Cleaning Validation - A lifecycle approach.pdf
- 1. Cleaning Validation - A Lifecycle Approach Ned Wyman Principal Scientist, MS&T/Validation AstraZeneca Operations - Supply Biologics 20 Aug 2015
- 2. 2 Objectives Cleaning Validation Overview Process Validation Lifecycle Cleaning Validation Life Cycle – Stage 1 Small-scale studies Biotech specific studies Cleaning Validation Life Cycle – Stage 2 The initial cleaning validation acceptance criteria Biotech approach Cleaning Validation Life Cycle – Stage 3 Statistics and cleaning validation Trending of cleaning validation data Summary
- 3. 3 The AstraZeneca Biologics Frederick Manufacturing Center Manufacturing Facility (~350,000 ft 2) ISPE’s Facility of the Year - 2011
- 4. 4 Cleaning Validation Overview Guidance for Industry: Q7A Good Manufacturing Practice Guidance for Active Pharmaceutical Ingredients (Aug 2001) • Cleaning procedures should normally be validated. • Validation of cleaning procedures should reflect actual equipment usage patterns. • The cleaning validation protocol should describe the equipment to be cleaned, procedures, materials, sampling methods, acceptable cleaning levels, parameters to be monitored and controlled, and analytical methods. • Equipment cleaning/sanitation studies should address microbiological and endotoxin contamination for those processes where there is a need to reduce total microbiological count or endotoxins.
- 5. 5 Cleaning is a Process Cleaning validation plays an important role in reducing the possibility of product contamination from pharmaceutical manufacturing equipment. It demonstrates that the cleaning process adequately and consistently removes product residues, process residues and environmental contaminants from the manufacturing equipment / system, so that this equipment / system can be safely used for the manufacture of specific subsequent products (which may be the same or a different product). PDA Technical Report No. 29: Points to Consider for Cleaning Validation (2012)
- 6. 6 Process Validation Life Cycle • Stage 1 – Process Design: The commercial manufacturing process is defined during this stage based on knowledge gained through development and scale-up activities. - Demonstrate a Understanding of the Process • Stage 2 – Process Performance Qualification (PPQ) or traditional Process Validation: During this stage, the process design is confirmed as being capable of reproducible commercial manufacturing. - Demonstrate Process Robustness at Commercial-scale • Stage 3 – Continued Process Verification (CPV): Assuring that during routine production the process remains in a state of control. - Demonstrate Understanding of Process Variability at Commercial-Scale FDA Guidance for Industry: Process Validation: General Principles and Practices (2011)
- 7. Stage 1: Cleaning Process Parameters: • Remember TACT: Time Action Concentration Temperature • Varying the cleaning process attributes directly impacts cleaning effectiveness • Varying these attributes directly impacts cleaning effectiveness. 7
- 8. Stage 1: Equipment Related: 8 • Materials of construction are identified for all pieces of equipment. • Equipment is designed/built with proper slopes to allow for free draining. • Spray coverage tests are performed for all systems/equipment with spray devices installed. • Identify areas that are difficult to reach or hard to clean. Varying these attributes directly impacts cleaning effectiveness.
- 9. Stage 1: Product Related: • Cleaning solutions are selected based upon the chemical and physical properties of the in-process materials, buffers and actives to ensure adequate removal of the residuals. • Analytical methods used to test cleaning samples are validated. • Product specific small-scale studies are performed to ensure the cleaning process will adequately remove process related residues as well as demonstrate adequate swab recovery. 9
- 10. Stage 1: Biotech Inactivation Approach • Demonstrate that active ingredient is pharmacologically inactive as a result of the cleaning process: • Extreme pH (<2 and >13) • Temperature (60 - 80°C) • Known degradation/denaturation/inactivation • Demonstrate that the inactivated product has been removed below a predefined acceptance limit. • 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. • Removes the need to develop product specific immunoassays for cleaning validation 10
- 11. Guidance for Designing Denaturation Studies • Design bench scale experiments to simulate full scale conditions • Conditions should be the least conducive (worst-case) for inactivation • Simplify the bench scale studies: • If product inactivation increases with shear rate, then it can be eliminated • Ratio of cleaning solution to process soil to be reduced • Acid wash and rinse steps can be eliminated to minimize dilution of the process soil • Include a neutralization step to demonstrate that the denaturation is not reversible • Ensure modifications do not result in experimental artifacts 11
- 12. Analytical Methods for Denaturation Studies • Methods should • Evaluate fragmentation and inactivation of the API at bench scale • Detect impurities in the cleaning validation sample • Additional Method Considerations: • Sodium Dodecyl Sulfate Polyacrylamide gel electrophoresis (SDS-PAGE) - 4 to 20% gradient corresponds to a molecular weight range of 4 – 250 kDa, sufficient for most biological APIs. • Capillary Electrophoresis (CE) - Greater sensitivity, lower variability - Higher throughput capability • SDS-PAGE and CE - Adequate for demonstrating distinct, size-based separation of protein fragments • Size Exclusion High Pressure Liquid Chromatography (SE-HPLC) - Can also be used for size based separation - Difficult to obtain a distinct size based separation across a wide range of fragment sizes. 12
- 13. Product Denaturation Study Trace 13
- 14. Cleaning Validation Life Cycle – Stage 2 Equipment Related: • Performed on actual commercial-scale equipment / system. • Acceptance criteria initially set by equipment capability • Matrix / Family approach to optimize the number of trials. Product Related: • Assess biochemical properties of the product and process intermediates. • Verify that the acceptance criteria demonstrate adequate removal. Cleaning Parameters: • Verify that the TACT parameters are consistent • 14
- 15. 15 Dirty Equipment Removal of most of product Removal of any remaining residual Cleaning agents/ residuals Removed Initial water rinse (cold) Cleaning agents Final WFI rinse (hot) Swab worst-case locations Cleanliness verified Stage 1: Cleaning Process: Rinse, Clean, Rinse, Verify
- 16. Stage 2: Typical approaches to establish Product Specific Cleaning Validation Acceptance Criteria Product Specific approaches: • 0.1% (1/1000th reduction) • Health Based Limits: • LD50 • NOAEL • ADE/PDE • TTC • Minimum therapeutic dose 16
- 17. Stage 2: Cleaning Limit Determination 17 Clean Equipment Last rinse = HWFI Swab worst-case locations for residual carbon Assume residual is uniformly distributed over total surface area < Final TOC Action Limit
- 18. Stage 2: Cleaning Limit Determination 18 Clean Equipment Last rinse = HWFI Swab worst-case locations for residual carbon Assume residual is uniformly distributed over total surface area < Final TOC Action Limit TOC AC ≤ X ppm x Volume of Desorption Solution (mL) x 1µg/mL Surface Area Swabbed (cm2) ppm
- 19. Calculation of a Permitted Daily Exposure (PDE) 19 PDE = NOAEL x Weight Adjustment F1 x F2 x F3 x F4 x F5 EMA: Guideline on setting health based exposure limits for use in risk identification in the manufacture of different medicinal products in shared facilities (20 Nov 2014) F1: A factor (2-12) to account for extrapolation between species F2: A factor of 10 to account for variability between individuals F3: A factor of 1-10 to account for toxicity studies of short duration F4: A factor (1-10) that may be applied in cases of severe toxicity F5: A variable factor that may be applied if the no-effect level was not established.
- 20. Typical approach to establish Multi-Product Cleaning Validation Acceptance Criteria A common approach to determine the cleaning validation acceptance criteria has been the Maximum Allowable Carryover (MAC). 20
- 21. Typical approach to establish Multi-Product Cleaning Validation Acceptance Criteria A common approach to determine the cleaning validation acceptance criteria has been the Maximum Allowable Carryover (MAC). The MAC calculation is as follows: Where: PDE a = Permitted Daily Exposure for Product A Volume b = Volume of first B batch (or minimum Batch Size of Product B) Dosage b = Maximum dose of Product B SSA e = Surface area of shared equipment 21 MAC = PDE a x Volume b Dosage b x SSA e
- 22. PDE/MAC Approaches from a Biotech Perspective • Both PDE and MAC based upon the assumption that the product remains active after cleaning. • In Biotech, API is typically denatured during the cleaning process • Calculated acceptance limits are often below the limit of quantitation (LOQ) of non-specific methods • Large Surface Areas, small batch sizes and low concentrations in Biotech while using non-product specific testing (e.g., TOC) to determine PDE and MAC, the AC could be below the LOQ for TOC • LOQ for TOC is typically between 0.05 and 0.2ppm • Use of product specific immunoassays (PSIA) • Recognizes epitopes; however, epitopes are known to be denatured by the cleaning process. Therefore, results can be misleading. 22
- 23. Biologics Cleaning criteria approach is different due to denaturation of the active material. 23 EMA “Guideline on setting health based exposure limits for use in risk identification in the manufacture of different medicinal products in shared facilities” (released 20-Nov-14) states: 5.3 Therapeutic macromolecules and peptides Therapeutic macromolecules and peptides are known to degrade and denature when exposed to pH extremes and/or heat, and may become pharmacologically inactive. The cleaning of biopharmaceutical manufacturing equipment is typically performed under conditions which expose equipment surfaces to pH extremes and/or heat, which would lead to the degradation and inactivation of protein-based products. In view of this, the determination of health based exposure limits using PDE limits of the active and intact product may not be required. EudraLex, Volume 4, “EU Guidelines for GMP for Medicinal Products for Human and Veterinary Use, Annex 15: Qualification and Validation” (released 30-Mar-15) states: 10.6.1 Therapeutic macromolecules and peptides are known to degrade and denature when exposed to pH extremes and/or heat, and may become pharmacologically inactive. A toxicological evaluation may therefore not be applicable in these circumstances.
- 24. Biologics approaches to establish Cleaning Validation Acceptance Criteria is different Biopharmaceutical approaches: • Proteins are degraded/denatured/inactivated by cleaning solutions at elevated temperatures • As each facility has unique characteristics and products manufactured, variables to consider at each facility are also unique. • These approaches are not intended to be inclusive of all acceptable methods to determine cleaning limits • Examples of alternative approaches that are scientifically based: • WFI Limits • Safety Factor • Toxicology Threshold 24
- 25. Cleaning AC: WFI Specs • Equipment can not be cleaner than the last solution to contact their surfaces. • The last solution of the cleaning process is Purified Water (typically WFI). • Purified Water contains an allowable and known/monitored amount of organic carbon (≤ 500 ppb TOC ). Maximum Surface Residual TOC (ng TOC/cm2) = Equipment Volume (mL) x Purified Water Alert Limit (ng TOC/mL) Equipment Surface Area (cm2) Residual TOC Swab Limit (µg TOC/swab) = Maximum Surface Residual TOC x SSA (cm2/swab) x 1 µg /1000 ng Note that the approximate amount of carbon in protein is 50% 25
- 26. Cleaning AC: WFI Specs (example) Maximum Surface Residual TOC (ng TOC/cm2) = Equipment Volume (mL) x Purified Water Alert Limit (ng TOC/mL) Equipment Surface Area (cm2) = 25,842 mL x 250 ng TOC/mL 3,916.45 cm2 = 1649.58 ng TOC/cm2 Residual TOC Swab Limit (µg TOC/swab) = Maximum Surface Residual TOC x SSA (cm2/swab) x 1 µg /1000 ng = <1649.58 ng TOC/cm2 x 25 cm2/swab x 1 µg /1000 ng = <41 µg TOC/swab (assuming 25 cm2 area swabbed) Remember that the approximate amount of carbon in protein is 50% 26
- 27. Cleaning AC: Safety Factor • The cleaning verification limit determined by the Safety Factor Approach calculates the reduction of the inactivated product at the acceptance criteria level as an organic impurity in the Drug Substance or Drug Product. • This organic impurity limit is 0.10% which is the equivalent to a Safety Factor of 1,000 Safety Factor = Concentration (mg/mL) x 1 ppm x 1000 µg x 50 % TOC Limit (ppm) x 1 µg/mL x 1 mg Where the approximate amount of carbon in protein is 50% Residual TOC Swab Limit = TOC Limit (ppm) x 1 µg x Volume desorption solution (mL) ppm x Surface Area swabbed (cm2) 27
- 28. Cleaning AC: Safety Factor (example) Safety Factor = Concentration (mg/mL) x 1 ppm x 1000 µg x 50 % TOC Limit (ppm) x 1 µg/mL x 1 mg = 100 mg/mL x 1 ppm x 1000 µg x 50 % 2 ppm x 1 µg/mL x 1 mg = 25,000 Where the approximate amount of carbon in protein is 50% Residual TOC Swab Limit = TOC Limit (ppm) x 1 µg x Volume desorption solution (mL) ppm x Surface Area swabbed (cm2) = 2 ppm x 1 µg x Volume desorption solution (mL) ppm x 25 cm2 = 60 µg / 25 cm2 28
- 29. Cleaning AC: Threshold of Toxicology Concern • Denatured biopharmaceutical product fragments may be considered to be Class I chemicals with a residual soil threshold of 100 µg / day. • Compounds that are not likely to be potent, highly toxic or carcinogenic a Acceptable Residual Limit (ARL) (µg/cm2) = 100 µg/day x minimum batch size (mg) Dose (mg/day) x surface area (cm2) Residual TOC Swab Limit (µg TOC/swab) = Acceptable Residual Limit (µg/cm2) x SSA (cm2/swab) x 50% Where the approximate amount of carbon in protein is 50% a Reference: JVT Nov 2012 Vol 18 Methodology for Assessing Product Inactivation During Cleaning Part II: Setting Acceptance Limits of Biopharmaceutical Product Carryover for Equipment Cleaning 29
- 30. Cleaning AC: Threshold of Toxicology Concern (example) Acceptable Residual Limit (ARL) (µg/cm2) = 100 µg/day x minimum batch size (mg) Dose (µg/day) x surface area (cm2) = 100 µg/day x 400,000 mg 50,000 µg/day x 28,573 cm2 = 28 µg/cm2 Residual TOC Swab Limit (µg TOC/swab) = Acceptable Residual Limit (µg/cm2) x SSA (cm2/swab) x 50% = 28 µg/cm2 x 25 cm2/swab x 50% = 700 µg/swab (assuming 25 cm2 area swabbed) Where the approximate amount of carbon in protein is 50% 30
- 31. Cleaning Validation Life Cycle – Stage 3 Equipment Related: • Maintain matrix / family approach. • Periodic re-validation against established acceptance criteria to verify the system is still within the validated state. • Assess process variability based upon historical data • Develop Alert (Control) Limits as appropriate. Product Related: • Evaluate new products against worst-case soilant: • Small-scale coupon studies • Large-scale verification studies • Perform additional validation studies as required. Cleaning Parameters: • Maintain parameters in validated state • Modifications handled through the change control system 31
- 32. CPV and Statistics in General • Statistics are to be used to determine the amount of variability within a process. • Process Capability (Cp, Cpk) and Process Performance (Pp, Ppk) are typically good tools used to perform this assessment: • Cpk is “short-term” in that assesses data within a sub-group • Ppk is “long-term” in that assesses data within the complete group • Establish Control Limits and follow trending rules (e.g., Western Electric) • These statistical tools are associated with a normally distributed dataset. 32
- 33. Cleaning AC: Performance Control Limit • Performance Control Limits may be considered once cleaning validation studies have been completed and routine cleaning consistently meets established acceptance limits. • Performance Control Limit approach should not change the rationale for acceptable cleaning validation/verification acceptance criteria. • Performance Control Limit establishes a limit that may be more reflective of the performance of the cleaning process. • The Performance Control Limit may be considered an Alert Limit: Enables detection of a change in the performance of the cleaning process Enables proactive investigation into a potential cleaning process issue. 33
- 34. Cleaning Validation and Statistics – Can they co-exist? The challenge with the data typically from effective cleaning processes is that the data are not normally distributed. 34
- 35. Western Electric Rules don’t really apply 35 • One data point outside ± 3 s • Two out of last three points outside on same side of ± 2 s • Four out of last five points outside on same side of ± 1 s • Eight points in a row on same side of center line • Six consecutive points in an increasing or decreasing direction • Fifteen consecutive points within one sigma zone • Fourteen consecutive points alternating direction • Eight consecutive points outside one sigma zone
- 36. Can the Cleaning Data be made “Normal” • Even though the cleaning validation data are not normal; theoretically, the data may be made normal. The data maybe transformable (i.e., other statistical models may be used). • A control limit at three standard deviations from the mean ensures a false out of tolerance (OOT) rate of 0.27% or 0.0027 (the alpha rate). • Since dataset contains an excessive number of zero values (or <LOQ), the “0” values should be removed and the alpha rate adjusted accordingly. • The Box-Cox method computes the lambda value to optimize normality using the following equation: Ytransformed = Yoriginal^ lambda – 1 lambda Where, Yoriginal is each TOC value, which must be greater than 0 (e.g., LOQ) 36
- 37. Normal Assessment of TOC Swab Data Data from: Methodology for Assessing Product Inactivation During Cleaning and Setting Cleaning Verification Limits – BPOG Multi-Product presentation The top is a histogram of the original TOC dataset. A normal probability plot of the same non-normal phenomenon (bottom)
- 38. Box-Cox Assessment of TOC Swab Data Performance Limits are then back-calculated to the original scale using the transformed dataset and the equation below: Yoriginal = (Ytransformed * lambda +1)(1/ lambda) The Box-Cox transformed data are presented in the top plot. The Box-Cox transformed data are normally distributed (bottom)
- 39. Transformation of real-life data 39
- 40. Does Transformation work in this example? 40 The data were evaluated by the Anderson-Darling normality test, which indicated that the data significantly deviate from each of the theoretical distributions (all p-values < 0.05)
- 41. Where to go from here: • Since the data do not follow a Normal, Exponential, Weibull, or Gamma distribution, one-sided parametric tolerance limits (95% and 99%) could not be estimated. • Instead, estimate potential Control (and potentially the Alert) Limits using the empirical distribution function such as the 99.9th and the 99th percentiles. • Evaluate the potential Limits against the process and data to determine if appropriate and adjust as necessary. 41
- 42. Develop Control (Alert) Limits 42 Acceptance Criteria: ≤ 150 µg/25 cm2 99.9% percentile: ≤ 95 µg/25 cm2 99 % percentile: ≤ 15 µg/25 cm2 Do either of these Control Limits appear to be appropriate?
- 43. Refine Control (Alert) Limits 43 Acceptance Criteria: ≤ 150 µg/25 cm2 99.9% percentile: ≤ 95 µg/25 cm2 99.7% percentile: ≤ 60 µg/25 cm2 99.5% percentile: ≤ 37 µg/25 cm2 99 % percentile: ≤ 15 µg/25 cm2 These additional percentiles provide more options for the analysis
- 44. What’s the next step? Now that the Control Limit has been accepted (99.5% percentile which equates to ≤ 37 µg/25 cm2), data can be evaluated from various perspectives • Maintain the scientifically derived Acceptance Criteria • Perform further assessment of historical data to identify trends • Assess future results against the Control Limit to identify potential trends before exceeding the Cleaning Validation Acceptance Criteria • Potentially apply modified trending rules such as: • One data point above the Control Limit • A number of data points from same data set above the Alert Limit • A number of points from consecutive data sets for a specific piece of equipment or family/matrix above the Alert Limit 44
- 45. Trend Cleaning Data by Time 45 Acceptance Criteria: ≤ 150 µg/25 cm2 99.9% percentile: ≤ 95 µg/25 cm2 99.7% percentile: ≤ 60 µg/25 cm2 99.5% percentile: ≤ 37 µg/25 cm2 99 % percentile: ≤ 15 µg/25 cm2 2010 2011 2012 2013 2014 2015
- 46. Trend Cleaning Data by Campaign 46 Start-up Campaign 1 Campaign 2 Campaign 3 Campaign 4 Campaign 5 Campaign 6 Campaign 7 Acceptance Criteria: ≤ 150 µg/25 cm2 99.9% percentile: ≤ 95 µg/25 cm2 99.7% percentile: ≤ 60 µg/25 cm2 99.5% percentile: ≤ 37 µg/25 cm2 99 % percentile: ≤ 15 µg/25 cm2
- 47. Trend Cleaning Data by Product vs. Time 47 Start-up Product A Product A Product B Product A Product C Product D Product A Acceptance Criteria: ≤ 150 µg/25 cm2 99.9% percentile: ≤ 95 µg/25 cm2 99.7% percentile: ≤ 60 µg/25 cm2 99.5% percentile: ≤ 37 µg/25 cm2 99 % percentile: ≤ 15 µg/25 cm2
- 48. Trend by Sample Location 48 • Sample location 7 as well as locations 9, 13 and 15 are of interest (and maybe 6, too) • Sample locations 1, 3, 8, 11 and 14 are well controlled • Sample locations 5 and 10 are not included in this cleaning family
- 49. 49 Summary Cleaning is a process • Cleaning is a process; therefore, utilizing the Lifecycle Approach for cleaning validation is appropriate. • Discussed Equipment, Process and Cleaning Parameters over the course of the Cleaning Validation Lifecycle: • Examined how the need of the small-scale studies support the at-scale validation studies. • Stressed the importance of the tracking/trending of the cleaning data following the initial validation studies. • Highlighted the need to be open to assess the cleaning validation data from various perspectives. • Traced the Biotech approach over the entire cleaning validation lifecycle.