Principles of Parametric Release: Emphasis on Data Collection and Interpretation

Abstract

Parametric release, which relies on use of process data for product release, provides many benefits. However, adoption by the sterilization industry has been slow, with release typically involving biological indicator (BI) growth responses/ dosimetry readings. The current article highlights how the data provided by the process (described through examples for ethylene oxide [EO], vaporized hydrogen peroxide [VHP], and radiation) may be better used to inform parametric release implementation. The examples involving EO and VHP demonstrated the ability of the sterilization equipment to deliver validated parameters repeatedly after the load presented was validated. For instances in which load variability has not been addressed in performance qualification, BI testing or even measurement of EO concentration cannot reliably or fully inform the impact of such variance on the validated process. “Direct” monitoring of EO concentration is a current requirement in ISO 11135:2014. Nonetheless, the findings presented here show that EO and VHP concentrations can be determined by the calculated method, rendering the use of a concentration measurement probe somewhat superfluous. In alignment with European Union good manufacturing practice Annex 17, a key requirement of parametric release is to have sufficient data to demonstrate the repeatability of the validated process. Similar to gas technologies, radiation processing strives to implement parametric release but is limited by the currently available means of measuring all critical parameters, such as photon delivery.

The release of products to the market based only on parametric data is advantageous in many ways. It eliminates the time, risks, and costs associated with biological indicator (BI) sterility testing or dosimetry analysis, reduces the amount of unreleased inventory, and allows for continuous demonstration of process control. According to ISO 11139:2018, parametric release is defined as “a declaration that product is sterile based on records demonstrating that the sterilization process variables were delivered within specified tolerances."¹ Requirements of ISO 11135:2014 for product release from sterilization include documenting "the criteria for designating conformance of the sterilization process used for a particular sterilization load," including "confirmation that the data recorded during routine processing meet the sterilization process specification.”²

As highlighted by Sadowski and Langille,³ such definitions “continue to focus primarily on the control and achievement of critical process controls for the sterilization process to support sterile product release.” That is, the specification, including process and cycle variables, are established during performance qualification (PQ), and the requirement for release (according to ISO 11135) and definition of parametric release (ISO 11139) both indicate a need to confirm delivery of variables to established tolerances validated during PQ.

A crucial element for achieving the sterility assurance level (SAL) is ensuring that a routine process is a replica of the validated process. Product definition is a requirement of sterilization standards. ISO 11135, section 7, describes product definition as a validation input to PQ and a requirement for ongoing monitoring and control during routine processing.² Thus, if the requirements of the standard are to be fulfilled during routine processing, product loading configuration is assumed to be validated. That is, ISO/TS 21387:2020 allows for mixed loads and different load configurations, but these variations should be addressed in the PQ.⁴ Clause 9.4.1.5 of ISO/TS 21387 states that a critical relationship among the product, packaging, load density, and configuration, relative to conditions in the sterilizer, should be established during PQ and shown to be reproducible. These relationship data will be used to create the load configuration parameter for routine production cycle and parametric release. Moreover, procedures should be established to ensure that each load meets the defined load configuration parameters.⁴ The importance of maintaining load consistency (i.e., evaluating load variables as part of PQ) to ensure that product SAL is not compromised is reiterated in clause 9.4.1.6 of ISO/TS 21387.

To release products parametrically, the requirement is to provide sufficient process data to demonstrate the reproducibility of the sterilization process compared with that used in the original PQ. Demonstration of reproducibility involves determining those parameters and process outputs critical to the delivery of safe and sterile products. Further, measuring and controlling parameters at a sufficient level and frequency provides confidence that a process is in a state of control.

The current view of parametric release is one of redundancy and replacement. For example, external process challenge devices (EPCDs) are made redundant in ethylene oxide (EO) and replaced with a "direct" measurement of EO and relative humidity (RH). Conceptually, dosimetry (output measurement) is made redundant in radiation processing and replaced with critical input parameters (Table 1). As Shintani⁵ described, the replacement strategy is sometimes considered "risky" and requires more stringent validation and control. Similarly, Mittendorfer and Gallnböck-Wagner⁶ highlighted that in radiation processing, dose monitoring by dosimeters provides information on whether a specification is met but not necessarily whether a process is in a state of control.

Table 1.

Critical parameters in an X-ray process.

Whether related to dosimeters in radiation or BIs in EO processing, these means of routine monitoring already are a replacement for a test for sterility, which continues to be used in pharmaceutical manufacturing. Sadowski and Langille³ discussed the testing and detection challenges when using a pharmacopeial test for sterility to release products as sterile. For similar reasons, the use of BIs in EO or reference dosimeters in radiation should never be relied on as a sole means of product release. This is because BIs and dosimeters simply are additional data helping to confirm the repeatability of a validated process. If either is removed, assessing the risk and ensuring sufficient data to demonstrate the delivery of a validated process is incumbent.

Lead indicators (e.g., pressure increment of EO) provide a means of measuring a primary machine input to ensure its operation as per validation. Product load inputs are to be controlled in accordance with ISO 11137:2014, section 7,⁷ and lack of such control should not rely on lag indicators because these only are valid based on a sterilant distribution relationship established during validation with a controlled load (regardless of whether its parametric). As highlighted by the requirements of European Union good manufacturing practice Annex 17 (hereafter referred to as Annex 17),⁸ which align with the requirements of ISO 11135,² demonstration of capability and control thereafter is a factor of many things: the quality management system (QMS), parameter risk assessment, process qualification and parametric data generation, and review for release.

Considering that the parametric release approach offers considerable benefits but has not been widely used in the industry, this article provides a framework for understanding the principles of parametric release. Three examples are described in this article:

1. An example involving VHP is used to demonstrate which data are of importance in determining the microbicidal efficacy of a process.

2. A radiation example is used to demonstrate an abundance of data with accelerator-based processes but that key data are needed to fully eliminate dosimeters from routine processing.

3. An EO example is presented in which sufficient data demonstrate the repeatability of a validated process. Therefore, an additional "direct" measurement of EO concentration may not be required.

The implementation of parametric release may require a review of current thinking, approaches, and normative requirements of process standards.

Critical Parameters for Parametric Release

The approach of conducting a process release using parametric data may be subdivided as follows. Examples also are provided.

Parameters Affecting Microbicidal Efficacy: VHP Example

By identifying essential prerequisites (e.g., equipment qualification, demonstration that required conditions for sterility are achievable) and using routine monitoring of critical parameters, the required conditions can be confirmed. One critical parameter that affects the microbicidal efficacy of a process directly is the quantity of the sterilant input. In the example of vaporized hydrogen peroxide (VHP), the sterilant input (concentration) is controlled through RH.⁹ Other parameters essential for achieving sterility include pressure and temperature. The sterilizer controls and reports the pressure using two probes, and a separate probe measures chamber humidity and temperature. These data, combined with a secondary confirmation that the injection is the sterilant (e.g., chemical indicator) could be used for parametric release.

Despite not being directly measured, the concentration of the sterilizing agent can be calculated using the ideal gas law, as previously shown by McEvoy et al.⁹ Those authors have used the calculated VHP concentration to generate the survival curve when investigating the inactivation kinetics of Geobacillus stearothermophilus spores treated with an industrial-scale VHP process. Initially, the surviving spore fraction was plotted against the number of VHP pulses and a log-linear inactivation was demonstrated (R² = 0.91). According to ISO 11138-7:2019,¹⁰ the inactivation follows first-order log-linear kinetics when the value for the coefficient of determination (R²) is no less than 0.8. However, because the amount of injected sterilant was not consistent among all pulses, the surviving spore fraction was plotted against the calculated concentration (assuming 100% sterilizing agent) and an improved R² value was obtained (R² = 0.98).

The improved R² value emphasizes the direct relationship of the calculated VHP concentration to VHP inactivation. Moreover, this example demonstrates the possibility and validity of using the calculated method to express the sterilant concentration. The survivor curve can be replotted using the actual sterilizing agent concentration of 35% (Figure 1, left), with the R² value remaining unchanged (R² = 0.98). Pressure is also included Figure 1 to demonstrate the relationship between pressure and concentration. The inactivation curve can also be replotted using humidity (Figure 1, right; R² = 0.98). Therefore, Figure 1 illustrates a relationship of the three parameters (sterilant concentration, sterilant injection pressure increase in the VHP sterilizing chamber, and humidity) to inactivation. In the chamber used in that study, RH was a controlling parameter and pressure a monitoring parameter (which can be reported as pressure or converted to concentration), and any of the three parameters could be used in a parametric release operation. However, considering that sterilant injection is a defined, programmable, and controllable parameter, it is more appropriate for validation as a parametric release parameter, with sterilizing agent concentration as a reportable output from the process.

Figure 1.

Left: Spore surviving fraction (log₁₀ N/N₀) plotted versus H₂O₂ concentration in vapor and sterilant injection pressure increment. Right: Spore surviving fraction (log₁₀ N/N_o) plotted versus H₂O₂ concentration in vapor and relative humidity (RH) increment. N_o represents untreated spores, and N represents the surviving fraction of vaporized hydrogen peroxide (VHP)-treated spores. The results are shown as a mean value of three independent runs. The inactivation follows first-order log-linear kinetics when the value for the coefficient of determination (R²) is no less than 0.8 (per ISO 11138-7:2019¹⁰).

Radiation: Current Gap in Critical Parameters

As specified above, identifying critical parameters is essential in a number of ways: (1) compliance with section 5 of many standards (e.g., ISO 11135.² ISO 22441:2022¹¹), (2) validation of a process and demonstration of achievement of SAL, and (3) measurement for sterile release. Although efforts have been made to address the feasibility of using parametric release in radiation sterilization,¹² the industry has been reluctant to eliminate dosimetry measurements from routine processing. As part of the parametric release implementation, the critical parameters (direct) need to be identified and differentiated from the contributing parameters (indirect) and a risk assessment should be conducted.³

Using X-ray as an example, a risk assessment was conducted. Critical parameters were identified (Table 1). Temperature, time, and energy were identified as contributing parameters. The next step was evaluating whether a means of measurement existed and if redundancy was present in measurement and data capture so that the parameter output could be used for product release. Current-day control systems in accelerator-based radiation processing directly monitor average current and conveyor speed at a frequency much higher than the routine dosimeter monitoring method. Radiation field characteristics (e.g., beam width and length) are more challenging to monitor directly, but the parameters that will affect the radiation field (e.g., scanning magnet current, radiation levels, and vacuum) may also be monitored.

For X-ray technology, an additional challenge is monitoring the electron-to-photon conversion and, most importantly, the mechanical characteristics of the conversion device. However, work is ongoing to develop photon flux measurement devices that may confirm the continuous delivery of parameters within defined specifications (unpublished observations from 2022 Kilmer conference) Thus, a future state of parametric release for X-ray will be a combination of accelerator and conveyor inputs and intermediate measurements, combined with direct and continuous measurement of the photon field.

Statistical Process Control and Reliability in Measurement

The VHP sterilizer used in the work of McEvoy et al.⁹ monitored parameters, including temperature, RH, and pressure, that may be used to demonstrate the repeatability of the validated process. Statistical process control (SPC) tools may be used to create control limits for such critical parameters. As identified by Mittendorfer and Gallnböck-Wagner,⁶ measurement of such parameters allows the definition of measurement intervals required for statistical analysis. These authors identified how dosimetry measurement in radiation processing is not ideal for routine SPC analysis, as it depends on processing batch size.

Measurement is only one aspect of the monitoring and control strategy required for safe product release. To learn from the pharmaceutical industry, a real-time release strategy should encompass the following: “The combined process measurements (process parameters and material attributes) and any other test data generated during the manufacturing process should provide a robust foundation for Real Time Release Testing and the batch release decision.”⁸

In a similar manner, Sharnez et al.¹³ subdivided the elements of parametric release of a cleaning program as ‘Characterization’, ‘Validation’ and ‘Monitoring’ and ‘Support Systems’ of Change Control, Managing Nonconformance, Training, Knowledge Management, Continual Improvement and Preventive Maintenance. Annex 17 provides guidance and identifies key criteria required for parametric release of which many are addressed in the requirements of ISO 11135 (Table 2).

Table 2.

Requirements of European Union good manufacturing practice Annex 17⁸ and related sections in ISO 11135:2014.²

As shown in Table 2, fulfilling the requirements of ISO 11135 satisfies many requirements for an Annex 17 approach to real-time release. The elimination of one measurement parameter, namely BI growth from a limited, statistically questionable sample sizing, necessitates simply greater control of those parameters deemed critical to sterility. An example involving EO is described in the following section.

Data Needed to Inform Parametric Release

EO Example

When switching from BI release to parametric release, current thinking with EO processing focuses on measuring EO concentration and RH. EO concentration typically is measured indirectly using equipment that converts an infrared signal to an EO concentration or by simply quantifying the input of sterilant. Similarly, RH may be quantified using data from the sterilizer. Parametric release may be considered an improvement, as EPCDs offer a very limited sample size that should only be used as an additional data set for release.

On the other hand, parametric release uses continuously available parameters that may be controlled, measured, validated, and specified. In this work, the data collected from 90 routine cycles were analyzed to compare different means of quantifying the delivery of variables to validated tolerances (e.g., EO gas analyzer probe versus EO concentration calculated by pressure increment). All cycles were identical in terms of cycle specifications and load configuration, and no growth of BIs was detected.

As shown in Figure 2, EO concentration was quantified and monitored using an EO gas analyzer (Sensor Electronics, Minneapolis, MN), in which EO absorbs infrared light at specific wavelengths due to molecular resonance. In addition, EO concentration was measured indirectly by using pressure differential and the following equation:

Figure 2.

Ethylene oxide (EO) concentration for each of the 90 routine cycles, measured with an EO gas analyzer probe and calculated by pressure differential. Validated EO concentration tolerances = 180–580 mg/L. EO concentration range detected with EO gas analyzer probe = 319. 5–363 mg/L; pressure differential = 392–409 mg/L. The dashed lines represent minimal and maximal values detected using EO gas analyzer probe.

$$\text{Concentration}=\frac{K\times P}{R\times T}$$

where K is constant for 100% EO (4.4 × 10⁴), P is pressure increment (mbar), R is gas constant (0.08314), and T is chamber temperature during exposure (°K).

Validated EO concentration tolerance is in a broad range (between 180 and 580 mg/L). The average EO concentration in routine cycles (n = 90) measured with EO gas analyzer probe was 347 mg/L, while the average concentration calculated based on pressure was 402 mg/L. The calculation based on pressure was consistently higher, which potentially is a matter of the location of measurement devices within the chamber (Figure 2). When determining the EO concentration by EO gas analyzer probe, the continuous readings can detect EO oscillations in the chamber. However, when concentration is determined by pressure differential, a single result is obtained: EO concentration in the chamber at the end of the EO injection. Nevertheless, both methods (EO gas analyzer probe and calculated method) only provide a reading at predetermined locations and are not necessarily representative of all areas within the chamber.

When a parametric release is used, ISO 11135 requires that value and tolerances for the EO concentration be "determined from direct analysis of chamber atmosphere using analytical methods to establish the process specification for routine processing."² However, it could be argued that (1) perceived "direct" analysis is in itself a conversion of an infrared signal to a stated concentration and, (2) supported by the data described here, calculating EO concentration (combined with chemical indicators as a confirmation that sterilant was injected) should also be acceptable and considered in future revisions of ISO 11135.

With regard to RH measurement, the validated tolerances are 30% to 100%. However, the average RH range detected in the 90 routine cycles was much narrower, spanning from 60.75% to 72.80% (Figure 3). When minimum and maximum values are compared, the minimal RH detected in a cycle was 53.1% and maximal was 83.0%. The theoretical added RH was relatively constant (between 36.8% and 42.3%). Another critical parameter (achieved exposure chamber temperature) was also consistent between the cycles and well within the validated tolerances (42–50°C), with a lower average temperature of 44.75°C and higher average temperature of 46.95°C. The minimal temperature detected in a cycle was 43.5°C and maximum was 47.4°C.

Figure 3.

Relative humidity (RH) and theoretical RH obtained from the routine cycles (n = 90). The results are shown as average values for each individual cycle, calculated from minimum and maximum achieved value. Validated humidity tolerances = 30–100%; detected RH range = 53.1–83%. Theoretical RH was between 36.8% and 42.3%.

In this example, Figure 2 shows that multiple data sources may inform a parametric release process. Although sources may give slightly different results by nature of how, where, and when they are measured, the most important aspect is that they are reliable and sufficiently sensitive to confirm the reproducibility of the validated process. The advantages of using input variables such as sterilant and humidity injection measured by pressure increment are that they are (1) measurable as an output for control and monitoring but also (2) programmable for validation purposes and process challenges during validation. However, similar to BIs, EO concentration and RH results from routine processing should not be viewed in isolation; instead, they must be assessed against all specifications to confirm the repeatability of the validated process.

Conclusion

In EO, parametric data can come from many sources: temperature and pressure transducers, EO gas analyzer probe concentration monitors, and humidity sensors. As described in this article, such data generated by gas processes (EO and VHP) can be used to inform parametric release (e.g., that sterilant concentration can be derived by calculation). Therefore, when performing the risk assessment, it is possible to conclude that the measurement of an EO injection and an EO gas analyzer probe measurement provide relative data from different measurement devices and by different means. However, either measurement equally contributes to demonstrating the repeatability of a validated process.

Maintaining product load consistency, as required by ISO 11135, is important. Also, to achieve the SAL, it is important to ensure that routine process is a replica of the validated process. Based on the example presented in this article, in can be concluded that EO sterilization processes provide such an abundance of data that risk assessments indicate sufficient data sets for confirming repeatability. A similar review of X-ray radiation processes identifies the critical parameters but also confirms the need for additional measurement devices to potentially eliminate the need for routine dosimetry.

To that end, X-ray equipment manufacturers are developing additional measurement devices (e.g., photon flux detectors) (unpublished observations from 2022 Kilmer conference). Ultimately, by identifying critical parameters, measuring and collecting output data, and using these data in combination with an all-encompassing QMS approach, a process that has been confirmed using real-time parametric data can be released. With the correct processes in place, release of products to the market based only on parametric data would benefit manufacturers, sterilization providers, and patients.