EtO Sterilization: Microbiological Aspects of Process Validation

logical indicators can provide a validation process for EtO sterilization that ensures accuracy, the absence of microbes, and a smooth testing procedure.

Susan Edel Satter and Paul J. Sordellini

A companion to this article, EtO Sterilization: Principles of Process Design, discussed the components of each phase of two 100% EtO with nitrogen processes, focusing on the engineering aspects of designing EtO cycles.¹ This article focuses on various approaches to medical device sterilization cycle validation from a microbiological standpoint. The discussion assumes that the first two stages of the validation—the commissioning and the physical performance qualification of the sterilization chamber—have been successfully completed. Therefore, this discussion deals solely with the third stage of EtO validation: the microbiological performance qualification (MPQ). The topics that are discussed include how to choose the appropriate microbial challenge for an EtO sterilization process, approach to developing the EtO cycle, product load, placement of biological indicators in/on the product load, method for determining cycle lethality, and calculations to determine the D-value.

Information is also given on the documentation for the report on validation and certification of the process, and revalidation is discussed briefly. Some suggestions exceed the current requirements presented in international standards, but they can enhance a validation process, resulting in a more thorough and accurate study.

International standards ISO 9001 and 9002 present the quality system requirements for the design, development, production, installation, and servicing of healthcare products. The ISO 9000 series treats medical device sterilization as a special manufacturing process because the results cannot be verified by inspecting and testing 100% of the product at the conclusion of the cycle. Sterilization processes must, therefore, be assessed for special considerations, validated prior to use (or during use in certain situations), and routinely monitored.

To design an effective validation and routine control program for a sterilization process, the bioburden on the product and packaging must be considered. Bioburden is defined as "the population of viable microorganisms on a product and/or a package" and is characterized in terms of number, identity, and resistance. A validated test method must demonstrate that it can consistently and adequately remove the bioburden from the product and packaging. There are various bioburden test methods and associated validation procedures from which to choose.²

Many factors can contribute to the bioburden on product and packaging, including the origin of the raw materials and components, transit and storage conditions, and the manufacturing environment in which the finished products are assembled and packaged. European standards place particular emphasis on controlling the processes used to manufacture sterile products.³

Sterility is defined as the "state of being free from viable microorganisms."⁴ Microbial death relating to the gaseous sterilization of healthcare products is an exponential function typically defined as the probability of a nonsterile item existing per a given number of units in a batch. This probability, called the sterility assurance level (SAL), defines the "probability of a viable microorganism being present on a product unit after sterilization."⁴ While sterilization can reduce the bioburden on a given product to a very low number, that probability can never be reduced to zero. Therefore, in order to achieve the desired bioburden levels, it is critical to design a validation program that provides a high degree of confidence for consistent sterilization.

DETERMINING THE APPROPRIATE MICROBIAL CHALLENGE

A biological indicator (BI) is an "inoculated carrier contained within its primary pack providing a known resistance to the relevant process."⁴ There are many different types of BIs but the most common include:

• Commercial units supplied with the manufacturer's certification, such as bacterial spore strips individually packaged in glassine envelopes or pieces of filter paper impregnated with a certified population of a challenge organism that has a known resistance to EtO, such as the spores of Bacillus subtilis var. niger.

• A liquid bacterial spore suspension commercially sold and certified by the manufacturer. The suspension is placed on or in the product, which is then referred to as inoculated product.

• A liquid bacterial spore suspension made by a device manufacturer using a commercial, certified strain of a bacterium.

• Strains of microorganisms that have been isolated from a device manufacturing facility. These strains represent the most resistant organisms found in or on the devices and are typically used for the combined BI/bioburden method of validation.

It is crucial to ensure that the type of BI used to validate or routinely monitor a given sterilization process is the most appropriate indicator for that process. In addition to identity, quantitation, resistance, storage, general directions for use, and disposal conditions, the manufacturers of BIs are required to provide information regarding the optimal culturing conditions, such as temperature and type of growth media.

Irrespective of which BI is chosen, the methods used to recover the challenge organism must be validated. This recovery is expressed in terms of the percent recovery of the original inoculum. These recovery studies can be especially challenging when using liquid spore suspensions because of the interaction between the suspension and the material onto which it is inoculated. The material substrate can alter the resistance characteristics of the inoculum because of such anomalies as spore clumping or the physical sheltering of spores in certain sites within the product.

The goal is to kill the microbes, which means disabling their ability to reproduce even in their most favorable growth conditions as described by their manufacturer. There have been records of seemingly killed microbes that regenerated when conditions become favorable.⁵ The user must validate that the incubation time under the prescribed conditions is sufficient to recover delayed growth of the organisms after exposure to a given sterilization process. For routine processing, this time period is typically 7 days unless validated for a shorter time period in accordance with current national requirements.⁶ In such cases, periodic checks should be run to confirm that the shorter time period yields equivalent recoveries to those obtained from the longer incubation period. It is also important to ensure that the incubation time is sufficient to recover growth from injured organisms exposed to sublethal cycles. In some cases, this may mean using a 14-day incubation period. This incubation period is also required by the U.S. Pharmacopeia for product sterility testing.⁷

Just as the BI must provide a defined resistance to a specified process, it is necessary to prove that the inherent bioburden on the product does not have a greater resistance than the BI. Characterizing bioburden involves quantitation, identity, and resistance of the bioburden. Several methods can be employed to determine which BI is appropriate for a specific situation.

• Challenge organisms that make up BIs typically resist a particular sterilization process far more than do common bioburden organisms. The resistance of the bioburden cannot be adequately evaluated by quantitation only, yet it must be determined if the products to be processed might contain organisms that are more resistant than the challenge organism. After the microbial identifications and quantitation of the bioburden have been completed and analyzed, comparative resistance determinations of the most resistant bioburden component or product must be calculated. A literature review is also required.

• When microbial identifications of the bioburden are not performed, at least one sublethal cycle should be run to compare the relative inactivation rates of the bioburden with that of the challenge organism. Product sterility testing, after exposure to at least one sublethal cycle under appropriate experimental conditions, can ensure that the product's bioburden is not more resistant than the challenge organism.

• If the quantity, identity, and resistance of the product's bioburden are known, it might be possible to validate and routinely monitor the sterilization process by combining BI and bioburden methods. It must be demonstrated that the BI's degree of challenge to the sterilization process is adequate to ensure that the process will attain the desired SAL for the bioburden. Combining the BI and bioburden methods to determine the appropriateness of the BI can be time-consuming and result in additional testing costs. However, the required sterilization parameters can be more accurately determined, which can result in reduced processing time and reduced exposure to the sterilant.

BIs can be configured in many different ways depending on the cycle development method chosen.

Inoculated Product. The actual product, configured and packaged as it is intended to be sold, can be inoculated with spores of a microorganism such as Bacillus subtilis var. niger. Direct inoculation usually uses spores suspended in liquid, then placed on the product and dried. The product's surface characteristics will affect the distribution of spores and may lead to a difference in resistance behavior compared with other challenge systems.⁴ It is, therefore, important to achieve an even distribution of spores on the product's surface. Indirect inoculation involves placement of a carrier, such as filter paper that has been impregnated with spores, in the product or its package.

Inoculated Unit. A carrier, such as a filter paper strip or disk, can be inoculated with a population of a resistant organism, such as Bacillus subtilis var. niger, that has been extensively characterized and certified by the manufacturer. The resistance of this inoculated carrier must be compared to that of the inherent bioburden of the product being validated or the equivalent simulated product. An inoculated unit is usually used when there is the potential that the bioburden on the product is more resistant than the indicator organism and is required for the combined bioburden and BI cycle development method.

Inoculated Simulated Product. A simulated product that comprises the most difficult to sterilize portions of a device or that configuratively represents a device family can also be directly or indirectly inoculated. This simulated product must present the greatest challenge to the process in order to be considered an adequate microbial challenge. Each unit must contain a certified inoculum either in liquid form or on a carrier.

Natural Product. The inherent bioburden on the product can also be used as the microbial challenge during validation and for routine monitoring when the absolute bioburden method is employed for cycle development (see "EtO Cycle Development Approaches", below).

All validation methods for EtO sterilization require that the BI used for validation and to monitor routine cycles must be more resistant than the bioburden of the product and be placed in a location that is more difficult to sterilize. Comparative resistance testing is an effective means of selecting the BI and its location in the product that presents the greatest challenge to the sterilization process. Such an assessment should be made prior to validation as part of determining the appropriateness of the BI. These studies are usually carried out in small chambers that are capable of delivering rapid ramp rates, e.g., the times required to achieve specific pressure set points.

Products should be exposed to cycles in which the only variable is the gas exposure time period. The data obtained from this testing can be used to justify the choice of a specific actual or simulated product to inoculate and use for the BI. If the design of the product is such that a BI unit cannot be placed in the part of the device that is the most difficult to sterilize, the product should be inoculated with a liquid spore suspension to provide a known number of viable spores. The spore suspension, materials, and techniques used should comply with ISO 11138, parts 1 and 2.^8,9

Many device manufacturers include an additional objective in their validation plan that involves the use of external BI monitoring systems. Often referred to as process challenge devices (PCDs), they assess the lethality of the EtO process after the cycle has been designed. The PCDs are geometrically distributed around the load rather than in internal locations in the case cartons. Direct comparisons can then be made between the sterility test data obtained from these external PCDs and the BIs placed in internal locations. A PCD must be shown through comparative resistance studies to provide more of a challenge to the process when it is placed in external locations in the load than do the the BIs placed in internal locations. They usually, therefore, bear no resemblance to the product. Examples of external PCDs are spore strips double-packaged in plastic bags, in sealed plastic tubing, or in syringes. There are also commercially available PCDs that are sold as ready-to-use packaged systems. It is advisable during the validation studies to evaluate different PCD configurations during the comparative resistance studies to determine the best candidate. To monitor routine sterilization cycles, it must be shown at the time of validation that the PCDs in the external locations comply with the same requirements for resistance to sterilization.

ETO CYCLE DEVELOPMENT APPROACHES

There are three basic approaches to developing EtO sterilization cycles—the overkill method, the combined bioburden and BI method, and the absolute bioburden method.

The overkill method is probably the most widely used because it is relatively easy to use and it results in a robust SAL. The method ensures that the sterilization process will inactivate a specific number of microorganism spores known to be resistant to the EtO sterilization process. The organism most commonly used to monitor the overkill process is Bacillus subtilis var. niger. A certified preparation consisting of a stated population of Bacillus subtilis var. niger spores is inactivated through exposure to specific cycle parameters that have been assessed to be significantly higher than those required to kill the inherent bioburden on the product. The parameters are increased on a routine basis to provide the desired SAL (see "Methods for Determining Cycle Lethality," below).

The combined bioburden and BI method is used when the two are equally resistant. This method requires routine bioburden and BI testing in addition to a considerable amount of routine sterility testing to develop a cycle that will inactivate the BI challenge population. The BI must be sufficiently resistant to ensure that the EtO process will deliver the desired SAL relative to the bioburden on the product.

The absolute bioburden method is used less frequently in cycle development because it requires extensive testing in both the development phase and routine processing. However, it must be used when the product's bioburden is more resistant than the BI. Such bioburden resistance to the EtO process can be caused by any number of factors, such as the configuration of the product, the quantity or location of the microorganisms, or the bioburden's intrinsic resistance. Since the bioburden on the product constitutes the essential microbial challenge for the process, the bioburden test method must be validated and strictly controlled. The resistant microorganisms are screened through bioburden testing and may be isolated and propagated for use in cycle development studies. One negative of this method is that the microorganisms' resistance can change as a result of how they are cultured, which can adversely affect the results of the cycle development studies. The absolute bioburden method also requires extensive controls of the manufacturing environment in addition to routine product bioburden monitoring and resistance studies.

DETERMINING THE APPROPRIATE PRODUCT LOAD CHALLENGE

Microbiological performance qualification (MPQ) should be performed using specified products and packaging configured in the same manner in which they will be routinely sterilized. For the cycle to be accurate, the product load must represent the greatest challenge intended for future routine sterilization. If a device manufacturer intends to use multiple load configurations on an ongoing basis, the densest configuration should be used for the MPQ.

Each type of configuration must be documented in terms of the number of product units per case, the number of cases per pallet, the stacking patterns on the pallet, and the density. This documentation should be included with the validation data. Some testing should also be conducted on the least dense configuration, which, theoretically, presents less of a challenge to the process. This testing can be as simple as placing thermocouples throughout the least dense load on a routine cycle and comparing the temperature distribution with that of the densest load. In other cases, additional microbial challenge studies might be required. Changes in the product load must be evaluated carefully because seemingly innocuous changes, such as changing the shrink wrap or corrugate on the load, can have a significant effect on the cycle's efficacy from the perspective of product sterilization.

BI PLACEMENT IN THE PRODUCT LOAD

After the product load challenge has been identified, the BI positioning and placement can be determined. BIs should be distributed throughout the product load and, as much as possible, in the same orientation (e.g., vertical). The placement must include those locations that are considered to present the greatest challenge to the process and can be the same as those used for temperature monitoring. The ANSI/AAMI/ISO 11135-1994 standard suggests placing two BIs at each location with a temperature-monitoring device in order to obtain additional information on process efficacy. It also provides the following recommendation for the number of BIs to be included in each validation cycle:

• At least 20 BIs for usable chamber volumes up to 5 m³.

• Increase the number of BIs by two for every additional 1 m³ of usable sterilizer chamber volume between 5 and 10 m³.

• Increase the number of BIs by two more for every additional 2 m³ of usable sterilizer chamber volume above 10 m³.

The AAMI technical information report "Contract Sterilization for Ethylene Oxide" can provide additional information on the number of BIs and monitoring devices recommended based on product load volume.¹⁰

METHODS FOR DETERMINING CYCLE LETHALITY

Results obtained from commissioning and physical performance qualification and monitoring devices should be used to identify critical features of the equipment or process that can be investigated during the MPQ. For example, it is critical that the sterilant injection time is consistent among the MPQ cycles to ensure a uniform delivery from one cycle to the next. Even minor changes in the sterilant injection time can result in significant differences in lethality.

The MPQ should be performed in the industrial chamber that will also be used for routine processing unless equivalency can be demonstrated between the industrial chamber and whatever chamber is used for the qualification. Maintaining the precise and consistent delivery of the sterilization cycle parameters is more difficult to accomplish in large industrial chambers than in small test chambers. It is also important to conduct these studies using the actual product load intended for routine sterilization. Hence, these studies are usually conducted in large industrial chambers rather than in small test chambers.

The MPQ can be performed by determining the lethality of the cycle on the basis of the number of D-values applied. The D-value is defined as "the time required to reduce a specific microbial population by 90% or one logarithm."⁴ The survivor curve construction or fraction-negative methods (described below) may be used as outlined in current standards.⁴ Another means of evaluating the MPQ is the half-cycle method, based on the number of times required to completely inactivate the BI microorganisms with an added margin of safety. The ultimate objective of each method is to determine the full cycle to which the product load must be exposed.

Survivor Curve Construction Method. The survivor curve construction method involves the direct enumeration of survivors in terms of colony forming units (CFUs) recovered after exposure to graded amounts of the sterilization cycle. A CFU is defined as "a visible outgrowth of a population of organisms arising from a single or multiple cells."² A minimum of five cycles should be run, each using different graded time exposures to EtO.⁴ The parameters used, with the exception of the gas exposure time, must be kept consistent. The first cycle is a time zero study in which the initial CFU survival count of the BI is determined by exposing the BIs to all stages of the process, including preconditioning if used, prior to the EtO injection phase of the cycle. All BIs should survive because they will not be exposed to the sterilant.

After each of the four or more additional cycles, all employing different gas exposure time periods, the number of BIs that survive the processes are counted. The BIs should be removed from the chamber and the load as soon as possible within the confines of worker exposure policies. The BIs should also be tested as soon as possible after being removed to reduce their exposure to EtO residuals, which can affect BI survival rates. In all cases, the time intervals between when the load is removed from the chamber, when the BIs are removed from the load, and when they are subjected to the enumeration process must be consistent among the cycles. Ideally, a final enumeration of the BIs from one cycle should be obtained before the next cycle is initiated to more accurately assess the exposure time to use. This is not always feasible because the enumeration process can take days or weeks to complete.

The number of BIs used in each of the cycles in the study should be statistically significant to ensure obtaining dependable data. The number can also be based on the size of the chamber or the size of the product load. The data acquired in the study are used to calculate an EtO exposure time and the minimum process parameters expected to elicit a specific probability of survival of the challenge organism expressed in CFUs (see "The Full Cycle," below). Theoretically, because all other process parameters are the same, the statistical evaluation of the data should result in a plotted survivor curve or regression analysis that demonstrates a consistent relationship between the EtO gas exposure time and the number of survivors (positive BIs). Unfortunately, this is not always the case. In addition to the issue of consistency between the BIs, the integrity of the data is dependent on the consistent delivery of the process parameters for each cycle, including precise control of gas injection times that can be considered part of the gas exposure phase for the purposes of this study. Controlling the gas injection time in large industrial EtO chambers necessitates ensuring that the head space pressure in the gas tanks is consistent from one cycle to another and that the gas delivery lines are uniformly either full or purged of gas. A few minutes difference in gas inject time can significantly change the results in this study. Other variables to consider include the temperature of the gas volatilizer, evacuation rates and times, gas makeups, and air exchanges.

If the survivor curve study is conducted in an industrial chamber, the data should elicit an accurate probability of survival of a specific challenge organism from which routine cycle parameters can be determined. The BI should be an inoculated carrier. For example, if the BI is a spore strip that contains a population of at least one million (10⁶) spores of Bacillus subtilis var. niger and the construction of the survivor curves demonstrates that total kill was obtained at 2 hours of gas exposure time, the full cycle used to routinely sterilize the product to an SAL of 10^–6 could employ 4 hours of gas exposure. To achieve a smaller SAL, for example 10^–3, would require adding the appropriate additional gas exposure time to the original 2 hours.

The BI testing process should be validated prior to initiation of the survivor curve study to obtain an acceptable and documented recovery method. The process begins with macerating the BIs in sterile water. The suspension is evaluated by plating specific dilution aliquots of serial dilutions of the suspension onto a selected agar medium and counting the number of CFUs after incubation. The number and extent of dilutions needed will be based on the duration of gas exposure and concentration in relation to the number of BI organisms. In other words, more dilutions are required when more survivors are expected.

A statistical analysis made from each cycle of survival data should show the log₁₀ of the surviving population plotted against EtO exposure time intervals. The best-fit rectilinear curve through the data can be drawn or determined by regression analysis using the method of least squares.⁸

The survivor curve method is complicated by the number of serial dilutions that must be prepared and the quality of the dilutions being dependent on the skill of the person performing the test as well as the precision of the equipment used. Only trained personnel who can adequately practice aseptic technique should conduct this test. Calibrated pipettes and dilution controls help ensure the test's accuracy. The dilutions should also be chosen to yield counts between 30 and 300 CFUs. It is generally assumed that numbers ranging from 30 to 100 CFUs should be used because it is thought that higher numbers of CFUs per plate could result in inaccurately low counts and that numbers lower than 10 CFUs per plate could give unreliable counts. For practical purposes, counts between 30 and 300 are generally acceptable.

Fraction-Negative Method. The fraction-negative method also involves exposing BIs to multiple cycles of graded exposures to EtO. The differences between the two methods are the number of cycles recommended and the number used to enumerate the survivors (in terms of the positive BIs).⁴ A minimum of seven cycles should be employed in this study, each utilizing different gas exposure time periods. These seven cycles should elicit the following survivor data:

• At least one sample set that elicits all survivors (growth in all BIs tested).

• At least four sample sets that elicit fractional data, i.e., a fraction of the BIs in each set demonstrates growth or survival.

• At least two sample sets in which there is neither growth nor survivors.

The method used to enumerate the number of positive BIs is more straightforward than in the survivor curve method. In the fraction-negative method, the BIs are immersed directly into the appropriate media and incubated. Results are recorded in terms of the total number of BIs demonstrating growth and the total number eliciting no growth for each set of test samples.

Calculating D-Values. The D-value corresponding to the survivor curve construction results is determined by either reading it from a graph or calculating from the data the relevant time interval for reducing the count by one log (Table I). The performance of the fraction-negative method will generate data that are then utilized to calculate a D-value using a method described by Pflug and Holcomb.¹¹ This method is also referred to as the full or generic Spearman-Karber procedure. It does not require the use of the same number of replicates nor the same gas exposure time intervals because it involves spore strips rather than counting CFUs.

Time of Exposure to Sterilant (t)	Number of Test Samples Exposed (n)	Number of Test Samples Showing No Growth (r)
t1	n1	r1
t2	n2	r2
t3	n3	r3
t4	n4	r4
t5	n5	r5
t6	n6	r6
t7	n7	r7

Table I. Variables used in calculating D-values during MPQ.

In Table I, t₁ represents the shortest exposure time to sterilant, and all test samples run through this short time frame should show growth. The variables t₂ to t₅ correspond to increasing exposure times in the fraction-negative area, otherwise known as the quantal region. Exposure times t₆ and t₇ should represent tests in which none of the tests show growth.

In the following equations, r₁ is the number of test samples out of the number exposed (n_i) that show no growth at exposure time t₁. For each period of exposure to sterilant t₁ to t₆, the factors x and y are calculated as shown:

At t_i, all test samples show growth, therefore,

From the calculated values of x_i and y_i above, the value µ_I (mean time to attain no growth) can be calculated for each period of exposure (t₁ to t₆) as follows:

The mean time to attain no growth from all of the test samples, µ, can be calculated from the sum of µ_I for each time of exposure t₁ to t₆:

Where the interval between exposure times (d) is constant and the same number of test samples (n) is used at each exposure time, the mean to attain no growth (µ) can be calculated from the equation:

The mean D-value (D) can be calculated from the equation:

where N₀ = the initial number of challenge organisms per test sample.

For the purposes of calculating the sterilization period using the D_calc method below, the upper 95% confidence level for D should be used. This can be calculated from the equation:

where V is derived as follows:

In addition to the method described above, there are three other commonly used means of calculating D-values. The limited Spearman-Karber procedure requires the same number of replicates for each exposure time and equal intervals between the various gas exposure times. For example, the exposure times of 9, 12, 15, and 18 minutes provide an arithmetic series of 3-minute time intervals.

The Stumbo-Murphy-Cochran procedure requires one result in the fraction-negative range, consisting of time (t), the number of units negative for growth (r) and the number of replicates (n) at each exposure time as well as the initial number of microorganisms per replicate (N₀). The D-value for each exposure time is calculated from the following equation. The confidence limits of the mean may be calculated by conventional procedures using natural logs.

The factors in the equation represent the following:

U = exposure interval.

N₀ = initial number of organisms per replicate carrier unit.

N_U = ln (n/r).

n = total number of replicate units at exposure U.

r = number of units negative for growth at U.

The D-values are averaged to obtain the overall D-value for the experiment.

The limited Stumbo-Murphy-Cochran procedure requires only one exposure time in the fraction-negative range.

It is important to choose the most appropriate procedure of the three described to calculate the D-value and its upper confidence limit. Certain methods have limitations. For example, Graham and Boris have noted that a confidence interval cannot always be determined from the Stumbo-Murphy-Cochran procedure.¹² Shintani et al. have observed that the Stumbo-Murphy-Cochran procedure can yield invalid results if r is close to n.¹³ They also state, however, that this procedure may be superior in situations in which the sample population is at least 50, r is at least 1, and r/n is less than 0.9.

Half-Cycle Method. The half-cycle method determines the minimum time a specific product load must be exposed to an EtO process to guarantee that no survivors exist from the BIs used to monitor the cycle's efficacy. The cycle chosen is based not only on the elements discussed in the companion to this article but also on the microbiological data obtained prior to determining the parameters for the half-cycle.¹ The minimum ranges of the half-cycle are based on this predetermined microbiological data, whereas the maximum ranges are dictated by the product and packaging tolerances established through functional testing.

If the half-cycle method is chosen, an additional cycle should be performed, preferably before the first half-cycle is conducted. In some cases, other proportionate cycles, referred to as sublethal or fractional cycles, may be used. The gas exposure time is shortened in order to obtain survivors from the BIs used to monitor the cycles. The process parameters in a sublethal cycle are the same as those in the half-cycle with the exception of the gas exposure time period. Such cycles are conducted to demonstrate that positive BIs can be recovered and to prove the adequacy of the BI, the sterility test method, and associated equipment and materials. The sublethal cycle is not required if either the survivor curve construction or fraction-negative method is used because these methods require positive sterility test results as part of the study.

A second objective in conducting a sublethal cycle is to attempt to demonstrate that the resistance of the product's bioburden is less than that of the BI. This is accomplished by using actual product test samples that have been manufactured, packaged, and handled in the same manner intended for routine production. It is imperative that these test samples were made in accordance with routine manufacturing procedures because the bioburden on the product must represent what the product would normally contain. Sometimes using a number of different product samples can yield useful information when there is concern about not only the quantity of bioburden but also where it is located. If a number of different samples have been sterility tested, those with the highest bioburden should be used for validation. After the completion of the sublethal cycle, the product samples are sterility tested. Since the objective of the test is to demonstrate sterility of the bioburden, the data can be compared to the positive sterility test results obtained from the BIs to demonstrate that the resistance of the product's bioburden is less than that of the BI.

A third objective of the sublethal cycle study is to determine that a PCD used as an external monitoring device for routine sterilization cycles yields more positive sterility tests than an internal BI does. It has been suggested that it is acceptable to obtain positive sterility test results from those external PCDs exposed to the half-cycles if previous comparative resistance studies have proven that they are as resistant or more resistant than the internal product BI and if no positive sterility test results are obtained from the BIs placed in internal load locations. If total kill is obtained from all PCDs on all half-cycles, it can be assumed that the resulting validated full cycle will provide an additional safety factor that goes beyond the minimum requirements.

Typically, the sterilization efficacy of the first half-cycle is determined before any subsequent cycles are conducted. If the first half-cycle experiment elicits all negative BIs, two additional half-cycles are conducted to confirm the data. All three half-cycles must employ the same parameters to demonstrate reproducibility and reliability. All three half-cycles must demonstrate the ability to elicit all negative BIs. If a SAL of 10^–6 is required, a gas exposure time that is at least double that used in the half-cycle becomes the minimum gas exposure time used routinely in the full cycle.

THE FULL CYCLE

The data obtained from the survivor curve construction method, the fraction-negative method, or the half-cycle method are used to design the full cycle that will be used to routinely sterilize the product. The full cycle must be capable of reliably demonstrating a required SAL that consists of the minimum time to obtain all negative BIs or CFUs with an additional margin of safety. Normally this can be expressed as 10^–n where n is the cumulative probabilities of the log minimum time to sterilize and the log margin of safety. For example, if the half-cycle method was chosen to validate a given sterilization cycle with a required SAL of 10^–6, the half-cycle must demonstrate the ability to ensure that there are no survivors from BIs that have a certified population of 10⁶. Since the BIs have a greater resistance than the product bioburden, it can be concluded that the time is sufficient to achieve product sterility. However, adding an equivalent sterilization time period increases the margin of safety. The additional 6 log reduction in the population (10^–6) is theoretical and is obtained by doubling the half-cycle gas exposure time. In other words, if the half-cycle gas exposure time period was 2 hours, the full-cycle gas exposure time would be a minimum of 4 hours.

The processing ranges established for the critical parameters in the full cycle should be based on the ranges determined in the half-cycle. The minimum process parameters in the full cycle, such as temperature, sterilant concentration, pressure, and humidity, may also be greater than those used in the half-cycle to ensure greater lethality with the full cycle. The first full cycle should be included in the validation to complete the profile of the product load. The same type and amount of monitoring should be conducted during the first full cycle as was conducted on each of the half-cycles. Comparisons of temperature distribution throughout the chamber and the load can justify a determination of reliability and repeatability for the process. These data can also be useful for determining the significance of any minor changes that are made to the load.

REPORT ON VALIDATION AND CERTIFICATION

The documentation on the validation and certification should include but not be limited to the following:

• References to the maintenance and calibration procedures for the processing equipment.

• Specifications for the EtO chamber.

• References to the commissioning data.

• An indication that all gauges, monitoring devices, etc., were calibrated prior to initiation of and at the conclusion of the validation.

• The validation protocol.

• Comparative resistance test data and reference to the protocol.

• A complete description of the products used as test samples, including packaging.

• BI certification and all data, according to current standards, required in the labeling.

• The physical and biological records from all of the validation cycles.

• Placement diagrams and data for the monitoring devices (including temperature, gas concentration, and relative humidity).

• Documentation on and verification of the positioning and location where the BIs and product test samples were placed on the product load.

• BI population determination test results and statement of retention samples.

• All biological test data associated with the validation.

• EtO residual test data.

• References to the test procedures used during the validation.

• Documentation of operating procedures, including process control limits.

• A test report summarizing and analyzing the validation data.

ANNUAL REVALIDATION

An annual revalidation of the process should be conducted to verify the integrity of the original validation data. It may be advisable to conduct this revalidation within 13 months of the initial validation and each year thereafter until sufficient data are obtained to allow that time period to be extended. Afterwards, revalidation should be conducted at least every 2 years.

Data from each revalidation should be compared with that from the original validation and any subsequent revalidation to confirm that the original performance specification remains valid. A revalidation consists of a thorough review of all the factors that could affect the sterilization process, including changes to the product or its packaging, the manufacturing methods and facility, and component and materials suppliers. Product bioburden test data should be analyzed and trends noted to ensure that there have been no substantial unacceptable changes in quantity or characterization. The testing conducted of the preconditioning room, the sterilization chamber and ancillary equipment, and the aeration room and equipment should be compared with the data obtained during the original validation as well as to the commissioning data. All associated programs, such as preventive maintenance and calibration, should be reviewed for effectiveness.

CONCLUSION

A well-designed and scientifically sound MPQ helps ensure that the process is safe, efficacious, and efficient. Performed well, the MPQ should deliver the required SAL for the product in an economical process and should enable the process to continue without having to be repeated. It should also prevent product reprocessing and delays in product release to the marketplace. A knowledgeable approach that considers all of the various aspects of the MPQ will ensure the success of the process.

ACKNOWLEDGMENT

The authors would like to express their gratitude to Aubrey S. Outshoorn, PhD, of the United States Pharmacopeial Convention Inc. for his contributions to this article.

REFERENCES

1. PJ Sordellini and SE Satter, "EtO Sterilization: Principles of Process Design," Medical Device & Diagnostic Industry 20, no. 12 (1998): 47–59.

2. "Sterilization of Medical Devices—Microbiological Methods—Part 1: Estimation of Population of Microorganisms on Products," ANSI/AAMI/ISO Standard 11737-1 (Arlington, VA: Association for the Advancement of Medical Instrumentation, 1995).

3. "Guidance on the Application of EN 29001 and EN 46001 and of EN 29002 and EN 46002 for Non-Active Medical Devices," BS/EN 724 (London: British Standards Institution, 1995).

4. "Medical Devices—Validation and Routine Control of Ethylene Oxide Sterilization," ANSI/AAMI/ISO Standard 11135 (Arlington, VA: Association for the Advancement of Medical Instrumentation, 1994).

5. EW Nester et al., Microbiology, 2nd ed. (Chicago: Holt, Rinehart, and Winston, 1978).

6. The Center for Devices and Radiological Health, FDA Guide for Validation of Biological Indicator Incubation Time (Rockville, MD, FDA, 1985).

7. United States Pharmacopeia 23, 8th supp., monograph <71>, 1998.

8. "Sterilization of Health Care Products—Biological Indicators—Part 1: General Requirements," ANSI/AAMI/ISO Standard 11138-1 (Arlington, VA: Association for the Advancement of Medical Instrumentation, 1994).

9. "Sterilization of Health Care Products—Biological Indicators—Part 2: Biological Indicators for Ethylene Oxide Sterilization," ANSI/AAMI/ ISO Standard 11138-2 (Arlington, VA: Association for the Advancement of Medical Instrumentation, 1994).

10. "Contract Sterilization for Ethylene Oxide," AAMI TIR no. 14-1997 (Arlington, VA: Association for the Advancement of Medical Instrumentation, 1997).

11. IJ Pflug and RG Holcomb, "Principles of Thermal Destruction of Micro-Organisms," in Disinfection, Sterilization and Preservation, ed. SS Block, 3rd ed. (Philadelphia: Lea and Febiger, 1983).

12. GS Graham and CA Boris, "Chemical and Biological Indicators," in Sterilization Technology: A Practical Guide for Manufacturers and Users of Health Care Products (New York: Van Nostrand Reinhold, 1993).

13. H Shintani et al., "Comparison of D₁₀-Value Accuracy by the Limited Spearman-Karber Procedure (LSKP), the Stumbo-Murphy-Cochran Procedure (SMCP), and the Survival-Curve Method (EN)," Biomedical Instrumentation Technology 29, no. 2 (1995): 113–124.

Susan Edel Satter of Satter & Associates Inc. (Boulder, CO) and Paul J. Sordellini of Quality Solutions Inc. (Annandale, NJ) are consultants for the medical device manufacturing industry.

Illustration by Brad Hamann