The ethylene oxide (EO) product test of sterility (ToS) can be conducted to comply with ANSI/AAMI/ISO 11135:2014 for the generation of data to demonstrate the appropriateness of the biological indicator (BI) that is used to develop and qualify the EO sterilization process. Clause D.8.6 of 11135 provides an option to perform a sublethal EO process, followed by conducting a product ToS, performing sterility testing of BIs from the process challenge device, and comparing the test results. Certain limitations for the EO product ToS should be considered when conducting studies that feature the use of this test, in order to support compliance with this requirement. Limitations for any sterility test include sample size, testing frequency, detection sensitivity, and/or the potential for false-positive/false-negative results, each of which must be recognized and well understood in order to support compliance with the standard. In addition, the experimental design of any study featuring the use of a sterility test should be carefully developed to ensure the generation of scientifically sound results and conclusions to support the study objective.
ANSI/AAMI/ISO 11135:20141 requires demonstrating the appropriateness of the biological indicator (BI) that is used to develop and qualify ethylene oxide (EO) sterilization processes. As with any BI used to support sterilization processes, the challenge of the BI and its respective placement must be demonstrated to be equivalent to or greater than the challenge of the product bioburden, in order to support the appropriateness and validity of the BI.
A product test of sterility (ToS) commonly is conducted during studies to support compliance with this requirement. A ToS is performed on product during development, validation, or requalification, and this differs from a test for sterility, which is performed on product following an aseptic process or exposure to a sterilization process.2 This article will focus exclusively on the application of the ToS with EO sterilization. ISO 11135 provides an option for compliance that includes the performance of a sublethal EO process, followed by conducting a product ToS and performing a ToS of BIs from process challenge devices (PCDs). The results from these tests of sterility then are compared to determine if the appropriateness of the BI has been demonstrated. Although this comparison can be used to provide the support for the appropriateness of the BI, 11135 does not provide clear guidance on experimental design and the interpretation of results. In addition, product sterility tests inherently have well-known limitations3 that should also be considered for the evaluation of the ToS results.
This article summarizes and evaluates the limitations of the EO product ToS, including experimental design attributes, in order to provide recommendations for the interpretation of results and the development of scientifically valid conclusions.
Requirements and Approaches
The requirements and approaches used to demonstrate the appropriateness of the BI vary across the traditional terminal sterilization processes, including radiation, moist heat, and EO.
With radiation sterilization, for processes that are conducted in compliance with ANSI/AAMI/ISO 11137-1:2006,4 a BI is not required and therefore demonstration of the appropriateness of the BI is not applicable or required for validating the radiation sterilization process. The most common validation approaches that use BIs are the overkill and the combined BI bioburden approaches.
With the moist heat overkill approach,5,6 the product bioburden population and resistance represent an exceedingly low risk to product sterility, as this approach uses a high level of physical lethality or F0 (typically F0 ≥ 12 min). In addition, the BI used to develop and qualify moist heat sterilization processes typically consists of 106 or greater Geobacillus stearothermophilus spores with a D-value of more than 1.5 minutes, which far exceeds the population and moist heat resistance levels of microorganisms typically associated with medical device and pharmaceutical manufacturing processes. Therefore, there are no requirements to demonstrate the appropriateness of the BI for this approach, as little scientific necessity exists for even routine bioburden monitoring.5
The moist heat product-specific approach uses a lower level of heat history and includes the use of a BI with lower resistance level than the overkill approach. However, the resistance level of the BI must be demonstrated to be equal to or greater than the resistance level of the product bioburden. This is confirmed through heat resistance characterization of bacterial spores or spore-formers, which is performed during routine product bioburden testing without reliance on a product ToS. Product bioburden heat resistance characterization is not difficult to conduct and often includes heat shocking of product bioburden samples, followed by further resistance characterization of any detected bacterial spores in a boiling water bath and/or a moist heat resistometer.
EO sterilization processes may also use an overkill approach with demonstration of the ability of a Bacillus atrophaeus BI to adequately represent product bioburden in the development and qualification of EO overkill sterilization processes, which are more complex compared with those associated with moist heat sterilization. This increased complexity with the EO BI is at least partially related to detection of the EO-resistant mold Pyronema domesticum7 in some EO-sterilized cotton-based products in the 1990s. Also, unlike moist heat, where a boiling water bath can be routinely used to evaluate the resistance level of spores, EO resistance testing can only be conducted following exposure to an EO cycle in a BI evaluation resistometer (BIER) vessel or production sterilizer, which makes it difficult and impractical to conduct on a routine basis. In addition, there are a number of EO-critical process parameters/variables for temperature, EO concentration, and relative humidity, therefore adding complexity to the EO bioburden D-value, which cannot be monitored without a biological challenge. This has resulted in a limitation on EO bioburden resistance data availability in the literature. Therefore, the comparison of product ToS and PCD ToS results after exposure to a sublethal EO process has become a common approach to demonstrating the appropriateness of the EO BI.
The 11135 Product EO ToS: Limitations and Interpretation of Results
Prior to the microbiological performance qualification (MPQ), the product ToS is performed in conjunction with a BI/PCD ToS (see clause D.8.6, approach 2, in 11135) after exposure to a sublethal exposure cycle, including parameters that typically are intended to yield at least one positive BI. The product ToS is conducted to ensure detection of any surviving product bioburden organisms from all surfaces of the product claimed to be sterile. For the BI/PCD ToS, BIs are placed within the PCD, exposed to a sublethal cycle, retrieved, cultured, and then assessed for growth.
Several limitations of the product ToS must be well understood to effectively interpret and apply the results from this test in support of the MPQ in an EO sterilization program. These limitations include testing frequency, sample size/detection sensitivity, false-positive results, false-negative results, experimental design, and interpretation of results. Of note, the limitations and interpretation of results covered in this section should be carefully considered to assess if these apply to other sterilization modalities.
Testing Frequency of the Product ToS
There is no stated frequency in 11135 for the performance of the product ToS after the initial qualification, and this leads to a wide variety of frequencies, as confirmed by the results from the 2019 Best Practices for EO Sterilization (BPEOS) survey (available in the supplemental material for this article at www.aami.org/bit), ranging from not at all to annually for the subsequent performance of this test. Seasonal influences and other inherent variations can affect bioburden population and resistance characteristics. These influences and variations must be well understood with the use of bioburden-based sterilization processes (e.g., dose audits per ANSI/AAMI/ISO 11137-2:20138 ). However, with overkill approaches, an ongoing bioburden monitoring program that considers organism type and distribution can also be effectively used to identify and mitigate risks, as indicated by unfavorable shifts in product bioburden.
Sample Size/Detection Sensitivity
The effect of the sample size used for the product bioburden ToS must be considered to understand the limitations of the results from this test. In most cases, it is possible that organisms that are highly resistant to EO could be present in low numbers on the product prior to being subjected to the sublethal EO exposure cycle. The level of survivorship of these microorganisms is even lower after processing, and the probability of detecting these survivors with the ToS can be calculated.
Considering the frequency and distribution of product bioburden organisms also is important, especially in the areas of the product that represent the greatest challenges for the penetration of EO, heat, and water vapor from the process. Because of automated assembly processes (often yielding low bioburden) and the potentially small surface areas that comprise some of these locations (e.g., mated surfaces of a stopcock), it is probable that the average number of microorganisms/device in the hard-to-reach spaces is low and often less than 1 (i.e., some of the presterilization samples have no bioburden at all in the hard-to-reach space). If no bioburden survivors are observed from the ToS in this example, the shortcoming of the interpretation of these results is that it could be erroneously concluded that an adequate level of lethality from the EO process was delivered to this area, when in actuality there may have been few, if any, microorganisms present in this hardest-to-sterilize area before the sterilization process was applied. The application of risk-based approaches to address this situation is discussed later in the article.
For example, if a product ToS is conducted with the following assumptions: 10 samples are included in the product ToS with a true survivor rate of one positive unit per 10 units tested, the probability of detecting that positive unit in the product ToS is only 66%, with a 34% probability of getting all negative ToS results with this sample size.9
In summary, the sample size selected for use with the ToS should include representation of the types, numbers, and locations of bioburden on the product. Recommendations for the ToS sample size are provided in various International Organization for Standardization (ISO) sterilization standards. For example, 11135 E2.4 states that “the number of samples selected for the product ToS shall not be less than that used for bioburden determination.”1 ANSI/AAMI/ ISO 11737-1:2018 indicates that “it is common practice to use a sample size of between three to ten items for routine monitoring of bioburden levels.”10 The results of the BPEOS survey indicated that it is common to use a sample size of 10 or 20 samples for this test. However, if the probability of an organism being in the hardest-to-reach space is very low (e.g., 1:1,000 or less), the probability of detecting surviving organisms with sample sizes of 10, 20, or even 100 becomes exceedingly remote, thus reducing the sensitivity and overall value of the test. This further highlights the importance of understanding the distribution of bioburden on the product when determining the sample size for the ToS.
False-Positive Product ToS Results
For products with sterile label claims, surviving microorganisms from all internal fluid path surfaces and external surfaces must be directly exposed to the microbiological growth media during the performance of the ToS. Depending on the size and complexity of the device, a limitation of the test method is that it might be difficult to completely immerse the entire device and to ensure that all surfaces are in contact with the microbiological growth media. An example of this would be when testing long lengths of tubing (sometimes exceeding 100 feet in length), complex multicomponent kits, or large containers (sometimes 50-L bags). Oftentimes, the tubing and other components of the medical device must be cut into smaller segments to simplify and optimize the process of ensuring the immersion and direct media contact with all internal and external surfaces with the microbiological growth media. With EO processes, these additional manipulations of the medical device typically are performed after exposure to the sublethal cycle, followed by testing in a highly controlled and confined area, such as in an isolator or a laminar flow hood in a sterility test suite. Depending on the testing scenario, the testing technicians must also be appropriately garbed with sterile gloves (typically two pairs), sterile gown, sterile face mask, and bouffant to protect the product samples from cross-contamination during testing. Although these controls may mitigate some risks for false-positive contamination during the performance of the ToS, these controls may also reduce the dexterity and agility of the technicians during the testing procedure, which can increase the risk of false-positive contamination during testing.
The false-positive rate for the sterility test using cleanroom technology has been estimated to be between 0.1% and 2%.11 Of important note, this false-positive rate is based on sterility testing of parenteral solutions, and the sterility test procedure associated with these products can be considerably less challenging than the sterility test procedure used with complex medical devices. Therefore, the false-positive rate is likely higher for complex medical devices. Certainly, the false-positive rate can be reduced through performing the ToS in an isolator, but some medical devices are too large to be tested in an isolator and not all sterility testing operations have access to an isolator. The potential for false positives and the resulting incorrect dispositions of product sterility tests underscore the need to proactively mitigate potential root causes associated with these situations wherever possible.
Because of the ever-increasing need of new medical devices to support more complicated configurations used in state-of-the-art medical treatments and therapies, EO-sterilized medical devices have, in some instances, also grown from simple devices (e.g., intravenous sets) to be quite complex (e.g., large multicomponent customer kits). For example, an EO-sterilized device that is used for certain therapies may contain, for example, large lengths of tubing (sometimes exceeding 100 feet in length), valves, connectors, and closure systems. This increased level of complexity of medical devices can lead to an associated increase in the level of procedural complexity, which can potentially raise the incidence of both false-positive and -negative results.
False-Negative Product ToS Results
After completion of the sublethal exposure cycle, it is imperative that all surviving microorganisms be effectively recovered and provided the opportunity to demonstrate growth. During the ToS, the device may be filled with and/or submerged into microbiological culture media or organisms are extracted from the device using a recovery fluid, which is subsequently tested for growth from surviving organisms.
In the case of devices filled with or submerged in media, the microbiological growth medium must come in contact with all surviving organisms to provide for growth with visual indication after incubation. In some medical devices, because of inadequate recovery methodologies, one limitation is that surviving microorganisms from all surfaces may not come in contact with microbiological growth media or be extracted with recovery fluid. Therefore, they may never be provided with the opportunity for growth. An example of this situation could be with a stopcock where a surviving microorganism could be isolated in the mated surface (between the housing and core pin) that is not in an open fluid path and never in contact with growth media during the ToS, potentially leading to a false-negative result. This situation can be mitigated by aseptically separating the core pin from the stopcock housing with both parts fully immersed into microbiological growth media.
Another example of a potential false negative is where a substance in the product that has not been properly neutralized leaches out into the microbiological growth medium and inhibits or prevents microbial growth. The application of validated neutralizing agents as part of demonstrating the method suitability (i.e., the absence of bacteriostatic/fungistatic activity) can be used to mitigate this risk.
In addition, it is important to recognize another limitation of the ToS that might lead to a false negative. Although the typical microbiological recovery media and associated incubation conditions are meant to be conducive for the growth of most typical microorganisms, a single set of media and incubation conditions would not be capable of recovering all types of microorganisms. For example, soybean casein digest broth is not capable of absolute recovery of anaerobic or acidophilic organisms, while the conventional incubation temperature range of 30°C to 35°C will not recover psychrophilic or thermophilic organisms. It should never be expected, and it would not it be practicable to expect, that a ToS would be able to detect all viable microorganisms present on a product.
Experimental Design and Interpretation of Results
In addition to the limitations previously summarized for the ToS, it is important to consider the conclusions that can or cannot be made based on the experimental design and associated results from the ToS. Clause D.8.6 of 11135 requires that BIs used as part of establishing the sterilization process shall be shown to be at least as resistant to EO as is the bioburden of the product to be sterilized.
The resistance or D-value typically is determined for a homogeneous population of microorganisms after exposure to a homogeneous level of lethality imparted by a sterilization process. It is important to understand that the current experimental design limitation is that it only provides data following exposure to a single sublethal EO cycle.
An estimate of the BI/PCD resistance or D-value can be determined from the BI starting population level (typically available from the BI certificate and/or enumeration during the study) and the BI survivor population level (fraction negative or survivor curve from the test results) by plotting these data on a semilogarithmic graph to generate a lethality curve. However, it may not be possible to estimate the D-value for the product bioburden because, whereas the bioburden survivor level is provided by the ToS, the starting population from which the survivors originated is difficult to determine because of the heterogeneity of the product bioburden and distribution of the product bioburden. This D-value estimation typically is not conducted. Therefore, although theoretically possible, it can be very difficult to accurately determine the resistance of the bioburden unless the starting population is known for each organism that survived exposure to the sublethal EO process.
Clause D.8.6 in 11135 provides three approaches that can be used for demonstrating the appropriateness of the BI. Approach 1 is focused on demonstrating that most of the microorganisms found on product represent a challenge that is lower than the BI/PCD challenge used to develop and qualify the EO sterilization process. Approach 2 is the primary focus of this article; it recommends the performance of a fractional cycle followed by a comparison of ToS survivor data from the product bioburden and from the BI/PCD. Finally, approach 3 provides potential risk mitigation options that can be used when the product bioburden challenge has been determined to be greater than the PCD/BI challenge.
With the use of approach 2, although it is stated that the typical intention of the study is to achieve no growth in any product ToS samples with the presence of survivors for the BI/PCD, this is not mandated and no further details are provided for the study design, interpretation of results, or minimum acceptance criteria.
After exposure to a single sublethal EO cycle, the ToS only provides a single data point for survivor data for the product bioburden and the BI/PCD subject to the limitations detailed above. A two-point EO lethality curve can provide an approximate estimate of the BI/PCD D-value, which can be generated based on the BI starting population and the level of BI survivors from the ToS. However, although the overall level of product bioburden survivors can be determined, the specific bioburden starting population and level from each species from which the ToS survivors originated typically is not known. An example, including a summary of this information, is depicted in Figure 1, with the following assumptions:
A total of 20 ToS replicates for BI/PCD and product bioburden, all of which are exposed to a homogeneous level of EO processing conditions (i.e., product bioburden and BI/PCD test articles placed adjacent to each other in locations that ensure equivalent exposure conditions)
A BI/PCD starting population of 1 × 106 spores/BI
The product bioburden starting population is unknown, but viable product bioburden organisms have been confirmed to be present at the hardest-to-sterilize location(s) for the product
ToS survivors for BI/PCD = 19 positives
ToS survivors for product bioburden = 10 positives
The calculation of the survivor level12 for the product bioburden includes the assumption that the surviving population is homogenous and present in the quantal region of the lethality curve. Based on the information provided and the associated survivor results, it can be stated that the 20-minute sublethal EO cycle provided a level of survivors (NF) for the BI/PCD (NF = 3.0 average survivors/device) that was greater than the level of survivors for the product bioburden (NF = 0.7 average survivors/device). Based on the two-point lethality curve presented, the D-value of the BI/PCD can be estimated to be approximately 3.6 minutes, while the D-value for the product bioburden cannot be determined from the single survivor data point presented. Based on the results from this 20-minute sublethal EO cycle and in consideration of responses from the BPEOS survey, some companies would consider these results to be an acceptable demonstration of the appropriateness of the BI, without further investigation or action, because the number of BI positives was greater than the number of product bioburden positives. Of important note, 11135 does not mandate complete inactivation of the product bioburden for these comparative studies. Because a single data point cannot be used to generate a lethality curve, and because survivors were detected for both test articles in this example, this is a limitation of the experimental design criteria, as no valid conclusions can be made regarding the resistance or challenge level of the product bioburden based on the information summarized in Figure 1.
Figure 2 is based on the same assumptions and depicts the same information from Figure 1, except that a second sublethal cycle with an increased exposure time of 26.5 minutes was performed. One positive was detected for the BI/PCD, and five positives were detected from the product bioburden in the ToS.
Based on the information provided and the associated survivor results, it can be stated that the 26.5-minute sublethal EO cycle provided a level of survivors for the BI/PCD (NF = 0.05 average survivors/device) that was less than the level of survivors for the product bioburden (NF = 0.3 average survivors/device). With the addition of the previous data for the 20-minute sublethal EO cycle, a two-point lethality curve can now also be generated for the product bioburden, which can be used to estimate a D-value of 17.6 minutes. In this example, this exceeds the D-value for the BI/PCD (3.6 min). Therefore, a minimum of two sublethal exposure cycles are needed, as a single sublethal exposure cycle is a limitation for this experimental design criteria for evaluating the resistance levels for the product bioburden and the BI/PCD. As shown above, the relative survivor counts can depend on the exposure time tested and the comparative resistance can only be accurately quantified if multiple points are known so that the inactivation rate for each can be determined. In cases where positives are detected for the product bioburden and BI/PCD ToS, and especially when the starting types of natural product bioburden are not homogeneous, an additional sublethal EO exposure cycle could be used to more accurately evaluate the resistance levels of the product bioburden and the BI/PCD for demonstrating the appropriateness of the BI.
Recommendations for Demonstrating Appropriateness of an EO BI
Based on the limitations summarized in the previous section, and before finalizing any decisions that involve the use of the ToS to evaluate the appropriateness of the BI, a risk-based analysis should be performed to support any conclusions made. This risk analysis should include the safety factors (e.g., minimum EO concentration/temperature/exposure time parameters for sterilization process validation, product characteristics, and the difficulty of delivering process parameters to the most difficult site to achieve microbiological inactivation) for the EO overkill approach used. In addition, the bioburden risk evaluation considerations summarized in approach 1, as detailed in clause D.8.6 of 11135, should be included to evaluate whether most of the microorganisms found on the product present a lesser challenge to sterilization compared with the BI:
The BI used in the PCD should be in compliance with clauses 5 and 9 of ANSI/AAMI/ISO 11138-2:2017.13
The product bioburden should be consistent and not likely to contain highly resistant microorganisms.
Bioburden trending data should be available to demonstrate the consistency of the bioburden regarding the number and types of microorganisms.
Manufacturing processes and product contact materials should have been evaluated to ensure that potential sources of bioburden are identified and controlled.
In addition to the risks outlined thus far, it also should be confirmed that the hardest-to-sterilize location in the product has been properly evaluated. In cases where these stated risks have been addressed, it may not be necessary to use the ToS to support the appropriateness of the BI. However, in cases where it is still appropriate to utilize the ToS to demonstrate the appropriateness of the BI, options are available for mitigating ToS limitations through the inclusion of risk-based scientific approaches to improve the experimental design and its capability to provide objective and scientifically valid conclusions. As microbial resistance to the sterilization process represents the primary focus for demonstrating the appropriateness of the BI, it is important to understand the two microbial resistance factors that are applicable to the achievement of this objective.
Intrinsic and In Situ Resistance Levels
With EO sterilization processes, both the intrinsic and in situ resistance of the product bioburden should be considered during development and qualification. These two resistance types are defined as follows.
Intrinsic resistance. The resistance or D-value of a population of microorganisms that is induced by the natural state, including microbial genetics, previous growth conditions, and environmental exposure conditions. Of note, with intrinsic resistance testing, the substrate upon which test microorganisms are located may also affect the overall level of resistance.
In situ resistance. The resistance or D-value of a population of microorganisms that is induced by intrinsic resistance factors, the substrate upon which the microorganisms are located, and any localized factors, including where the microorganisms are located on the device that could inhibit, in any way, direct exposure to a sterilant and/or optimal sterilant exposure conditions.
In the studies depicted in Figures 1 and 2, the in situ resistance was calculated to provide a basis for comparison to evaluate the appropriateness of the BI. As stated in its definition, the in situ resistance includes the contribution of intrinsic resistance. However, if it can be demonstrated that the intrinsic resistance level of the BI is greater than or equal to the product and in cases where qualification studies have been successfully completed with the BI placed in the hardest-to-sterilize location(s) of the product, it may be unnecessary to perform sublethal EO cycle studies to compare the in situ resistance of the product bioburden with the resistance of the BI/PCD.
Evaluating Relative Product Bioburden Intrinsic Resistance Levels
In situations where the BI/PCD in situ resistance has already been confirmed (see following section) to be greater than or equal to the in situ resistance of the product bioburden, or when it can be demonstrated that there is a low risk that the product bioburden intrinsic resistance is greater than the intrinsic resistance of the BI (e.g., 11135 clause D.8.6, approach 1, discussed earlier in this article), the associated support and rationale for this conclusion should be formally documented. In addition, it may not be necessary to provide characterization of the intrinsic resistance of the product bioburden. Therefore, the product ToS may not be required in this instance.
Where the in situ resistance of the BI has not been confirmed to be greater than or equal to the in situ resistance of the product bioburden, where the product has a high level of bioburden, and/or where the bioburden potentially contains microorganisms that are highly resistant to EO, a screening study can be performed to compare the intrinsic EO resistances of the product bioburden and the BI.
This determination should include a representative number of samples of the product for which the product bioburden population is known. These product samples should be at a microbiological state that represents the product at the time of EO sterilization. Because the focus of this study is evaluating the product bioburden intrinsic resistance and not the in situ resistance, the product samples may be specifically prepared prior to being subjected to a sublethal EO cycle to reduce the potential for false positives (e.g., aseptically cut, configured into “easy-to-handle” segments, and sealed in EO-permeable packaging). A quantity of BIs (with 106 spores/carrier) that is identical to the quantity of natural product bioburden test articles should be used for this comparative study.
Depending on the physical size of the product, the product and BI test articles can be processed with a sublethal EO cycle in a BIER vessel or a small research-and-development (R&D) sterilizer to maximize the homogeneous exposure of all of the test articles to the EO sterilizing conditions. Because this is a comparative study, all test articles should be located adjacent to each other within the sterilizer. The product test articles and BIs will be processed in a sublethal EO cycle followed by performing the ToS on the product bioburden test articles, along with enumeration of survivors performed on the BI test articles. The log reduction value (LRV) for both test article types then is calculated for comparison purposes.
To demonstrate that the intrinsic resistance of the BI is greater than or equal to that of the product bioburden, the product bioburden must be completely inactivated with an LRV that is greater than or equal to the LRV for the BI. If the product bioburden has survivors, and/or if the product bioburden LRV is less than the LRV for the BI, the intrinsic product bioburden resistance may be greater than the intrinsic resistance for the BI. In this case, further investigation, including additional studies, may be warranted to corroborate the initial data. In addition, a risk-based approach should be considered to set requirements for the frequency of future sterilization resistance evaluations, including linkage to product change control.
Evaluation of In Situ Product Bioburden Resistance Levels
In the study depicted in Figure 2, the in situ resistance was calculated to provide a basis for comparison to evaluate the appropriateness of the BI. As stated in its definition, the in situ resistance includes the contribution of intrinsic resistance factors. However, in cases where qualification studies have been successfully completed where the BI was placed in the hardest-to-sterilize location(s) of the product, and where it has been demonstrated that the intrinsic resistance level of the BI is greater than or equal to the intrinsic resistance level of the product bioburden, performing studies to compare the in situ resistances of the product bioburden and the BI/PCD may not be necessary. Therefore, the use of the product ToS to assess in situ resistance may not be required in this instance.
In consideration of the limitations of the ToS and in situations where evaluation of the in situ resistances of the product bioburden and the BI/PCD are still necessary, improvements can be made to strengthen the approach that was used to generate the data presented in Figure 1. Because this study is focused on the evaluation of the resistance of product bioburden, knowledge about the bioburden population and its distribution within the product should be known. The sample size for this study should be adequate to provide a high level of confidence that bioburden for the test articles is at a microbiological state representative of product with adequate population levels present in areas of interest, including the hardest-to-sterilize locations for the product. As this study focuses on an evaluation of the in situ resistances, BI/PCDs will also serve as test articles to support this comparison.
Product ToS test articles should be paired with and placed adjacent to the BI/PCD test articles. As this is a comparative study, both types of test articles should be located adjacent to each other within the sterilizer. Based on the size and quantities of the test articles, this study could be performed in a BIER vessel, R&D sterilizer, or a production sterilizer. A BIER vessel and/or R&D sterilizer may be able to provide a tighter control of sterilant conditions (e.g., temperature, relative humidity, EO concentration) to reduce variability of the test results. However, if the product size is too large, conducting these studies in a production sterilizer may be necessary. The product ToS test articles and BI/PCDs should be processed with a sublethal EO cycle, then subjected to sterility testing to compare the level of survivors from the product ToS versus the BI/PCDs. For this study to be valid, there must be at least one product ToS sample showing growth and/or one BI/PCD showing growth for the run, and there cannot be all positives for both the product ToS and the BI/PCD test article types in the same run.
Figure 3 depicts a scenario where product bioburden and BI/PCD test articles (20 of each) were exposed to a sublethal EO process with an exposure time of 20 minutes. After processing in the sublethal EO cycle, the test articles were subjected to the ToS. There were no positives (NF < 0.05) detected for the product bioburden test articles. For the BI/PCD test articles, there were 19 positives (NF = 3.0) detected after the 20-minute exposure time. From these data, it can be stated that the resistance level for the BI/PCD is greater than the resistance level for the product bioburden because if subsequent longer exposures were performed generating at least one BI/PCD positive, then no further product bioburden positives would be expected to be generated in this study.
If positives for both test articles are detected in the first study, a second sublethal EO exposure run is required. The number of positives for both test articles for each of the two runs is determined. The most resistant test article type is the one with the greatest number of positives for each of the two runs. Ideally, the same test article type will have the greatest number of ToS positives for both runs, which then would support its greater level of in situ resistance. However, in situations where this is not true, an investigation should be performed to ensure the validity of study results. In some cases where the in situ resistance levels of the product bioburden and the BI/PCD are similar, an additional sublethal EO study may be warranted.
Similar to intrinsic resistance testing, a risk-based approach should be used to set requirements for the frequency of future evaluations for in situ resistance, including linkage to product change control.
The results from the BPEOS survey indicated that the overkill approach is the most common cycle design approach used for EO sterilization processes. Although the overkill approach includes the use of multiple safety factors and is the most conservative option, product sterility should be supported with the application of scientifically valid approaches for process development and validation, and this includes assessment of product bioburden risks.
To support the demonstration of the appropriateness of the BI, product bioburden and PCD/BI test articles can be exposed to a sublethal EO sterilization process followed by a comparison of the ToS results to assess the survivor levels for each test article type. With the performance of any sterility test, particularly the ToS in this application, limitations must be recognized in the performance, interpretation, and use of results from this test. Because of these limitations, the use of a risk-based assessment and supporting scientific rationale may be leveraged to support the appropriateness of the BI wherever possible without reliance on the product ToS.
In cases where the ToS remains necessary, the experimental study design recommendations that have been provided may be considered to ensure the generation of scientifically valid results and conclusions. It is also critical that an extreme level of diligence should be exercised to ensure the proper execution of this test, including effective mitigations that reduce the probability of a false-positive, false-negative, and/or any invalid result.
Publication of a future article featuring a decision tree aid will be sought in order to provide additional background and recommendations of the best demonstrated approaches, including scientifically valid approaches that are not reliant on the ToS.
About the Authors
Michael Sadowski, BS, is the lead scientist at Baxter Healthcare in Round Lake, IL. Email: email@example.com
Clark Houghtling, BS, is the vice president of business development and technical affairs at Cosmed Group in Jamestown, RI. Email: firstname.lastname@example.org
Sopheak Srun, BS, MPH, is a principal sterilization specialist at Quality Tech Services in Bloomington, MN. Email: email@example.com
Tim Carlson, BS, is an engineering specialist at Baxter Healthcare in Mountain Home, AR. Email: firstname.lastname@example.org
Jason Hedrick, BS, is a senior principal sterilization microbiologist at Medtronic in Moundsview, MN. Email: email@example.com
Andrew Porteous, BS, is a principal engineer at Baxter Healthcare in Round Lake, IL. Email: firstname.lastname@example.org
Ken Gordon, BS, is a principal scientist of innovation and industry representation at STERIS in Spartanburg, SC. Email: email@example.com