity varies among people and therefore among counts. These human-generated counts are therefore subject to error, and those errors can be costly in our business and undermine confidence in the laboratory operations. The original data (the plate itself) is ultimately discarded, and the recorded value is also subject to human error in transcription. These errors may be considered by a health authority investigator to be an indication of data integrity issues. We now have technology that can reduce or eliminate sources of human error-derived inaccuracy, return results faster to decision-makers, and greatly improve data integrity controls for microbial colony counting assays.
Approach to Adoption of Automated Colony Counting (ACC) Systems: Applying a 9-Step Framework
There are variations in the approaches taken to implement automated systems, such as automated colony counting systems. The process can be broken up into nine steps, from initial evaluation up to final implementation of the method, similar to the implementation of alternative and rapid microbiological methods. This process was described by a BioPhorum collaborative team in a previous publication (2) and suggested as a general framework for the evaluation and implementation of alternative and rapid microbiological methods, addressing the questions and considerations relevant in this context.
This general framework is applied in this article to automated colony counters. It will also be applied by a BioPhorum collaborative team to alternative rapid sterility tests; the relevant paper has been submitted for publication (3). A paper on biofluorescent particle counters using this framework is planned for 2022, and a paper on alternative mycoplasma detection methods is scheduled for publication in 2022.
The nine steps described in this framework (2) and applied for automated colony counters here are as follows:
Step 1: Identify operational/business need
Step 2: Define the application
Step 3: Assess requirements
Step 4: Compare options and technologies—landscaping and candidate(s) selection
Step 5: Develop a business case: technical, quality, and business evaluation and justification
Step 6: Perform proof-of-concept studies/feasibility studies/prevalidation studies
Step 7: Validate at pilot or primary site
Step 8: Deploy global/company-wide qualification of additional laboratories
Step 9: Define regulatory filings and implementation strategy
Please note that all or some of these steps may be taken, as necessary. This may not be the only approach, but it is based upon the input from members of the industry who have experience with these instruments.
The application of these nine steps to the implementation of an automated colony counting system is described next.
Step 1: Identify Operational/Business Need
The first step is the evaluation of the benefits to implementation of an automated colony counter and how the system can be applied to specific operation(s).
The benefits of automated colony counting systems are myriad. Some are more compelling to certain stakeholders than are others, but all contribute to a superior system for faster and precise detection of microbial colonies. Perhaps the most beneficial aspect is that automated methods allow for greater control of data, data integrity, and data security. Data integrity issues have become a major concern in pharmaceutical manufacturing. There are several guidance documents that discuss implementation of systems to assure data integrity (4, 5). Regarding the discerning, interpreting, documenting, and verifying of results, all stakeholders should appreciate the difference between a validated automated system and the current manual colony count based on the less sensitive and more variable human performance. Automating these aspects of a traditionally manual process allows for greater optimization, standardization, and regulation of practices and provides highly controlled data management capabilities. The degree of automation can vary depending on the system. With some instruments, only the counting of colonies is replaced, whereas other systems automate the complete chain of custody in the laboratory from sample handling, incubation, counting, and data recording to migration of data into a laboratory information management system (LIMS) (6). With the new systems, counts are performed and results are determined objectively and consistently, eliminating human errors and subjectivity. Data are managed via 21CFR Part 11-compliant software in which data can be annotated with appropriate audit trails. User roles can be identified and managed to prevent the possibility of data modification, deletion, or repeat-testing without proper approval. Any modifications to the methods or procedures can be confirmed by the review of audit trails, even if the changes are made by an operator with the proper permissions.
Another business benefit of the automated systems may be the ability to generate results considerably faster than the traditional means of counting colonies (human eye). For example, some automated systems can discern bacterial colonies at around 100 cells and fungal colonies at around 20–50 cells (6), whereas the human eye requires colonies of around 105 cells for detection (1). Depending on the microorganism, the duration of growth between these colony sizes could be days (6). The power of the earlier result is clear whether it is a desired “passing” result or a limit-exceeding or “excursion” result. Faster results can allow, for example, an earlier return of equipment to service after cleaning verification, faster product change-over, shorter times in commissioning or remediation, and earlier mitigation activities including prevention of cross-contamination of other equipment/facilities, earlier response to perform equipment cleaning, earlier investigation launch, and earlier corrective action generation.
A compelling good laboratory practice (GLP) advantage is that the mechanisms for automated colony counting have proven to perform at least equivalently and, in many cases, more accurately and precisely than traditional counting (1, 6⇓–8). For instance, there are cases in which microbial colonies grow together or on top of each other, making accurate counting by the human eye difficult even by highly trained and experienced analysts. The instruments we will discuss can detect and enumerate these colonies before they become confluent. Traditional human counting accuracy is also subject to human error due to differences in analyst visual acuity, method of tallying colonies, and transcription of observed results to the final data record.
Finally, there are also obvious gains in analyst time and analytical redirection of laboratory resources. Time normally spent by analysts performing repetitive and labor-intensive tasks such as colony counting, data transcription, and data review can be avoided with validated automation. For example, some sites may have environmental monitoring (EM) samples in which 90% of the plates have no colonies (0 CFU) detected. Automated systems will free up analysts from having to assess all of these 0 CFU plates. Finally, these systems are easier to standardize across labs, sample types, and data management systems.
In summary, automation of colony counting delivers:
Data Integrity through automation, direct data capture, access-controlled software, traceability of data, and audit trails
Possible automation of the complete sample chain of custody at the laboratory
Equivalent or improved accuracy, precision, and consistency of counts by introducing an objective readout
No need for contemporaneous/second analyst verification
Faster results for both negative or passing results and positive or failing results
Efficiency given the ability to redirect analyst time away from repetitive tasks and towards tasks that truly benefit from human operations
Standardization of systems and, subsequently, harmonized equipment qualification and method validation packages
Faster lot release that allows products with limited shelf life and also a more flexible supply chain, both of which support patients’ needs
Step 2: Define the Application
Where might an instrument such as an automated colony counter be used to improve a quality control workflow? The answer is simple: in any setting where microbial colonies are recovered and counted on solid agar or on membranes. The applications of automated colony counters for which the previously mentioned benefits can be realized include:
Environmental monitoring (EM)
Utility monitoring (UM): water, compressed gas
Product testing (in-process control samples and bulk drug substance bioburden)
Cleaning validation and monitoring of cleaning efficiency
Raw material bioburden
Root cause investigations including bioburden
Growth promotion testing (GPT) of media
It is important to evaluate carefully the exact application of the system and the critical parameters that need to be evaluated. Some examples are time to result (TTR), ease of use, good manufacturing practices (GMP) compliance, and the type of data output; this is not a comprehensive list and individual companies should define their own requirements.
Automation advantages and return on investment (ROI) would be greater where there is high throughput (e.g., EM program) or a high value attached to a rapid result, such as raw material release, change-over or new equipment release, and product release testing. Accuracy, data integrity, and standardization gains apply to all cases.
Step 3: Assess Requirements
Tables I and II may be used as guides for defining the user requirement specifications and for assessing and rating the automated colony counters that best suit company needs and purposes. These tables present information provided by members of the authoring team, based upon their companies’ evaluations and needs. Please note that both types of systems use a high definition (HD) camera and illumination in conjunction with an algorithm that detects viable colony expansion in real time (or across a time series). One system uses the intrinsic autofluorescence to detect microbial colonies, whereas the other technology improves the first-generation colony counters.
Comparison of Candidates for Automated Colony Counting Systems Using Mandatory Core Requirements
Comparison Table of Candidates for Automated Colony Counters Using Application-Specific Requirements
Step 4: Compare Options and Technologies—Landscaping and Candidate(s) Selection
Each prospective user should conduct a more detailed analysis of the topics in the preceding tables, but a cursory readthrough reveals that some systems are more suited than others depending on the user’s particular need. Major differences between platforms include TTR, flexibility regarding the incubation temperatures (one or two incubators per equipment and the flexibility to move media plates from one incubator to the other incubator, which allows biphasic incubations at 20°C–25°C and 30°C–35°C, for example), the extent of automation, connectivity with LIMS or electronic notebook systems, data management options, initial and sustained cost, and compatibility with existing microbial growth media. If there is a limited number of full-time employees (FTEs) but sufficient budget available for goods and services and interest in cutting-edge systems, the user may benefit the most from a fully automated option. If staffing is not a concern, but there is a desire to gain accuracy, precision, speed, or data integrity, one may opt for a system that allows for the use of existing media but that must still be manually loaded by analysts. In addition, the storage and retrieval of data must be taken into consideration. For example, will the unit be a stand-alone unit, or will it be connected to a general data retrieval system? If there is no need for faster results, one could forego the demonstration of equivalency of rapid result generation and simply assess the accuracy and precision of the automated colony counter and perform reads after the compendial minimum duration (e.g., 3–5 days for product bioburden testing). If data integrity, data storage, and data retrievability are a priority, this must be considered during the computerized system validation of the instrument. After careful assessment, it may be realized that several options could be utilized at individual companies, tailored to specific sites or tests, and which system to investigate will require further evaluation.
Step 5: Develop a Business Case: Technical, Quality and Business Evaluation and Justification
Once a suitable system has been determined for an application, subject matter experts and decision-makers may need to demonstrate an estimated ROI to obtain approval to move forward with procurement or budget proposal for the future. There are several areas of costs and benefits that could be subjected to theoretical and projected calculations, whereas other areas are more appropriately presented through case studies. Some obvious costs are procurement costs, validation, warranties, extended maintenance plans, and training. The difference in consumable costs (if any) may be substantial and should also be factored in.
Benefits can be realized through automation, accuracy, precision, data integrity, faster result generation, and communication. Automation aspects can be assessed against available information such as:
Current FTE time spent on all manual tasks per sample
Future FTE time after implementation on all tasks per sample
Number of samples, both currently and projected
Cost of the relevant FTE time
Return on Investment
A ROI can be determined in multiple ways, such as through actual experience of side-by-side use if a facility has the actual automated colony counter to compare with the traditional method or via a paper exercise to identify the potential cost of implementation of the automated colony counter versus the traditional method. A few considerations (Table III) may help identify the cost differences between the automated colony counter and the traditional method.
Financial Business Case Considerations
Often the manufacturer of the instrumentation will be able to help with the ROI estimation. As part of the ROI evaluation, care should be taken to create and maintain a contingency plan. If the new system is unavailable either temporarily or indefinitely at some point in the future, how are samples, agar plates, and data to be processed? The laboratories should be prepared with a backup plan (for instance by maintaining the traditional way of testing, qualified incubators, and skilled analysts for manual plate reading), especially in cases in which there are limitations with single sourcing of proprietary consumables. It is important to understand and mitigate single points of failure in quality and ensure there is backup equipment/a business continuity plan for disaster mitigation to further derisk supply.
Accuracy Case Study
Consider a critical-to-release sample with a result of “Too-Numerous-To-Count” (TNTC). Many microbial colonies grow quickly and occlude each other to the extent that counting is difficult or impossible. If the microbial specification is for example <10 CFU/10 mL, batch disposition may be at risk as the exact number of colony-forming units cannot be determined. In fact, there are several known examples of TNTC results prompting the rejection of batches. If; however, automated techniques had been applied, it is reasonable that some subset of these cases would not lead to batch rejection. This assumption can be made because earlier reads could be analyzed to determine how many colony origins are on the plate and interim results and plate images may be retrievable depending on the system and software involved. Depending on batch cost analysis, one could be saving hundreds of thousands if not millions of dollars per instance. These examples may be infrequent, but if it happens only once in the first 5 years of deployment, the savings could pay for several instruments.
Data Integrity Considerations
There have also been many examples of undocumented or misplaced data leading to batch rejection and or regulatory observations and fines. With manual data entry and management systems, whether paper or electronic, it is much more likely that result determination, documentation and transcription, and raw data retention can go awry. In order to mitigate these data integrity deficiencies, many companies rely on contemporaneous verification (5), that is, on having a second analyst count the colonies and verify the correct data transfer, which results in additional costs. With validated, automated result recording, storage (potentially including data back-up systems), audit trails and user configuration, it is clear how some failure modes would be removed. The costs of data loss or mismanagement leading to batch rejection or regulatory action should be considered. Specific batch costs and potential or historically identified regulatory fines and indirect impact from regulatory findings should be weighed.
Rapid Result Case Studies
Perhaps a company performs testing that is required to meet a GMP requirement before releasing equipment or facilities. This may even happen many times per year over several sites or equipment process trains. If faster “passing” results could be generated, there would be fewer days of downtime. The cost of downtime for a facility or part of a facility could be calculated per day. If earlier off-testing is sought and demonstrated, savings due to eliminated downtime could be estimated. Traditional plate reading for product samples is generally performed after a minimum of 72 h of incubation (9, 10) and is subject to the availability of analysts to read and report results. With automation, this time could be cut in half and notification potentially performed in real time. Each case would allow for release up to 2 days earlier. Depending on the number of iterations of tests such as these across the facility network, the savings in annual product yield could easily be used to defend system upgrade costs.
Like passing results, failing results can also yield savings when generated and communicated earlier than results generated via current manual methods. For instance, earlier detection and notification may avoid contamination of some equipment downstream or secondary contamination of a subsequently processed lot using contaminated equipment. It might be possible to mitigate the issue without batch loss. Savings might be possible on raw materials that would otherwise have been wasted, column resins that would otherwise have to be replaced, the cost of sanitization, clean-in-place (CIP) and again, though some facility or equipment downtime is unavoidable, there should be less with earlier detection. Equipment and facilities with contaminations detected earlier will have intuitively earlier immediate and corrective actions due to an improved root cause analysis of the event and; therefore, experience an earlier return to service of affected equipment or classified space.
Because ROI may not be solely a financial gain but a gain in different aspects like the ones listed previously, a holistic business case utilizing a decision-making matrix (Table IV) that ties finance, efficiency gains, user/industry feedback, and user requirements (see Column 1) could help identify the right instrument for application in a company. In such an approach, one should determine which categories (finance, efficiency gains, and so forth) are the most important. This will help prioritize that category and allow a higher weighted score to highlight its importance (see Column 2). Weighted scores are arbitrary values and can be determined per the discretion of a cross-functional team. Each company’s values are different; hence the holistic business cases will likely differ between each company even for the same instrument. As with any risk assessment, parameters (see Column 3) need to be identified within each category as topics of evaluation and a high, medium, and low score should be assigned to each parameter with the appropriate definitions (see Column 4). The document owner or cross-functional team may develop as many parameters as needed to produce a thorough evaluation. Once a detailed system is in place, it could be used to rank the instrument under evaluation. A hypothetical example is listed next:
Decision-Making Matrix Using Some of the Aspects Listed Previously (To Be Completed by the End User)
Example Calculation
URS, user requirement specification.
A risk calculation score and outcome could be determined by a cross-functional team.
This example of a holistic business case is one of many methods useful in determining the value of the instrument. There are many more business case approaches, and there is no one size that fits all. However, a thorough business case should be developed with specific local examples, based on actual numbers (samples, facility certifications, change-over qualification, batch losses and other experiences) and/or theoretical savings (if needed during start-up for example) based on the experiences of other companies’ contamination rates or regulatory fines incurred in the past with manual systems in place.
Step 6: Perform Proof-of-Concept Studies/Feasibility Studies/Prevalidation Studies
Once the systems suited to purpose(s) have been determined, it is recommended that proof-of-concept exploration be performed. The following is a list of questions that should be answered. This list is not necessarily exhaustive.
Does the method as well as media type used work for the test matrix?
Does the system detect colonies of specific local microflora?
Is the system and software (if applicable) compatible with the network in the company and LIMS?
Do the instrument and software (if applicable) meet the data integrity program needs within the company?
Can the instrument involved recognize labels or barcodes used in the company?
A way to determine some of these answers may be available through the supplier, who may analyze samples representative of the normal requirements. In addition, benchmarking can be performed with earlier adopters via industry forums and networks such as BioPhorum. If the vendor will not or cannot provide proof-of-concept testing for a company’s specific needs, perhaps beta-testing or system rental is a possibility before full-scale investment is made.
Media Qualification -Growth Promotion Properties
If the media used for the microbial enumeration tests must be purchased from the vendor of the automated colony counter (e.g., specific media cassettes/petri dishes required for the robotic handling), the media plates/cassettes must be evaluated for growth-promoting properties and must be compared with the routine media plates. Typically, the growth promotion test and equivalency evaluation are performed by the receiving company. It is expected that sites will have protocols with predefined acceptance criteria (e.g., percent of recovery of inoculated microorganisms with colony count of the traditional media plates as reference [= 100%]) available for the qualification of a new media supplier, so those can be followed directly.
The media qualification can be performed without the automated colony counting system, as the testing is primarily to evaluate media growth promotion and for contact plates, disinfectant neutralization, and organism recovery from surfaces. As the fundamental growth promotion characteristics of the media result in colony formation, and recovery depends more on the operator’s performance to dilute and inoculate the test organisms, testing can be performed before receipt of the automated colony counting unit itself.
Step 7: Validate at Pilot or Primary Site
The key message for automated compendia method validation is that: “Only the automated aspect of the method requires validation” (11). It is important to note that this is a perspective of certain regulatory agencies and may not be the consensus view of all health authorities. In the case of colony counting that would be the counting instrument itself: accuracy, precision and, IF seeking an earlier read, equivalence to accuracy and precision of the human read at the compendia minimum incubation period needs to be validated. Many aspects of software and firmware validation will take place during equipment qualification (installation and operational qualification (IOQ) including data generation, storage, migration, and retrieval. Performance qualification (PQ) would include demonstration that the instrument and method perform as intended in the assay.
A modular equipment qualification and method qualification approach is shown in Figure 1.
Modular concept of equipment qualification and method qualification of an automated colony counter. EM, environmental monitoring; IOQ, installation and operational qualification; MQ, method qualification; PQ, performance qualification; TTR, time to result.
Not all regulatory agencies have provided detailed expectations, but five countries (EU, US, Canada, Australia, and China) already approved this qualification approach.
General Considerations for the Modular Concept in Figure 1
To claim automated compendia status, all other aspects of the compendia requirements must be maintained (i.e., media formulation, plating method, minimum volumes, incubation temperature, etc.). If earlier readout is desired, “non-inferiority” per USP <1223> (12) and “equivalency” per Ph. Eur. 5.1.6 (13) must be demonstrated for the shorter incubation automated read compared with the compendial duration manual read via a TTR study.
Care should be taken that the analyte matrix itself does not obscure colonies from the reading mechanism.
A minimum of three replicates should be performed for each combination of microorganism and agar during validation.
A target load of 30–70 CFU is recommended (low enough to show sensitivity but high enough to facilitate recovery accuracy and precision calculations).
For qualifying instruments for automated compendial testing the following should be performed:
IOQ.
Post installation, the system will undergo the standard IOQ regardless of the application. With technical transfers, the standard IOQ will be performed on all systems installed.
PQ.
The PQ verifies the accuracy of the automated colony counter algorithms following guidance in USP <1223> (12) and Ph. Eur 5.1.6 (13). It is proposed to use three organisms as indicators of different morphologies/growth characteristics challenging the vision algorithms, that is, E. coli (fast grower), B. subtilis (variable feathered edge colonies), and A. brasiliensis (hyphal mold). Organisms will be analyzed in triplicate with multiple analysts (e.g., three analysts) counting the colonies for each media plate. Secondary publications (6, 7) supporting the vision performance are available to support the reduction in test organism species.
The accurate detection of colonies can be performed on any media plate of the supplier, as colony shape is independent of media formulation and the test organisms cover the range of shapes that are frequently seen.
Accuracy
A minimum acceptance criterion should be established based on the application and corresponding percent recovery guided by the pharmacopoeia. Generally, for microbial enumeration test, harmonized Ph. Eur. 2.6.12 (9)/USP <61> (10) and the nonharmonized USP <1227> (14) provide an acceptance criterion of 50%–200%.
Precision
Differences between counts generated by the automated colony counting system and multiple analysts for each replicate should be analyzed rather than variance in counts across replicates1. To be clear, do not compare counts from separate plates! Variance between plates is not driven primarily by the reading technique but by sample preparation differences (especially when it comes to EM). Remember that the counting techniques are being compared, so the differences in counting mechanisms for the individual plates are sufficient.
Minimum acceptance criteria have not been solidified but allowance for higher counts from the automation than from humans should be considered in the criteria.
TTR Equivalence
If, for the intended use of automation, there is a desire to take advantage of the increased acuity of the system and the ability therefore to discern and quantify colonies at a time earlier than the minimum incubation duration described in the compendia, a noninferiority study (12)/equivalency study (13) must be performed. Results generated via the automated method after the desired minimum incubation must not be inferior to results generated for the same plates incubated to the compendia minimum duration and read via the human eye. There is a possibility of higher counts from the automation than from humans as the colonies might merge during the prolonged incubation times needed to perform the visual counting.
The TTR is linked to the method qualification (MQ) or method suitability tests depending on the number of organisms found in the test samples. For water and EM, sample positions are chosen that carry some microbial load so that suitable numbers can be recovered to give confidence of detection. For water samples, larger volumes than routine can be taken to increase cell numbers.
Sample incubation will be for the duration of the control method, 3–5 days according to Ph. Eur. 2.6.12 (9) or USP <61> (10) or per validated duration for water samples (e.g., 7 days), and visual counts of the colonies will be made by multiple (e.g., three) analysts. The TTR must be determined—especially if the intended incubation time is shorter than the incubation period recommended in regulations (e.g., an incubation of only 36 h for product bioburden samples).
EM samples from class C and D sites and water systems with lower microbiological bioburden requirements sites may have naturally stressed organisms present, whereas water for injection and in process samples are generally 0 CFU. For the latter, microorganisms of an in-house EM collection should be used and should include some suitably stressed microorganisms. A range of microorganism species, including in-house slow growers, should be used for setting TTR. There needs to be a rationale for organism selection for the validation panel.
Sites should address whether to include all types of water samples. The number of water sample types can be reduced as most have minimal variation; however, on occasion, certain in-process samples can demonstrate background fluorescence, so bracketing may not be suitable in all cases.
Note: For EM the incubation strategy (single vs. dual temperature incubation) needs to be defined before the TTR study can be performed.
MQ Modules
The MQ modules are divided by applications and by media. The two bioburden methods, water (using R2A agar) and product (using TSA and SDA agar) use filtration for sample preparation, whereas EM is a direct contact test. The media in the MQ can be different from the media used for the PQ test, as the PQ was testing morphology detection and enumeration, which is generic across media types. All MQ modules rely on the media qualification stage to have proven that the media used by the automated colony counter and the standard in-house media are equivalent for the recovery of organisms in the facility (see PQ previously). The MQ verifies that the interaction of the sample has no effect on the detection of colonies in the sample matrix. For most applications, the TTR and MQ data are generated from the same experimental data set.
Product Bioburden MQ.
The product bioburden MQ may require spiking with pharmacopoeial microorganisms or microorganisms from the local collection if no bioburden is present in the product samples. Using stressed microorganisms with prolonged lag-phase (e.g., caused by treatment at low pH or higher temperatures) is recommended. The test verifies that the process sample type has no effect on recovery, accuracy, and precision of the system count (details see following). After incubation to the TTR defined earlier, manual colony counts will be compared with system colony counts. Equivalent numbers demonstrate no matrix interference with the system detection of microorganisms.
Water MQ.
Samples from water systems with lower microbiological bioburden requirements, for example, systems for water for injection, may be run directly, whereas sterile water types require spiking with pharmacopoeia organisms or organisms from the local collection. The test verifies that the water type has no effect on recovery and accuracy of the count. Following incubation to the TTR defined earlier, manual colony counts will be compared with system colony counts on the same growth plate. Equivalent colony counts demonstrate no matrix interference with the system detection of organisms.
EM MQ.
An in-situ qualification approach is being used. Data for the EM MQ will be taken from active air and surface sample cassettes to verify the colony detection with the two methods. Following incubation to the TTR defined earlier, manual colony counts will be compared with system colony counts on the same growth plate. Equivalent colony counts demonstrate no matrix interference with the system detection of microorganisms. The microorganisms in the test will be examples of stressed cells due to dehydration for air samples and disinfectant effects for surface samples.
Note: if proprietary media plates are used for settling plate application, the surface area of the proprietary media plates must be considered. One system uses a media plate with a relevant area for colony counting of 55 mm in diameter. The limits in for example Annex 1 (current revision (15) but also the Consultation Document December 2020 (16) are based on the use of media plates with a diameter of 90 mm. There are several ways to solve this problem, each with advantages and disadvantages:
Two of the plates (55 mm dimension), may be used. The disadvantage of this option is that this procedure requires more manipulation in the critical hygiene zone—and this should be avoided as this poses a risk to the filling process. And—last, but not least—this procedure reduces the capacity of the automated colony counter.
The proposed use of one 55 mm cassette or media plate as a settle plate is generally acceptable but needs to be justified. For this option, it is recommended to establish a lower action limit to compensate for the decrease in sensitivity due to the smaller surface area of the 55 mm cassette. It should be noted that even if the Food and Drug Administration (FDA) accepts this approach, this does not mean that other health authorities would accept it. For products that are marketed globally, this option needs to be discussed and requires acceptance from other health authorities prior to routine implementation.
Subsequent Matrices
Once an instrument is qualified, automated results generated for subsequent products or formulations need not be constantly compared with traditionally generated results. Instead, the current expectations of the harmonized compendia should be met using the automated method during method verification: 50%–200% recovery of microbial levels <100 CFU should be confirmed for the microorganisms described in the harmonized Ph. Eur. 2.6.12 (9) and USP <61> (10) to demonstrate the ability of the enumeration test to detect microorganisms in the presence of product.
Step 8: Deploy Global/Company-Wide Qualification of Additional Laboratories
Generally speaking, it is advantageous to pilot at one facility when new systems and strategies for equipment qualification are used because
The regulatory perspective is evolving so subsequent qualification iterations can account for new expectations
Resource constraints may limit implementation at several sites at once
Subsequent sites can take advantage of strategy and document precedents
Lessons learned throughout the process can be applied after the initial implementation. Established subsequently adopting sites will require IOQ and PQ considering microflora at a minimum. Entirely new sites with no established microflora, analysts, and manual colony counting history should consider performing a complete Installation, operation, performance qualification (IOPQ)
In other words, each site should have an instrument qualification package considering microflora and local test performance, but TTR studies could be shared across sites if all method parameters are identical (instrument/software versions, incubation temperatures and durations, and so forth). Under these circumstances, experiments performed at the primary site using compendial microorganisms need not to be repeated from secondary sites.
Step 9: Define Regulatory Filings and Implementation Strategy
Regulatory surveillance information for automated colony counters will happen in many ways: under inspection, in annual reports, in new submissions, and in post approval change filings. There is still a debate as to whether or not regulatory filing is necessary for transitioning from manual to automated compendial performance of colony counting for product bioburden samples. As part of the BioPhorum collective experience, one facility has implemented a colony counter for environmental monitoring. The transition from manual to automated counting had no deviations from the compendial test method (e.g., same incubation conditions and incubation time). For this facility, a regulatory decision was made that filing was not needed and as part of their biennial GMP inspection, the colony counter qualification was reviewed and accepted by the FDA. Depending on the language in each company’s current filings, it will need to be determined if amendments to the filings are needed. If there are changes to the compendial test method such as the use of a single temperature incubation condition, a shorter incubation period, the use of different media, and so forth, it is recommended to assess the equipment qualification and method validation strategy and obtain feedback from health authorities for alignment on filing expectations. Recently, another company successfully filed an automated colony counter in Europe and the USA, Switzerland, Canada, and China for product bioburden (IPC as well as Drug Substance release samples) as an “automated counting machine” and NOT as an alternative microbiological method. The equipment qualification was reviewed and accepted by the respective health authorities. The important thing to note is that this may not reflect the view of other health authorities. There is diversity in experience within the member companies and the industry, as another company has validated the automated colony counting system as an alternative microbiological method. Currently there is not enough collective experience within the industry to recommend a standard accepted pathway for regulatory notification and approval, but there is hope that guidance will become clearer as more facilities begin to implement these automated systems.
For new facilities, automated methods may be used from the beginning. In this case, when applying early off-testing, care should be taken to verify the automated colony counting at the desired minimum incubation time and the manual counting method at the compendial minimum so that both counting mechanisms are covered. In this way, counting can be performed manually, if and when the automated system is unavailable.
For new sample matrices tested on previously validated platform methods, only the requirements for compendia verification described in the harmonized compendia need be satisfied. Results read earlier than the compendia minimum must be demonstrated to satisfy these requirements at that earlier time.
Disclaimer
This document represents a consensus view and, as such, it does not fully represent the internal policies of the contributing companies. Neither BioPhorum nor any of the contributing companies accept any liability to any person arising from their use of this document.
This draft article has been developed by a collaborative group of subject matter experts from ten biopharmaceutical companies. It represents a current consensus view also based on the feedback from health authorities. Users of the systems are of course free to deviate from the described equipment qualification and implementation approach.
Conflicts of Interest Declaration
The authors declare that they have no competing interests.
Acknowledgements
Many improvements were suggested by many subject matter experts who reviewed the manuscript. Any remaining errors are our own and should not tarnish the reputations of these reviewers. Special thanks go to Phil Villari (Merck & Co., Inc., Kenilworth, NJ, USA) and Mousumi Paul (Merck & Co., Inc., Kenilworth, NJ, USA) for their helpful insights and comments.
The work was facilitated by BioPhorum, which since its inception in 2004 has become an open and trusted environment where senior leaders of the biopharma industry come together to openly share and discuss the emerging trends and challenges facing their industry. The BioPhorum Operations Group’s mission is to create environments where the global biopharmaceutical industry can collaborate and accelerate their rate of progress, for the benefit of all.
More information can be found at www.biophorum.com.
Footnotes
↵1 Equivalence (or comparability) to the compendial method is shown in terms of “performance equivalence” per USP <1223> (12). This requires the demonstration of equivalent or better results with respect to validation criteria. For the validation of the automated colony counting method, the validation parameters “accuracy” and “precision” are considered.The demonstration of “equivalence” per Ph. Eur. 5.1.6 (13) is carried out through a statistical hypothesis test (which is strongly related to the concept of a statistical confidence interval). In the case of a positive validation outcome, “equivalence” can be claimed with “statistical evidence.” In other words, if equivalence was incorrect, the maximum risk of a false positive validation decision would be only 5%.Statistically, “equivalence” means “non-inferiority” as it includes the two situations “equivalence” and “(much) better.” With respect to “accuracy,” “non-inferiority” means that the automated colony counter generates, on average, equivalent or higher counts than the compendial reference method. With respect to “precision", “non-inferiority” means that the intermediate precision of the automated colony counter is equivalent to or much better than a (fixed) compendial reference precision.
- © PDA, Inc. 2022
In This Issue
Jump to section
- Article
- Abstract
- Introduction
- Approach to Adoption of Automated Colony Counting (ACC) Systems: Applying a 9-Step Framework
- Step 1: Identify Operational/Business Need
- Step 2: Define the Application
- Step 3: Assess Requirements
- Step 4: Compare Options and Technologies—Landscaping and Candidate(s) Selection
- Step 5: Develop a Business Case: Technical, Quality and Business Evaluation and Justification
- Step 6: Perform Proof-of-Concept Studies/Feasibility Studies/Prevalidation Studies
- Step 7: Validate at Pilot or Primary Site
- Step 8: Deploy Global/Company-Wide Qualification of Additional Laboratories
- Step 9: Define Regulatory Filings and Implementation Strategy
- Disclaimer
- Conflicts of Interest Declaration
- Acknowledgements
- Footnotes
- References
- Figures & Data
- References
- Info & Metrics