IVT Cleaning Validation IV 0 [PDF]

  • Author / Uploaded
  • Noemi
  • 0 0 0
  • Gefällt Ihnen dieses papier und der download? Sie können Ihre eigene PDF-Datei in wenigen Minuten kostenlos online veröffentlichen! Anmelden
Datei wird geladen, bitte warten...
Zitiervorschau

SPECIAL EDITION

CLEANING VALIDATION Published by:

CLEANING VALIDATIoN Validation Case Study: Erroneous Negative Cleaning Validation Results | IVT ...................................................... 1 Experimental Parameters for Small-scale Cleaning Characterization Part I: Dilution of Process Fluids During Cleaning | IVT ....................................................................................... 7 Experimental Parameters for Small-Scale Cleaning Characterization. Part II: Effect of Fluid Velocity on the Kinetics of Cleaning | IVT ........................................................................ 11 Methodology for Assessing Product Inactivation during Cleaning Part I: Experimental Approach and Analytical Methods | IVT ............................................................................. 16 Methodology for Assessing Product Inactivation During Cleaning Part II: Setting Acceptance Limits of Biopharmaceutical Product Carryover for Equipment Cleaning | IVT ........... 20 Aseptic Transfer Risk Assessment: A Case Study | IVT ...................................................................................... 27 People in Cleanrooms: Understanding and Monitoring the Personnel Factor | IVT .............................................. 32 PAT: Using PAT to Support the Transition from Cleaning Process Validation to Continued Cleaning Process Verification | IVT ............................................................................................................................................. 39 Translating Laboratory-Developed Visual Residue Limits to Process Area Applications | IVT .............................. 45 New Perspectives on Cleaning: Cleaning Validation of Multiproduct Equipment—Acceptance Limits for Inactivated Product | IVT ................................................................................................................................................... 51 Biopharmaceutical Cleaning Validation: Acceptance Limits for Inactivated Product Based on Gelatin as a Reference Impurity | IVT .................................................................................................................................. 60 Multiproduct Cleaning Validation: Acceptance Limits for the Carryover of Inactivated API Part I–The Comparable Quality Approach ............................................................................................................................................. 67 Validation of a Cleaning Process for Medical Devices | IVT ............................................................................... 72 Ensuring Sterility: Autoclaves, Wet Loads, and Sterility Failures | IVT ............................................................... 79

©2015 Advanstar Communications Inc. All rights reserved. Reproduction in whole or part is prohibited without prior written permission of the publisher.

CLVQ4005

Paul L. Pluta

Validation Case Study: Erroneous Negative Cleaning Validation Results | IVT Paul L. Pluta, Ph.D.

“Validation Case Studies” provides a forum for validation practitioners to share information about actual validation experiences. Previous discussions addressed a wide range of activities. Previous case study titles discussed in this series include the following:

1. Visual Observations, Journal of Validation technology ( JVT), Volume 16, #1. 2. Equipment Qualification, JVT, Volume 16, #1. 3. Identical mixing Tanks, JVT, Volume 16, #3. 4. Cleaning HPLC Peaks, JVT, Volume 16, #4. 5. Documentation Practices, JVT, Volume 17, #1. 6. Yield, JVT, Volume 17, #2. 7. Like-for-Like Changes, JVT, Volume 17, #2. Readers are invited to participate and contribute manuscripts for this series -- we encourage sharing successful practices with others. Please contact journal editor-in-chief Paul Pluta at [email protected] or content specialist Dustin Henderson at [email protected] with comments or submissions for publication. ABSTRACT This case study describes a cleaning validation event in which failing results for API residue from a small molecule extended release tablet dosage form were observed. The initial two lots in the cleaning validation were successful. The third lot failed acceptable residue limits. Investigation of the failure comprised cleaning process development and performance; residue sampling, sample handling, sample analysis, and evaluation of the analytical method. Investigation of this event initially involved interviews of relevant personnel and reviews of associated documentation. Two areas were identified for further evaluation – residue sampling and the cleaning process. Regarding sampling, a newly trained technician, working alone, sampled the first two acceptable lots, while an experienced technician working with a colleague sampled the third failing lot. Evaporation of sampling solvent occurred causing residue to be insufficiently recovered from the equipment surface resulting in erroneous false negative test results. Regarding the cleaning process, manufacturing operators commented that the new extended release formulation was more difficult to clean than the original immediate release formulation although the same cleaning procedure was utilized for both products. Evaluation of the cleaning process indicated that the process parameters were not optimal to clean the new extended release product. An improved cleaning process with increased cleaning agent concentration, increased cleaning time, and higher temperatures was developed, implemented, and ultimately validated. Special edition: Cleaning Validation

1

Paul L. Pluta

Cleaning validation sampling personnel must have good technical understanding of their work, and must know the technical reasons for the procedures they perform, and potential problems if procedures are not correctly performed. Sampling personnel training for cleaning validation should include a quantitative demonstration of acceptable cleaning by means of analytical testing. Training exercises must also include worst-case sampling such as with volatile solvents, multiple equipment, and other potential variations in sampling. SOPs must be carefully written to describe potential problems and include performance constraints to minimize variation and risks. There is an inherent danger when variation is not deliberately introduced into a validation project – material variation, manufacturing operator variation, and in this case study, sampling personnel variation. Sampling by two different technicians enabled erroneous results to be discovered. Regarding the cleaning process, inactive ingredients in a formulation may have very significant effects on cleaning processes. Cleaning of residues does not depend solely on the properties of the API. INTRODUCTION The compliance event involved cleaning validation for cleaning of residue from a new small molecule extended release tablet dosage form. The active ingredient in the tablet was a potent drug – dosage of the active ingredient was low and cleaning was expected to be successful. The site already had a successful history of cleaning a marketed immediate-release tablet containing a lower dose of the same active drug. No changes in the cleaning method were required for the new product. After cleaning, the target residue level was below “visually clean” and required the residue level to be determined by swab sampling and chemical analysis. Three lots were required for cleaning validation. The following are discussed: • Compliance event. A description of the cleaning validation event • Investigation. Interviews and actions conducted to investigate the event • Discussion. Key information, activities, and analysis • CAPA. Actions and improvements implemented in the cleaning process, sampling process for cleaning validation, and training of sampling personnel • Cleaning validation of modified cleaning process. Implementation of the new cleaning process and subsequent validation. 2 Special edition: Cleaning Validation

BACKGROUND The process of validation typically comprise the following sequence of activities: 1. Change desired. An equipment or process change is needed or required. This may be a necessary change or a desirable improvement. 2. Development work. Appropriate Stage 1 development work is completed in support of the change. 3. Validation plan. A formal request to initiate the validation process is submitted to the validation approval committee (VAC). Development reports may be included in the request in support of the change. The change request includes a proposed level of work to confirm the acceptability of the change. The level of work is based on risk to the patient and to the organization. The VAC approves the change request. The approved change request document is stored in the validation library. 4. Protocol. A protocol is written specifying detailed sampling and testing to confirm the acceptability of the change. The VAC approves the protocol. The approved protocol is stored in the validation library. 5. Validation work. Stage 2 PPQ validation work is performed according to the protocol. Sampling and testing are completed. Data and other results are generated and recorded. 6. Validation report. A report containing all test results with discussion and conclusions is prepared and submitted to the VAC for approval. The report is approved and the process or equipment change is implemented. The approved validation report is stored in the validation library. 7. Validation closure. If no other work is needed, the validation project initiated by the change request is closed. 8. Continued verification. Stage 3 post-validation monitoring confirming acceptability of the change continues throughout the product / process lifecycle. The issue addressed in this case study occurs in #5 and #6 above. The actual work conducted to confirm the acceptability of the validation project is performed by technical people. In this case study, samples were removed and tested for residue content. VALIDATION EVENT A small molecule pharmaceutical company conducted initial cleaning validation on a new extended release tablet product containing a water-insoluble API as active ingredient. The new product was a line exten-

Paul L. Pluta

sion -- an extended release formulation of a marketed immediate-release tablet product. The new product contained a polymeric matrix to enable prolonged release and once-daily dosage to patients. The cleaning validation exercise was expected to be successful. Although the product contained a highly potent active drug which required low residue levels on cleaned equipment, the company had extensive experience with the cleaning procedure over several years. The original immediate-release product cleaning was relatively easy and had a long successful history of performance. Several previous cleaning validations had been successfully accomplished. The analytical method for residual API from swab samples was easily performed and very reliable. Sampling of three lots of new product was planned for cleaning validation. The manufacturing process comprised several unit operations. Sampling of unit operations for cleaning validation was performed on multiple days for each lot. The first lot was manufactured and cleaning completed on all equipment. Equipment was visually clean. Swab sampling was done by the sampling technician. Cleaning validation analytical test data indicated no active drug present in all swab samples – all acceptable results. A second lot was manufactured. Cleaning was completed. Swab sampling was done. Cleaning validation analytical test data again indicated no levels of residual drug in all swab samples. A third lot was manufactured. Cleaning was completed. Swab sampling was done. Cleaning validation analytical test data indicated extremely high residue levels significantly above the required acceptance criteria. Test results on the third lot indicated a significant failure of the cleaning process. This event prompted multiple questions to be investigated and answered. 1. Cleaning process performance. Did manufacturing personnel correctly perform the cleaning process in the third (failing) lot? Which operator cleaned the equipment? Were manufacturing personnel adequately trained in the cleaning procedure? What was past history with use of this cleaning process? Were repeat cleanings required in past cleaning? Were deviations required? 2. Cleaning process development. How were the cleaning process parameters developed? What was the history of this method with the immediate release product? We any changes made for cleaning the extended release product? 3. Sampling. Did the sampling technician correctly sample the recommended equipment surfaces? Were sampling personnel adequately trained?

4. Residue samples. Was the integrity of residue samples adequately protected during transport to the lab? Could samples have become contaminated causing the test failures? Were samples correctly and quickly transferred to the lab for analysis? Were samples handled during transport and storage according to procedures? Were samples exposed to high temperatures during transport and storage? 5. Analytical laboratory. Was the analysis correctly performed? Which technician performed the analysis? Were laboratory technicians adequately trained? Was analytical equipment qualified for use for the API analysis? Was system suitability below required limits? 6. Analytical R&D. Were there any problems with the analytical method? Was the analytical method correctly developed? Was the analytical method validated? INVESTIGATION Investigation of this compliance event initially involved interviews of relevant personnel and reviews of associated documentation. Personnel related to the compliance event included manufacturing personnel, QC personnel, cleaning sampling technicians, and the technical personnel responsible for product formulation and process, cleaning method development, technical support, and analytical testing. There were many details and variables that needed to be investigated and/or confirmed. Personnel from all groups interacted to address the above issues. Documentation reviews included manufacturing documentation, cleaning documentation, equipment inspection records, laboratory records, analytical method development reports, validation reports, and other records. All applicable manufacturing SOP’s and analytical SOPs were reviewed. Cleaning Process Performance Manufacturing personnel correctly performed the cleaning process in all three lots. Cleaning procedure documentation for all lots was reviewed and found to be perfectly executed. No deviations were issued. Different operators executed cleaning of multiple equipment in the validation lots. An experienced operator executed cleaning in the third (failing) lot. Training records for all operators were reviewed and found to be acceptable. Operators commented that the new extended release product was more difficult to clean than the original immediate release product. The extended release polymer made removal of the product residue more difficult that was typical with the original immediate-release product. This observation reflected operator experience with manual cleaning of small parts. Product was able to be removed from Special edition: Cleaning Validation

3

Paul L. Pluta

equipment surfaces and yield visually-clean surfaces. No repeat cleanings were required. No deviations were issued. Cleaning Process Development The cleaning process for the immediate release product had been previously developed. An alkaline cleaning agent that was used on several other products in the plant was used. Because the API in the new extended release product was the same as in the immediate release product, no changes were made to the cleaning method. Technical personnel were unaware of any difficulty in cleaning the extended release product. Sampling Two different sampling technicians performed sampling in the three lots. A newly-trained technician sampled lots #1 and #2, both of which had acceptable low residue levels. The newly-trained technician worked alone to accomplish sampling since no other sampling technicians were available. Lot #3 was sampled by an experienced technician. The experienced technician worded with a colleague who helped sample the recommended equipment surfaces and complete required documentation. Both sampling personnel were adequately trained as evidenced by training documentation. Residue samples Residue samples were packaged in protective wrapping for transfer to the lab. Samples were immediately closed and not contaminated. Transport to the lab was rapid and without exposure to unusual environmental conditions or excessive heat. Analytical laboratory The laboratory analysis was correctly performed. Experienced technicians performed the analysis. All technicians were adequately trained. Analytical equipment was qualified for use for the API analysis. Lab documentation indicated acceptable execution of the analytical procedure. Analytical R&D The same analytical method was used for the original irradiate-release product and for the new extendedrelease product. Analytical R&D verified that the test method performed acceptably for the new product. There were no problems with the analytical method. The analytical method was validated. DISCUSSION Interviews and discussion of the above questions did not clearly indicate an obvious cause for the problem. Manufacturing personnel confirmed that they 4 Special edition: Cleaning Validation

performed cleaning as required by procedure. Equipment was cleaned by automated methods wherever possible. All associated small parts were manually cleaned was cleaned according to procedure. The manufacturing supervisor verified that procedures were followed and that the equipment was visually clean. Quality unit personnel who inspected the equipment also verified that all equipment and small parts were visually clean. All inspections were conducted after the equipment was dry. Samples were transported to the lab quickly and according to procedure. Samples were also quickly stored in the laboratory upon receipt and under specified security conditions. Laboratory personnel confirmed acceptable performance of analytical procedures. Analytical standards over a range of concentrations tested along with the actual cleaning validation samples yielded accurate results. Analytical R&D scientists confirmed acceptable performance of the validated test method. Two areas were identified for further investigation. These included: 1. Swab sampling. A newly trained technician, working alone, sampled the first two acceptable lots. All samples in these lots were acceptable. An experienced technician, working with a colleague, sampled the third failing. Was something different about the third lot, or was the failing data due to the difference in sampling personnel? 2. Cleaning process. Manufacturing operators commented that the new extended release formulation was more difficult to clean than the original immediate release formulation. The same cleaning procedure was utilized for both products. Swab Samplingy Swab sampling for the three lots was done by two different sampling technicians. The first two lots were sampled by a newly-trained person. Data for these lots indicated minimal or no residual soil – acceptable results. The third lot with failing data was sampled by an experienced technician who worked with a colleague. The sampling method required wetting of the swab with organic solvent to dissolve residue from the equipment surface. The new technician did all sampling alone. The experienced technician performed sampling with a colleague to accomplish the sampling procedure in minimum time. She explained the necessity of the rapid sampling technique because evaporation of the sampling solvent must be minimized. The new technician was not aware of the time limitation in sampling. Although not conclusively proven, it was suspected that evaporation of solvent

Paul L. Pluta

occurred causing residue to not be adequately recovered from the equipment surface. The new technician worked slowly and carefully, and completed all necessary steps. However, the time required for performance, especially since she worked alone, may have caused residue recovery to be incomplete or minimal. The analytical lab confirmed that if sufficient solvent was not present on the swab, residue recovery would be unsuccessful. Cleaning Process Technical personnel responsible for the cleaning process had no previous experience with the cleaning method. The cleaning method for the original product had been established many years ago and never required new technical evaluation. Manufacturing management decided to use the well-established cleaning method without involvement of technical personnel. Management’s rationale was that since the API in the original product had been reliably cleaned for many years, there was no need to evaluate the cleaning process for the new product. Technical personnel had not been requested to evaluate the cleaning process used in the failed cleaning validation. In light of the cleaning failure, technical personnel recommended laboratory studies to evaluate available cleaning agents, cleaning process parameters, and related factors in a systematic way. Evaluation of the cleaning process indicated that the process parameters were insufficient to clean the new product. The polymeric matrix in the new product (methylcellulose mixture) was much more difficult to clean than the original immediate release product. Technical personnel conducted studies to establish new cleaning process parameters suitable for the extended release product. A new cleaning method with increased cleaning agent concentration, increased cleaning time, and higher temperatures was developed. CORRECTIVE ACTION / PREVENTIVE ACTION (CAPA) Two CAPA activities corrected the problems experienced in the original cleaning validation. These involved new training of swab sampling personnel and a modified cleaning process for the extended release product. Swab Sampling Training Personnel who perform cleaning residue sampling using swabs wetted with volatile solvents were taught the importance of rapidly performing swab sampling. Many of the swab sampling technicians did not have a technical background and did not understand solvent volatility and the consequences for swab sampling. Studies confirmed that the new technician, who worked alone in the sampling activity, did not perform swab sampling quickly. When sampling

was not performed quickly, solvent evaporated and surface residue was not able to be dissolved. Analytical results on evaporated swab samples indicated extremely low or no levels of residue which erroneously passed cleaning validation acceptance criteria – a false negative due to solvent evaporation. Future training of swab sampling technicians included new test procedures to require rapid performance of sampling procedures. The previous qualification test did not utilize a volatile solvent and did not require rapid performance. The new qualification test required technicians to demonstrate rapid sampling in order to become a qualified sampling technician. Sampling teams (two technicians) were required when volatile solvents were used in sampling. Technicians were required to quantitatively recover residue in training to be qualified for residue sampling. Training was repeated on an annual basis. SOPs describing cleaning sampling methods using volatile solvents were strengthened to require rapid sampling and working in teams. The combined emphasis of new training and new procedures that both underscored the risks and potential variation in residue sampling strongly addressed the issues described in this case study. Modified Cleaning Process for Extended Release product Technical personnel evaluated the cleaning process and determined that process parameters were not optimal to reliably clean process residues. The cleaning agent concentration was increased, the temperature was increased, and the cleaning time was increased in the new procedure. These parameters enhanced the cleaning process to more effectively and more efficiently remove the polymeric residue. CLEANING VALIDATION OF MODIFIED CLEANING PROCESS The new cleaning process was implemented. Manufacturing operators confirmed that new cleaning process parameters significantly improved the cleaning process. Three product lots were manufactured. Cleaning was performed on required equipment in three lots. Worst-case locations on equipment were swab sampled by two-person teams of sampling personnel. Two-person teams ensured minimal solvent evaporation and rapid sampling procedures. All test results passed the acceptance criteria. SUMMARY AND FINAL THOUGHTS A case study describing a compliance event in which erroneous false negative analytical data was generated in cleaning validation. These data caused a mistaken conclusion that a cleaning process for a new modified release dosage form was acceptable. The cause Special edition: Cleaning Validation

5

Paul L. Pluta

of the problem was not easily determined – all test data were acceptable. Initial investigation of potential problem areas indicated that everything was done according to procedure – nothing was done incorrectly. It was ultimately determined that the sampling process for product residue was not sufficiently controlled, and that the equipment cleaning process was not adequate for the modified release formulation. The sampling error, i.e., loss of solvent in sampling, had a major effect on cleaning validation. The sampling technician did not understand the importance of working quickly to minimize solvent loss. This lack of understanding resulted in a false negative test result and an erroneous conclusion that the cleaning process was acceptable. Fortunately the error was discovered when a different technician correctly and rapidly sampled the equipment surfaces. The combined emphasis of new training and new procedures that both emphasized the risks and potential variation of sampling strongly addressed the sampling issues described in this case study. Observations by manufacturing personnel caused the cleaning process to be evaluated by technical personnel, and a new cleaning process with optimized process parameters was developed. The new cleaning process was ultimately validated. Lessons Learned Several important lessons may be learned from this case study. • Sampling personnel understanding of sampling process and training. Sampling personnel must have good technical understanding of their work. They must know the reasons for

6 Special edition: Cleaning Validation

the procedures they perform. They must know potential problems if procedures are not correctly performed. • New procedures. SOPs describing cleaning sampling methods using volatile solvents were strengthened to require rapid sampling and working in teams. Time constraints were added to all affected procedures. SOPs must be carefully written to identify potential risks and minimize variation. • Sampling personnel training. Training of cleaning validation sampling technicians is a critical activity. Training exercises must include a quantitative demonstration of acceptable cleaning by means of analytical testing. Training exercises must also include worst-case sampling such as with volatile solvents, multiple sampling equipment, and other potential variations used in sampling. Retaining of technicians at some defined and reasonable frequency should be considered. • Inactive ingredients effects on cleaning. Inactive ingredients may have very significant effects on cleaning processes. Cleaning of residues does not depend solely on the properties of the API. Formulation ingredients may significantly affect the cleaning process. In this case study, an extended release polymer in the formulation caused difficulty in the cleaning process. Inactive ingredients such as dyes and flavors may also greatly influence cleaning, and may actually be the most difficult ingredients to clean in a formulation. All components in a formulation must be considered when developing a cleaning process.

Rizwan Sharnez

Experimental Parameters for Small-scale Cleaning Characterization Part I: Dilution of Process Fluids During Cleaning | IVT Rizwan Sharnez, Ph.d., Angela to, Arun tholudur, Ph.d.

CLEANING VALIdAtIoN Methodologies for estimating experimental parameters for small-scale cleaning characterization studies are described in this series: dilution of process fluids (e.g., process soil, cleaning solution, or rinse water) during cleaning is discussed in this paper; worst-case fluid velocity and soil load will be discussed in subsequent parts. Dilution of the process fluid during cleaning was estimated to be on the order of 1016 for a typical cleaning cycle and 105 for intermediate cleaning steps. These dilution factors are used to estimate the concentration of impurities in the final rinse and to simulate worst-case cleaning conditions for cleanability and inactivation studies.

INtRodUCtIoN Small-scale cleaning characterization (CLC) studies are used to identify suitable cleaning chemistries (1, 2), optimize cleaning parameters and processes (3, 4), establish cleaning times for manufacturing equipment (5, 6), and streamline validation requirements for multiproduct equipment (7, 8). Experimental models for small-scale CLC have been described in the literature (9-12). In these experiments, the kinetics of soil removal (mass transfer) from a surface is measured under simulated cleaning conditions. A critical step in the development of these models is to identify and scale down the hardest-to-clean (worst-case) location in the equipment (13). Note that it is not necessary to simulate the entire cleaning process; instead, it is only necessary to simulate the location within the equipment that is the hardest to clean. If the process soil can be adequately removed from the worst-case location, it follows that it can also be adequately removed from the other locations in the equipment. Further, with this approach, the cleaning times obtained at small-scale would be indicative of those at full scale assuming that there is adequate coverage at the surfaces that need to be cleaned. The scalability of small-scale CLC data depends on the accuracy with which the experimental parameters are estimated. Experimental parameters for simulating the worst-case location fall into two categories (14): • Parameters that can be readily determined from process data such as: o Material of construction and surface smoothness of coupons or parts used to simulate large-scale equipment o Post-soiling parameters such as hold time, temperature, and humidity o Cleaning parameters such as rinse or wash time, temperature of rinse solvent, and temperature and concentration of cleaning solution. • Parameters that typically need to be determined from first principles such as: o Dilution of the process fluid during cleaning Special edition: Cleaning Validation

7

Rizwan Sharnez

o Velocity of rinse solvent or cleaning solution at the worst-case location o Soil load at the worst-case location. Residual Process Fluid The volume of residual process fluid (VR) can be estimated from the surface area of the equipment that makes contact with the process fluid (SA) and the residual volume of the process fluid per unit surface area (R): VR = SA × R

[Equation 1a]

R can be determined experimentally by measuring the amount of process fluid that remains on a surface after it is drained. An experimental method for estimating R is described in the next section. VR for a CIP circuit can be estimated by dividing the equipment into the following sections: vessel walls (W); bottom dish of vessel (D); and associated piping, filter housings, and other miscellaneous parts (P). Thus, VR = SAw × Rw + SAD × RD + SAP × RP

[Equation 1b]

For a tank, the surface area of the sidewall that makes contact with the process fluid is SAw = πdh

worst-case process soil from the standpoint of drainage. Also, residual volumes obtained with curved surfaces (pipes) were significantly greater than that for flat surfaces (plates). Consequently, pipes were used to estimate the worst-case residual volumes.

[Equation 2]

Figure 1: Experimental Setup for Estimating Residual Process Fluid.

The experimentally determined residual volumes for the configurations that were tested are given in Table I. Note that impellers, filter housings, and other components associated with the vessel may contain surfaces at a range of angles, typically between 5° and 90°. The residual volume for these components may be set to the R values for the smaller angle of repose (5°) because it represents a worse case from the standpoint of drainage.

Where d is the diameter of the tank and h is the height up to which the process fluid wets the sidewall. The bottom dish surface area (SAD) is estimated from ASME engineering tables (15). The surface area of piping, filter housings, and other miscellaneous components (SAP) is estimated with a 15% overage factor. Thus, SAP = 0.15 (SAw + SAD)

[Equation 3]

Estimation of Residual Process Fluid The experimental method for estimating the residual volume of a process fluid (R) is described in this section. Two flanged stainless steel sections of piping and an end cap were connected as shown in the Figure. The pipes were one inch in diameter and four inches in length. The bottom section of piping was used to model the continuous nature of surfaces at full scale; it was not part of the surface area that was used to estimate the residual volume. The piping assembly was filled with the process fluid, turned to the appropriate angle (5° or 90°), and then drained by removing the end cap. The amount of residual process fluid in the top section was measured by gravimetry. Residual volumes were estimated for rinse water and process soil. These fluids were simulated with deionized (DI) water (< 1 µS/cm) and a 56% glucose solution (w/w), respectively. Due to its very high concentration and viscosity, the glucose solution was representative of a 8 Special edition: Cleaning Validation

table I: Residual Volumes per Unit Surface Area (R) for Deionized Water ( δ, the thickness of the film; and (3) the distance over which the solute diffuses into the film (d) is > δ; and (2) when viscous forces are large enough to prevent continued acceleration of the liquid along the length of the plate – i.e., at a low Reynolds Number, when the flow is laminar. Under these conditions, V is independent of the distance traversed along the incline (L). Note that at the worst-case location the above conditions would be satisfied because (a) the surfaces being cleaned are relatively large, and thus L and W would be >> δ; and (b) the flow would be laminar. The Reynolds number (Re) is used to classify a falling film into three flow regimes: (a) laminar flow with negligible rippling (Re < 20); (b) laminar flow with pronounced rippling (20 < Re < 1500); and (c) turbulent flow (Re > 1500). When Re is less than 20, the ripples are very long and grow slowly down the surface of the film. As Re increases above 20, the ripple growth increases rapidly. Because of the assumptions made in developing the above model (21), the error in using Equation 5 to estimate velocity increases with ripple growth and Re. The velocities estimated using Equation 5 have been shown to be in good agreement with experimentally observed velocities when Re is less than 1000 (22, 23). The average velocity of the cleaning solution is estimated from the flowrate per unit width (Q/W) and the angle of inclination (α) using Equation 5. The estimates are based on the kinematic viscosity (ν) of water at 20°C (0.01 cm2/s), acceleration due to gravity (g) of 981 cm/ s2, and a range of values of Q/W and α. The results, summarized in Table 2, indicate that for 5° < α < 90° and Q/W < ~1 gpm/ft (~2 mL/s/cm), the flow would be laminar and the thickness of the film ( would be < ~1 14

Special edition: Cleaning Validation

table II: Average velocity, film thickness, Reynolds Number and relative mass transfer rate in a laminar falling film for a range of values of flowrate and angle of inclination.

CoNCLUSIoN Small-scale experimental models are used to determine worst-case soils for cleaning validation, and estimate cleaning times and other performance parameters. A critical step in the development of these models is to identify and scale down the hardest-to-clean or worstcase location in the equipment. For cleaning operations, the worst-case location is typically an area within the equipment with poor circulation, such as a shadowed or occluded area. Examples of such locations include the underside of a probe or an impeller blade where there is no direct or indirect impingement of the cleaning solution during CIP operations. Instead, the flow of the fluid at the worst-case location is in the form of a laminar falling film. A mathematical model for diffusion-controlled mass transfer in a laminar falling film was used to characterize the effect of flowrate per unit width (Q/W) and fluid velocity (V) on the kinetics of cleaning under worst-case conditions. The results indicate that the effect of these parameters on the rate of mass transfer – and therefore the kinetics of cleaning – is relatively weak. The mass transfer rate increases only by a factor of 8% and 12%, respectively, when Q/W or V is doubled. An experimental approach for optimizing Q/W and V for evaluating relative cleanability at small scale was described. Relative cleanability was shown to depend on the solubility and diffusivity of the soils being compared. Further, for diffusion-controlled mass transfer – which typifies worst-case cleaning conditions for CIP opera-

Rizwan Sharnez

tions – relative cleanability was found to be independent of Q/W and V. Thus, if the objective of the study is to rank coils based on cleanability, Q/W and V can be set to any reasonable value, provided that the resulting flow is laminar. In practice, however, a lower value of Q/W is preferable because the absolute difference between the cleaning times of the soils (∆tAB) is amplified, and as a result, the ability to differentiate between the soils based on the larger magnitude of ∆tAB is enhanced commensurately. If necessary, ∆tAB can be amplified by reducing Q/W up to the point where the film is still intact and uniform – i.e. it does not disintegrate into slower-moving unsteady drops. The laminar falling film model was also used to estimate fluid velocity (V), film thickness () and Reynolds Number (Re) from Q/W and the angle of inclination (α). The results indicate that for 5° < α < 90° and Q/W < ~1 gpm/ft (~2 mL/s/cm), the flow would be laminar and would be < ~1 mm, if the flow was stable. Further, under these conditions, V would be < ~52 cm/sec, which is substantially less than the design criterion for VMIN in pipes and hoses – viz. 150 cm/sec (5 ft/s). The calculated values of V have been shown to be in good agreement with experimentally observed velocities when Re is less than 1000. SYMBoLS ANd ACRoNYMS CIP Clean-in-place D Diffusivity g Acceleration due to gravity L Length of object being cleaned N Mass or molar flux Q Volumetric flow rate Re Reynolds number S Solubility t Time V Average velocity of cleaning solution W Width of laminar falling film α Angle of inclination (slope) δ Film thickness µ Dynamic Viscosity ν Kinematic viscosity ρ Density ω Mass flow rate SUBSCRIPtS A Product A B Product B MIN Minimum WCL Worst-case location X Fluid X

REFERENCES 1. R. Sharnez, A. To, and A. Tholudur, “Experimental Parameters for Small-scale Cleaning Characterization Part I: Dilution of Process Fluids during Cleaning,” Journal of Validation Technology 19 (3), 2013. 2. R. Sharnez, “Leveraging Small-Scale Models to Streamline Validation,” Journal of Validation Technology 14 (4), 2008. 3. R. Sharnez, “Taking the Guesswork out of Validation,” Journal of Validation Technology 14 (3), 2008. 4. R. Sharnez, “Master Soils for Cleaning Cycle Development and Validation: A Case Study,” Cleaning and Cleaning Validation 2(2013) Davis Healthcare & PDA. 5. R. Sharnez, et al., “Industry Comes Clean at PDA Annual Meeting,” PDA Letter XLVIII (7), 28-32, July/Aug 2012. 6. B. Hoist, “Developing a Cleaning Process: Cleaning in Development,” Journal of GXP Compliance 10 (3), 2006. 7. R. Sharnez, “Strategies for Developing a Robust Cleaning Process – Part I: Application of Quality by Design to Cleaning,” American Pharmaceutical Review 13 (5), 77-80, 2010. 8. R. Sharnez, “Validating for the Long Haul,” Journal of Validation Technology 14 (5), 2008. 9. R. Sharnez and M. VanTrieste, “Quality-by-Design for Cleaning Validation,” in Cleaning and Cleaning Validation 1, (2009) Davis Healthcare International & PDA. 10. R. Sharnez and L. Klewer, “Strategies for Developing a Robust Cleaning Process – Part II: Demonstrating Cycle Effectiveness,” American Pharmaceutical Review - Digital Edition 15 (3), 2012. 11. Canhoto, “A Novel Bench Scale Apparatus to Model and Develop Biopharmaceutical Cleaning Procedures,” Journal of Validation Technology 11 (4), 2004. 12. R. Sharnez et al, “In Situ Monitoring of Soil Dissolution Dynamics: A Rapid and Simple Method for Determining Worst-case Soils for Cleaning Validation,” PDA Journal of Pharm. Sc. & Tech. 58 (4), 203-214, 2004. 13. P. Pluta, “Laboratory Studies in Cleaning Validation,” Journal of Validation Technology 13 (4), 2007. 14. R. Sharnez and L. Klewer, “Parametric Release for Cleaning, Part I: Process Characterization,” Journal of Validation Technology (14) 8, 30, 2009. 15. R. Sharnez and M. Monk, “Strategies for Enhancing the Performance of Cleaning Processes – Part I: A Framework for Assessing Performance,” Journal of Validation Technology 17 (1), 36-39, 2011. 16. ASME Bioprocessing Equipment (BPE) Standard, 2012. 17. Kramers, H., and P. J. Kreyger, “Mass Transfer between a Flat Surface and a Falling Film”, Chemical Engineering Science, Vol. 6, pp. 42-48 (1956). 18. R. Bird, W. Stewart, and E. Lightfoot, “Transport Phenomena”, 2nd Ed., 2007. p. 562-563. 19. M. Blount, “Aspects of advection-diffusion-reaction flows of relevance to decontamination”, KTN Internship Report, (2010). 20. J. Landel, H. McEvoy, and S. Dalziel, “Cleaning of Viscous Drops on a Flat Inclined Surface Using Gravity-Driven Film Flows”, Food and Bioproducts Processing, Vol 93, p. 310-317 (2015). 21. R. Bird, W. Stewart, and E. Lightfoot, Transport Phenomena. 2nd Ed., 2007. pg.46. 22. West and Cole, “Surface velocities of thin liquid films,” Chemical Engineering Science, 22, 1388-1389, 1967. 23. S. Portalski, “Velocities in film flow of liquids on vertical plates,” Chemical Engineering Science 19, 575-582, 1964.

Special edition: Cleaning Validation

15

Rizwan Sharnez

Methodology for Assessing Product Inactivation during Cleaning Part I: Experimental Approach and Analytical Methods | IVT Rizwan Sharnez, Ph.D

ABSTRACT For multiproduct cleaning validation, the conventional approach for setting an acceptance limit for the process residue is based on the maximum allowable carryover (MAC) of the active pharmaceutical ingredient (API) (depending on the process soil, API refers to the active pharmaceutical ingredient in the drug product, drug substance, or drug substance intermediate). However, if the API becomes pharmacologically inactive during cleaning the acceptance limit does not need to be based on active product. This is an important consideration in biopharmaceutical manufacturing because the cleaning conditions are generally aggressive enough to inactivate the product. The experimental approach and analytical methods for assessing inactivation of the API during cleaning are described in Part I. A rational approach for setting safety-based acceptance limits for inactivated product and process residuals is described in Part II. The scope of this paper is limited to biopharmaceutical cleaning processes; nonetheless, the underlying concepts may be useful in designing inactivation studies and setting acceptance limits for other types of pharmaceutical cleaning processes. INTRODUCTION An important regulatory expectation for multiproduct cleaning validation is to demonstrate that potential carryover of the previously manufactured API (Product A) into the subsequently manufactured product (Product B) is below an acceptable level. This criterion is often assessed through a maximum allowable carryover (MAC) calculation for the previously manufactured API (1-5). The MAC calculation is typically based either on the minimum therapeutic dose (1), or the acceptable daily exposure (ADE) (2) of the previously manufactured API. Limitations of the MAC Approach A limitation of the conventional MAC approach is that it is based on the assumption that the product is active after the cleaning. This has important implications for biopharmaceutical manufacturing because the API is often inactivated by the cleaning process (6, 7). Another limitation of the MAC approach is that the calculated acceptance limits are often below the limit of quantitation (LOQ) of nonspecific analytical methods, such as total organic carbon (TOC). The LOQ of TOC- based methods is typically between 0.05 and 0.2 ppm. The large surface areas and small batch sizes involved in biopharmaceutical manufacturing further exacerbate this issue. Product specific immunoas16

Special edition: Cleaning Validation

Rizwan Sharnez

says (PSIA) such as ELISA and EIA have been used to address this issue; the LOQ of most PSAs is on the order of 10 ppb. PSIAs detect activity indirectly by recognizing specific epitopes (short sequences of amino acids ) in the API; however, epitopes are known to degrade during cleaning, and thus the results can be misleading (8, 9). Other limitations of the MAC approach are discussed in the literature (9). PRODUCT INACTIVATION APPROACH With the product inactivation approach, if the API is inactivated during cleaning, the acceptance limits may be set based on the inactivated product instead of the API. The product inactivation approach is therefore more reflective of the phenomenological aspects of the cleaning process. Additionally, the acceptance limits based on inactivated product are very unlikely to be below the LOQ of TOC (refer to Part II). Thus, the product inactivation approach alleviates the limitations of the MAC approach described in the previous section. The methodology described in Part I includes experimental simulation of the cleaning processes at small scale and analytical methods to evaluate inactivation of the API during cleaning. A rational approach for setting safety-based acceptance limits is described in Part II. PROPOSED METHODOLOGY Inactivation of the product during cleaning has important implications for cleaning validation of multiproduct equipment. If it can be demonstrated that the product becomes pharmacologically inactive during cleaning, there is limited value in verifying removal of the active ingredient. Instead, it is more appropriate to demonstrate that the inactivated product has been removed below a predefined acceptance limit. This is consistent with the expectation that the carryover of an extrinsic impurity into a subsequent batch should be justified from the standpoint of the safety and efficacy of the product. It also obviates the need to develop PSIAs for cleaning validation. Biopharmaceutical cleaning cycles are generally designed to expose product contact equipment to extremes of pH (13) and temperature (6080°C) for several minutes. Under these conditions monoclonal antibodies, therapeutic proteins, and other biopharmaceuticals are known to degrade and denature rapidly, and are therefore likely to become pharmacologically inactive (6, 7). The product inactivation approach should therefore be considered for biopharmaceutical cleaning validation. GUIDANCE FOR DESIGNING INACTIVATION STUDIES Fragmentation and inactivation of an API during cleaning can be assessed by exposing the process soil to

worst-case cleaning conditions at bench scale (10, 11). The results of the bench-scale studies can justifiably be extrapolated to the full-scale cleaning process. That is because under worst-case cleaning conditions of laminar flow and low shear rate fragmentation and inactivation are independent of scale (i.e., they depend on cleaning parameters that are not a function of the of spatial coordinates of the system, such as time, temperature, concentration, and the ratio of cleaning solution to process soil). The bench-scale experiments are typically performed in a vial or dialysis cassette, and are designed to simulate full-scale cleaning conditions that are least conducive (worst-case) for inactivation. For example, for a chemical wash, the lowest applicable concentration of cleaning agent, temperature, duration, and ratio of cleaning solution to residual process soil should be considered in simulating the cleaning cycle at bench scale. Other operating parameters that can contribute to product inactivation include dirty hold time (DHT) and associated drying conditions (humidity and air circulation rate), and shear rate due to impingement and turbulence. An operating parameter or step can be eliminated from the experimental design if its elimination represents a worst-case scenario from the standpoint of inactivation. This approach can be leveraged to simplify the bench-scale studies. For instance, if it is reasonable to assume that product inactivation increases with shear rate, then it can be eliminated from the experimental design (i.e., the shear rate need not be simulated in the experiment). Similarly, the ratio of cleaning solution to process soil can be reduced, and the acid wash and rinse steps can be eliminated to minimize dilution of the process soil, and facilitate detection of the process residue in the sample. When making such changes, unexpected effects such as aggregation of the API can occur. It is therefore important to make sure that the modifications do not result in experimental artifacts. If the cleaning cycles are being developed or modified, the inactivation study should be designed to evaluate the effect of key operating parameters on the fragmentation and inactivation rate of the API. This information, together with data from cleanability studies (12, 13), can be used to identify cycle parameters that are effective in inactivating the API. For existing cleaning cycles, the cleaning conditions for the inactivation study should be based on worstcase operating parameters for all systems involved. For instance, if different systems are cleaned with different cleaning solutions and at different temperatures, then the study should be performed with the mildest cleaning solution, at the lowest cleaning agent concentration, and the lowest temperature, if these conditions are least conducive for inactivation. Further, for cleanin-place (CIP) systems with multiple toggle paths, the Special edition: Cleaning Validation

17

Rizwan Sharnez

duration of cleaning should be based on the toggle path with the shortest cleaning time. After exposing the process soil to worst-case cleaning conditions, the samples are titrated to a neutral pH, and cooled to 4°C to minimize further fragmentation and inactivation of the API. The samples are then subjected to analytical testing as described in the next section. If the API is inactivated when exposed to worst-case cleaning conditions, then the acceptance limit for the inactivated product can be set based on the approach described in Part II. If the results indicate that the API is not inactivated during cleaning, then the acceptance limits should be set based on the acceptable carryover of the API (1-2). If the API is partially inactivated, then the acceptance limits should be determined for the API, as well as for the inactivated product, and the lower of the two limits should be used. Alternatively, the cleaning parameters can be modified to ensure inactivation of the API. This can be facilitated by running additional studies to characterize the effect of specific cleaning parameters on the API. RECOMMENDED ANALYTICAL METHODS Analytical methods commonly used to evaluate the effect of cleaning parameters on the previously manufactured API are described in this section. These methods are used to evaluate fragmentation and inactivation of the API at bench scale, and to detect target impurities (in this case, the previously manufactured API and/or inactivated product in the process residue) in cleaning validation samples. The analytical results are used to understand the impact of the cleaning conditions on the process soil, and to set rational safety-based acceptance limits for the target impurities (Part II). The detection methods are used to verify that the concentrations of target impurities in cleaning validation samples are below their respective acceptance limits. Sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) or capillary electrophoresis (CE) are generally used to characterize fragmentation of the API during cleaning. For SDS-PAGE, a 4 to 20 percent gradient corresponds to a molecular weight range of 4 to 250 kDa, which is sufficient for most biological APIs. While CE provides greater sensitivity, lower variability due to the absence of staining, and high throughput capability as compared to SDS-PAGE, both methods are adequate for demonstrating distinct, size-based separation of fragmented protein. Size exclusion high-pressure liquid chromatography (SEHPLC) can also be utilized for size-based separation of protein fragments; however, it can be difficult to obtain a distinct size-based separation across a wide range of fragment sizes. Inactivation of the API can be evaluated by methods that measure loss of biological activity or function 18

Special edition: Cleaning Validation

(binding sites that are functionally intact), such as a bioassay. These methods measure the relative amount of biologically active product. Thus, they can be used to evaluate the degree of inactivation of the API during cleaning. A non-specific method such as total organic carbon (TOC) can be used to detect target impurities in cleaning validation samples. TOC has an LOQ of approximately 0.2 ppm, which is sufficient for most cleaning applications. The use of a non-specific method allows for the detection of both intact API and inactivated product. With the above methods, appropriate standards and untreated controls should be included to provide a basis for comparison, and to assess the impact of any experimental artifacts and potential matrix effects. For SDS-PAGE, appropriate MW markers should be included to facilitate comparison of the fragments to the untreated controls. CONCLUSION An important consideration in multiproduct cleaning validation is to demonstrate that the carryover of the previously manufactured API into a batch of the subsequently manufactured product is below an acceptable limit. This criterion is often met through a MAC assessment for the API; however, if the previously manufactured API becomes pharmacologically inactive during cleaning, the acceptance limit does not need to be based on active product. This is an important consideration in biopharmaceutical manufacturing because the cleaning conditions are generally aggressive enough to inactivate the product. Fragmentation and inactivation of an API during cleaning can be assessed by exposing the process soil to simulated cleaning conditions at bench scale (10, 11). The bench scale experiments are typically designed to simulate full-scale cleaning conditions that are least conducive (worst case) for inactivation. The degree of inactivation is evaluated by subjecting the sample and untreated controls to the appropriate assays, (e.g., SDS-PAGE and bioassay are used to evaluate fragmentation and biological activity, respectively). The results of the study are used to set appropriate acceptance limits for cleaning validation. If the API is inactivated during cleaning, the acceptance limits for the process residuals can be set based on the approach described in Part II. ACKNOWLEDGEMENTS We thank Arun Tholudur, Aine Hanley, and Sam Guhan of Amgen; Rich Kemmer and Paul Whetstone of Bayer; James Crawford and Michael Maurer of GlaxoSmithKline; John Krayer of Janssen; Josh Getchell of Lonza; David Barabani and Michael Parks of Pfizer; and Markus Bluemel of Novartis for their help and support.

Rizwan Sharnez

REFERENCES 1.

2.

3.

4.

5.

6.

7.

G.L. Fourman and M.V. Mullen, “Determining Cleaning Validation Acceptance Limits for Pharmaceutical Manufacturing Operations,” Pharmaceutical Technology 17 (4), 54-60, 1983. ISPE, Risk-Based Manufacture of Pharmaceutical Products: A Guide to Managing Risks Associated with Cross-Contamination 7, International Society for Pharmaceutical Engineers (ISPE). Tampa, FL, First ed., 2010. R. Sharnez, “Strategies for Setting Rational MAC-based Limits – Part I: Reassessing the Carryover Criterion,” Journal of Validation Technology 16 (1), 71-74, 2010. R. Sharnez, A. To, and L. Klewer, “Strategies for Setting Rational MAC-based Limits – Part II: Application to Rinse Samples,” Journal of Validation Technology 17 (2), 43-46, 2011. R. Sharnez, and A. To, “Strategies for Setting Rational MACbased Limits – Part III: Leveraging Toxicology and Cleanability Data,” Journal of Validation Technology 17 (3), 24-28, 2011. R. Sharnez, and A. To, “Cleaning Validation of Multiproduct Equipment: Acceptance Limits for Inactivated Product, Part I – The Comparable Quality Approach,” Journal of Validation Technology 17 (4), 32-36, 2011. R. Sharnez, E. Aisenbrey, J. Bercu, D. Binkley, and A. Tholudur, “Cleaning Validation of Multiproduct Equipment: Acceptance Limits for Inactivated Product, Part II- Application of the Com-

8.

9.

10.

11.

12.

13.

parable Quality Approach to Intrasite Assessments,” Journal of Validation Technology 18 (2), 17-25, 2012. Health Canada, Cleaning Validation Guidelines (GUIDE-0028); Section 8.3, Health Products and Food Branch Inspectorate, 2008, available at: http://www.hc-sc.gc.ca/dhp-mps/compliconform/gmp-bpf/validation/index-e.... R. Sharnez, M. Horner, A. Spencer, and A. Tholudur, “Leveraging Acceptable Exposure of Host Cell Protein to Set Acceptance Limits for Inactivated Product,” Journal of Validation Technology 18 (3), 38-44, 2012. K. Kendrick, A. Canhoto, and M. Kreuze, “Analysis of Degradation Properties of Biopharmaceutical Active Ingredients as Caused by Various Process Cleaning Agents and Temperature,” Journal of Validation Technology 15 (3), 69, 2009. N. Rathore, W. Qi, C. Chen, and W. Ji, “Bench-scale characterization of cleaning process design space for biopharmaceuticals,” Biopharm International 22 (3), 2009. R. Sharnez, “Don’t Bet on Quality-by-Chance: Part II – Leveraging Small-Scale Models to Streamline Validation,” Journal of Validation Technology 14 (4), 2008. R. Sharnez and L. Klewer, “Strategies for Developing a Robust Cleaning Process – Part II: Demonstrating Cycle Effectiveness,” American Pharmaceutical Review - Digital Edition 15 (3), 2012.

JVT

Special edition: Cleaning Validation

19

Rizwan Sharnez

Methodology for Assessing Product Inactivation During Cleaning Part II: Setting Acceptance Limits of Biopharmaceutical Product Carryover for Equipment Cleaning | IVT By Adam Mott, Bill Henry, Edward Wyman, Greg Randall, Kathleen Bellorado, Markus Blümel, Mary Ellen Clark, Michael Parks, Ronan Hayes, Scott Runkle, Wendy Luo

ABSTRACT For multi-product biopharmaceutical facilities, setting the acceptable level of process residues following equipment cleaning is an important regulatory, business, product quality, and patient safety consideration. Conventional approaches for setting an acceptance limit for process residues have been based on the assumption that the active pharmaceutical ingredient (API) (depending on the process soil, API refers to the active pharmaceutical ingredient in the drug product, drug substance, or drug substance intermediate) is chemically or functionally intact following the cleaning process. These approaches include Maximum Allowable Carryover (MAC) Health Based Exposure Limits and other “dose” or Permissible Daily Exposure (PDE)based limits. The concept for cleaning acceptance limits based on intact product originated from the manufacturing of small molecule pharmaceuticals (1). In contrast to pharmaceutical small molecules, biopharmaceutical products are large molecules that are likely to degrade and become inactive when exposed to cleaning conditions. Therefore, an alternative approach to setting cleaning acceptance limits for biopharmaceutical products based on the actual process residues that could potentially be present on production equipment should be considered. Part I described the methodology to assess and verify API inactivation during cleaning (2). In Part II, alternative approaches for setting acceptable levels of process residue will be described building upon the basis that API inactivation by the cleaning process has been demonstrated. INTRODUCTION When multiple products are manufactured using the same equipment, it is important to ensure that potential product or process residues from the previously manufactured batch are removed to an acceptable level to ensure the subsequently manufactured product will not be impacted. The acceptable level of carryover has often been based on the active, intact API. However, for biopharmaceutical products, the API typically degrades and becomes pharmacologically inactive during cleaning, and therefore the cleaning acceptance criteria do not need to be based on the concept of intact and active product. Rather, the cleaning acceptance limit should be based on potential process residues that have a greater carryover potential founded on phenomenological aspects of the cleaning process. The scope of this paper tar20

Special edition: Cleaning Validation

Rizwan Sharnez

gets biopharmaceutical APIs; nonetheless, the underlying concepts may be useful in setting acceptance limits for other types of pharmaceutical products where inactivation during the cleaning process can be demonstrated. This paper will include a review of product inactivation, information on product detection using total organic carbon (TOC), and alternative approaches for setting acceptance limits for equipment cleaning. The intention of this paper is to propose acceptable approaches for setting cleaning limits for biopharmaceutical process equipment that may be considered. However, it should not be considered prescriptive for what approach is most appropriate or should be used since every production facility, processes, and products manufactured are unique. PRODUCT INACTIVATION Biopharmaceuticals are large molecule drug products (e.g., monoclonal antibodies, therapeutic proteins, etc.) that are made in processes using living organisms rather than extracted from a native source or by synthesizing compounds. The equipment cleaning cycles are designed to expose product contact areas to cleaning detergents that include alkaline and acidic chemicals. Under these exposure conditions, the high pH in alkaline chemicals (typically pH >11) and low pH in acidic chemicals (typically pH 0

Figure 2: Example Histogram of Cleaning Verification Data.

Rizwan Sharnez

number of “0” results relative to the total number of results as described in the equation below:

After the review and adjustment for the “0” data results, the Box-Cox transformation is performed using the adjusted alpha rate (in the example dataset, the adjusted alpha rate is 0.0046). Figures 3 and 4, below, depict the example TOC data that have been transformed using the Box-Cox method. Figure 3: Box-Cox Transformation.

The top-left histogram describes the distribution of the original TOC dataset. This dataset is non-normal, being truncated at zero. The same non-normal phenomenon is displayed in the associated normal probability plot in the lower-left. The top-right histogram describes the same data after applying the Box-Cox transformation. In this case, the data are normally distributed as evidenced with the normal probability plot in the lower-right. The Performance Limits are then back-calculated to the original scale using the transformed dataset and the equation below: Yoriginal = (Ytransformed * lambda + 1)(1/lambda)

Figure 4: Effect of Using the Box-Cox Transformation.

The Box-Cox transformed Performance Limit from the example data is 4876 ppb TOC and is shown in Figure 5. Finally, as further cleaning studies are conducted, additional TOC data will be collected. An appropriate review of the overall dataset should be conducted, and the performance limits adjusted if performance changes for reasons that should be well understood.

Figure 5: Performance Control Limit from Example Dataset.

If the dataset contains an excessive number of zero values, the “0” values should be removed and the alpha rate (e.g., 0.27% or 0.0027) adjusted accordingly prior to transforming the data with the Box-Cox method. In the example dataset, 428 of 1034 results are “0.” The alpha rate (0.027) is therefore adjusted according to the

CONCLUSION Setting acceptable limits for process residue following equipment cleaning in multiproduct biopharmaceutical facilities requires an understanding of each product’s composition and the effects of the cleaning process on the API. The degrading and denaturing effects of chemical detergents should be studied for each product manufactured within the facility. Setting acceptance limits for product carryover based on TOC can be accomplished with the Cleaning Process Capability, Safety Factor, or Toxicology Threshold approaches. As on-going cleaning studies collect TOC data, these data can be evaluated with the Performance Control Limit approach to ensure control of the equipment cleaning process is maintained.

Special edition: Cleaning Validation

25

Rizwan Sharnez

26

ACKNOWLEDGEMENTS We thank Kristina Conroy, Mariann Neverovitch, Michael Hausladen of Bristol Myers Squibb, Rob Lynch of GlaxoSmithKline, Jim Heimbach and Ben Locwin of Lonza, Stephanie Donat, Gareth Sanderson of Novartis, Martin Hammarström of Pfizer and David Bain of BioPhorum Operations Group for their help and support.

TOC TTC

ACRONYMS AND DEFINITIONS

REFERENCES

Action Limit An empirical limit that the cleaning process cannot exceed without potential impact to product quality or patient safety. PDE Permissible Daily Exposure (also called ADE, Acceptable Daily Exposure) which represents a dose of a drug to which a human may be exposed per day or per dose (for biologics) without any anticipated pharmacologic or toxicological effects, so in the event of potential carry-over of one API to another, there would be no risk to the patient. Alert Limit An empirical limit, statistically established from study data, which is used to monitor the quality of the cleaning process. API Active Pharmaceutical Ingredient Degrade To cause the cleavage and hydrolysis of chemical bonds within peptides and amino acid strings, such that the biological activity is diminished or eliminated. Denature To cause the tertiary structure of a biological product to unfold, as with heat, alkali, or acid, so that some of its original properties, especially its biological activity, are diminished or eliminated. MAC Maximum Allowable Carryover Peptides A chemical compound containing two or more amino acids (amino acid polymers) that are coupled by a peptide bond. Peptides are often classified according to the number of

1.

Special edition: Cleaning Validation

amino acid residues. Oligopeptides have 10 or fewer amino acids. Molecules consisting from 10 to 50 amino acids are called peptides. The term protein describes molecules with more than 50 amino acids. Total Organic Carbon Toxicological Threshold of Concern

G.L. Fourman and M.V. Mullen, “Determining Cleaning Validation Acceptance Limits for Pharmaceutical Manufacturing Operations,” Pharmaceutical Technology 17 (4), 54-60, 1993. 2. “Methodology for Assessing Product Inactivation during Cleaning, Part I: Experimental Approach and Analytical Methods,” Journal of Validation Technology 17 (4), 2012. 3. “Solubility of Proteins,” Journal of Protein Chemistry 5 (6), 1986. 4. ICH, Q3A Impurities in New Drug Substances, 2008. 5. AR.C Beavis, “Chemical mass of carbon in proteins,” Analytical Chemistry 65, 496-497, 1993. 6. Kroes et al., “Structure-based thresholds of toxicological concern (TTC): Guidance for Application to Substances Present at Low Levels in the Diet,” Food and Chemical Toxicology 42 65-83, 2004. 7. D.G. Dolan, B.D. Naumann, E.V. Sargent, A. Maier, and M. Dourson, “Application of the Threshold of Toxicological Concern Concept to Pharmaceutical Manufacturing Operations,” Regulatory Toxicology and Pharmacology 43, 1–9, 2005. 8. J.P. Bercu and D.G. Dolan, “Application of the Threshold of Toxicological Concern Concept When Applied to Pharmaceutical Manufacturing Operations Intended for Short-term Clinical Trials,” Regulatory Toxicology and Pharmacology 65, 162–167, 2013. 9. ICH, Assessment and control of DNA reactive (mutagenic) impurities in pharmaceuticals to limit potential carcinogenic risk. M7 Step 2 version (2013). 10. Statistics for Experimenters: Design, Innovation, and Discovery, 2nd ed., 2005.

Tim Sandle

Aseptic Transfer Risk Assessment: A Case Study | IVT Tim Sandle, Ph.D.

ABSTRACT This paper uses one example of a risk assessment approach to illustrate how risk assessment can be incorporated into good manufacturing controls. The specific activity assessed involves transferring a set of sterilised stoppers from an autoclave to a filling machine within a sterile manufacturing facility. The risk assessment approach adopted is a form of HACCP (Hazard Analysis Critical Control Points). Approaches to risk assessments are discussed. Key activities are described. Documentation is critical. There is no such thing as “zero risk.” A decision is thus required as to what is “acceptable risk. Key components of HACCP including hazard analysis and critical control points. The “seven pillars” of HACCP are described. Stepwise activities needed to accomplish the risk assessment are discussed. This case study described a low risk activity. Rationale for the assessment is described. This discussion described a risk assessment approach using a relatively simple case to illustrate its potential for other applications. INTRODUCTION Risk analysis and risk management are significant considerations in current pharmaceutical manufacturing practices. All areas and functions in the pharmaceutical manufacturing plant assess level of risk to a process and then take steps to eliminate that risk. This paper uses one example of a risk assessment approach to illustrate how risk assessment can be incorporated into good manufacturing controls. Risk assessment involves identifying risk scenarios either prospectively or retrospectively. The former involves determining what can go wrong in the system and all the associated consequences and likelihoods; the latter this looks at what has gone wrong. Risk assessment is then used to assess the process, product, or environmental risk and to aid in formulating the appropriate actions to prevent the incident from re-occurring (1). The case examined here is a prospective risk assessment. The case study involves transferring a set of sterilised stoppers from an autoclave to a filling machine within a sterile manufacturing facility. The risk assessment approach adopted is a form of HACCP (Hazard Analysis Critical Control Points). APPROACHES TO RISK ASSESSMENT There are various approaches to risk assessment being used in the pharmaceutical industry. Each has their respective merits. The use of risk assessment is commonly expected by regulatory authorities (2). There are various tools that can be drawn upon for conducting risk assessments. Some common methods include FMEA (Failure Mode and Effects Analysis), FTA (Fault Tree Analysis), and HACCP (Hazard Analysis Critical Control Points). Many of these employ a scoring approach. No definitive method exists for all applications and different Special edition: Cleaning Validation

27

Tim Sandle

approaches are useful for different situations (3). The various approaches differ in their structure and with the degree of complexity involved. Nonetheless, there are some broad similarities. The different analytical tools are similar in that they generally involve: • Constructing diagrams of work flows, • Pin-pointing areas of greatest risk, • Examining potential sources of contamination, • Deciding on the most appropriate sample methods, • Helping to establish alert and action levels, • Taking into account changes to the work process / seasonal activities. Documenting risk assessments is important. When a risk assessment is presented to an auditor, it is likely that the reviewer will be concerned with whether the risk assessment is traceable to a risk methodology and that the process is clear, understandable, and that it has been performed consistently. In this sense, the reviewer will want to understand how the decisions were made and that the risk (or problem) was defined upfront. With the question established, the reviewer will also wish to see if the process performed actually answered the question in a meaningful way using scientific principles. Important points to remember for any risk assessment are the following: • There is no such thing as “zero risk.” A decision is thus required as to what is “acceptable risk.” • Risk Assessment is not an exact science - different people will have different perspectives on the same hazard.

Before embarking upon risk assessment it is important to establish and to define: • Develop and agree on the risk question. A clearly defining the risk question facilitates selection of the appropriate risk assessment tool, identifies relevant data, information and assumptions; assists in the identification of resources, responsibilities and accountabilities; and ensures that appropriate focus on the business objective is maintained. • Select the most appropriate risk assessment tool. The selected risk assessment method or tool will be used to organise collected data, understand what steps can be taken to reduce or control risk, and to help make appropriate decisions. APPLICATIONS TO CLEANROOM ENVIRONMENTS AND PHARMACEUTICAL PROCESSES The primary risk assessment approaches can allow for a complete review of operations within the cleanroom ensuring those facilities, operations and practices are also satisfactory. The approaches recognise a risk, rate the level of the risk, and then develop a plan to minimise, control, and monitor the risk. For example, the monitoring of the risk will help to determine the frequency, locations, and level for environmental monitoring (4). This case study outlined in this paper uses a modification of the HACCP method, which is sometimes called the “Lifecycle Approach”. The approach adopted has been used for similar assessments of pharmaceutical processes (5). There are numerous risk factors to be considered when analysing aseptic processing. The HACCP approach can help to identify these risk factors. The following factors shown in Figure 1 are useful to note

Risk assessments should always: • Be based on systematic identifications of possible risk factors • Take full account of current scientific knowledge • Be conducted by people with experience in the risk assessment process and the process being risk assessed • Use factual evidence supported by expert assessment to reach conclusions • Do not include any unjustified assumptions • Identify all reasonably expected risks-simply and clearly, along with a factual assessment and mitigation where required • Be documented to an appropriate level and controlled/approved • Ultimately be linked to the protection of the patient • Should contain objective risk mitigation plan. 28

Special edition: Cleaning Validation

Figure 1: ASEPTIC PROCESSING RISKS

Tim Sandle

when undertaking HACCP analysis in pharmaceutical manufacturing. HACCP HACCP was developed during the 1960’s from collaboration between NASA, a food company (the Pillsbury Company), and the US Army Natick Laboratories. The objective was to provide a zero-defect food supply for astronauts. HACCP was derived from Failure Mode Analysis (6). There are two key components of HACCP: • Hazard Analysis: Determining what microbiological, physical, or chemical risks are associated with a process. • Critical Control Point: A point, step, or procedure at which control can be applied. HACCP generally involves an assessment of the following conditions – known as the “seven pillars.” 1. Conducting a hazard analysis. List all potential hazards associated with each step, conduct a hazard analysis, and consider any measures to control identified hazards. 2. Determining the Critical Control Points (CCPs). Critical control points have to be defined as processing steps at which necessary action can be applied to ensure and maintain compliance with specified conditions. Those points are identified with appropriate measures that can be applied to control each hazard. If no such critical control points can be established, product specific validation should eliminate potential risks of a certain process step. 3. Establishing critical limit(s). Establish critical limits for each CCP. 4. Establishing a system to monitor control of the CCP. 5. Establishing the corrective action to be taken when monitoring indicates that a particular CCP is not under control. 6. Establishing procedures for verification to confirm that the HACCP system is working effectively. 7. Establishing documentation and record keeping. Aspects of these have been utilized in the case study. In general, the advantages of the HACCP approach are that it allows for a systematic overview of the process for the evaluation of each processing step, allows each step to be examined the possible risks, and allows for the specification of measures required for controlling each risk. The

primary disadvantage of HACCP is that, unlike FMEA, HACCP cannot be used to rank or prioritize risks. DECONSTRUCTING THE PROCESS Four stepwise activities are needed to accomplish the risk assessment: • A route map. The facility is drawn and the route indicated • Identification of hazards, which can be divided into biological; physical; equipment; transport and chemical. This will allow an assessment of existing control measures • Process flow • Assessment of environmental monitoring. This will determine if the activity is safe to proceed (7). Route Map The first step was to outline a route map. This focused on transporting the stoppers by the shortest and safest route possible as illustrated in figure 2.

Figure 2: ROUTE MAP. Arrows indicate stopper transfer route from autoclaves to filling room

Identification of Hazards The second step was to understand the hazards and the process. The primary hazards with an aseptic process are biological, physical, and equipment related (8) as displayed in Table 1. Process flow The third step was to outline the process flow as shown in Figure 3. Environmental monitoring The fourth step involved assessing dynamic state viable microbiological environmental monitoring and particle counts. If data have been satisfactory and there are no indications of adverse trends, the proposal can be advanced. If there is a level of risk, this must be addressed first using appropriate Special edition: Cleaning Validation

29

Tim Sandle

Table 1a: IDENTIFICATION OF HAZARDS

Figure 3a: Process flow with CCPs (critical control points) marked. Process flow identifies hazards.

Table 1b: IDENTIFICATION OF HAZARDS

corrective and preventative action. Undertaking an unusual event in a high-risk situation will compound the problem. 30

Special edition: Cleaning Validation

Figure 3b: Process flow with CCPs (critical control points) marked. Process flow identifies hazards.

Tim Sandle

RISK ASSESSMENT – PERFORM A SIMULATION Before undertaking the activity, a simulation should be performed so that any previously unforeseen problems can be noted and further preventative measures taken. If any variability is expected, the simulation should be repeated. The simulation should be timed and the number of staff required noted. The focus should remain on the ease of transit. In this case study, the duration of the activity is five minutes and the number of staff involved is two.

REFERENCES 1. Sandle, T. (2011): Risk Management in Pharmaceutical

2.

3.

CONCLUSION From three possible options -- high risk, medium risk or low risk -- the conclusion of this risk analysis was that the risk level was low. This evaluation is based on the successful identification of the hazards and control points. The strongest evidence for this came from the following:

4.

5.

• The process being performed by trained clean room personnel operating in an aseptic manner • The autoclaved stoppers being contained within a sealed box. The seal remains intact throughout the transfer • The outlet at the bottom of the box has an integrity tested vent filter • An additional vent filter is placed on the top of the box • The stoppers are wheeled on a trolley which minimises human intervention • The microbiological environmental monitoring for the rooms is satisfactory.

6.

7. 8.

9.

It is important to note the impact of change and to review any risk assessments on a regular basis in developing any risk assessment. This case study may or may not be typical. It may or may not be wise for an organisation to undertake such an activity. The purpose of this discussion was to demonstrate an example of the risk assessment approach using a relatively simple case so that the wider application can be appreciated.

10.

Microbiology. In Saghee, M.R., Sandle, T. and Tidswell, E.C. (Eds.) Microbiology and Sterility Assurance in Pharmaceuticals and Medical Devices, New Delhi: Business Horizons, pp553588 Sandle, T. and Lamba, S. S. Effectively Incorporating Quality Risk Management into Quality Systems. In Saghee, M.R. (2012) Achieving Quality and Compliance Excellence in Pharmaceuticals: A Master Class GMP Guide, New Delhi: Business Horizons, pp89-128 Sandle, T. (2013). Contamination Control Risk Assessment. In Masden, R.E. and Moldenhauer, J. (Eds.) Contamination Control in Healthcare Product Manufacturing, Volume 1, DHI Publishing, River Grove: USA, pp423-474 Sandle, T. (2003): The use of a risk assessment in the pharmaceutical industry – the application of FMEA to a sterility testing isolator: a case study, European Journal of Parenteral and Pharmaceutical Sciences, 8(2): 43-49 Munro, M. J., Millar, B. W. and Radley. A. S. (2003): A Risk Assessment of the Preparation of Parenteral Medicines in Clinical Areas, Hospital Pharmacist, Vol. 10, pp303-305 Gavin, A., and Weddig, L.M. (Eds.). (1995): Canned Foods: Principles of Thermal Process Control, Acidification and Container Closure Evaluation, Food Processors Institute, Washington, D.C., pp5-10 Sandle, T. (2006) Environmental Monitoring Risk Assessment, Journal of GXP Compliance, 10(2): 54-73 Whyte, W. and Eaton, T. (2004): Assessing microbial risk to patients from aseptically manufactured pharmaceuticals, European Journal of Parenteral and Pharmaceutical Sciences, 9 (3): 71-79 Whyte, W. and Eaton, T. (2004): Microbiological contamination models for use in risk assessment during pharmaceutical production, European Journal of Parenteral and Pharmaceutical Sciences, 9 (1): 11-15 Sandle, T. (2004): General Considerations for the Risk Assessment of Isolators used for Aseptic Processes, Pharmaceutical Manufacturing and Packaging Sourcer, Samedan Ltd, Winter 2004, pp43-47

Special edition: Cleaning Validation

31

Tim Sandle

People in Cleanrooms: Understanding and Monitoring the Personnel Factor | IVT Tim Sandle, Ph.D.

ABSTRACT Cleanroom contamination can arise from a number of sources. Most contamination within the pharmaceutical facility can be traced to humans working in cleanrooms. The paper discusses staff gowning and personnel behavior in pharmaceutical cleanrooms, and how cleanroom risk can be minimized. The human skin ecosystem is discussed. The Human Microbiome Project (HMP) from the US NIH characterized microorganisms found in association with both healthy and diseased humans. Information from this project has great impact on cleanroom activities including gowning practices. Topics associated with cleanroom garments are discussed including fabric types, garment lifespan, recycling, laundering, human changing procedures, training, behavior, hand sanitization, ongoing assessments, and associated topics. INTRODUCTION Cleanroom contamination can arise from a number of sources. These may vary depending upon the type of cleanroom, its geographic location, the types of products processed, and so on. Nevertheless, these sources can generally be divided into the following groups1: • People • Water • Air and ventilation • Surfaces • Transport of items in and out of clean areas Most contamination within the pharmaceutical facility can be traced to humans working in cleanrooms2. This is, in some way, evidenced from the association of microorganisms transient or residential to skin being the primary isolates from environmental monitoring in controlled environments3. Human personnel shed high numbers of skin cells mostly as skin flakes. The cleanroom garments worn by personnel cannot contain all human detritus. The paper discusses staff gowning and personnel behavior in pharmaceutical cleanrooms. Further, it considers how cleanroom risk can be minimized. Basic training for all cleanroom staff including activities such as cleanroom entry and gowning practices is examined. HUMAN SKIN Before proceeding to look at gowning, it is worthwhile to consider the human skin ecosystem. The human body is an intricate system that hosts trillions of microbial cells across the epithelial surface and within the mouth and gut. These microorganisms play a role in human physiology and organ function including digestion and immunity. The microorganisms also affect the outside environment as they are shed 32

Special edition: Cleaning Validation

Tim Sandle

AREA Scalp Saliva and nasal fluid Back Groin Forehead Hand Feet

NUMBER OF MICROORGANISMS/cm2 1 million 10 million/gram 100 1 – 20 million 100 – 1000 10,000 – 100,000 1 million

Table 1: HUMAN BODY SITES AND TYPICAL NUMBERS OF MICROORGANISMS WITHIN AREA

from the skin or deposited through different orifices. This latter issue has important implications for cleanrooms. The outer layer of human skin can host up to one million microorganisms per square centimeter4. The population, as well as the diversity, varies according to anatomical locale. Research suggests that a typical person sheds 1,000,000,000 skin cells per day of a size 33 µm x 44 µm x 4 µm -- equivalent to a rate of 30,000 to 40,000 dead skin cells shed from the surface of the skin every minute. Of these, Whyte indicates that approximately 10% of particles carry microorganisms5. There are, on average, four microorganisms per skin cell. A term commonly used to describe skin flakes with adhered microorganisms is “microbial carrying particles.” The significance of this is that people are not only a source of contamination, but also are an agent for transferring contamination possibly to locations that could pose a product risk. Microorganisms are spread from sneezing, coughing, and touching. While microorganisms suspended in the air are less of a concern, should such organisms gravitate towards a product or critical location, they may present a significant risk. HUMAN MICROBIOMe PROjeCT Our understanding of the risk the people pose to cleanrooms has been advanced by the findings from the Human Microbiome Project. The Human Microbiome Project (HMP) is a United States National Institutes of Health initiative. The project goal is to identify and characterize the microorganisms found in association with both healthy and diseased humans (the human microbiome). The human microbiome describes the aggregate of microorganisms and their genetic interactions that reside on the surface and in deep layers of skin, in the saliva and oral mucosa, in the conjunctiva, and in the gastrointestinal tract. Some of the most illuminating HMP research has been with the human skin. The skin is a complex ecosystem, supporting a range of microbial communities that live in distinct niches. These niches are

affected by available nutrients as well as by several non-nutritional factors such as pH, humidity, and temperature. With the skin, however, as epithelial cells are continually being shed, many microbial communities on the external surface are rarely stable6. The outcomes of the HMP research have shown that there is a high population on, and a considerable diversity of microbial species across, the outer layer of the skin. There are approximately 1000 species of bacteria from 19 phyla on human skin. Of these, most bacteria can be categorized into four phyla: • Actinobacteria (51.8%). Actinobacteria are a group of Gram-positive bacteria with high guanine and cytosine content, such as Micrococcus, Corynebacteria and Propionibacteria. • Firmicutes (24.4%). This includes the genera Clostridia and Bacillus. • Proteobacteria (16.5%). This is a major phylum of bacteria that includes a wide variety of pathogens such as Escherichia, Salmonella, Vibrio, Helicobacter, and many other notable genera. • Bacteroidetes (6.3%). The phylum Bacteroidetes is composed of three large classes of Gramnegative, nonspore-forming, anaerobic, and rodshaped bacteria7. Reasons for the topographical variations relate to the physicochemical properties of the skin. This is especially so for temperature, pH, amounts of oil, and moisture8. From this, there are three main ecological areas of the skin: sebaceous, moist, and dry. Examples of microbial divergence include Propionibacteria and Staphylococci species dominating the sebaceous areas (with a high oil content). Dry, calloused areas (arms and legs), Gram-positive cocci (primarily the Micrococcaceae) are found on the arms and legs and Gram-positive rods are found in high numbers on the torso. Staphylococci and Corynebacteria are found together with some Gram-negative bacteria in moist areas9. These types of microorganisms generally reflect the types recovered from cleanrooms10. The reason that Gram-positive bacteria predominate across the skin is because the skin is generally a dry environment, and any fluids present on the surface generally have a high osmotic pressure11. Thus Gram-positive bacteria (especially the Staphylococci and Micrococci) are better adapted for such environments, not least to being resistant to desiccation. Where other species occur this is due to variations in temperature and with areas of higher sweat production12. For example, this can lead to higher levels of fungi on the feet13. In relation to pharmaceutical manufacturing, the presence of any such organisms remains problematic. Special edition: Cleaning Validation

33

Tim Sandle

A further observation is that the ratio of the microorganisms recovered from the skin is relatively evenly divided between the aerobic and the anaerobic. The aerobic microorganisms tend to live on the outermost layers of the skin and the anaerobic microorganisms live in the deeper layers of the skin and hair follicles14. The information from the human microbiome project about the rich depth of variety of microorganisms on the skin introduces several implications for cleanroom environmental monitoring. The most important question of whether gowning practices are adequate to exclude all microorganisms from the richest areas of the skin microbiome. This is a pertinent point given that most bacteria free-floating in cleanroom air current are not free-living but are instead the result of direct particle shedding of desquamated skin cells and subsequent re-suspension of skin detritus in the air stream. The answer to this question should lead to a consideration of: 1. The types of cleanroom undergarments used and an examination as to whether these provide an effective barrier, especially for the more moist parts of the body. 2. The importance of the outer gown covering all parts of the body, including the forehead. 3. The quality of cleanroom certified undergarments. 4. The level of training required for operators in relation to gowning and the way that gowning qualification as conducted. 5. How long a cleanroom suit should be worn for in relation to material integrity against operator perspiration. 6. The environment in which gowns are donned, where higher air-change rates might prove effective. 7. How often gowns should be recycled which involves washing and irradiation. At some point the material fibers will weaken, thereby reducing the bacteria filter efficiency of the gown. The wearer of the gown should know what types of testing is conducted on recycled gowns and what the procedures are for rejecting gowns where a loss of integrity is detected. Cleanroom microbiologists may wish to consider how concerned they are with each of the items listed in excluding microorganisms found on all regions of the skin. There must be good understanding of the environmental monitoring methods used to assess cleanrooms. These may not show how good or bad gown changing and gown wearing is. These concerns are best addressed through good gowning practices. 34

Special edition: Cleaning Validation

PROTeCTING PRODUCTS FROM PeOPLe Despite some advances with automation and robotics, in most situations people cannot be eliminated from cleanrooms. Control of contamination from people in cleanrooms is achieved by application of two principles: • We “wrap” the people to minimize the amount of “shedding” of microorganisms. • We put localized protection around the product to minimize the amount of contact with the people. The localized protection issue is achieved through local air protection, such as unidirectional airflow cabinets and isolators. With clothing, personnel working in cleanrooms are required to wear special clothing designed for the clean environments. Such clothing is necessary, as indicated above, because the human body creates its own micro-environment of potentially damaging particulate contamination. To be effective, cleanroom clothing must: • Form a particulate barrier for the human microenvironment. • Allow freedom of movement and be comfortable. • Address any specialist requirement, e.g. static dissipation. • Avoid being a significant particulate contributor in itself. Cleanroom garments must meet specific protection criteria. Not all cleanroom garments are of the same quality. This involves manufacturing the garments from special materials, following particular construction methods, and then tailored for individual styling. The gowns must be comfortable, easy to apply and practical in use. Some gowns are disposable and others are made to be re-laundered and sterilized depending on the cleanroom grade. CLeANROOM GARMeNTS AND FABRICS Fabrics used in the manufacture of cleanroom garments must have the following features15: • Be low in particulate shedding. • Permit the body to breathe while trapping particles within the garment. The contaminant should be retained within the garment and not released into the surrounding atmosphere. • Be flexible enough for comfortable wearing. • Withstand repeated cleaning and sterilization cycles. • Meet any specific requirements such as control of static.

Tim Sandle

• Meet opacity requirements. • Look and feel as good as possible. • Be cost-effective. Fabric Categories There are three broad categories of fabric used in the construction of cleanroom garments: • Woven fabrics. Woven or re-usable fabrics are the most commonly used fabrics in cleanroom environments. Such garments are woven on sophisticated looms from yarns of continuous filaments of polyester. The thickness of the yarn and filaments is important -- the finer the yarn, the tighter the weave can be made, and the better the filtration. The pattern and tightness of the weave is important to reduce the pore size to a minimum. The use of continuous filament polyester means that there are few loose ends from which particles may be shed. • Laminated or membrane fabrics. Laminated fabrics are favored for some high-grade microelectronic environments. These types of garments are not commonly used in the pharmaceutical sector. • Disposable or limited life materials. The most common of these non-woven fabrics are from spun bonded olefin and polypropylene. Comprising a densely interlinked matt of fibers, these fabrics can provide good results for a limited period. Garments from such materials need to be processed and decontaminated before use in the cleanroom. Disposable or limited use garments are mainly used in those environments where protection of the wearer against potentially hazardous products is required. Garment Considerations Garments are designed to provide protection for the head, body, hands and feet. In establishing a system for garment selection, it is important to consider the broader aspects of cleanroom use: suitability of fabric, garment style, layers, the nature of the tasks involved, costs, regulatory requirements, and any specific customer requirements. The classification of the cleanroom will inevitably be the major factor in determining the degree of personnel protection required and the fundamental choice of garments. One important issue with gowns is the maximum length of time that a gown can be worn. As people perspire, the integrity of the gown will weaken. Complicating factors are the temperature and humidity of the cleanroom and the variations between people. The length of time will also depend upon the grade of the cleanroom. In aseptic areas, such as ISO 14644 class 7 / EU GMP Grade B areas, gowns are typically

worn only for the length of the shift (normally four hour periods to enable operators to take breaks). In lower grade cleanrooms, a gown might be worn for several sessions during the course of the working day. Other factors affecting the lifespan of the gowns that are subject to recycling are repairs and the number of permitted washing cycles. With repairs, it is prudent to have a repair policy. This will vary across facilities, and again, it will be affected by the cleanroom class. With aseptic areas, if a gown becomes torn it is normally discarded. In other grades of cleanroom, a gown can be repaired depending upon the size of the hole and the impact on the material. Some organizations set a maximum size for any hole or tear and for the number of times a gown can be repaired. Gowns that are recycled are subject to laundering. Gowns are washed by special washing machines with suitable detergents, dried, folded, and then wrapped in cleanroom packaging. For gowns that are to be used in aseptic areas, such gowns are irradiated. A policy should be in place outlining how often a gown can be processed -- typical times range between 20 and 40 times. To make the tracking task easier, many gowns sterilized by irradiation or gassing are fitted with barcodes and scanned. It is further important to establish the extent that the sterilization process affects the integrity of the gown material. In order to assess the contamination risks from re-laundering, gowns are subject to particle counting. There are different ways to do this, although the most common means is the Helmke Drum particle emission test. With this, the test method simulates particle shedding of clothing under movement. The garment under test is tumbled in a rotating drum (approximately 10 revolutions per minute) to release particles from the surface of the cleanroom garment in a controlled manner. An automatic particle counter is used to sample the air within the drum to determine the average particle concentration of the air during the initial ten minutes of the test. The common standard is the IEST “Recommended Practice RP-CC003.3: Garment System Considerations for Cleanrooms and Other Controlled Environments”. An alternative measure is the Body Box test. This method simulates particle filtration and release under real wear conditions. As a consequence it measures the contamination of the cleanroom by the clothing/wearer. For this, particle counters determine the quantity of particles generated by the wearer/garment that are emitted into the chamber. CHANGe PROCeDURe Cosmetics, such as powder, rouge, eye liner, mascara, and lipstick must be banned in cleanroom environments. Jewelry, such as rings, watches, necklaces, Special edition: Cleaning Validation

35

Tim Sandle

bracelets, earrings and other items, together with all forms of visible piercing, are commonly not allowed in cleanrooms. The best method of changing into cleanroom garments is one that minimizes contamination getting onto the outside of the garments. Change areas can vary in design, but it is common to find them divided into three zones: 1. Pre-change zone. Outside of changing rooms ‘tacky mats’ or polymeric flooring can be positioned to help reduce the level of particles carried on footwear16. 2. Changing zone. The changing room design contributes to the assurance of appropriate personnel access and microbial contamination control. The changing room should be provided with filtered air. Intermediate (bag) filters will typically be suitable for this purpose, though High Efficiency Particulate Air (HEPA) filtration may be used. The air pressure should be negative with regards to the manufacturing area corridor, but positive relative to external adjacent areas. 3. Cleanroom entrance zone. This must be of the same grade or class as the main cleanroom into which the area leads. Ideally there should be separate routes through airlocks for material required in cleanrooms. Taking items through personnel change areas should be discouraged. TRAINING Personnel training in gowning is an important function. Gowning practices must be assessed periodically and monitored frequently. Training programs should ideally include visual assessment and microbiological assessment. The microbiological assessment varies, but can include the exposure of settle plates during the change process and the assessment of gown cleanliness through post-use suit contact plates. The results of the cleanroom sampling should not exceed those of the room class. If results are exceeded, the individual may be an unusually high shedder of skin particles. Training required for staff who work in cleanrooms should include: • Introduction to micro-organisms and microbiological contamination control. • Entry and exit of production facilities (including gowning). • Personal hygiene training. • Microbiological risks associated with specific production tasks. 36

Special edition: Cleaning Validation

Training must be documented and regularly reviewed. Training must be effective. Actual performance of personnel competency in gowning should be demonstrated on a regular basis. CLeANROOM PeRSONNeL BeHAVIOR Working in clean environments demands knowledge, discipline, motivation as well as a thorough understanding of contamination risks among all personnel involved. Each individual cleanroom should have its own documented rules and procedures. Training includes reminding personnel that they must not be allowed to touch critical products and equipment with their naked hands. All critical work must be undertaken wearing gloves. Critical activities requiring personnel contact such as aseptic processing or sampling must be done through the use of clean utensils such as tweezers, forceps, and the equivalent. All devices and gloves used must fully comply with the cleanliness demands of the cleanroom and work undertaken in the cleanroom. They must be cleaned, disinfected, or sterilized as appropriate for the criticality or activity and risk of contamination. Another aspect of best practice is in instructing personnel in the appropriate behaviors within the cleanroom. The generation of contamination is proportional to activity conducted. A person with head, arms, and body moving can generate about 1,000,000 particles ≥ 0.5 µm/min. A person who is walking can generate about 5,000,000 particles ≥ 0.5 µm/min. However a person in motionless position can generate only 100,000 particles ≥ 0.5 µm/min. In addition, personnel should reduce activities like talking, singing, whistling, coughing, sneezing etc., especially when being close to the handled products and production equipment. People working in cleanrooms and other forms of controlled environments must be physically healthy. Diseases in the upper respiratory tract as well as stomach disorders can create problem in hygienic applications. Another factor that can impact upon the environment is the number of people in the cleanroom. Only necessary and limited number of persons should be allowed in a cleanroom at the same time. The more persons simultaneously present in a cleanroom, then the higher the contamination level will be, i.e., the higher concentration of particles in the air). This is particularly important in relation to changing rooms. HAND SANITIZATION Good personal hygiene is a requirement of all pharmaceutical cleanroom activities. However, studies show poor compliance is common in relation to basic hand washing technique. Hand hygiene and glove

Tim Sandle

Figure 1: HAND SANITIZATION (IMAGE: TIM SANDLE)

hygiene are important given the high numbers of microorganisms found on the human body including the hands and the risks posed by hands as a means of contamination transfer. Microorganisms associated with hands are found mainly on the surface of the skin and under the superficial cells of the stratum corneum. The dominant species is Staphylococcus epidermidis that is found on almost every hand, together with other species of Staphylococcus and species of the genera Micrococcus17. Hands must be washed with soap and water prior to entry to the cleanroom. Hand washing facilities should not be located in an actual cleanroom, but rather in an area leading to the cleanroom changing room. As an alternative, a hygienic handrub can be used. Where gloves are required these should be put on using a method designed to prevent the ungloved hand from touching the clean or sterile outer part of the glove. Once in the cleanroom, gloved hands should be subject to periodic hand sanitization18. When decontaminating hands with an alcoholbased antiseptic hand rub, apply product to palm of one hand and rub hands together, covering all surfaces of hands, fingers and wrists, until hands are dry (alcohol-based hand rubs are not to be used with water). The process typically takes between thirty seconds and one minute. Follow the manufacturer’s recommendations regarding the volume of product to use. The technique for applying alcohol to gloved hands is similar to applying a handrub to skin. It is important to ensure that all surfaces are covered. With glove sanitization, there are two alcohols of choice: ethyl alcohol (ethanol) and isopropyl alcohol (IPA). Other alcohols, such as methyl alcohol (methanol) are unsuitable19. Of the two alcohol forms, IPA is slightly more bactericidal than ethanol, although ethanol has better viricidal properties20. Another factor is application to the skin, and here IPA can be quite harsh.

Thus, ethanol is more often applied to bare skin (often in a denatured form) whereas IPA is more often applied to gloves. These sanitizers have bactericidal action against vegetative cells but not spores. The concentration of alcohol to water varies, although the optimal range is 60 to 90% (volume/volume). Below 60%, bactericidal action drops, and above 90% there is insufficient water for the bacterial cell to absorb water. The alcohol does not enter the cell and is unable to denature the bacterial proteins21. Most preparatory concentrations are 70%. While most bacteria are killed after ten seconds of contact with alcohol22, contact times in practice are longer due to the variability of hand rubbing. Typical contact times are thirty seconds. It is important that the selection of a hand sanitizer is qualified. There are different approaches that can be taken for qualification. Most of these require individuals to wear gloves and to place their hands into broth containing a high concentration of a nonpathogenic microorganism. Disinfectant is then applied, and the bacterial reduction is assessed through placing the treated hands into broth and performing dilutions. ON-GOING ASSeSSMeNT In higher-grade cleanrooms such as those used for aseptic processing, it is common practice to assess the risk of personnel to the process by taking suit contact plates of the gown as worn by the person as they leave the aseptic area. The gown must be discarded after the plates have been taken due to the potential effect on the gown integrity when an agar plate contacts the gown. It is good practice to begin suit sampling with a higher number of samples. These can then be reduced over time. Some facilities perform more samples from the gown during gowning test qualifications compared with routine sampling. Sites considered for selection include the top of the head, the face mask, both arms, middle torso, and both legs. In terms of limits, for EU GMP Grade B/ISO class 7 areas, the aim is often to adopt the same limits as per the limits applied to finger plates. The action level for gowns is ordinarily 5 CU/25cm 2. Experience has shown that higher counts are obtained from the top of the head, perhaps because this is the warmest region of the body. Care must be undertaken when sampling as so not to break the integrity of the gown. In addition to gowning control, a procedure should be in place for the notification of health conditions by staff. Staff who are ill (coughs, colds, and so on) should not enter cleanrooms. This is because the Special edition: Cleaning Validation

37

Tim Sandle

illness may affect product quality. It is important to control the potential risks from personnel carrying • Infectious disease. • Open lesions on any exposed part of body. • Shedding skin conditions, such as eczema or psoriasis, dermatitis, and dandruff (skin scales may harbor objectionable micro-organisms that may impact pharmaceutical products and patients). • Gastric upsets. Personnel with any of the above conditions must be excluded from working within cleanrooms for the duration of their illness. CONCLUSION This paper has considered the personnel factor and the relationship between people and cleanrooms. It addressed why people are a risk in relation to the skin microbiome, and how good gowning practices can help to minimize that risk. The paper has also considered other factors that can affect contamination risks from people, including the importance of good behaviors and the necessity of hand sanitization. Capturing these various issues through procedures and imparting the key concepts through training are a necessary part of cleanroom management. ReFeReNCeS 1. Reinmüller, B. (2001). “People as a Contamination Source - Clothing Systems”. In: Dispersion and Risk Assessment of Airborne Contaminants in Pharmaceutical Cleanrooms. Royal Institute of Technology, Building Services Engineering, Bulletin No. 56, Stockholm, pp. 54-77 2. Hyde, W. (1998). Origin of bacteria in the clean room and their growth requirements. PDA J Sci Technol; 52:154–164 3. Sandle, T. (2011). A Review of Cleanroom Microflora: Types, Trends, and Patterns, PDA Journal of Pharmaceutical Science and Technology, 65 (4): 392-403 4. Proksch E.; Brandner J.M.; Jensen J.M. (2008) The Skin: An Indispensable Barrier, Exp. Dermatol. 17 (12): 1063–72 5. Whyte, W. (1981) Setting and impaction of particles into containers in manufacturing pharmacies, J. Paren. Sci. Technol., 36: 255-268 6. Roth, R.R. and James, W.D. (1988): Microbial Ecology of the Skin. Annu. Rev. Microbiol. Vol. 42, pp. 441–64 7. Grice, E.A., Kong, H.H., Renaud, G., Young, A.C. (2008). A diversity profile of the human skin microbiota. Genome Research.18:1043-50. (PMID:18502944)

38

Special edition: Cleaning Validation

8. Costello, E.K., Lauber, C. L., Hamady, M., Fierer, N., Gordon, J.I., Knight, R. (2009). Bacterial community variation in human body habitats across space and time, Science, 326: 1694–1697 9. Chen YE, Tsao H. (2013). The skin microbiome: current perspectives and future challenges, J Am Acad Dermatol. 69(1):143-55 10. Grice, E.A., Kong, H.H., Conlan, S. et al (2009). Topographical and Temporal Diversity of the Human Skin Microbiome, Science, 324: 1190 – 1192 11. Gao, Z., Tseng, C.H., Pei, Z., and Blaser, M.J. (2007). Molecular analysis of human forearm superficial skin bacterial biota. Proc. Natl. Acad. Sci. 104: 2927–2932 12. Kong, H.H. and Segre, J.A. (2012). Skin Microbiome: Looking Back to Move Forward, Journal of Investigative Dermatology, 132: 933–939 13. Findley, K., Oh, J. Yang, J., Conlan, S. et al (2013). Topographic diversity of fungal and bacterial communities in human skin, Nature, 498(7454):367-70. doi:10.1038/nature12171 14. Cogen A.L, Nizet, V. and Gallo, R.L. (2008): Skin Microbiota: A Source of Disease or Defence?, Br. J. Dermatol., 158 (3): 44255 15. Ramstorp, M. (2011) Microbial Contamination Control in Pharmaceutical Manufacturing. In Saghee, M.R., Sandle, T. and Tidswell, E. (Eds.) Microbiology and Sterility Assurance in Pharmaceuticals and Medical Devices, Business Horizons: New Delhi, pp615-701 16. Sandle, T. (2006) The use of polymeric flooring to reduce contamination in a cleanroom changing area, European Journal of Parenteral and Pharmaceutical Sciences, 11 (3): 7 17. Kampf, G., Kramer, A. (2004). Epidemiologic Background of Hand Hygiene and Evaluation of the Most Important Agents for Scrubs and Rubs, Clinical Microbiology Review, p. 863–893 18. Sutton. S., (2009). Hand Washing-A Critical Aspect of Personal Hygiene in Pharma, Journal of Validation Technology, 15 (4): 50-55 19. Spaudling, E.H. (1964) Alcohol as a surgical disinfectant, AORN J., 2: 67-71 20. Klein, M. and DeForest, A. (1963) The inactivation of viruses by germicides, Chem. Specialist Manufact. Assoc. proc., 49: 116118 21. Morton, H.E. (1950) The relationship of concentration and germicidal efficiency of ethyl alcohol, Ann N.Y Acad. Sci, 50: 191-196 22. Morton, H.E. (1983) Alcohols. In Block, S.S. (Ed.) Disinfection, Sterilisation and Preservation, Philadelphia: Lea and Febiger, pp225-239

Xxxxxxx

PAT: Using PAT to Support the Transition from Cleaning Process Validation to Continued Cleaning Process Verification | IVT XXXXXX

ABSTRACT In accordance with the 2011 FDA guidance for industry, Process Validation General Principles and Practices, there is a requirement to continue to verify equipment cleaning processes. This continued verification would reduce the risk of cross contamination, and consequently, the acceptable residue limits could be increased to reflect this reduced risk. A combination of visual residue limits and rapid analytical technologies for the quantification of residues may lead to a reduced risk of cross contamination for the patient while at the same time result in significantly reduced manufacturing cycle times. This paper discusses the current cleaning paradigm and the changes continued cleaning process verification could bring for small molecule pharmaceutical manufacturers. Technologies that offer potential for increased automation of cleaning validation are introduced. INTRODUCTION Cleaning can generally be defined as the removal of unwanted contaminants. The objective of a cleaning process is primarily to ensure safety, efficacy, and quality of product subsequently manufactured in the same equipment. Cleaning validation is the documented evidence demonstrating the effectiveness of a cleaning procedure based on pre-determined acceptance criteria. It is performed to assure that production materials that come into contact with manufacturing equipment surfaces are not contaminated or adulterated. A graphical representation of the typical cleaning process development and validation process is detailed in Figure 1. It summarizes the typical process and demonstrates the extensive workload associated with cleaning process development and validation. Figures 2 and 3 summarize the current cleaning procedure development and validation approach. This approach is time-consuming and costly. It can significantly inhibit clinical trial manufacture activities because of once-off swabbing requirements following each mini-batch manufactured. It can also inhibit on-going commercial manufacturing activities due to the burden associated with cleaning procedure validation and routine monitoring. Within the FDA guidance document on process validation (1), process validation is defined as the collection and evaluation of data, from the process design stage throughout production, which establishes scientific evidence that a process is capable of consistently delivering quality products. Process validation involves a series of activities taking Special edition: Cleaning Validation

39

Xxxxxxx

Figure 3: Current cleaning procedure validation process flow. Figure 1: Typical cleaning process development and validation process.

Figure 2: Current cleaning procedure development process flow.

place over the lifecycle of the product and process. The guidance describes the process validation activities in three stages: process design, process qualification, and continued process verification. While the greatest focus of this guidance documentation has been related to manufacturing process validation, the same three stages of validation have relevance to cleaning process validation. It is likely that the demonstration of continued cleaning process verification is expected by regulatory bodies in order to demonstrate sufficient control of the cleaning process. With the increased frequency of verification associated with such a philosophy, it is logical that the level 40 Special edition: Cleaning Validation

of risk of cross-contamination of subsequent formulations would reduce accordingly. As a result, the acceptable residue limits (ARLs) currently associated with cleaning validation may no longer be appropriate. This is because the safety factors built into such calculations may have been included as a result of the “snapshot-in-time” nature of a cleaning validation activity. VISIBLE-RESIDUE LIMITS DURING CLEANING PROCEDURE DEVELOPMENT AND VALIDATION A visible-residue limit (VRL) is the quantity of target residue (usually API) remaining on manufacturing equipment surfaces when it has reached a visually detectable level. An ARL is the amount of target residue (usually API) that remains on manufacturing equipment surfaces and carry over to the next formulation with no pharmacological or adulteration risk. This is typically calculated using 10ppm or 1/1000th minimum daily dose as outlined in Figure 1. With a shift to continued verification of a cleaning procedure following every clean and the reduction of cross-contamination risk, the associated ARLs will increase, leading to greater numbers of ARLs exceeding VRLs. This may subsequently enable greater use of a VRL approach to continued verification. The use of visual inspection as an assessment criterion for equipment cleaning effectiveness has always been a component of cleaning programs. Mendenhall proposed the use of only visual examination to assess equipment cleanliness in 1989 (2). He summarised that visible cleanliness criteria were more rigid than quantitative calculations and, therefore, adequate. FDA acknowledges the use of visually clean criteria for product-dedicated equipment. LeBlanc has also reviewed the use of visual examination as the sole acceptance criteria

Xxxxxxx

development, process qualification, and continued verification. Considerable quality assurance, time and resource, and financial benefits may be realized from adopting such an approach.

Table 1: Advantages and limitations of visible-residue limits (VRLs) during cleaning procedure development and validation.

ALTERNATIVE TECHNOLOGIES FOR CONTINUED VERIFICATION OF Some examples of alternative technologies currently employed or that show potential for continued verification of equipment cleaning processes either as lab, at-line, or in-line approaches are detailed as follows. Lab-based Technologies Lab-based technologies include ion-mobility spectrometry (IMS), total organic carbon (TOC), and liquid chromatography-mass spectrometry-mass spectrometry (LC-MS-MS).

Figure 4: A process flow of proposed cleaning procedure development and validation activities.

for cleaning validation (3). Forsyth et al. (4, 5) propose that VRLs could be adopted as the primary acceptance criteria during cleaning evaluation activities if the VRL is calculated to be less than the ARL. They cited the following advantages and limitations detailed in the Table. As is outlined in the Table, there are still vulnerabilities of such an approach, in particular the personto-person variability of “visually clean” equipment and the variable lighting conditions associated with commercial-scale equipment. Even with considerable training and experience these risks may remain. There is an opportunity to utilize rapid at-line and on-line automated technologies to compensate for the deficiencies within the VRL approach. A combination of both approaches (i.e., VRLs and automated residue detection systems) could support the pharmaceutical manufacturing sector to realize a significantly robust cleaning process aligned with reduced manufacturing cycle times. This proposed paradigm is outlined in Figure 4. Figure 4 provides an overview of the proposed alternative approach to cleaning procedure

Ion-mobility spectrometry. IMS is a separation technique, similar to time-of-flight mass spectrometry, that distinguishes ions of a given compound based on their velocities through a drift tube under the influence of a weak electric field. To perform the IMS analysis, a sample solution containing a compound of interest is injected by an autosampler onto a polytetrafluoroethylene (PTFE) substrate and allowed to dry. The sample is introduced by heating the substrate to circa 290°C, which results in desorption or vaporization of the sample into an inlet tube (6). Primary ion formation occurs through atmospheric pressure chemical ionisation (APCI) using nickel-63 as the radioactive source. Following many collisions, product ions are formed and gated into the drift tube (6). These ions based on size, shape, and charge travel through the drift tube toward the detector at different velocities. In contrast to mass spectrometers, separation of ions is based on a size-charge relationship rather than a mass-charge ration. It is a fast and specific off-line tool for verifying the cleanliness of pharmaceutical equipment. Recovery percentages and standard deviations for IMS samples are consistent with those obtained with HPLC analysis, but sample throughput of IMS is about 50 times faster (6). The disadvantages associated with such a method are that it is generally an off-line analytical tool that requires sample swabbing and preparation activities. In addition, significant method development time is required prior to implementation. However online, real time applications are becoming available. Total organic carbon. TOC analysis is a method that has been used successfully for monitoring water quality, in particular the quality of water for injection used in pharmaceutical manufacturing. It offers a number of distinct advantages over other commonly used residual testing methods. A typical method may Special edition: Cleaning Validation

41

Xxxxxxx

involve acidification of a swab sample followed by sparging with purified air to remove inorganic carbon from solution as carbon dioxide gas. The organic compounds remaining in solution are then oxidized to carbon dioxide in, for example, a combustion tube packed with a catalyst at 680°C (7). The carbon dioxide produced in this manner passes through a dehumidifier and halogen scrubber before it reaches a cell where it is quantified by a non-dispersive infrared detector tuned to the highly specific carbonyl frequency in the 1-700cm-1 range (7). The signal response is compared with a calibration curve generated by analysis of solutions of potassium biphthalate corresponding to known levels of organic carbon. The disadvantages associated with such a method are that it is non-specific and an at-line analytical tool that requires sample swabbing and preparation activities (7). LC-MS-MS. For low-dose compounds, equipment requiring low residue limits, and compounds lacking strong chromophores, the sensitivity and selectivity of LC-MS-MS facilitates rapid detection of low levels of residues of active pharmaceutical ingredients (8). Like IMS, the disadvantages associated with LCMS-MS are that it is an off-line analytical tool that requires significant method development, sample swabbing, and preparation activities. At-line Technologies ATP luminescence. Adenosine triphosphate (ATP) luminescence-based technologies may be used to evaluate the level of microbial contamination following the completion of an equipment cleaning process. ATP is a marker for cell viability because it is present in all metabolically active cells where the concentration declines rapidly when the cells undergo necrosis or apoptosis. The approach is based on the production of light caused by the reaction of ATP with added luciferase and D-luciferin (9-11). The emitted light is proportional to the ATP concentration within certain limits. The limitation associated with common luciferase assay technology is the short half-life of the light emission. This flash-type signal requires luminometers with reagent injectors to measure the quick reaction. Prior to addition of luciferase, there is a requirement to lyse the cells so that the ATP is released. This is a rapid method of swab analysis that is accepted within the food and beverage industries. It may have application within sterile and aseptic operations in particular. However, it is a non-specific method and has the potential for contamination by human somatic cells if sample-handling procedures 42

Special edition: Cleaning Validation

are not sufficiently rigorous (9-11). 0n-line Technologies UV-Vis spectrophotometry. UV-Vis technologies monitor rinse effluent to quantify the level of API contained. UV-Vis spectroscopy is the measurement of the wavelength and intensity of absorption of near-ultraviolet and visible light by a sample. Ultraviolet and visible light are energetic enough to promote outer electrons to higher energy levels. UV-Vis spectroscopy is usually applied to molecules and inorganic ions or complexes in solution. The UV-Vis spectra have broad features that are of limited use for sample identification but are useful for quantitative measurements. The concentration of an analyte in solution can be determined by measuring the absorbance at some wavelength and applying the Beer-Lambert Law. The light source is usually a hydrogen or deuterium lamp for UV measurements and a tungsten lamp for visible measurements. The wavelengths of these continuous light sources are selected with a wavelength separator such as a prism or grating monochromator. Spectra are obtained by scanning the wavelength separator and quantitative measurements can be made from a spectrum or at a single wavelength. When the level of API in the rinse effluent reaches a pre-determined acceptable low level, the cleaning process end-point is reached. This is a novel approach that has significant benefits for identification of cleaning process endpoint. One disadvantage identified would be the challenges of wavelength accuracy. Wavelength accuracy is defined as the deviation of the wavelength reading at an absorption band or emission band from the known wavelength of the band. The wavelength deviation can cause errors in the qualitative and quantitative results of the UV-Vis measurement, thereby affecting the accuracy and the sensitivity of the method (12). A second disadvantage, similar to other rinse studies, is that it may not detect API if it is adhered to the vessel wall. Single point near infrared (NIR). In situ infrared reflection-absorption spectroscopy (IRRAS) has the potential to provide a more rapid analysis than the existing methods. Modern Fourier Transform-Infrared (FT-IR) instruments are quite stable and have sufficient sensitivity. Infrared fibreoptics based on chalcogenide glasses can allow sampling up to several meters from the spectrometer (13-16). Multivariate chemometric tools, such as partial least squares (PLS) regression, allow implicit modelling of unquantified interfering species and instrumental artifacts. For cleaning validation, the physical presentation of the contamination will depend on the solvent, the chemi-

Xxxxxxx

cal nature and quantity of the residual material, and the substrate surface. Mehta et al. (14), Hamilton et al. (14), and Teelucksingh et al. (16) demonstrated that IRRAS calibrations can be established for individual surfactants and APIs on metallic and glass surfaces, for an API in the presence of a surfactant on a glass surface, and for an API in the presence of excipients on a stainless steel surface (13-16). Single point NIR has potential applications for cleaning verification purposes. Its main advantage over traditional swabHPLC techniques is that it samples those contaminants that are present on the surface, rather than those that can be removed by the swab coupled with its speed of analysis. The primary limitation associated with the method is the restricted field of analysis and whether it is sufficiently large and representative to assess the success of the cleaning activity. NIR-chemical imaging. Chemical imaging (CI) is an emerging platform technology that integrates conventional imaging and spectroscopy to attain both spatial and spectral information from an object (17). It has not to date been evaluated as a technology to quantify residual levels of active ingredients and detergents from equipment surfaces. However, based on its broad applicability and sensitivity, it has potential for cleaning validation. Near infrared-chemical imaging (NIR-CI) is the fusion of near-infrared spectroscopy and image analysis. It can be used to visualize the spatial distribution of the chemical compounds in a sample providing a chemical map of a region (Figure 5). Each sample measurement generates a hyperspectral data cube containing thousands of spectra. An important part of a NIR-CI analysis is the data

Figure 5: Schematic representation of hyperspectral imaging hypercube showing the relationship between spectral (λ) and spatial (X, Y) dimensions (17).

processing of the hyperspectral data cube (18). The large number of individual spectra acquired across the spatial dimension of heterogeneous compounds provides a basis from which relative concentrations can be determined for each spatial location. Alternatively, these individual concentrations may be added together to give the total concentration of a specific material within the sample area. Chemical images are made up of hundreds of contiguous wavebands for each spatial position of a target studied. Consequently each pixel in a chemical image contains the spectrum of that specific position. The resulting spectrum acts like a fingerprint that can be used to characterize the composition of that particular pixel. There are two basic methods to construct the chemical image. One method involves acquisition of simultaneous spectral positions. The object is moved underneath an imaging spectrograph—this is termed pushbroom acquisition. The other method involves keeping the image field of view fixed and obtaining images one wavelength after another—this is termed staring imager configuration (17). This method is of particular interest within the pharmaceutical industry for sample analysis and would be considered to pose the greatest potential for use as a cleaning verification device. This is primarily because the sample can remain stationary and the field of view would be comparable with that currently used during conventional swabbing techniques (circa 25cm2). Point-source spectroscopic assessments do not provide information on spatial distribution of different constituents. In other words, NIR and Raman spectroscopy can only provide information on a very narrow sample site. NIR-CI can provide information on a substantial sample site thereby making it applicable to analysis of equipment surfaces during cleaning validation. Challenges with regards to method sensitivity for such residual levels will need to be overcome before this technology becomes commercially viable. CONCLUSIONS Rapid technologies would remove some significant variables identified with VRLs and enable an analytical technique that could significantly reduce the time and cost associated with the cleaning procedure. They could also support the transition away from once-off cleaning validation towards continued cleaning process verification. This would be in-line with regulatory expectations and pharmaceutical manufacturing needs. The increased frequency of equipment cleaning process verification will reduce the risk of active and detergent cross-contamination and thereby enable higher acceptance criteria for active and detergent Special edition: Cleaning Validation

43

Xxxxxxx

carryover. As the pharmaceutical manufacturing industry transitions towards these continued verification philosophies, the requirement for rapid mobile analytical technologies has become essential in order to sustain robust and lean equipment cleaning processes. REFERENCES FDA, Guidance for Industry, Process Validation General Principles and Practices, Current Good Manufacturing Practices, January 2011. 2. D.W. Mendenhall, “Cleaning Validation,” Drug Development and Industrial Pharmacy 15 (13), 2105 – 2114, 1989. 3. D.A. Le Blanc, “Visually Clean’ as a Sole Acceptance Criteria for Cleaning Validation Protocols,” PDA J. Pharm Sci. And Technology, 56 (1), 31-36, 2002. 4. Richard J. Forsyth et al., “Risk-Management Assessment of Visible-Residue Limits in Cleaning Validation,” Pharm. Technol. 30, 104–114 September 2006. 5. Richard J. Forsyth, “Ruggedness of Visible Residue Limits for Cleaning Validation,” Pharm. Technol. 33, 102 – 111 March 2009. 6. Elizabeth Galella et al., “Cleaning Verification: Method Development and Validation using Ion Mobility Spectrometry,” Pharm. Technol. 33, 60 – 63 July 2009. 7. A. Strege, Terry L Stinger, Brett T. Farell and Avinash L Lagu, Biopharm International, April 1996. 8. Kevin J. Kolodsick et al., “Enhancing Drug Develoment by Applying LC-MS-MS for Cleaning Validation in Manufacturing Equipment,” Pharm. Technol. 30, 56 – 71 February 2006. 9. Kangas L., Grönroos M. and Nieminen A.L., “Bioluminescence of cellular ATP: a new method for evaluating agents in vitro,” Medical Biology, 62,338 – 343, 1984. 10. Lundin A., Hasenson M., Persson J. and Pousette A., “Estimation of biomass in growing cell lines by ATP assay,” Methods Enzymol. 133, 27 – 42, 1986. 11. Crouch S.P.M., Kozlowski R., Slater K.J. and Fletcher J., “The use of ATP bioluminescence as a measure of cell proliferation and cytotoxicity,” J. Immunol. Methods, 160, 81 – 88, 1993. 1.

44 Special edition: Cleaning Validation

12. Lam, Herman, “Performance of UV-Vis Spectrophotometers,” Laboratory Focus–Gazette Edition, Issue, 8, April 2000. 13. Hamilton et al., “Grazing-angle fiber-optic IR reflection-absorption spectrometry (IRRAS) for in situ cleaning validation,” Org. Process Res. & Devel., 9(3), 337-343, 2005. 14. Mehta, N.K.; Goenaga-Polo, J.; Hernandez-Rivera, S.P.; Hernandez, D.; Thomson, M.A.; Melling, P.J. Biopharmacology, 15, 36-71, 2002. 15. Hamilton, M.L.; Persoton, B.B.; Harland, P.W.; Williamson, B.E.; Thomson, M.A.; Melling, P.J.; Org Process Res. Dev., 9, 337-343, 2005. 16. Teelucksingh, N.; Reddy, K.B., Spectroscopy, 20, 16-20, 2005. 17. Gowen.A, O’Donnell.C.P, Cullen.P.J, Bell.S.E.J, “Recent applications of Chemical Imaging to pharmaceutical process monitoring and quality control,” European Journal of Pharmaceutics and Biopharmaceutics 69, 10–22, 2008. 18. Ravn,C. Skibsted,E. Bro, R. “Near-infrared chemical Imaging (NIR-CI) on pharmaceutical solid dosage forms-Comparing Common calibration approaches,” Journal of Pharmaceutical and Biomedical Analysis 48 554–561, 2008. JVT

ARTICLE ACRONYM LISTING APCI Atmospheric Pressure Chemical Ionisation API Active Pharmaceutical Ingredient ARLs Acceptable Residue Limits ATP Adenosine triphosphate CI Chemical Imaging FT-IR Fourier Transform-Infrared IMS Ion-Mobility Spectrometry IRRAS Infrared Reflection-Absorption Spectroscopy LC-MS-MS Liquid Chromatography-Mass Spectrometry-Mass Spectrometry NIR Near Infrared NIR-CI Near Infrared-Chemical Imaging PTFE Polytetrafluoroethylene TOC Total Organic Carbon UV Ultraviolet VRL Visible Residue Limit

Keith Bader & Kelly Scalva

Translating LaboratoryDeveloped Visual Residue Limits to Process Area Applications | IVT Keith Bader and Kelly Scalva

ABStrAct While considerable attention has been given to the development of visual residue limits (VRLs) in a laboratory setting, translating the bench scale values to the assessment of process surfaces has not yet been thoroughly assessed. However, knowledge of both the critical parameters that impact the determination of VRLs and the influence of those parameters on visual inspection can provide a framework for the development of a robust visual inspection program. Development of such a program first entails the determination of constraints imposed by equipment geometries and facility lighting. VRLs can then be determined for post-productions residues of concern, which, of course, carries its own specific challenges. Once VRLs have been determined, they cannot be immediately applied without considering certain strategic cleaning program approaches and potential sources of variability. Many factors influence how visual inspection will be conducted in a manufacturing facility. Among the most critical are inspection conditions in the facility, the condition of existing equipment surfaces, and the physical characteristics of post-production residues deposited on product contact surfaces. While many in industry embrace the importance of visible residue limits (VRLs), few have a clear pathway to translate VRLs determined in the laboratory to the manufacturing floor. The intent of this paper is to provide the background and information that will allow industry to formulate a plan of attack to practically integrate VRLs into the visual inspection program. To begin, the best starting place is to determine the constraints imposed by the manufacturing floor. EquipmEnt inSpEction conditionS Conducting a visual residue limit (VRL) determination study in the laboratory must be completed such that important process equipment parameters from the manufacturing floor are captured. To wit, the first step is characterization of the conditions on the manufacturing floor where visual inspection is put into practice. For any residue, there are three parameters that must be considered for proper translation and laboratory analysis: viewing angle, distance, and light intensity at the product contact surface. The most critical variable for visual inspection is, of course, the room lighting, so it is important to understand and characterize lighting in the facility to properly later simulate it at the bench scale. The interior task lighting generally conforms to standards set by a professional organization such as the Illuminating Engineering Society of North America (IESNA). In the past, “over-lighting” was the norm with standard luminance regularly reaching 1200 lux (1). Lighting was Special edition: Cleaning Validation

45

Keith Bader & Kelly Scalva

figure 1: Determination of Viewing Distance and Angles.

also primarily done with less efficient incandescent and full-spectrum lamps that inherently provided good color rendering. However, the advent of current sustainable design, legislated energy conservation requirements, and project engineers that target seemingly non-critical lighting designs in value engineering exercises means that manufacturers can no longer assume that room lighting conditions will be consistent or even adequate for visual inspection. Specifically, current legislation limits energy use for lighting to 1.1–2.0 w/ft, depending on the space function, forcing less experienced lighting designers to select options that may not provide equivalent performance to historical lighting choices. Indeed, Forsyth et al noted that lighting levels can vary from 400 to 1200 lux in a facility; such a range is representative of that found in most manufacturing facilities (2). Such variability is often unavoidable as facility and process layout can change over time to meet manufacturing needs, leading to shadows or reflections unaccounted for in the original design. Accordingly, it is recommended that the lighting be assessed and recorded through the use of commonly available light meters to acquire custom-tailored ranges for developmental VRL studies. The most convenient way of recording these numbers is to record them on a layout drawing of the facility. Furthermore, it is useful to note the style of 46 Special edition: Cleaning Validation

light fixtures and whether the color temperature of the lamp closely replicates sunlight. Lighting at the product contact surfaces within process vessels is a unique and involved set of measurements to acquire, as the distances and angles from which operators view the product contact surface is dependent upon available viewing ports and the differences in tank configuration, shown in Figure 1. Measurement of the distances and angles is most reliably performed in the field using a laser distance meter and clinometers; however, this information can also be determined from mechanical drawings of the equipment. As process equipment is surveyed, it is advisable to develop a baseline record of known blemishes, marks, imperfections, and visual obstructions to ensure that they can be distinguished from post-production residues. Establishing this baseline for each process tank and training inspectors on the location and appearance of these fingerprint imperfections are critical to prevent the development of a viewing bias that may impact the care used when inspecting process equipment surfaces. Characterized and appropriately restricted viewing angles developed during the survey will allow for better translation of the VRL from the laboratory to the floor. Conditions in areas where manually cleaned equipment or small parts are inspected after they are cleaned out of place are important in that many pieces of manufacturing equipment used for closed processing operations do not allow for reliance on ambient lighting and require the specification of an independent light source for visual inspection. Accordingly, it is important to determine the maximum inspection distance based on the size and configuration of vessels and other closed processing equipment. Once this is determined, it can be used not only as a selection criterion for a light source but also as one of the constraints for conducting laboratory VRL studies. thE SElEction of A light SourcE for ViSuAl inSpEction Before conducting the laboratory study or sending technicians into the field to inspect equipment, it is important to specify a hand-held light source. Specifying a hand-held or tank-mounted light source carries some additional considerations. This requires understanding some lighting design terminology and heuristics. The hand-held light should be selected with three primary characteristics taken into consideration: color temperature, color rendering index (CRI), and light output or luminous flux. Color temperature, or chromaticity, refers to the color of the light emitted from the light source. Colors similar to the warm yellow orange of a candle flame

Keith Bader & Kelly Scalva

Source

CRI

Chromaticity (°K)

Incandescent

100

2800K

Halogen

100

2900K

Fluorescent

62–90

2700–6000K

Mercury

15

5700K

High Pressure Sodium

22

2000K

Metal Halide

65–70

2700–4000K

Pulse Start Metal Halide 65–75

3000–4000K

are assigned a temperature of 1800 Kelvin; up the scale is the bluish light cast by metal halide and the fluorescent light sources that are assigned a color temperature of 4200 Kelvin (3). The ability of observers to accurately perceive colors is also somewhat dependent on the chromaticity of the light source based on the second characteristic noted above, CRI. CRI is quantified on a scale that assigns the light source a rating from 0 to 100, with the best rendition of color achieved with a CRI of 100. For detail-oriented physical inspection work and accurate evaluation of residues on process equipment surfaces, the CRI of a light source should, at a minimum, be at a value of 65 or greater (4). The general relationship between color temperature, CRI, and various light sources is shown in the Table. Finally, the light output of the source should be considered. Many are also tempted to get the brightest light they can find; however, this can lead to problems with detecting residues because excessive glare can actually inhibit detection. The inspection of process vessels is thus facilitated by selecting a tank light or handheld light source that replicates levels of lighting equivalent to the recommended for ambient task lighting within the vessel. Most light manufacturers, however, do not provide the luminous intensity at various distances from the source. Handheld light output, when rated, is often quantified using lumens out-of-the-front (OTF). On occasion, a specific intensity may be provided in Lux (lumen/m 2) for some lamps at a specific distance. To convert intensity at a given distance to one matching the distances determined for the production equipment in the field, it is important to know that light intensity follows an inverse square relationship relative to distance, as shown in Figure 2. As the distance from the light source increases, the light must diffuse over a larger surface thereby reducing the surface brightness in accordance with a “one over d squared” relationship. Alternatively, an inspection light of appropriate intensity can be determined empirically in the laboratory by using a light meter at the appropriate distance. For the most part, an acceptable CRI and chromaticity are provided in handheld light sources by either incandescent or halogen

figure 2: Light Intensity vs. Distance.

lamps as shown in the Table. Furthermore, a flashlight with an adjustable spot size can also be used to vary the intensity and provide greater flexibility over a range of distances. lABorAtory Vrl dEtErminAtion StudiES Translating the angle, distance, and lighting parameters from the onsite equipment to the confines of the laboratory is best examined within a range of values. Most commonly tested in the lab are three angles (30°, 45°, 90°), two to three distances (closest and farthest away), a minimum of two different lighting intensities (standard room task lighting and the intensity from a hand-held light source), and a variety of materials of construction (MOC). Assuming the viewing variables were tightly and accurately measured at the manufacturing site, the only other test variables to be investigated in the laboratory are the process soils (in varying concentrations), materials of construction, and the assessment operators. As the potential variability in technicians is the most difficult parameter to control and replicate, they are discussed in greater detail in a subsequent section. ExpErimEntAl dEtAilS And logiSticS First, coupons composed of materials of construction representative of process equipment should be obtained (see Figure 3) and cleaned thoroughly to remove any possible residue on the coupons; such as dust, mechanical cutting oils; or, if the coupons are not new, residue from previous tests. Typically, a cleaning regimen consistent with United States Pharmacopeia (USP) for cleaning glassware employed for total organic carbon analysis will effectively remove materials that can potentially confound results. Cleanliness can then be confirmed using the American Society for Testing and Materials (ASTM) water break free test (5, 6). Clean coupons are then rigorously inspected for visual cleanliness, surface imperfections, water spots, discoloration, or oxidation before use in the VRL Study. Coupons that appear to have any noticeable imperfections are rejected from the study. Special edition: Cleaning Validation

47

Keith Bader & Kelly Scalva

figure 4: Determining the Visibility Inflection Point.

figure 3: Coupons of Representative MOCs

Following a visual inspection of the coupons to ensure their cleanliness, the process soils of interest are diluted into a range of concentrations. These concentration ranges are then spiked with equivalent volumes onto the center of the coupon surface and allowed to dry. Coupons commonly are dried at ambient conditions overnight to ensure dryness, but they can be dried at elevated temperatures depending on process conditions. A visual target range is then determined to find the appropriate loading density and loading solvent for use in VRL testing. Purified water is the common solvent used to make the residue dilutions; however, surface tension effects combined with some process soils can create a very visible ring around the periphery of the original spiking area. As water dries at the edges of the spiked solutions, it concentrates the residue, making them more visible and quantification of the mass per unit surface area more difficult. This concentrates the residue to a greater degree in the laboratory setting than that found on process equipment. Thus, it is sometimes nessary to create a dilution mixture that 48

Special edition: Cleaning Validation

includes a mixture of residue, purified water, and isopropyl alcohol (IPA) to reduce the solution surface tension and more uniformly distribute the residue over the spiking area as the liquids dry, thereby preventing concentration around the outer perimeter or center of the spiking area. The exact ratios of IPA to water must be determined on a per soil basis as well as the solubility because each residue is different. Furthermore, chemical interactions or protein denaturation that alters the visual characteristics of the soil should also be considered. Once drying effects have been effectively contended with, the threshold concentration of each process soil must be determined before multiple technicians are screened. Specifically, the approximate concentration range visibility inflection point, when a residue goes from visible to not visible, can be determined, as shown in Figure 4. Creating the initial spiking solutions in order of magnitude intervals works well to initially narrow the range. For example, a serial dilution of 1 ml residue solution in 10 ml solvent followed by 1 mL of the second solution in 10 mL of water, and so on, allows a simple way to determine the range of concentrations to be used for the VRL intermediate precision testing assessment. To prepare the coupons for testing, a consistent volume of spiking solution is then spiked onto coupons over a reproducible surface area. The concentrations applied to the coupons span the range of Visible, Slightly Visible, and Not Visible as determined in the initial bulk VRL screening exercise described above. At least three replicates of each concentration per material of construction should be used to ensure statistical significance within the results. Four operators with 20/20 or corrected 20/20 vision are then selected to establish the VRL intermediate precision testing at different viewing angles and distances under ambient and additional hand-held lighting. To better simulate visual inspection conditions on the manufacturing floor, it is important that the coupons are placed on a background made of visually similar material. The visually similar material reduces operator bias during the laboratory study so that contrast from the viewing background will not be an issue. Thus, when testing the viewing of different materials of construction, the background for these assessments must either be constructed of the same material as the coupon or must be consistent with the

Keith Bader & Kelly Scalva

in the lab will not have the geometric restrictions imposed by ancillary equipment, viewing ports, and manways, thereby necessitating artificial limitations on the amount of movement they may instinctively employ as they try to observe spotted coupons. A representative arrangement of coupons is shown in Figure 5.

figure 5: VRL Determination Conditions Based on Manufacturing Area Assessment.

appropriate viewing parameters employed for the process equipment. For example, if the major material of construction is 316L SS EP20Ra with minor materials such as borosilicate glass (commonly site glasses) and/or PTFE (primarily tested to simulate valves), the minor materials should have viewing backgrounds equivalent or comparable to the major material of construction. For example, 316L, 316, and 304 stainless steel would be visually comparable materials. If glass were the major material of construction, matching the background could also be somewhat complex since interactions between process equipments and the manufacturing area floors or walls could lead to a more difficult situation to emulate in the laboratory. Other variables to consider are the impact of personal protective equipment (PPE) worn in the manufacturing space, such as goggles, safety glasses, laboratory coats, coveralls, or scrubs. The spiked coupons are then arranged in a viewing area with controlled lighting and background. To control for variations in visual acuity, operators were screened to ensure that they had either corrected or uncorrected 20/20 vision. Those without 20/20 vision were excluded from the study. Screening was conducted by first asking each operator if they had had a professional eye exam within the last twelve months. If so, the operator was asked to view two different Snellen eye charts from distances of 6 ft. and 10 ft., respectively. A passing result for this portion of the test required that the operator could correctly recite the line on the chart corresponding to 20/20 without reading any of the letters incorrectly. This practice is recommended to qualify technicians both in the laboratory and on the manufacturing floor. When transferring the measured angles and distances to the laboratory, keep in mind that technicians

rEcommEndAtionS for tESting In any quality system, the integration of qualityby-design principles starts with the education of all personnel involved in the production of a therapeutic substance. Specifically, when developing training materials for visual inspection of the process equipment, operators should be informed of the importance of the activity as an important orthogonal method for the prevention cross contamination. This will help prevent technicians from treating the activity as “just another task” and rushing to complete it at the potential peril of patients receiving a particular therapeutic compound. In the laboratory setting, the study was most influenced by the level of training of the visual assessment operators. For example, operators who had experience cleaning coupons in the lab were much more adept at dismissing imperfections on the surface of coupons that may have otherwise confused those who had never before conducted visual inspection or VRL testing with residues. Not only is it important for operators to know and understand why they are completing a task, but it is just as important for the operators to have the chance to study a dirty vessel directly after a process run and at the end of the dirty hold time in addition to having historic areas of concern and vessel imperfections pointed out to them. A well-educated operator trained to recognize the appearance of process equipment at both its worst and best relative to cleanliness is much more likely to properly ascertain the presence of residues on product contact areas. Specifically, operators should also have the opportunity during training to visually inspect the vessel directly after the cleaning cycle. If possible, they should also be able to observe the equipment in a clean, wet state and a completely dry state since residue is typically better seen on dry process surfaces. Knowledge of how the vessel appears when the surfaces are covered in process residues typically makes the cleanliness easier to assess if an idea of the range of cleanliness is known. Bias can also be avoided in the laboratory evaluation by arranging the coupons both in a random loading concentration order and in rows of, or greater than, five coupons to prevent operators from memorizing which coupons they identified as soiled under a previous lighting, distance, and viewing angle combination. Based on the authors’ experience, more Special edition: Cleaning Validation

49

Keith Bader & Kelly Scalva

consistent results are also obtained if the operators verbally relay whether the residue is visible rather than be required to fill out protocol worksheets. Overall, it was found that slight movement of the flashlight was necessary for the operators to attempt to see the residues; directing the light towards the coupon led to excessive glare that hindered the ability of the operator to view residue. Residues were typically noted when they were on the periphery of the hand-held light source spot. It was also noted that operators who viewed the surfaces longer generally saw more residue than operators who quickly glanced at the coupons. However, with the use of the flashlight, operator-viewing duration increased by about 50%. Interestingly enough, visibility and detection of dirty coupons decreased when operators used the hand-held light over diffuse ambient lighting. While the procedures described above may seem fairly complete and extensive, there are still many related topical permutations warranting further investigation. For example, most studies, to date, focus on the intact API; however, cleaning processes employed for biopharmaceuticals often employ alkaline clean-

50

Special edition: Cleaning Validation

ing agents that degrade post-production residues. As with any chemical change to a compound, the physical properties can change, including visibility of the compound on process surfaces. Future studies will investigate this and other similar considerations. rEfErEncES 1.

2.

3.

3.

4.

E. Teicholz, “LIGHTING” in Facility Design and Management Handbook, (McGraw-Hill Professional, 2001), AccessEngineering. R.J. Forsyth, V. Van Nostrand, and G. Martin, “Visible-Residue Limit for Cleaning Validation and its Potential Application in a Pharmaceutical Research Facility,” Pharmaceutical Technology 28 (10), 58–72, 2004. T. Croft, W.I. Summers, F.P. Hartwell, “PRINCIPLES AND UNITS” in American Electricians’ Handbook, Fifteenth Edition, (McGraw-Hill Professional, 15th ed., 2009 2002 1996 1992 1987 1981 1970 1961). R.C. Rosaler, “Lighting” in Standard Handbook of Plant Engineering, Third Edition (McGraw-Hill Professional, 3rd ed., 2002 1995 1983) USP29–NF24 , Cleaning Glass Apparatus, 2896

Rizwan Sharnez

Cleaning Validation of Multiproduct Equipment— Acceptance Limits for Inactivated Product Part II—Application of the Comparable Quality Approach to Intrasite Assessments Rizwan Sharnez, Elizabeth Aisenbrey, Joel Bercu, David Binkley, and Arun Tholudur “New Perspectives on Cleaning” is an ongoing series of articles dedicated to cleaning-process development, validation, and monitoring. This column addresses scientific principles, strategies, and approaches associated with cleaning that are relevant in everyday work situations.   Reader questions, comments, and suggestions are requested for future discussion topics. These can be submitted to Rizwan Sharnez at rsharnez@ amgen.com.

SUMMARY For multiproduct cleaning validation, acceptable carryover of the previously manufactured active pharmaceutical ingredient (API) (Product A) into the subsequently manufactured API (Product B) is determined through a maximum allowable carryover (MAC) calculation (1). A limitation of the conventional MAC approach is that if the previously manufactured API becomes therapeutically inactive during cleaning, then there is limited value in verifying removal of the API. Instead, it is more appropriate to demonstrate that the carryover of the inactivated product between lots of different products (i.e., A → B, or B → A) is acceptable. If the API is inactivated when exposed to worst-case cleaning conditions (i.e., conditions that are least conducive for inactivation), the comparable quality (CQ) approach can be used to set acceptance limits for carryover of the inactivated product (IP) (2). The CQ approach is designed to ensure that the amount of inactivated Product A in the largest dose (LD) of Product B (M LD IP A→B ) is less than or equal to the amount of inactivated Product A in the largest dose of Product A (M LD IP A→A), M LD IP A→B



M LD IP A→A

[Equation 1]

To apply the CQ criterion to cleaning validation, the above inequality must be expressed in terms of measureable equipment and product parameters. An approach for expressing this criterion in terms of analytical data (e.g., total organic carbon [TOC] swab or rinse data), product contact surface area, and batch and dose sizes is described in this paper. The resulting expression for the CQ criterion for intrasite assessments (i.e., where Product A is being manufactured at the same site where Product B is being manufactured) is given by Equation 2. When this inequality is satisfied, the amount of inactivated Product A in the largest dose of Product B will be limited to the amount of inactivated Product A in the largest dose of Product A.

Special edition: Cleaning Validation

51

Rizwan Sharnez

If the characterization data indicate that the previously manufactured API (i.e., API in Product A) is inactivated when exposed to cleaning conditions, and the CQ criterion is met, then the acceptance limits for the process residue can be set based on the lowest of the following two limits: process capability and 10 ppm limit for carryover of process residue. A flowchart summarizing the methodology described in this paper is shown in Figure 1. The CQ criterion is derived in Appendix I and a worked example is given in Appendix II. The CQ approach was described in Part I of this series. In Part II, the CQ approach is applied to intrasite assessments; application of the CQ approach to intersite transfers is described in Part III. INTRODUCTION Biopharmaceutical cleaning cycles are generally designed to expose product contact equipment to extremes of pH (pH 2–13) and temperature (60-80°C) for several minutes. The equipment may also be steam sterilized or sanitized. Under these conditions, monoclonal antibodies, therapeutic proteins, and other biological APIs are known to degrade and denature rapidly, and become therapeutically inactive (2-4). Inactivation Studies Inactivation of the API during cleaning can be assessed by exposing the process soil to simulated cleaning conditions at bench scale. The inactivation studies are designed to simulate full-scale cleaning conditions that are least conducive (i.e., worst case) for inactivation (e.g., lowest ratio of cleaning solution to protein and shortest exposure). The objective of these studies is to ascertain whether the API in the process sample is inactivated when exposed to worst-case cleaning conditions. After exposure to the worst-case cleaning conditions, the samples are titrated to a neutral pH and cooled to 4°C to minimize further degradation and inactivation. Appropriate untreated controls are included to assess the impact of any experimental artifacts and potential matrix effects. The samples are then subjected to SDS-PAGE and bioassay to determine the degree of degradation and inactivation, respectively. The samples may also be subjected to TOC analysis to determine whether the inactivated product can be adequately recovered by TOC. The results of the small-scale studies can justifiably be extrapolated to the full-scale cleaning process, because degradation and inactivation are essentially scale-independent phenomena (i.e., they depend on cleaning parameters that are effectively independent of spatial coordinates, namely time, temperature, concentration, and the ratio of cleaning solution to process soil). It should be noted that degradation and inactivation studies are typically not performed at full scale 52

Special edition: Cleaning Validation

because the concentration of the degradants in cleanin-place (CIP) samples is generally well below the limit of detection (LOD) of SDS-PAGE and most bioassays. In the small-scale studies, this issue is addressed by using a soil to solution ratio (R) that is high enough to ensure that the concentration of degraded product in the sample is above the LOD of the assay. Note that a higher R represents a milder condition from the standpoint of degradation (i.e., higher R implies that a smaller fraction of the total soil would degrade). Implications for Cleaning Validation The inactivation of the API during cleaning has important implications for cleaning validation of multiproduct equipment. Demonstrating that the product is inactivated during cleaning obviates the need to set limits based on a conventional MAC assessment for the API (5). It also eliminates the need to develop product specific assays for cleaning validation (6). The Comparable Quality Approach If the API is inactivated when exposed to worst-case cleaning conditions, the CQ approach can be used to set acceptance limits for cleaning (2). The CQ approach ensures that the amount of inactivated Product A in the largest dose of Product B is limited to the amount of inactivated Product A in the largest dose of Product A. Thus, the CQ approach provides assurance that in a single exposure the amount of inactivated Product A that a patient taking Product B receives is less than or equal to the amount of inactivated Product A that a patient taking Product A receives. The rationale for this approach is that if one set of patients (i.e., those taking Product A) are being exposed to a certain amount of inactivated Product A, then it is acceptable for another set of patients (i.e., those taking Product B) to be exposed to a lesser or equal amount of inactivated Product A. The CQ approach is based on the premise that an appropriate product quality assessment has been completed on Product A to show that Product A is of acceptable quality. Differences in Patient Subpopulations and Frequency or Duration of Dosing For pharmacologically-active substances, therapeutic effects and toxicity that occur from a single peak exposure or continually over time are important considerations for setting acceptable limits for exposure. In order to study the effects of key variables such as dosing frequency, length of exposure, and sensitivity, active substances are extensively tested in the patient subpopulations taking the drug. While these endpoints are important for the study of active substances, they are not relevant to degradants that are pharmacologically inactive. Instead, the comparable CQ approach is more appropriate for establishing limits for inactive degradants.

Rizwan Sharnez Perform inactivation study a Is API in process sample inactivated after exposure to cleaning conditions? b

Yes

No

If CQ criterion is satisfiedc set acceptance limit (AL) for inactivated product (IP) d per Eqn. 2

Use conventional MAC approach to set acceptance limit (AL) for API e

Can IP be recovered by TOC?

Can API be recovered by TOC?

Yes Is AL ≥ LOQ of TOC? Use TOC to demonstrate removal of IP

Yes

No No

Is AL ≥ LOQ of TOC?

Develop alternative assay to demonstrate removal of IP

Use TOC to demonstrate removal of API

No No

Develop alternative assay to demonstrate removal of API

Inputs: process sample, reference standard, cleaning cycle, and equipment parameters Based on bioassay analysis c If CQ criterion is not satisfied, the approach described in this paper is not applicable d Inputs: dose and batch sizes of product, surface area of dedicated and shared equipment, and surface concentration of IP (refer to Appendix 1.2) e Inputs: dose and batch sizes of product, surface area and rinse parameters of shared equipments [8] a

b

Figure 1: Flowchart for setting acceptance limits for multiproduct cleaning validation.

Effect of Cleaning Process Note that for a given cleaning process for Product A, the molecular weight distribution (MWD) of Product A that is carried over into a subsequent lot of Product A (intracampaign cleaning) will be similar to the MWD of Product A that is carried over into a subsequent lot of Product B (intercampaign cleaning). It is also important to note that no additional inactivated product is introduced into either product as a result of implementing the CQ approach. A literature search did not uncover methods for calculating cleaning validation acceptance limits for the carryover of inactivated product. Setting Acceptance Limits Based on Protein Inactivation A methodology for setting acceptance limits for multiproduct equipment is summarized in Figure 1. Degradation and inactivation studies are first performed under simulated worst-case cleaning conditions. If the sample is shown to have no detectable activity (based on the

results of the bioassay), then the CQ approach is used to set acceptance limits (AL) for the inactivated product (left side of Figure 1). However, if the sample is shown to have biologically active product, the conventional MAC approach is used to set the AL for the previously manufactured product (APIA) (right side of Figure 1). If the process residue (inactivated product or API, as the case may be) can be recovered by TOC and the corresponding AL ≥ the LOQ of TOC, TOC may be used to verify removal of the process residue at full scale; otherwise, an alternative assay may need to be developed. COMPARABLE QUALITY CRITERION IN TERMS OF MEASURABLE PARAMETERS If the product becomes therapeutically inactive during cleaning, then the acceptance limit for the inactivated degradants of the previously manufactured product (Product A) in the subsequently manufactured product (Product B) can be set based on the CQ criterion. In terms of the carryover of inactivated product, the CQ criterion can be stated as follows. Special edition: Cleaning Validation

53

Rizwan Sharnez

The amount of inactivated Product A in the largest LD dose (LD) of Product B (M IP A→B ) should be less than or equal to the amount of inactivated Product A in the LD largest dose of Product A (M IP A→A) (2). Thus, M LD IP A→B



M LD IP A→A

[Equation 1]

To apply this criterion to set acceptance limits for inactivated product, the above inequality must be expressed in terms of measureable equipment and product parameters. An equation for expressing this criterion in terms of analytical data, product contact surface area, and batch and dose sizes is derived in Appendix I. For intrasite assessments, where the intracampaign cleaning cycle (A→A) is similar to the intercampaign cleaning cycle (A→B), the result is as follows (Equation AI-5):

∑Ni=1 [ CA¤A,i ⋅ SA A¤A,i] ≥ ∑Mj=1 [CA¤B,j ⋅ SA A¤B,j ]

SBS A LD B ⋅ [Equation 2] SBS B LD A

where, • SAA→A,i is the total product-contact surface area for the ith system (CIP circuit) of the equipment train used to manufacture Product A • N is the total number of systems that are used to manufacture Product A • SAA→B,j is the product-contact surface area for the jth system of the equipment train that is shared between Products A and B • M is the number of systems that are shared between the two products • CA→A,i and CA→B,j are average surface concentration of inactivated Product A on SAA→A,i and SAA→B,j. The average surface concentrations can be estimated from swab and rinse data as described in Section AI.2 of Appendix I. • SBSA and SBSB are the smallest integral (7) batch size of Product A and B, respectively • LDB and LDA are the largest doses of Products A and B, respectively. If inactivated Product A can be recovered by TOC assay, then TOC is used to verify removal of the inactivated product; otherwise, an alternative assay is developed to demonstrate removal of the inactivated product at full scale. The CQ criterion can be applied to smaller sections of the total equipment train. For instance, if the drug substance (DS) and drug product (DP) are manufactured in different facilities, it may be more efficient from a change management perspective to decouple the CQ assessments for the DS and DP equipment trains. SETTING ACCEPTANCE LIMITS FOR PROCESS RESIDUE If the characterization data indicate that the API degrades when exposed to worst-case cleaning conditions, 54

Special edition: Cleaning Validation

and the CQ criterion per Equation 2 is met, then the acceptance limit for the process residue can be set based on the lowest of the following two limits: • Process capability. This is based on either the capability of the equipment or the capability of the CIP circuit to clean a specific process residue. Note that the CQ approach is based on the use of historical cleaning data to support the baseline amount of inactivated Product A in a dose of Product A. Thus, the use of historical cleaning data to set acceptance limits for cleaning is justified. If the historical data are below the limit of quantitation (LOQ) of the assay, the process capability limit can be set to the LOQ of the assay. This approach is consistent with industry practices. It is justified because the cleaning cycle is validated and continually monitored, and has appropriate controls in place. Additionally, adopting this approach does not impact the relative amount of degraded product that is carried LD over (i.e., M LD IP A→A vs. M IP A→B), which is consistent with the CQ criterion. • 10 ppm limit for carryover of process residue. This is a widely used default limit (8) that is based on the criterion that the cumulative carryover of process residue from the previously manufactured batch into the subsequently manufactured batch should be ≤ 10 ppm. Note that the raw data from the analytical instrument should be divided by the appropriate swab or rinse recovery factor when estimating the residue on the surface. Additionally, if TOC is used to analyze the sample, the default limit should be multiplied by the fraction of carbon in the process soil to account for the amount of carbon in the residue. CONCLUSION An important consideration in multiproduct cleaning validation is to demonstrate that the carryover of the previously manufactured API into a batch of the subsequently manufactured product is below an acceptable limit. If, however, the previously manufactured API becomes pharmacologically inactive during cleaning, then there is limited value in verifying removal of the API. Instead, from a patient safety standpoint, it is more meaningful to demonstrate that the inactivated product is adequately removed. Inactivation of the API during cleaning can be assessed by exposing the process soil to simulated cleaning conditions at bench scale. The degree of inactivation is evaluated by subjecting the sample and untreated controls to the appropriate assays (e.g., SDSPAGE and bioassay for evaluation of degradation and biological activity, respectively).

Rizwan Sharnez

If the protein is inactivated during cleaning, then the acceptance limit for the inactivated degradants of the previously manufactured product (Product A) in the subsequently manufactured product (Product B) can be set based on the CQ criterion. The CQ approach provides assurance that the maximum amount of inactivated Product A that a patient taking Product B receives (M LD IP A→B) is limited to the maximum amount of inactivated Product A that a patient taking Product A receives (M LD IP A→A). The rationale for this approach is that if one set of patients (i.e., those taking Product A) are being exposed to a certain amount of inactivated Product A, then it is acceptable for another set of patients (i.e., those taking Product B) to be exposed to a lesser or equal amount of inactivated Product A. In terms of the mass of the inactivated product, the CQ criterion is given by Equation 1. In order to use this criterion to set acceptance limits for inactivated product, the Equation 1 inequality was expressed in terms of measureable equipment and product parameters (Appendix I). The resulting expression for the CQ criterion for intrasite assessments, where the intracampaign cleaning cycle (A→A) is similar to the intercampaign cleaning cycle (A→B), is given by Equation 2. When this inequality is satisfied, the amount of inactivated Product A in the largest dose of Product B will be less than the amount of inactivated Product A in the largest dose of Product A. If inactivated Product A can be adequately recovered by TOC, then TOC is used to verify removal of the inactivated product; otherwise, an alternative assay is developed to demonstrate removal of the inactivated product at full scale. If the characterization data indicate that the API is inactivated when exposed to cleaning conditions, and the CQ criterion is met, then the acceptance limits for the process residue can be set based on the lowest of the following two limits: process capability and 10 ppm limit for carryover of process residue. Application of the CQ criterion to intersite transfers will be described in Part III. REFERENCES 1. Fourman, Gary L. and Michael V. Mullen, “Determining Cleaning Validation Acceptance Limits for Pharmaceutical Manufacturing Operations,” Pharmaceutical Technology, 17 (4), 54-60, 1993. 2. Sharnez, R., To, A., “Cleaning Validation of Multiproduct Equipment: Acceptance Limits for Inactivated Product, Part I–The Comparable Quality Approach,” Journal of Validation Technology, Autumn 2011, p. 32-36, 2011. 3. Kendrick, K., Canhoto, A., Kreuze, M., “Analysis of Degradation Properties of Biopharmaceutical

Active Ingredients as Caused by Various Process Cleaning Agents and Temperature,” Journal of Validation Technology, Vol 15, No. 3, p. 69, 2009. 4. Rathore, N., Qi, W., Chen, C., Ji, W., “Bench-scale characterization of cleaning process design space for biopharmaceuticals,” Biopharm Int, Vol 22, No. 3, 2009. 5. Parental Drug Association, Inc., PDA Technical Report 49 Points to Consider for Biotechnology Cleaning Validation, July 2010. 6. Health Canada, Cleaning Validation Guidelines (GUIDE-0028); Section 8.3, Health Products and Food Branch Inspectorate, 2008. 7. Note: i.e., before the original manufacturing batch is subdivided into smaller batches for processing (e.g., filling). 8. Note: i.e., if the calculated acceptance limit is above the default limit, then the default limit is used; conversely, if the calculated acceptance limit is below the default limit, then the acceptance limit is used. ARTICLE ACRONYM LISTING A A→A A→B ADE API B CQ DP DS LD LOQ M LD IP A→A M LD IP A→B MAC MWD SDS PAGE SBS TOC

Product A – Previously manufactured product Between batches of the same product (intracampaign) Between batches of different products (intercampaign) Acceptable Daily Exposure Active Pharmaceutical Ingredient Product B – Subsequently manufactured product Comparable Quality Drug Product Drug Substance Largest Dose Limit of Quantitation Mass of inactivated Product A in largest dose of Product A Mass of inactivated Product A in largest dose of Product B Maximum Allowable Carryover Molecular Weight Distribution Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis Smallest Batch Size Total Organic Carbon

ACKNOWLEDGEMENTS The authors thank Aine Hanly, Donna Corvese, Lillian Colon, Sam Guhan and Steve Hatke for their help and support.

Special edition: Cleaning Validation

55

Rizwan Sharnez

APPENDIX I Derivation of CQ Criterion

Derivation of the comparable quality (CQ) criterion in terms of measureable equipment and product parameters is described below. In terms of the carryover of inactivated product, the CQ criterion states that: The amount of inactivated Product A in the largest dose (LD) of Product B (M LD IP A→B) should be less than or equal to the amount of inactivated Product A in the largest dose of Product A (M LD IP A→A) (2). Thus, M LD IP A→B



M LD IP A→A

[Equation AI-1]

AI.1 CQ criterion in terms of measureable parameters In order to apply this criterion to set acceptance limits for inactivated product, the above inequality must be expressed in terms of measureable equipment and product parameters. This is demonstrated in Figure AI.1 with an example in which two products, A and B, are subsequently manufactured in a shared equipment train. The equipment train shown in Figure AI.1 is used to manufacture Product A and has a total surface area of SA A→A. The carryover of the residual inactivated Product A (that remains on the equipment surface after batchcleaning), into the next batch of Product A (M IP,A¤A) can be expressed as:

facture Product A (Figure AI.2). The surface area that both products are exposed to, also known as the “shared” surface area, is SA A→B. Similarly, the carryover of residual inactivated Product A (that remains on the equipment surface after batch cleaning), into a batch of Product B (M IP,A¤B) can be expressed as: M

∑ [C

batch

M IP,A¤B =

A¤B,j

∙ SA A¤B,j]

[Equation AI-2b]

j=1

where, SA A→B,j is the product contact surface area for the jth system of the equipment train (cm2), M is the number of systems that are shared between the two products, and CA→B,j is the average surface concentration of inactivated Product A on SA A→B,j (µg/cm2). Thus, the carryover of inactivated Product A into largest dose of Product B (M LD IP A→B) can be expressed as: N



M IP,A¤B= LD

[CA¤B,j ∙ SA A¤B,j] ∙

j=1

LDB

SBSB [Equation AI-3b]

where, LDB and SBSB are the largest dose and smallest batch size of Product B, respectively. Substituting for the left-hand and right-hand sides of Equation AI-1 with the right-hand side of Equations AI-3b and AI-3a, respectively, gives: M



[CA¤B,j ∙ SA A¤B,j] ∙

j=1

LDB SBSB

N



∑ [C

A¤A,i

∙ SA A¤A,i ] ∙

i=1

LDA SBSA

[Equation AI-4] N

∑ [C

batch

M IP,A¤A =

A¤A,i

∙ SA A¤A,i]

[Equation AI-2a]

i=1

where SA A→A,i is the product contact surface area for the ith system (CIP circuit) of the equipment train (cm2), N is the total number of systems that are used to manufacture Product A, and CA→A,i is the average surface concentration of inactivated Product A on SA A→A,i (µg/cm2). Thus, the carryover of inactivated Product A into the largest dose of Product A (M LD IP A→A) can be expressed as: N



M IP,A¤A= LD

[CA¤A,i ∙ SA A¤A,i] ∙

i=1

LDA SBSA [Equation AI-3a]

where LDA and SBSA are the largest dose and smallest batch size of Product A, respectively. Part or all of the equipment train that is used to manufacture Product B will also be used to manu56

Special edition: Cleaning Validation

Rearranging the terms in Equation AI-4 gives:

∑ ∑

N i=1 M j=1

[CA¤A,i ∙ SA A¤A,i ] [CA¤B,j ∙ SA A¤B,j ]



SBSA LDB ∙ SBSB LDA [Equation AI-5]

Equation AI-5 gives the CQ criterion for intrasite manufacturing (i.e., when Product A is already being manufactured at the same site where Product B is manufactured). Note that N is the total number of systems used in the manufacture of Product A, and M is the number of systems that are used in the manufacture of both Product A and Product B (i.e., shared systems). AI.2. Surface concentration in terms of swab or rinse data The surface concentration terms in Equation AI-5 (CA→A,j and CA→B,j) can be expressed in terms of TOC (or other analytical) results as follows:

Rizwan Sharnez

Upstream

Downstream

Fill/Finish

Figure AI.1: Equipment train used to manufacture Product A has a total surface area of SA A→A

Upstream

Downstream

Fill/Finish

Figure AI.2: The surface area that both products will be exposed to, also known as the shared surface

1 CA¤A,i= S

S



CA¤A,k,i

[Equation AI-6]

k= 1

where S is the number of data points (samples) for the ith system (CIP circuit) and CA→A,k,i are the measured surface concentrations of inactivated product for each of the S data points. The CA→A,k,i for the equipment train used to manufacture Product A can be estimated from TOC swab data as follows:

CA¤A,k,i =sTOCk,i

sRFi ⋅ cfi

[Equation AI-7a]

area, is SAA→B

swab recovery factor and carbon fraction for the process soil in the ith system. Similarly, CA→A,k,i can be estimated from TOC rinse data as follows:

CA¤A,k,i =rTOCk,i

rRFi ⋅ cfi

[Equation AI-7b]

where, rTOCk,i is the TOC rinse data for the kth rinse sample result for the ith system, and rRFi and cfi are the rinse recovery factor and carbon fraction for the process soil in the ith system. Similarly, CA→B,j can be estimated from TOC swab or rinse data, recovery factors and carbon fractions for the shared equipment.

where, sTOCk,i is the TOC swab data for the kth swab sample result for the ith system, and sRFi and cfi are the

Special edition: Cleaning Validation

57

Rizwan Sharnez

APPENDIX II Worked Example for Intrasite Assessment Is the CQ criterion satisfied for the following scenario?: Product A and Product B are currently being manufactured at the same site. The total surface area associated with the manufacture of Product A and relevant process parameters are given in Table AII-1. TOC rinse data for the above circuits are listed in Table AII-2. In this case, for each circuit there is one TOC rinse sample. Part of the equipment train used to manufacture Product A is also used to manufacture Product B. Circuits 2 and 3 are shared for the manufacture of Product A and Product B. The shared surface area SA A→B,j and relevant process parameters for the manufacturing Product B are as given in Table AII-3. Because the cleaning cycles for A→A are the same as the cleaning cycles for A→B, CA→A,i = CA→B,j for Circuits 2 and 3, and the TOC rinse data listed in Table AII-2 can be used for both surface concentrations. AII. Solution For this scenario (intrasite assessment), the CQ criterion is given in Equation AI-5:

∑ ∑

N i=1 M j=1

[CA¤A,i ∙ SA A¤A,i ] [CA¤B,j ∙ SA A¤B,j ]



SBSA LDB ∙ SBSB LDA

The calculated parameters for this example are as given in Table AII-4. 3

∑ [C

A¤A,i

∙ SA A¤A,i ]= (3.71×12.6)+(2.14×10.2)

i=1

+(3.15×11.6)=105.1

2

∑ [C

A¤B,j

∙ SA A¤B,j] = (2.14×10.2)+(3.15×11.6)=58.4

j=1

SBSA SBSB LDB LDA

=

=

4.5 5 75 50

= 0.9

= 1.5

Substituting the above results into the left and right hand sides of Equation AI-5 gives: Left hand side = 1.80 Right hand side= 1.35 Thus, the CQ criterion is satisfied for the above intrasite assessment. JVT

Table AII-3: Shared surface area for Products A and B and relevant process parameters.

Description of Parameter

Abbreviation

Value

M

2

SA A → B,1

10.2 m 2

SA A → B,2

11.6 m 2

Smallest batch size of Product B

SBSB

5 kg

Largest dose size of Product B

LDB

75 mg

Number of shared CIP circuits The shared surface area of circuit 2 (j=1) that is used to manufacture Product A and Product B The shared surface area of circuit 3 (j=2) that is used to manufacture Product A and Product B

Table AII-4: Calculated parameters for intrasite assessment.

Calculated Parameter

Equation

Result

C A → A,1

AI-7b

3.71 ppm

C A → A,2

AI-7b

2.14 ppm

C A → A,3

AI-7b

3.15 ppm

C A → B,1

C A → B,1= C A → A,2

2.14 ppm

C A → B,2

C A → B,2=C A → A,3

3.15 ppm

58 Special edition: Cleaning Validation

Worked Example for i=1 CA¤A,k,i =rTOCk,i

CA¤A,1,1=0.65

rRFi ⋅ cfi

(0.70x0.25)

=3.71

Rizwan Sharnez Table AII-1: Total surface area associated with the manufacture of Product A and relevant process parameters.

Description of Parameter

Abbreviation

Value

N

3

Surface area of circuit 1 (i=1) that is used to manufacture Product A

SA A → A,1

12.6 m 2

Surface area of circuit 2 (i=2) that is used to manufacture Product A

SA A → A,2

10.2 m 2

Surface area of circuit 3 (i=3) that is used to manufacture Product A

SA A → A,3

11.6 m 2

Smallest batch size of Product A

SBSA

4.5 kg

Largest dose size of Product A

LDA

50 mg

Number of CIP circuits

Table AII-2: TOC rinse data.

Description of Parameter TOC rinse sample data for the i circuit th

Rinse recovery factor for the i circuit th

Carbon fraction for the i circuit th

Abbreviation

Circuit 1 (j=1)

Circuit 2 (j=2)

Circuit 3 (j=3)

rTOCi,j

0.65 ppm

0.2 ppm

1.1 ppm

rRFj

0.70

0.55

0.92

cfj

0.25

0.17

0.38

Special edition: Cleaning Validation

59

Rizwan Sharnez

Biopharmaceutical Cleaning Validation: Acceptance Limits for Inactivated Product Based on Gelatin as a Reference Impurity | IVT Rizwan Sharnez, Ph.D., Abby Spencer, Jeanine Bussiere, Ph.D., Dan Mytych, Ph.D., Angela To, and Arun Tholudur, Ph.D.

Biopharmaceutical cleaning and sterilization processes denature and degrade the active pharmaceutical ingredient (API)i into fragments that are pharmacologically inactive. A rational approach for setting safety-based acceptance limits for inactive fragments is described. The approach is based on the use of gelatin as a reference impurity. It is designed to ensure that the carryover of inactive fragments between batches of different products is acceptable from a predictive safety standpoint. The scope of this paper is limited to process residues of non-conjugated human therapeutic proteins. Nonetheless, the underlying principles may be useful in setting acceptance limits for other types of inactive impurities.

INTRODUCTION An important regulatory expectation for multiproduct cleaning validation is that the potential carryover of the previously manufactured API into the subsequently manufactured product should be below an acceptable level. This criterion is often met through a maximum allowable carryover (MAC) assessment for the previously manufactured API (1-5). The MAC assessment is typically based either on the minimum therapeutic dose (1), or the acceptable daily exposure (ADE) (2) of the previously manufactured API.  A limitation of the conventional MAC approach is that it does not provide appropriate acceptance limits for pharmacologically inactive product. This has important implications for biopharmaceutical manufacturing because the API is inactivated during cleaning and sterilization. Another limitation of this approach is that the acceptance limit for the API is often below the process capability limit (PCL) of the cleaning processes and, in some instances, below the limit of quantitation (LOQ) of non-specific assays, such as total organic carbon (TOC). Other limitations of the conventional MAC approach have been discussed previously (6). Fragmentation and inactivation of an API during cleaning and sterilization can be assessed by exposing the process soil to simulated operating conditions at bench scale (7). The bench scale experiments are designed to simulate full-scale operating conditions that are least conducive (worst-case) for inactivation. The degree of inactivation is evaluated by subjecting the sample and untreated controls to the appropriate assays (e.g., sodium dodecyl sulfate polyacrylamide gel electrophoresis [SDSPAGE] and bioassay can be used to evaluate fragmentation and biological activity, respectively). The results of the study are used to set appropriate 60

Special edition: Cleaning Validation

Rizwan Sharnez

acceptance limits for the process residue. The product inactivation approach is therefore more science-based and reflective of the phenomenological aspects of cleaning processes.  Inactivation of the product during cleaning and sterilization has important implications for cleaning validation of multiproduct equipment. If the API degrades into pharmacologically inactive fragments, the acceptance limit for the process residue can be set based on the inactive fragments instead of the active ingredient. This is consistent with the expectation that the carryover of an impurity–in this case the inactive fragments–into the subsequently manufactured product should be justified from the standpoint of the safety and efficacy of the product.  The experimental approach and analytical methods for assessing fragmentation and inactivation of the API have been discussed in the literature (6-11). This paper describes a rational approach for setting safety-based acceptance limits for the inactive fragments. The proposed methodology builds upon previously published approaches (6, 10-11). ASSESSING THE SAFETY OF INACTIVE FRAGMENTS OF HTPS The safety of inactive fragments of human therapeutic proteins (HTPs) is assessed qualitatively in this section. Consider an equipment train that is used to manufacture Product A. The cleaning and sterilization cycles are known to denature and degrade any residual product in the equipment into fragments that are pharmacologically inactive. The inactive fragments of Product A are carried over into the subsequent batch of Product A. Thus, as a class of molecules, inactive fragments of HTPs do not present a new or unknown risk from a safety standpoint. In fact, these types of fragments have been present in biopharmaceutical products for decades. Further, a comprehensive literature search did not reveal any evidence of safety or efficacy issues directly attributable to the presence of inactive fragments in parenteral drugs (12). Implications for Multiproduct Cleaning Validation Consider the introduction of a new product (Product B) into the facility. Part of the equipment train is now used to process both products. The cleaning and sterilization cycles between batches of different products (A → B or B → A, [i.e., intercampaign processing or changeover]) are the same as those between batches of the same product (A → A or B → B, [i.e., intracampaign processing]). Thus, for a given set of cleaning and sterilization cycles, the molecular weight distribution (MWD) of inactive fragments of Product A (IFA) that are carried over into a subsequent batch of Product B (intercampaign processing) is the same as the MWD of IFA that are carried over into a subsequent batch of Product A (intracampaign processing). The same is true for the MWD of inactive frag-

ments of Product B (IFB). Thus, the IFA that are carried over during changeover into Product B do not present a new or unknown risk from a safety standpoint. This implies that the equipment train can be used to manufacture multiple products without introducing a new or unknown class of impurities into any of the products. Further, the carryover of IFA into Product B is significant only for the first lot of Product B that is manufactured following changeover.  COMPARABLE QUALITY APPROACH The acceptance limits for inactive fragments in the process residue can be set based on the Comparable Quality (CQ) approach (6, 10-11). With the CQ approach, the amount of the target impurity–in this case inactive fragments of Product A–that is carried over into the largest dose (i.e., largest dose that is administered to a patient in a day) of the subsequently manufactured product (Product B) is limited to the acceptable exposure per dose of a reference impurity. The reference impurity must be comparable to or worse than the target impurity from a predictive safety standpoint.  Predictive safety for inactive fragments is evaluated in terms of the key factors that determine toxicity and immunogenicity. For HTPs, toxicity is determined by pharmacological activity (13); thus, toxicity is generally not a concern for inactive fragments of HTPs. Immunogenicity is primarily determined by foreignness and chemical complexity (14). Chemical complexity increases with molecular weight (MW); thus, larger molecules tend to be more immunogenic (15). The most active immunogens tend to have a MW greater than 100 kilo Daltons (kDa) (14). HTP fragments with MWs less than 10 kDa are generally weak immunogens (14). Small polypeptides under 10 kDa usually need to be conjugated to large immunogenic carrier proteins or administered with adjuvants to ensure an antibody response (16). The suitability of gelatin as a reference impurity for setting acceptance limits for inactive HTP fragments is evaluated in the next section. GELATIN AS A REFERENCE IMPURITY FOR INACTIVE HTP FRAGMENTS The use of gelatin as a reference impurity for inactive HTP fragments is justified for the following reasons:     • Gelatin consists of a mixture of animal protein fragments derived from the hydrolysis of collagen, a protein that is commonly found in connective tissues (17). The collagen is hydrolyzed by exposing the connective tissues to pH and temperature extremes (18). HTPs in the process residue are exposed to similar operating conditions during cleaning and sterilization. Thus, in terms of chemical composition, the protein fragments in gelatin are comparable to the HTP fragments in the process residue. Special edition: Cleaning Validation

61

Rizwan Sharnez

• To elicit an immune response, a molecule must be recognized as nonself by the immune system (14). The protein fragments in gelatin are of animal origin whereas the HTP fragments in the process residue are of human origin. Thus, the peptide sequences in the HTP fragments are more likely to be recognized by human immune systems than the peptide sequences in the protein fragments in gelatin. Consequently, as compared to the protein fragments in gelatin, the HTP fragments in the process residue are less likely to elicit an immunogenic response in humans.  • The molecular weights of most of the HTP fragments are less than 10 kDa (19). HTP fragments

with MWs less than 10 kDa are generally weak immunogens (14). Protein fragments in gelatin range from 15 kDa to 400 kDa (18), which is substantially higher than the 10 kDa threshold for weak immunogens.  CQ APPROACH BASED ON GELATIN This section describes the application of the CQ approach based on the use of gelatin as a reference impurity to inactive HTP fragments.  Gelatin is used as a stabilizer in many parenteral products. The amount of gelatin in common parenteral products ranges from several hundred micrograms to over 15,000 µg per dose (refer to table).

Table: Gelatin Content of Common Parenteral Products. 

Product

Trade Name

DTaP (Sanofi Pasteur)1, 2

Tripedia

Gelatin Content per dose 28 µg in 0.5 mL

Influenza (Sanofi Pasteur)2 Fluzone

250 µg in 0.5 mL

Japanese Encephalitis (Sanofi Pasteur)2

500 µg in 1.0 mL

JE-VAX

Leuprolide acetate for Lupron Depot, 650 µg in 3.75 mg depot suspension (Abbot)3 3.75 mg Measles, Mumps, Rubella ATTENUVAX 14,500 µg in 0.5 mL (Merck)2 Measles, Mumps, Rubella, ProQuad 11,000 µg in 0.5 mL Varicella (Merck)2 Rabies (Novartis)2 RabAvert 12,000 µg in 1.0 mL

Varicella (Oka/Merck)2 Yellow Fever (Sanofi Pasteur)1, 2 Zoster (Oka/Merck)2 Isoplex Solution for Infusion4

VARIVAX (frozen) YF-VAX ZOSTAVAX ISOPLEX

Route of Dosing Schedule Administration Intramuscular Five dose series at two, four, injection six, 15 months, and four to six years of age Intramuscular Yearly injection Subcutaneous Three doses on days zero, injection seven, and 30. Booster dose of 1 mL can be given after two years Intramuscular Monthly

Subcutaneous injection Subcutaneous injection Intramuscular injection

12,500 µg in 0.5 mL Subcutaneous injection 7,500 µg in 0.5 mL Subcutaneous injection 15,580 µg in Subcutaneous 0.65 mL injection 4% w/v (20g in Intravenous 500 mL bag)

Two doses: one at 12 months of age and one at four years of age Two doses: one at 12 months of age and one at four years of age Post-exposure dosage: 1 mL doses on days zero, three, seven, 14, and 28 Two doses each given four weeks apart One dose every 10 years Single dose

Given for blood loss, 500 mL can be given in as little time as  five minutes in the case of rapid blood loss 1 Kelso, John M., Li, James T., Adverse reactions to vaccines; Ann Allergy Asthma Immunol., 2009 Oct;103(4 Suppl 2):S1-14. 2 Package inserts via www.fda.gov/BiologicsBloodVaccines/Vaccines/default.htm 3 Package insert via http://www.rxabbvie.com 4 Package insert via http://www.mhra.gov.uk/Safetyinformation/Medicinesinformation/SPCandPILs/index.htm 62 Special edition: Cleaning Validation

Rizwan Sharnez

Lupron serves as a good model for the CQ approach because it is administered in multiple doses over an extended period. Note that from the standpoint of immunogenicity, repeated dosing of an immunogen is important; however, dosing frequency is generally not critical provided that the time interval between exposures is relatively short (e.g., weekly, bi-weekly, or monthly). Each dose of Lupron contains 0.65 mg of gelatin; thus, the CQ Criterion based on the gelatin in this product is   ≤ 0.65 mg   [1a]  Where re  is the mass of inactive frag  ments of Product A (IFA) that is carried over into the largest dose (LD) of the subsequently manufactured Product (B). The concentration of the inactive fragments of Product A (CIFA) in the largest dose of Product B ( ) is,      [2a]

CIFA = 

  

       [2b]

 

 ≤ 

IFA 

[6b]

Rearranging Equation 6b gives SC ≤ 0.65 mg ( IFA 

 /

) / SSA   [6c]

The above inequality can be used to set an acceptance limit for SC  in terms of three measureable parameters: ,   , and SSA. IFA

Solved Example  Calculate the acceptance limit for TOC in the swab extractant for the following scenario:   = 20 L,   = 10 mL, SSA = 105 cm2, swab recovery factor for the process residue (SRF) = 0.5, carbon fraction in the process residue (CF) = 0.5, area swabbed (A) = 25 cm2, and mass of extractant (m) = 25 g.  Substituting for ,  , and SSA into Equation 6c gives:   SC  ≤ 650 (µg) • 20,000 (mL) / 10 mL / 105 cm2 = 13 µg/cm2 Based on the worst-case assumption that all the carbon in the process residue is from the IFA, the acceptance limit for TOC (ALTOC) in the swab extractant is ALTOC =  SC  • A • SRF • CF / m  ALTOC = 13 µg/cm2 • 25 cm2 • 0.5 • 0.5 / 25 g  = 3.25 ppm Thus, the cleaning is effective if the TOC in the swab extractant is ≤ 3.25 ppm above the baseline TOC of the control sample. Note that the above acceptance limit is based on relatively unfavorable system parameters (small batch size, large dose and shared surface area, and low recovery). Further, the acceptance limit is considerably higher than the process capability limit (PCL) of most cleaning processes, which is typically on the order of 1 ppm Carbon for TOC swab samples. It is also substantially higher than the LOQ of TOC, which is typically between 0.05 and 0.2 ppm Carbon. It is therefore unlikely that the acceptance limit based on this approach would be below the PCL of the cleaning process or the LOQ of TOC. IFA

Substituting for   from Equation 1a gives the acceptance limit (AL) for CIFA,  ≤

SC • SSA / 

IFA

A worst-case estimate of CIFA ( ) is obtained by assuming that all of the residual IFA on the surface of the shared equipment train after cleaning (RIFA) is transferred into the smallest batch of Product B ( ) that is manufactured in the equipment trainii. Thus,  = R / 

  [3]

Also,  RIFA = SSA • SCIFA   [4] Where SSA is the shared surface area of the equipment train, and SC is the average surface concentration of IFA on the shared surface area. Thus, IFA 

 = SC  • SSA /  IFA

 [5]

Note that any splits of the batch into smaller batches for filling or other operations should be appropriately accounted for in estimating SSA and  . Based on the definitions of   and C AL IFA, the cleaning is effective if   ≤ 

[6a]

Substituting for   and   from Equation 2b and Equation 5, respectively, gives

METHODOLOGY FOR SETTING ACCEPTANCE LIMITS The proposed methodology for setting acceptance limits for multiproduct equipment is summarized in the flowchart. Inactivation studies are first performed under simulated worst-case operating conditions (7). If the sample is shown to have no detectable activity, the CQ approach can be used to set acceptance limits for the inactivated product (right side of flowchart). However, if there is no detectable loss in activity, the conventional MAC approach can be used to set the acceptance limit Special edition: Cleaning Validation

63

Rizwan Sharnez

Figure 1: Flowchart for Setting Acceptance Limits (AL) for the Process Residue. a Based on results of bioassay

for the previously manufactured product (1-2). If the results indicate that the API is partially inactivated, the acceptance limit should be determined for the API as well as the inactivated product, and the lower of the two limits should be used. Alternatively, the operating parameters can be modified to ensure inactivation of the API. This can be facilitated by running additional studies to characterize the effect of the operating parameters on the API. Default Limits If the carryover of the target impurity into the subsequent batch based on the above acceptance limit (AL) is above 10 ppm, then the AL should be reduced to limit the carryover to 10 ppm. The 10 ppm carryover limit is a widely used default limit (20). It is based on the empirical observation that raw materials and intermediates commonly found in biopharmaceutical manufacturing process residues are safe to ingest at concentrations up to 0.1% (1000 ppm). In order to extrapolate this oral limit to other routes of administration, it is reduced by a factor of 100 to 10 ppm. If sufficient historical cleaning data are available to establish a statistically sound process capability limit (PCL), then the AL should be set to the PCL. If the historical data are below the LOQ of the analytical method, the PCL can be set to the LOQ of the analyti64

Special edition: Cleaning Validation

cal method. This approach is consistent with industry practices and is justified because the cleaning cycle is validated and continually monitored, and has appropriate controls in place. Additionally, adopting this approach does not impact the relative amount and MWD of the target impurity that is carried over between batches of different products, which is consistent with the CQ criterion. Thus, setting the acceptance limit to the PCL ensures that the MWD of the target impurity that is carried over into a batch of Product B is comparable to the MWD of that target impurity that was historically present in a batch of Product A. CONCLUSION Biopharmaceutical cleaning and sterilization cycles are designed to expose product contact equipment to pH and temperature extremes for several minutes. Under these conditions biological APIs degrade and denature rapidly thereby becoming pharmacologically inactive.  Inactivation of the product during cleaning and sterilization has important implications for cleaning validation of multiproduct equipment. If the API is inactivated, the acceptance limits for the process residue can be set based on the inactive product instead of the API.  A comprehensive literature search did not reveal any evidence of safety or efficacy issues directly attribut-

Rizwan Sharnez

able to the presence of inactive fragments in parenteral drugs (12). Further, these types of fragments have been present in biopharmaceutical products for decades. Thus, as a class of molecules, inactive fragments of HTPs do not present a new or unknown risk from a safety standpoint. The Comparable Quality (CQ) approach based on gelatin can be used to set acceptance limits for inactive fragments of non-conjugated HTPs. With this approach, the carryover of inactive fragments into the largest dose of the subsequently manufactured product is limited to the acceptable exposure of an appropriate reference impurity. The reference impurity–in this case gelatin–was shown to be comparable to or worse than the inactive fragments from a predictive safety standpoint.  If the product is not inactivated during cleaning and sterilization, the acceptance limit for the process residue should be set based on the acceptable carryover of the API (1-2). However, if the results indicate that the API is partially inactivated, the acceptance limits should be determined for the API as well as for the inactive fragments, and the lower of the two limits should be used for cleaning validation (refer to Figure). The acceptance limit for inactive fragments based on gelatin as a reference impurity was ascertained to be 0.65 mg per dose. At 0.65 mg of inactive fragments per dose, the acceptance limit for TOC swab samples was shown to be 3.25 ppm Carbon.iii This estimate was based on relatively unfavorable system parameters. Note that this acceptance limit is substantially higher than the LOQ of TOC, which is typically between

0.05 and 0.2 ppm Carbon. It also compares favorably to the process capability limit (PCL) of most cleaning processes, which is typically on the order of 1 ppm Carbon for TOC swab samples. Thus, with the CQ approach based on gelatin, it is unlikely that the acceptance limit for the process residue would be below the PCL of the cleaning process or the LOQ of TOC.  The methodology described in this paper is not applicable to allergenic ingredients, penicillin, cephalosporin, potent steroids, and cytotoxic compounds. Acceptance limits for process residues associated with these types of APIs are typically set to the limit of detection (LOD) of the best available analytical method (21). Depending on the process soil, API refers to the active pharmaceutical ingredient in the drug product, drug substance, or drug substance intermediate. ii i.e., surface area of equipment train that is subjected to cleaning and that comes into contact with both Product A and Product B. If inactive fragments of Product A are removed by the purification steps for Product B, then the part of the equipment train upstream of those purification steps can be excluded from the shared surface area.  iii Note that the use of a non-specific method such as TOC also allows for the detection of intact protein. i

ACKNOWLEDGEMENTS We thank Vijay Chiruvolu, Sam Guhan, Aine Hanly and Anthony Mire-Sluis for their help and support. 

Special edition: Cleaning Validation

65

Rizwan Sharnez

ACRONYMS A

Area swabbed

A→A

Between batches of the same product (intracampaign)

A→B

Between batches of different products (intercampaign)

ADE

Acceptable Daily Exposure

AL

Acceptance limit

ALCQ

Acceptance limit based on comparable quality approach

ALMAC

Acceptance limit based on MAC

ALTOC

Acceptance limit for TOC swab sample

API

Active pharmaceutical ingredient Acceptable limit for concentration of inactivated fragments of Product A

CIFA

Concentration of inactivated fragments of Product A Worst-case estimate of CIFA

CF

Carbon fraction in process residue

CQ

Comparable quality

HTP

Human therapeutic proteins

IFA

Inactive fragments of Product A

IFB

Inactive fragments of Product B

kDa

Kilo Dalton

LD

Largest dose

LOD

Limit of detection

LOQ

Limit of quantitation

m

Mass of extractant

MAC

Maximum allowable carryover Mass of inactive fragments of Product A in largest dose of Product B

MW

Molecular weight

MWD

Molecular weight distribution

PCL

Process capability limit

PR

Process residue

Product A

Previously manufactured product

Product B

Subsequently manufactured product

RIFA

Residual IFA on the surface of the shared equipment train after cleaning

SC

Average surface concentration of IFA on the shared surface area

IFA

SDS PAGE

Sodium dodecyl sulphate polyacrylamide gel electrophoresis

SRF

Swab recovery factor

SSA

Shared surface area (i.e., surface area of the equipment train that comes into contact with both Product A and Product B)

TOC

Total organic carbon Largest dose of Product B Minimum volume of the final batch of Product B (i.e., smallest batch size of Product B)

66

Special edition: Cleaning Validation

Rizwan Sharnez and Angela To

Multiproduct Cleaning Validation: Acceptance Limits for the Carryover of Inactivated API Part I–The Comparable Quality Approach Rizwan Sharnez and Angela To

“New Perspectives on Cleaning” is an ongoing series of articles dedicated to cleaning process development, validation, and monitoring. This column addresses scientific principles, strategies, and approaches associated with cleaning that are faced in everyday work situations. Reader questions, comments, and suggestions are requested for future discussion topics. These can be submitted to the column coordinator Rizwan Sharnez at [email protected].

SUMMARY An important consideration in multiproduct cleaning validation is to demonstrate that the carryover of the previously manufactured active pharmaceutical ingredient (API) into a batch of the subsequently manufactured product is below an acceptable limit. If, however, the previously manufactured API becomes therapeutically inactive during cleaning, then there is limited value in verifying clearance of the API. Instead, it is more appropriate to demonstrate clearance of inactivated product. This approach is gaining acceptance in the industry. A methodology for evaluating the degree of inactivation of a product during cleaning and setting acceptance limits for the carryover of inactivated product in multiproduct equipment is described. A new approach for justifying acceptance limits for inactivated product, known as the comparable quality (CQ) approach, is described in Part I; the application of this approach to biopharmaceutical cleaning will be described in Part II. The general principles of the CQ approach are applicable to most active pharmaceutical ingredients (APIs). INTRODUCTION For multiproduct cleaning validation, acceptable carryover of the previously manufactured API (Product A) into the subsequently manufactured API (Product B) is determined through a maximum allowable carryover (MAC) calculation (1-3). A limitation of the conventional MAC approach is that it does not account for the carryover of the inactivated molecule between lots of different products (i.e., A → B, or B → A). This is an important factor to consider when aggressive cleaning conditions are used. For example, biopharmaceutical cleaning cycles are generally designed to expose product contact equipment to extremes of pH (i.e., 2-13) and temperature (i.e., 60-80°C) for several minutes. The equipment may also be steam sterilized or sanitized after cleaning. Under these conditions, monoclonal antibodSpecial edition: Cleaning Validation

67

Rizwan Sharnez and Angela To

ies, therapeutic proteins, and other biological APIs are known to degrade and denature rapidly (4, 5) and are, therefore, likely to become therapeutically inactive (6). Inactivation of the product during cleaning has important implications for cleaning validation of multiproduct equipment. If it can be demonstrated that the product becomes therapeutically inactive during cleaning, a MAC assessment for the API would not be required. It also obviates the need to develop product specific assays (PSA) for cleaning validation. Inactivation of the product during cleaning can be assessed by exposing the process soil to simulated cleaning conditions at bench scale (4, 5). The bench scale studies are designed to simulate the conditions that are least conducive (worst-case) for inactivation. For example, for alkaline washes, the lowest applicable pH, temperature, duration, and ratio of cleaning solution to process soil is used to simulate the cleaning cycle at bench scale. The sample is then neutralized and cooled to minimize any further inactivation. The degree of inactivation is evaluated by subjecting the sample and an untreated control to the appropriate assays (e.g., Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis [SDS PAGE] and bioassay for biological products). A literature search did not uncover any scientific approaches or regulatory guidances for setting acceptable limits for the carryover of inactivated product for cleaning validation. LIMITATIONS OF THE MAC APPROACH The MAC approach is often used to set cleaning validation acceptance criteria for the carryover of the previously manufactured API (Product A) into the subsequently manufactured API (Product B) (1-3). A limitation of this approach is that it does not account for the carryover of the inactivated molecule between lots of different products (i.e., A → B, or B → A). Another limitation of the MAC approach is that the acceptance limits for cleaning validation are often below process capability limits and/or below the limit of quantitation (LOQ) of non-specific assays (e.g., total organic carbon [TOC]; the LOQ of TOC is typically between 0.05 and 0.2 ppm). This issue is further exacerbated by the low recovery of APIs from process soils. PSAs such as enzyme-linked immunosorbent assay (ELISA) and enzyme immunoassay (EIA) are sometimes used to address this issue because they have very low LOQs (typically below 10 ppb). However, PSAs are difficult and laborious to qualify, and can give inaccurate results if the API degrades during cleaning (7). That is because PSAs detect activity indirectly, by recognizing specific epitopes (i.e., short amino acid sequences that PSAs are designed to detect). The epitopes can be destroyed by buffer and cleaning agent components. However, a biological API can be therapeutically active even if the epitopes are destroyed, and this can lead to false negatives. 68

Special edition: Cleaning Validation

Similarly, it is possible for the API to be inactivated even though the epitopes are not completely destroyed, and this can lead to false positives. Another issue with PSAs is that it is difficult to get an accurate recovery factor. That is because the experimental conditions of the recovery study (e.g., direct spotting of coupons with the API) do not represent the actual sample matrix (i.e., a small amount of API in the presence of degradants and cleaning agent residues). For the above reasons, PSAs should be used judiciously for verifying clearance of biological APIs after cleaning (7). Another issue with the MAC approach is that every time a new product is introduced into a facility there is a risk that one or more of the new MAC limits for the previously validated products could be below the existing acceptance limits for cleaning validation. PROTEIN DEgRADATION AND INACTIvATION APPROACH Performing Degradation and Inactivation Studies A bench-scale approach for evaluating the bioactivity of the residual API and the molecular weight distribution of the degradants after exposure to cleaning conditions is described in this section. Full-scale cleaning conditions of shared product contact equipment are evaluated to determine the conditions that are least conducive for inactivation (e.g., lowest ratio of wash volume to protein, lowest cleaning agent concentration, lowest temperature, and shortest duration of exposure). The product is exposed to these conditions at bench scale. The objective of the study is to ascertain whether the API in the process sample is inactivated when exposed to cleaning conditions. Product is spiked into tubes containing alkaline cleaning solution, heated to the appropriate temperature, and allowed to incubate for the duration of the alkaline wash. Samples may also be titrated with the acidic cleaning solution to the pH of the acidic wash and held for the duration of the acidic wash. Samples are then titrated to a neutral pH and cooled to 4°C to minimize further degradation and inactivation. The samples are then subjected to SDS-PAGE and bioassay to determine the degree of degradation and TOC analysis to determine whether the degradants can be adequately recovered by TOC. Assays SDS-PAGE and bioassays are used to evaluate protein degradation and inactivation, respectively. The product is exposed to cleaning conditions at small scale and then analyzed with the above assays to determine degree of degradation (i.e., molecular weight distribution) and bioactivity. Additionally, samples are analyzed for TOC to determine the recovery of the inactivated protein and the applicability of the TOC assay for demonstrating clearance of the inactivated product at full scale.

Rizwan Sharnez and Angela To

SDS-PAGE solubilizes aggregated and degraded proteins and separates them based on molecular weight (MW). The inclusion of MW standards allows for estimation of the MWs of the degradants and any aggregates, and the inclusion of control samples at a defined protein load allows for the estimated quantitation of protein concentration by densitometry. Staining of gels is sensitive to 5-10 ng for Silver Staining and 100 ng for Coomassie Staining. SDS-PAGE has the advantage of providing a wide range of specificity for detecting proteins with unknown primary structures, size, charge, and hydrophobic states. This feature is particularly useful for protein degradation analysis because the level of protein degradation due to cleaning is highly unpredictable and can extend over a wide range of MW. SDS-PAGE also provides high sensitivity for detecting trace amounts of protein. Bioassays measure the relative amount of biologically active product present in a sample. Thus, bioassays can be used to determine the effect of cleaning conditions on the inactivation of biologicals. SETTINg ACCEPTANCE LIMITS BASED ON PROTEIN INACTIvATION The methodology for setting acceptance limits is summarized in the Figure. Degradation and inactivation studies are first performed under simulated cleaning conditions. If the sample is shown to have therapeutically active

product, the MAC approach is used to limit carryover of previous product to an acceptable level (1-3). If the MAC limit is higher than the LOQ of TOC, TOC is used to verify clearance of previous product at full scale. If the MAC limit is below the LOQ of TOC, an alternate assay may need to be developed to verify clearance of the API. The MAC approach limits the amount of API of the previously manufactured product (A) in a dose of the subsequently manufactured product (B) to the acceptable daily exposure (ADE) of A (8), or to 1/1000th of the minimum dose of A (9). COMPARABLE QUALITY APPROACH If the protein becomes therapeutically inactive during cleaning, then the acceptance limit for the inactivated molecule of the previously manufactured product (Product A) in the subsequently manufactured product (Product B) can be set based on the CQ approach. With the CQ approach, the amount of inactivated Product A in a dose of Product B is limited to the amount of inactivated API of Product A in a dose of Product A. Appropriate adjustments are made to account for differences in process parameters of the two products. If inactivated Product A can be recovered and detected by TOC, and its acceptance limit is greater than the LOQ of TOC, then TOC is used to verify clearance of the inactivated API. Otherwise, an alternative assay is developed to demonstrate clearance of the inactivated API at full scale.

Figure: Methodology for setting acceptance limits flowchart. Continued on page 70

Inputs • Process samples and reference standard • Cleaning cycle and equipment parameters

Perform inactivation study

Is API in process sample inactivated after exposure to cleaning conditions?1

1 Based on results of bioassay

Yes

No

Use CQ approach to set acceptance limit (AL) for inactivated API

Use MAC approach to set acceptance limit (AL) for API

continued

continued

Special edition: Cleaning Validation

69

Rizwan Sharnez and Angela To

Continued from page 69

Inputs • Dose and batch sizes of products • Surface area and rinse parameters of shared equipment • Recovery study for API • Carbon content of API

Use MAC approach to set acceptance limit (AL) for API

Is AL LOQ of TOC? No

Yes

Develop alternative assay to demonstrate clearance of API

Use TOC to demonstrate clearance of API

Inputs • Dose and batch sizes of products • Surface area of dedicated and shared equipment • Recovery study for inactivated API • Carbon content of inactivated API

Use CQ approach to set acceptance limit (AL) for inactivated API Can inactivated API be recovered by TOC? No

Yes

Is AL LOQ of TOC?

No

2 This

is done on a case-bycase basis

Yes Use TOC to demonstrate clearance of inactivated API

70  Special Edition: Cleaning Validation

Develop alternative assay to demonstrate clearance of inactivated API2

Rizwan Sharnez and Angela To

CONCLUSION The inactivation of a product during cleaning and steaming has important implications for cleaning validation of multiproduct equipment. Demonstrating that the product becomes inactivated during these operations obviates the need to perform arduos MAC assessments for the API. It also eliminates the need to develop PSAs for cleaning validation. PSAs are designed to detect specific epitopes; thus, if the API degrades, the result from the assay may not necessarily correlate to therapeutic activity. The CQ approach can be used to set acceptance limits for the carryover of inactivated product for multiproduct cleaning validation. This approach is designed to ensure that the amount of inactivated Product A in a dose of Product B is less than the amount of inactivated Product A in a dose of Product A. Application of the CQ approach to biopharmaceutical cleaning will be described in Part II of this series. REFERENCES 1. Sharnez, R., “Strategies for Setting Rational MACbased Limits–Part I: Reassessing the Carryover Criterion,” Journal of Validation Technology, Vol 16, No. 1, p. 71-74, 2010. 2. Sharnez, R., To, A., Klewer, L., “Strategies for Setting Rational MAC-based Limits–Part II: Application to Rinse Samples,” Journal of Validation Technology, Vol 17, No. 2, p. 43-46, 2011. 3. Sharnez, R., To, A., “Strategies for Setting Rational MAC-based Limits–Part III: Leveraging Toxicology and Cleanability Data,” Journal of Validation Technology, Vol 17, No. 3, p. 24-28, 2011. 4. Kendrick, K., Canhoto, A., Kreuze, M., “Analysis of Degradation Properties of Biopharmaceutical Active Ingredients as Caused by Various Process Cleaning Agents and Temperature,” Journal of Validation Technology, Vol 15, No. 3, p. 69, 2009. 5. Rathore, N., Qi, W., Chen, C., Ji, W., “Bench-scale characterization of cleaning process design space for biopharmaceuticals,” Biopharm Int, Vol 22, No. 3, 2009. 6. Martinez, J.E., “Immunogenic Potential of Therapeutic Protein Residues after Cleaning,” Bioprocess International, Vol. 9, P. 38-44, 2004. 7. Health Canada, Cleaning Validation Guidelines (GUIDE-0028); Section 8.3, Health Products and Food Branch Inspectorate, 2008. http://www.hc-sc. gc.ca/dhp-mps/compli-conform/gmp-bpf/validation/index-eng.php 8. ISPE, Risk-Based Manufacture of Pharmaceutical Products: A Guide to Managing Risks Associated with Cross-Contamination, 1st Edition, Vol. 7, ISPE, 2010. 9. Fourman, Gary L. and Michael V. Mullen, “Determining Cleaning Validation Acceptance Limits for Pharmaceutical Manufacturing Operations,” Pharmaceutical Technology, 17 (4), 54-60, 1993. JVT

ACKNOWLEDgEMENTS The authors are grateful to Arun Tholudur and Joel Bercu for their helpful suggestions and support. ARTICLE ACRONYM LISTINg A ADE AL API B

Product A–previously manufactured product Acceptable Daily Exposure Acceptance Limit Active Pharmaceutical Ingredient Product B–subsequently manufactured product CQ Comparable Quality EIA Enzyme Immunoassay ELISA Enzyme-Linked Immunosorbent Assay LOQ Limit of Quantitation MAC Maximum Allowable Carryover MW Molecular Weight PSA Product Specific Assay SDS PAgE Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis TOC Total Organic Carbon

Special edition: Cleaning Validation

71

Sebastian Clerkin

Validation of a Cleaning Process for Medical Devices | IVT Sebastian Clerkin, Ph.D.

AbStrACt Many medical device manufacturers find it a considerable challenge to plan and conduct a cleaning validation. The main challenges are an establishment of the cleanliness limits and an identification of the challenge conditions to be assessed during the process validation. This paper describes a logical risk-based approach to overcome these challenges. It begins with assessing the complete manufacturing process and identifying the manufacturing agents. It discusses risk tools to determine which manufacturing agents need to have cleanliness limits. It describes the manufacturing conditions to be considered when conducting the cleaning validation. The concepts described within this paper can be utilized by a medical device manufacturer to establish a cleaning process that will consistently provide clean medical devices and comply with the relevant regulations. INtrODUCtION Contamination of a medical device can have serious implications. Medical device manufacturers must ensure they have correctly identified all potential contaminants and have established controls. The United States Food and Drug Administration captures this requirement within the Quality System Regulations (QSR) by stating that each manufacturer shall (1, 2): Establish and maintain procedures to prevent contamination of product by substances that could be expected to have an adverse effect on product quality Establish and maintain procedures for the use and removal of manufacturing materials to ensure that it is removed or limited to an amount that does not adversely affect the device’s quality (2). International Organization for Standardization (ISO) 13485 requires that a medical device manufacturer establish documented requirements for the cleanliness of a medical device in the following circumstances (3): • Product is cleaned by the organization prior to sterilization and/or its use, or • Product is supplied non-sterile to be subjected to a cleaning process prior to sterilization and/or its use, or • Product is supplied to be used non-sterile and its cleanliness is of significance in use, or • Process agents are to be removed from product during manufacture. Therefore, to comply with the QSR and ISO 13485, a medical device manufacturer must establish documented cleanliness requirements. 72

Special edition: Cleaning Validation

Sebastian Clerkin

Figure 1: Example of a Manufacturing Process Flow.

However, these regulations do not explicitly state that a cleaning process validation must be completed. The Global Harmonization Task Force (GHTF) Study Group 3 does provide guidance on the requirement for a cleaning process validation. Their process validation guidance (4); which was written by regulators in the US, Europe, Japan, Australia and Canada; states a cleaning process may be validated or may be satisfactorily covered by verification. ISO 14696, which provides guidance on the application of ISO 13485, states that a cleaning process needs consideration of use and the controls in place to determine whether some or all of the elements of validation are required (5). Therefore, a decision to conduct a cleaning process validation is dependent on the outputs of the cleaning process and the controls in place. For example, Figure 1 shows a manufacturing process flow that has an intermediate cleaning step. The sole output of this intermediate clean step may be that the parts are visibly clean. In this instance, verification may be sufficient with no requirement for process validation. On the other hand, the manufacturer taking a risk-based approach has specified a level of acceptable residues after intermediate cleaning. In this instance, verification would not be sufficient and process validation would be necessary. In the above manufacturing process flow (Figure 1), the final cleaning step is the more critical cleaning step of the two, as this is the last step in the process to make sure that the medical devices are sufficiently clean prior to packaging. The cleaning validation approach described within this paper is more applicable to this final clean step. It consists of a number of logical steps from identifying the risks to establishing limits and from validation to process monitoring. revIew PrOCeSS FlOw AND IDeNtIFy the MANUFACtUrINg MAterIAlS In order to determine the cleanliness limits for manufacturing agents after the final cleaning process, it is imperative to first evaluate the complete manufacturing process. An overall process flowchart should be created to demonstrate that the process has been adequately assessed, and it should contain information on the manufacturing materials that come in contact with the product at each process step. Examples of manufacturing materials that must be considered include: lubricants, detergents, wipes used during inspection, and polishing agents. This manufacturing material maybe attributable to one of the following groups:

• An organic residual; these are mostly insoluble in water and include greases and oils • An inorganic residual; these are water soluble, and examples include metal ions • Particulate; an example would be metallic particles left over from a cutting process. Hazardous components of manufacturing materials can be obtained from the Material Safety Data Sheet (MSDS). The MSDS should be available for all the manufacturing materials used in the process. It must be noted, however, that the MSDS generally only lists the main components that are present in a mixture. For example, the United States Department of Occupational Safety and Health Administration (OSHA) only requires that hazardous constituents in excess of 1% be disclosed on a MSDS (6). For carcinogens it is 0.1%. This is due to the fact that MSDS are designed to protect the workers, not identify potential hazardous contaminants on a medical device. There is also a risk that multiple constituents below the 1% disclosure threshold could have a cumulative effect on the intended use of the device. Even though the MSDS does have these shortcomings, it is still a very useful initial tool in identifying potentially hazardous manufacturing agents. When dealing with complex manufacturing materials such as cutting fluids, it is a considerable challenge to identify the individual components as they contain many different chemicals. These chemicals may not be identified by the supplier or may be considered proprietary. In these instances, it is imperative to liaise directly with the manufacturing material supplier to identify any potential contaminants that could impact the intended use of the device. In addition, any process steps completed by an external supplier must also be evaluated. For example, manufacturing material may be present on a supplied component. These must be considered. This is why it is crucial that the medical device manufacturer has a written agreement in place with their supplier that no changes are made to their manufacturing materials without prior agreement. Only after the overall process flow has been completely reviewed and all the potential contaminants on the device identified can the manufacturer start to consider their impact. Their impact is assessed by conducting a risk analysis. Special edition: Cleaning Validation

73

Sebastian Clerkin

Process Step

Manufacturing Potential Material Hazards

Potential Effect

Potential Cause

S

Passivation

Passivation solution

Sodium dichromate (CAS#: 7789-12-0)

Carcinogen; Irritant; Mutagen

Residual 5 Sodium dichromate present on the finished device

Grinding

Grinding wheel

Metallic particulates

Third body wear of bearing surface

Metallic particulates present on finished device

4

L R

Mitigation

S

L R

2

10 Use a passiv- 5 ation solution that contains no Sodium dichromate

1 5

2

8

4

1 4

Establish a particulate specification for finished device

table 1: Hazard Analysis. Severity (S) is scored1 to 5, 5 being the most severe; Likelihood is scored 1 to 5, 5 being the most likely; Risk index (R) is calculated by multiplying the Severity score by the Likelihood score.

risk Analysis and Identification of Materials of Concern For the risk analysis, the impact of the contaminants from a hazardous perspective and from an intended functionality perspective must be considered. A useful tool in identifying which contaminants are of the most concern is to use a hazard analysis and to ask the following questions: • Will too much of this contaminant be harmful to the patient? • Will too much of this contaminant impact the proper functioning of the device?

calculate limits for the pre-identified manufacturing materials. However, in many instances, the NOAEL is not known, and so the manufacturer must rely on using the LD50 values. LD50 is the median lethal dose. In other words, the amount of a particular toxin that will kill 50% of the population over specified time duration. These LD50 values can be readily obtained from the MSDS. The LD50 values are then used to calculate the Acceptable Daily intake (ADI) using the following equation: ADI = LD50 x mB/CF Where:

An example of a hazard analysis is shown in Table I; the process steps, manufacturing material and particular agents are listed, and then the risk is considered. The mitigation from the hazard analysis can be used to establish a particular cleaning limit or use of an alternative manufacturing material. eStAblIShINg CleANINg lIMItS Now that the potential hazardous contaminants have been identified, the acceptable level of contamination on the medical device must be determined. These levels or limits must be documented and scientifically justified by the medical device manufacturer. For toxic contaminants where there is known toxicity data, ISO 10993-17 is very useful. It describes a method to determine the acceptable levels of leachable material from a medical device using the No Observed Adverse Effect Level (NOAEL) (7). The NOAEL is the highest concentration of a material that causes no significant adverse effects in the exposed population. The standard takes this value and uses it to calculate the tolerable intake (TI) for a specific leachable substance. This approach can be used to 74

Special edition: Cleaning Validation

LD50 = median lethal dose mB = is the body mass of the patient population and is generally defaulted to 70kg CF= conversion factor The conversion Factor (CF) is typically a factor between 100 and a 1000 and is derived to incorporate uncertainty factors (UF) such as: • Extrapolation from animal to human tolerances (typically defaulted to a factor of 10) • Inter-human variability (typically defaulted to a factor of 10) • Additional UFs can be based on the type of medical device (i.e., medical device class) and the duration of exposure. The weighting of each UF should be documented and justified (8). The UFs are then used to calculate CF: CF = UF1 x UF2 x UF3

Sebastian Clerkin

In most cases, a conversion factor between 100 and 1,000 is sufficient; however, there may be instances where significant risks have been identified and a CF as high as 10,000 may be appropriate (9). Refer to Kramer et al.for further information on conversion factors (10). Consider a real world example of the above approach using an alkaline cleaner: The CF has been established as 1,000 to account for no human toxicity data (factor of 10), inter-human variability (factor of 10), and a short exposure time of the device (factor of 10). The LD50of the alkaline cleaner is 365 mg/kgrat. The average human body weight is 70 kg. This would give: ADI/device = 365 mg/kg x 70 kg/1000 =25.55 mg/ device Therefore, the cleanliness limit for this chemical would be 25.55 mg per device. Using this approach, a cleanliness limit can be calculated for each specific toxin that was identified during the risk analysis. Obviously, this approach only identifies a cleanliness limit for known toxins. It is not suitable for calculating the cleanliness limit where there is a lack of toxicological data available or the contaminants have no associated toxicity but will impact the proper functioning of the device. For potential toxins where there is no readily available toxicological data, a series of spiking studies can be completed. This is where the device is artificially contaminated with known amounts of the potential toxin. Biocompatible studies can then be completed to determine the point of failure. The suite of ISO10993 standards provide a wealth of information that can be used to define the biocompatibility studies needed in establishing the failure point. Cytotoxicity, sensitization, systemic toxicity, and genotoxicity studies are examples of biocompatibility studies that could be considered. The established failure point can then be used to derive the cleanliness limit. Another approach to spiking studies is to approach the issue from the opposite end. In other words, instead of finding the failure point, the device is spiked with a known amount of the contaminant that is above the level expected to be observed after cleaning. If this higher level is established as safe for the patient, it can be defined as the cleanliness limit. Spiking studies can also be useful if the risk analysis has identified a potential cumulative effect of various contaminants. In other words, if each contaminant is treated independently of each other, a cleanliness limit may be established that does not take into account a potential cumulative effect. In this instance, the patient may be exposed to unacceptable risk. Spiking studies should also be considered for

contaminants where the LD50 value is known but the manufacturer has identified additional risks. For example, if a carcinogen has been identified as a potential contaminant, a spiking study may be instigated with a genotoxicity and/or carcinogenicity study as the endpoint. Working out the cleanliness limits for non-toxic contaminants, which have the potential to interfere with the proper functioning of the device, needs to be established by reviewing historical data or by spiking experiments. The spiking study is conducted in a similar manner as above, but instead of a biocompatibility study being the endpoint, the levels of contamination are now evaluated against a specific functional requirement for the device. For example, for a device that has a bearing surface and device failure attributed to particulate contamination, a spiking study may be developed to determine the level of particulate that will induce the failure. eStAblIShINg CleANINg teSt MethOD Now that the cleaning limits have been identified for specific manufacturing agents, the next step is to decide on a method to quantify the levels. There are two types of tests that can be developed: • A specific analytical test can be developed to quantify a contaminant (refer to Table II for examples). • A non-specific test can be developed to quantify many different contaminants at the same time (refer to Table II for examples). There are pros and cons for both these approaches. With respect to a specific test method, an accurate measurement of a particular residue can be evaluated. This can be very important when this residue has been identified as being highly hazardous. However, these specific methods are more difficult to implement and are more expensive; therefore, they are really only used when a specific risk has been identified during the risk analysis. Non-specific methods are more commonly used in validating cleaning lines; they are less expensive and easier to develop. However, due to their non-specific nature, they do not give an accurate quantification of any individual contamination, only a total level of a group of contaminants. Generally, this can be sufficient where the requirement is to demonstrate a certain level of overall cleanliness. When developing any test method, the following factors should be considered: • Detection Limit; for example, the test method must be sensitive enough to detect relevant levels of the contaminants. Special edition: Cleaning Validation

75

Sebastian Clerkin

Specific/NonSpecific

Test Method

Description

Specific

Infrared Spectroscopy

A spectroscopic technique. Useful to identify and quantify specific molecules

Grinding wheel

Metallic particulates

Grinding wheel

Metallic particulates

Gas Liquid Partition Chromatography (GLPC)

Useful for quantifying a specific contaminant that can be vaporized without composition

Specific

High-Performance Liquid Chromatography (HPLC)

A method that separates a mixture allowing individual components to be quantified

Non-Specific

Conductivity

Measure of ionic compounds

Non-Specific

Total Organic Carbon (TOC)

Quantitative measure of carbon contained within an organic compound. A very good method of showing overall cleanliness.

Non-Specific

Visual

Visual Assessment of cleanliness

Non-Specific

Gravimetric Analysis

Quantitative measure of mass of a solid within a solution

Specific

table 2: List of Test Methods. Each could potentially be used to assess outputs of a cleaning process.

• Percentage recovery; the amount of contaminants that can be recovered from the device must be determined. • Reproducibility and Repeatability • Linearity • Specificity. Once the test methods have been developed, they must be qualified prior to being used in a cleaning process validation. The test method validation must demonstrate that the analytical method and the extraction and/or sampling method is repeatable. With respect to the extraction method, it must be demonstrated that the contaminant can be consistently recovered from the medical device. There is no point in having a repeatable analytical method if the contaminant of interest cannot be consistently extracted off the device. Cleaning equipment Qualification (IQ) The cleaning equipment must be qualified prior to commencing the process validation. This will demonstrate that the equipment is installed correctly and functions as intended. Operational Qualification (OQ)—Challenge Conditions As part of the OQ phase, the critical process inputs should be identified. It must be established how these process inputs impact the cleanliness outputs. Brainstorming and tools such as fishbone diagrams can be very useful in identifying the critical process inputs. Some examples of potential process inputs that could be considered during a brainstorming are listed in Table III. 76

Special edition: Cleaning Validation

To adequately challenge a cleaning process during OQ, the worst-case product should be used. For medical device manufacturers that clean multiple different devices, identification of the worst-case product can be a challenge. To aid in this identification, a worstcase product matrix can be used to help determine the worst-case product. Each product is scored according to predefined criteria. These predefined criteria would have a potential impact on the cleaning ability of the process. Examples include the following: • Device geometry; more complex geometry is potentially more difficult to clean • Surface area; greater surface area could be a greater cleaning challenge. • Up-stream process flow; this could have an impact on the cleanliness levels of the parts precleaning. A greater level of contamination on the parts pre-clean would obviously be a greater challenge to the process. The scores against these criteria can then be tabulated in the form of a matrix and used to calculate a total risk score. See Table IV for an example. Once one has established the critical process inputs, these should be challenged within the OQ to prove that product can be cleaned that meets the predetermined cleaning specifications under all anticipated manufacturing conditions. PerFOrMANCe QUAlIFICAtION PQ means establishing by objective evidence that the process, under anticipated conditions, consistently produces a product that meets all predetermined requirements (11).

Sebastian Clerkin

Cleaning Process Inputs

Comment

Tank Temperature

Higher temperatures are generally more effective while lower temperatures are less effective.

Cleaning Time

Longer times are generally more effective while lower times are less effective.

Concentration

Usually the higher the concentration of cleaning agent the better; however, at higher concentrations there may be higher level of cleaning residual present on the parts post-cleaning.

Cleaning Solution Contamination

As parts are processed through the cleaning solution, it will become contaminated. There is a risk that as the cleaning tank becomes dirtier, its cleaning effectivity will decrease. To assess this, the change-out frequency of the tank, the level of contamination on the parts, and the number of parts processed between change-outs must be considered.

Flow Rate of Cleaning Agent

Higher flow rates are generally more effective while lower flow rates are less effective.

Agitation

Agitation should improve the cleaning effectiveness.

Ultrasonic

Ultrasonic should improve the cleaning effectiveness.

Load

A higher number of parts or dirtier parts may represent a greater challenge to the cleaning process.

Rinse Parameters

Rinse times, temperatures and volume are important in removing the cleaning agent from the product.

Drying Parameters

Time, temperature and airflow are critical parameters that can impact the dryness of the parts. The geometry of some parts, such as the internal diameter of a narrow tube, may make them difficult to dry. Another consideration is the temperature used for drying. Too high a temperature could have an impact on the material properties of the product.

table 3: Cleaning Process Inputs.

Worse Case Product Matrix Product Number

Surface Area

Device Geometry

Up-Stream Process Flow

Total Risk Score

1001

10

10

5

500

1002

7

5

10

350

table 4: Worst-Case Product Matrix. The product is scored between 1 and 10 for each criterion. These values are then multiplied together to give the total risk score. The product with the highest risk score is then considered the worse case product.

To establish this, the medical device manufacturer must demonstrate that the process in anticipated manufacturing conditions is stable and can consistently produce product to the predefined cleaning specifications. The manufacturer can use tools such as control charts to demonstrate that the process is stable as well as capability analysis to show that the process is consistent. With respect to identifying the anticipated conditions of a cleaning process, factors that could be considered are the following: • Material lots, for example, batches of cleaning agents • Environmental conditions • Multiple Shifts • Operators • Process operating windows defined in the OQ; for examples, see above.

PrOCeSS MONItOrINg After the cleaning process is validated, it is desirable to implement process monitoring of key process parameters and product outputs to maintain a state of control. The action levels can be established during the process validation. ISO 14969 specifically states that the process parameters used for cleaning should be routinely monitored in accordance with documented procedures (12). Examples of critical process inputs that could be monitored would be conductivity on a rinse tank or concentration of cleaning agent in a cleaning tank. The product outputs that would be monitored are risk-based and would relate to the cleaning specifications established previously. For example, the medical device manufacturer may consider monitoring the TOC of the product on a regular basis to maintain confidence Special edition: Cleaning Validation

77

Sebastian Clerkin

that the cleaning process is still in control. For newly developed cleaning processes, this monitoring can initially be at a higher frequency and reduced as the manufacturer gains more confidence in their process. revAlIDAtION Obviously, changes to the cleaning system and process should be assessed for their impact to the validation. In addition, any changes to the manufacturing materials used in the up-stream manufacturing processes may have an impact on the cleaning validation. This includes manufacturing steps completed by an external vendor. If the change is deemed to have an impact, the cleaning validation should be either repeated in part or in full. Again, this needs to be a risk-based decision. The addition of new products to the clean line should be assessed to determine if they are a new worst-case. If they are considered a new worst-case product, the cleaning validation should be repeated. A recommended approach to determining if revalidation is required with a new product is to update the worst-case product matrix described earlier with the new product. If the new addition scores higher than the original worst-case, then the cleaning validation must be repeated. CONClUSIONS The main steps in conducting a cleaning validation is to assess the manufacturing material for each manu-

facturing process step, establish cleanliness limits, and then validate the cleanliness test methods and the cleaning process itself (Figure 2). Cleaning validation does not need to be difficult. If medical device manufacturers take a methodical approach and base each decision on sound scientific rational, they will be able to establish a cleaning process that will consistently provide clean medical devices to the market. reFereNCeS Code of Federal Regulations, Title 21, Quality System Regulations, Part 820.70(e), 2013 2. Code of Federal Regulations, Title 21, Quality System Regulations, Part 820.70(h), 2013 3. ISO 13485:2003, Medical devices -- Quality management systems -- Requirements for regulatory purposes, section 7.5.1.2.1. 4. Global Harmonisation Task Force Study Group 3, Quality Management Systems Process Validation Guidance, January 2004 5. ISO 14969:2004, Medical devices -- Quality management systems -- Guidance on the application of ISO 13485: 2003, section 7.5.2.1.1.5. 6. Code of Federal Regulations, Title 29, Part 1910 - Occupational Safety and Health Standards, Subpart Z Toxic and Hazardous Substances, 2013. 7. ISO 10993-17:2009, Biological evaluation of medical devices -- Part 17: Establishment of allowable limits for leachable substances. 8. ISO 10993-17:2009, Biological evaluation of medical devices -- Part 17: Establishment of allowable limits for leachable substances. 9. ISO 10993-17:2009, Biological evaluation of medical devices -- Part 17: Establishment of allowable limits for leachable substances. 10. H.J. Kramer, W.A Van Den Ham, W. Slob, and M.N. Pieters, “Conversion Factors Estimating Indicative Chronic No-Observed-Adverse-Effect Levels from Short-Term Toxicity Data,” Regulatory Toxicology Pharmacology 23, 249-255, 1996. 11. Global Harmonisation Task Force, Quality Management Systems - Process Validation Guidance Edition 2, January 2004 12. ISO 14969:2004, Medical devices -- Quality management systems -- Guidance on the application of ISO 13485: 2003, Section 7.5.2.1.1.6. 1.

AbOUt the AUthOr

Figure 2: Overall Process Flow to a Cleaning Validation. 78

Special edition: Cleaning Validation

Sebastian Clerkin, Ph.D., B.Sc., is the Founder of GMP Advisory Services Ltd. and has more than 10 years experience in laboratory, pharmaceutical, and medical device manufacturing environments. He currently provides validation and regulatory expertise to the medical device industry. He has direct experience across a broad range of medical device types and process technologies. He has a keen understanding of FDA regulatory requirements, and has a strong background in process development and validation. He has a Ph.D. from the University of Bristol, which involved studying the toxicological and biological effects of prosthetic implants. He holds a B.Sc. in Biochemistry from University College Dublin, and is a member of AAMI and ASTM. E-mail: [email protected]

Tim Sandle

Ensuring Sterility: Autoclaves, Wet Loads, and Sterility Failures | IVT Tim Sandle, Ph.D.

ABSTRACT Steam sterilization is a critical process in the pharmaceutical and related industries. Modern autoclaves are computer-controlled and reliably provide a defined sterilization cycle. When steam enters the autoclave chamber and contacts with the item to be sterilized, steam collapses (condenses). Water formed must be discharged through condensate management or re-vaporized in order to prevent wet loads. Repeated occurrences of wet loads are indicative of a major fault with the sterilizer, potential non-sterilized materials, and other problems. This paper considers some of the potential causes for wet loads and addresses some of the measures that can be taken to address occurrences. Topics discussed include reasons for wet loads, causes including wet steam, inadequate condensate removal, steam trams, pressure control, and other causes; diagnosing problems by information collection, and corrective actions. Corrective actions may include vacuum drying, heating the load before steam introduction, an air in-bleed phase, and other approaches. Problems identified may be caused by a combination of factors requiring a multidisciplinary team to evaluate potential causes. INTRODUCTION The most widely used sterilization method in the pharmaceutical industry remains steam sterilization in autoclaves (moist heat in the form of saturated steam under pressure). This method is primarily applicable to the terminal sterilization of products, stainless steel items, and equipment not intended for single-use. With this method, sterilization occurs as the latent heat of condensation is transferred to the load causing it to heat rapidly (1). Modern autoclaves are computer-controlled and are generally very reliable. The autoclave acts as a pressure-cooker: Water boils at 100°C at atmospheric pressure; water boils at lower temperatures at lower pressures, and water boils at higher temperatures at higher pressures. At a steam over-pressure of one bar (a non-SI unit of pressure, exactly equal to 100,000 Pascals), water boils at approximately 121°C. This allows the autoclave to produce temperatures above those that can ordinarily be achieved. For sufficient time and with the correct conditions, such temperatures can destroy bacterial endospores (2). More importantly, the required temperature must be maintained for the required time in order for the sterilization cycle to be successful (3). Retrieving sterilized items from autoclaves only to discover that they are wet results in a need to repeat the autoclave cycle. A wet item is evidence of non-sterility. This is time consuming and expensive. This forms part of the general rule with packages removed from autoclaves: Any item that has been sterilized should not be used after the expiration date has been exceeded or if the sterilized package is wet, torn, or punctured (4). Special edition: Cleaning Validation

79

Tim Sandle

When steam enters the autoclave chamber and contacts with the product, it is important that the steam collapses (condenses) on the product. This is in order for the heat to be released to the load. However, the formation of water must be discharged through condensate management or re-vaporized in order to prevent contamination of the product. Removal of the excess water is important to prevent insulation of the load from the steam. Repeated occurrences of wet loads are indicative of a major fault with the sterilizer. Water or damp spots on the load insulate the intended product from achieving temperature. This means that the load has not been subjected to the required lethality and it is therefore not sterile. During sterilization, the wetness in the steam clogs the pores of packed loads and prevents the steam from properly penetrating wrapped loads or sealed pouches. Furthermore, there is a post-contamination risk. If the instruments or products absorb too much humidity, resulting in wet loads at process termination, this dampness is an optimal habitat for bacteria to thrive. Sterile items that become wet are considered contaminated because moisture brings with it microorganisms from the air and surfaces. Moreover, if the sterile barrier system is still wet, it has lost its bacterial barrier properties and there is a major risk of contamination. In addition to issues of sterility, wetness can cause corrosion or spotting on the instrument being sterilized. This can cause irreparable damage. Wet loads present a problem of non-sterility; a problem in terms of recontamination of a load; and damage can occur to equipment. This paper considers some of the potential causes for wet loads and addresses some of the measures that can be taken to address the occurrence of wet loads. REASONS FOR WET LOADS Water is heated up to the boiling point to generate steam. As water reaches its boiling point at approximately 100°C, the temperature will no longer rise; any energy added to the water will lead to evaporation of water thus creating steam. If steam touches a material or gas with a lower temperature than the temperature of the steam itself, the steam will give off energy and will condense and water is formed. In order to be able to evaporate the condensate energy is needed. The way to obtain sufficient energy is from its environment. The reason why wet loads occur is that the condensate becomes separated from the condensation energy and therefore cannot re-evaporate. To avoid wet loads, autoclave cycles must ensure that the condensate remains in contact with the materials it condensed on, or, alternatively, the process ensures 80

Special edition: Cleaning Validation

that all condensate that gets separated from its energy finds its way to the drain. Thus it is critical that the condensate made by each load item (including loading furniture and wraps) must not migrate to other objects. There are three possible scenarios for wet loads. These are: • Visible moisture on outside of packs • Moisture inside pack (e.g. moist towel) • Visible water inside the tray containing the load. Each of these is of equal seriousness. POTENTIAL CAUSES OF WET LOADS In examining the causes of wet loads there are several areas to consider. These are discussed below: Steam Supply Wet steam can also be associated with wet loads; therefore steam needs to have a certain level pf dryness (5). The moisture content of the steam (dryness fraction) is measured as the weight of dry steam present in a mixture of dry saturated steam and entrained water. Poor steam quality (insufficient dryness) can lead to excessive condensate formation within the autoclave. This can arise when excessive demands are placed on the steam supply (essentially the boiler). Wet steam is steam at saturation temperature containing more than 5% water (6). Wet steam lowers the heat transfer efficiency of steam, which results in an inefficient sterilization procedure. Steam quality can be assessed by measuring the condensate discharge just ahead of the sterilizer. The measurement of steam quality is known as the dryness fraction and this is dependent upon the steam flow rate (7) where ζ = dryness fraction ww = mass of water flow (kg, or lb / unit time) ws = mass of steam (kg or lb / unit time). For example, if the water content of the steam is 5% by mass, then the steam is said to be 95% dry and has a dryness fraction of 0.95. In this example, the steam quality would be considered unsatisfactory. A satisfactory dryness fraction is generally regarded to be 0.97 (8). Non-condensable gasses can also affect steam quality. Non‐condensable gases should be removed before the steam leaves the boiler. When they are not, they modify the steam from being pure water vapor to a mixture of steam and gas. These become an unwanted contaminant that does not allow steam to make

Tim Sandle

contact with the item. Further, there is the issue of superheated steam (discussed below), Areas of Condensate Formation Insufficient condensate being removed from the distribution system can result in in water lying around in the system and being drawn into the sterilizer when the cycle demands it (pulsing). Signs of condensation can sometimes be seen from an examination of the inside of the autoclave chamber. This may resemble water lines or circles of water droplets (9). When this occurs, the steam circuit and supply should be examined to determine where condensate formation might occur. As steam travels, there are many places and reasons why steam cools down and condenses. Reasons include dead legs in pipework or improperly trapped or insulated piping. When steam comes into contact with piping connections it can condense; moreover, poorly insolated piping will cause steam to condense. Some condensate is natural and is taken care of by steam traps which are placed throughout the piping system. This can also be due to inadequate maintenance of the system and steam traps (see below). The further away the steam line is from the heating source/medium, then the more likely condensate is to form. This is connected to insulation, for the further away the steam line is then the greater the need for insulation. Other Issues with Pipes and Steam Supply There are other factors relating to pipework and the supply of steam, which may need to be considered. These are: • Where steam pipes are not properly insulated steam will condense in the pipe • The water separator should be close to the autoclave • If pipes run down towards the sterilizer, this can cause steam to condense • The steam generator should be located close to autoclave, minimizing pipework length • If a steam generator does not have sufficient capacity for the autoclave, pressure drop may occur at times of peak demand for steam and a “carry over” of water can happen • If chemicals are added during the steam production, operational issues may occur. Steam Traps Steam traps are intended to discharge condensate while not allowing the escape of live steam. Problems can arise if the steam trap is not functioning correctly. To guard against this, the jacket steam trap should be inspected, along with the valve at the drain

port, to ensure that they are clear and functioning properly. Too much steam at too fast a rate can result in excessive water formation that might overwhelm or “swamp” steam traps. Therefore steam traps need to be properly sized during the autoclave design phase. Pressure Control Given that autoclaves destroy microorganisms by direct steam contact at the required temperature and pressure for a specified time, pressure control is important. Poor pressure control can cause variations in steam velocities. This often arises because maintenance of the pressure-regulating valves (PRV) in the system is not always reliable. PRVs can change the condition of the steam and, in theory, can actually dry the steam as it passes through the valve. Autoclave Loads What is being loaded into the autoclave can be a contributing factor. One mistake that some organizations make is to pack the chamber full of product or consumables in order to meet demand. However, dense packing is frequently a cause of excessive condensation. This cannot always be satisfactorily flashed off by subsequent steam injections. Furthermore, large quantities of materials and complex packaging can make effective steam circulation a challenge. It is also worth considering: • The assembly of instruments in the set. For instance, are the instruments well distributed in the set? Is there too much metal mass in the set? High density items wrapped in absorbent material can help to reduce condensate (10). • How the set was packaged. It is important to consider if the wrapper is too large (traps condensate inside set), whether the package was taped too tightly (traps condensate inside package), the size of peel pouches for load contents (too much metal mass in the pouch). In addition, before using wrapping material, it should be held at room temperature (20oC to 25oC) and at a relative humidity of 30 to 60%. • Loading techniques. Consider reducing the metal mass in the load and check whether linen items on the top rack and metal items on the bottom rack. • Rigid containers. Containers should be spaced about one inch apart from each other. Stacking should not be performed unless the container manufacturer gives specific information on this process. Place containers beneath other items since they produce condensate. • Sterilizers should never be overloaded. There should be sufficient space between items to allow for steam to permeate around the package. Special edition: Cleaning Validation

81

Tim Sandle

Packaging Materials Packaging materials, used to wrap the items to be sterilized, are probably one of the most important parts of the sterilization process. Packaging provides a protective barrier for the sterile item and prevents it from becoming re-contaminated. When used in aseptic processing, the packaging should be of low particle generation. Packaging needs to be well designed and of sufficient porosity so that steam can pass through packaging. In addition, packaging should be water repellent to provide a sterile “field.” Furthermore, the packaging material must not change significantly during sterilization or release any substances that might interfere with the action of the steam. Other Factors There are other factors that are linked with wet loads, although these are less common with a well-designed system. These are: • Autoclaves are equipped with heated jackets to assist in drying of the load and to ensure that condensate does not form on the chamber walls during the hold period. If the autoclave jacket has a different temperature than the chamber, wet loads may occur. To avoid this, larger steam jackets contain more heat and assist in the drying process. • The autoclave carriage can be a factor. A stainless steel carriage shelf creates more condensate than an equivalent aluminum carriage. To add to this, the shape and size of the chamber can greatly affect the outcome of the load. • The preheat process. Here the problems range from poor steam penetration, superheat, damaged sterile barrier systems and other problems. • When the autoclave doors have a different temperature than the chamber or walls. Or where the jacket temperature is not uniform or lower than intended. • When the item being sterilized touches the wall of the autoclave. • Stacking two trays for holding loads within the autoclave. The lower tray does not have sufficient energy to dry the condensation. • Items being loaded in wet (anything going in wet will come out wet.) When the steam makes contact with the wet instruments additional condensate and cause dripping on other instruments that may not dry. • Overloading the autoclave. Large quantities of hard goods or even complex packaging can make proper steam circulation a challenge. Other parameters that may influence drying are the density of the wraps and the design of the set 82 Special edition: Cleaning Validation

• The materials that make up the load, for example heavy metal mass can be a cause of wet packs (11). Further, some loads are erroneously designed where they become water traps or they are positioned in such a way the orientation leads them to retain water. • Condense drains can become blocked. It is important to clean these at least twice a year. If they become saturated they will not take the water out of the steam. • Sometimes the wet loads follow a pattern and may occur if the autoclave has not been used for a few days. Here the cause may be the build-up of condensate in the supply pipework. • When steam pipework is of an incorrect diameter. • If the drain valve is not closed, leading to drain water being returned. • Consistency of operator control of the sterilizers and loading patterns. There are some actions that mask the wet load issue rather than address it. One example is with the use of tray liners. Tray-liners do not solve the problem but disguises it. Tray-liners have a high absorption capacity and are placed in the tray under the items to be sterilized. This way the condense is absorbed and dispersed; however, non-sterility remains an everpresent risk. DIAGNOSIS Collecting meaningful information is important for establishing why wet loads occur. Whenever wet packs occur, some of the information that should be collected includes: • Date of wet pack(s) • Time of day • Load configuration (number of trays) • Number and description of trays that were reported as wet • Selected cycle time (gravity or dynamic air removal) • Cycle temperature and exposure time • Packaging material used for set(s) that were wet • Type of containment device used (e.g. Mayo tray, metal mesh basket) • Person(s) who prepared/wrapped sets • Inventory list for set (to consider set configuration and weight of set) • Vacuum integrity • Vacuum performance • Temperature/pressure calibration • Chamber level • Chamber/jacket trap performance • Chamber valves (solenoid and check)

Tim Sandle

• Jacket insulation integrity • Sterilizer exposure time • Sterilizer dry time. Analyzing such data can help to establish patterns and these can feed into the root cause analysis. CORRECTIVE ACTIONS In addition to running a diagnostic to address some or all of the above areas, there are other actions that can be considered in order to reduce the possibility of wet loads occurring. These include: Vacuum Drying In order to dry the load after the steam sterilization step, some users are tempted to employ excessively long deep vacuums at the end of the cycle. Because the chamber is heated via the jacketing and the fact that water will “flash off” into steam at a lower temperature under vacuum, this technique can be relatively successful. However, it will lead to an unnecessarily long cycle time, and a potentially a non-sterile load can still result since water can collect in certain areas. Reducing Condensate Formation A strong preventative measure to prevent the formation of condensate is to ensure that the load itself is brought up to a certain temperature prior to the introduction of steam. This can be accomplished by establishing a “load heat up” phase at the beginning of the cycle. The phase functions at the dictate of jacket set point, chamber drain (or load thermocouple) set point and a phase timer. The phase is not advanced until the set points are achieved and the phase timer elapses. From here it would then pass to a purge (or pre-vacuum) condition. Air-in Bleed Phase and the Problem of Superheated Steam An alternative approach is to add a heated air inbleed phase at the beginning of the sterilization cycle. For this, a heat exchanger can be installed to the air in-bleed assembly to enable a combination of vacuum and air in-bleed pulses. These would bring the chamber load up to temperature more quickly, reducing the condensate formation at the beginning of the cycle. While a heat exchanger is not absolutely necessary, it can be useful for particularly large chambers or dense loads. As an alternative to the heat exchanger, increasing the jacket temperature set point can be considered. Care must be taken to ensure that this does not interfere with maintaining the temperature uniformity within the chamber during the sterilize dwell phase. Merely increasing the jacket temperature will result in a “super-heating” of the steam.

Superheated steam occurs when the temperature of the steam is higher than its saturation pressure. This usually happens when the pressure of the steam drops as it moves through the steam supply system. This dries the steam and once there is a reduction in moisture, less energy is required to raise the temperature of the steam. An autoclave fed with superheated steam will function like a dry heat sterilizer, in which the killing of microorganisms is far less efficient than the optimal saturated steam required for sterilization. A further cause of superheated steam is if the reduction valve is too close to the autoclave. This valve should be at least 10 meters away. This is because where the distance is too short the temperature will not be reduced. A further reason relates to steam pressure, where this is too high. Although fundamental to the laws of physics, it should not be forgotten that temperature and pressure are related to each other. Here it follows that the higher the pressure then the higher the temperature. Charge Rate Control Some autoclaves do not have a proportional control valve. This provides the control system capability to control the “ramp up” (charge) rate of steam injection during the cycle. If a chamber does not have the means to control the rate that steam that is added to the chamber, this can lead to flooding of the chamber with steam. This, in turn, results in instant condensate formation. If the steam injection can be controlled so that it as it is “ramped up” more slowly, the amount of steam (and subsequent condensate) can be more easily controlled. Filter Sterilization Some autoclaves are fitted with an air in-bleed filter casing that can be sterilized in situ. Some manufacturers of autoclaves will sterilize this filter during each cycle, whereas other types require a special cycle to do this. This second approach reduces the wear on the filter element, resulting in a much longer life-time and prevents repeatedly saturating the filter with steam. This latter factor would result in excessive moisture being brought into the chamber at the completion of the cycle. Thus avoiding the in situ sterilization of the filter can be effective in avoiding excessive moisture. CONCLUSION This paper has addressed the issue of wet loads from steam sterilization through an autoclave. Although the probability of wet loads occurring can be avoided through good validation and effective cycle development, even autoclaves that have been in service for years can be subject to wet loads. This paper has outlined why wet loads are a major concern (nonsterility, the risk of re-contamination and damage to Special edition: Cleaning Validation

83

Tim Sandle

equipment) and it has presented some of the reasons why wet loads can occur. The paper has also outlined, in addition to addressing the potential causes of wet loads, some preventative measures that can be adopted. Humidity problems are a complex matter and are in the majority of cases created by a combination of different factors. Often wet load issues cannot be solved immediately and time is required, together with a multidisciplinary team, in order to address the problem. REFERENCES 1.

2.

3. 4.

84

Lemieux, P. (2006). Destruction of Spores on Building Decontamination Residue in a Commercial Autoclave, Appl. Environ Microbial, 72(12): 7687-7693 Sandle, T. (2013). Sterility, Sterilisation and Sterility Assurance for Pharmaceuticals: Technology, Validation and Current Regulations, Woodhead Publishing Ltd.: Cambridge, UK, pp93-110 Agalloco JP, Akers JE, Madsen RE. (1998) Moist heat sterilization-myths and realities. PDA J. Pharmaceutical Sci. Technol. 52:346-50 Dunkelberg H, Fleitmann-Glende F. (2006) Measurement of the microbial barrier effectiveness of sterilization containers in terms of the log reduction value for prevention of nosocomial infections. Am. J. Infect. Control 34:285-9.

Special edition: Cleaning Validation

5.

Baker, J. & Godfrey, G. (2009). Water quality systems in sterile processing. Healthcare Purchasing News, January 2009, 32‐35 6. Agalloco J. (2000). Steam sterilization and steam quality. Commentary. PDA J Pharm Sci. Technol.; 54(1):59–62 7. Shuttleworth K. (2000). The application of steam quality test limits, Eur J Parenter Pharm Sci.; 5(4): 109–14 8. Association for the Advancement of Medical Instrumentation. Steam sterilization and sterility assurance in health care facilities. ANSI/AAMI ST46. Arlington, VA, 2002:ANSI/AAMI ST46:2002 9. Brown, J.M. & Bliley, J. (2008). How to solve wet packs and evaluate water issues. Materials Management in Healthcare, July 2008, (50‐52). 10. Rutala WA, Weber DJ. (2000) Choosing a sterilization wrap. Infect. Control Today; 4:64,70 11. Chobin N, Furr D, Nuyttens A. (2004) Wet packs and plastic accessory cases. Infect Control Today, 24, 28-30

ABOUT THE AUTHORS Tim Sandle, Ph.D., Tim Sandle, Ph.D., is the head of the microbiology department at Bio Products Laboratory Limited, a pharmaceutical organization owned by the UK Department of Health. Dr. Sandle is, additionally, a visiting tutor at the School of Pharmacy, Manchester University.