Journal of Validation Technology , 02/01/2007 13 2
Validation of a method for the determination of polysorbate 20 residue for the support of the cleaning of pharmaceutical vial closures. Juarbe, Nieves *~|~*Strege, Mark *~|~*
COPYRIGHT 2007 Advanstar Communications, Inc.
ABSTRACT This report summarizes the development and validation of an analytical procedure to support cleaning validation of pharmaceutical closures (stoppers). Based on current guidelines by regulatory agencies, a sensitive and quantitative test is needed to demonstrate the post-cleaning removal of cleaning agents from vial closures prior to their use. Specifically, an analytical test was required for trace level quantitation of polysorbate 20 (PS20), a non-ionic surfactant used for stopper cleaning. Samples consisted of aqueous solutions from the extraction of rubber vial stoppers and swabs taken from surfaces of equipment such as stainless steel tanks and fiberglass reinforced polyester wash tubs. Total organic carbon (TOC) determination was employed for sample analysis because of ease-of-use, sensitivity, and fast analysis time. This investigation demonstrated the applicability and validity of using stopper extraction, swab sampling, and TOC analysis for the determ!
ination of residual PS20 in support of cleaning validation.
INTRODUCTION
Good manufacturing practice
(GMP) requires Pharmaceutical and Biopharmaceutical Industries to strive for the highest manufacturing standards. (1) All phases of the manufacturing process must be controlled for predictability and for delivery of a finished product that consistently meets predetermined quality standards and specifications. The cleaning processes for equipment and supplies in direct contact with the drug product are closely inspected, because an inadequate cleaning procedure can result in adulterated or contaminated product. Important factors to control include the cleanliness of both pharmaceutical closures used in parenteral drug products and the washing equipment (stainless steel tanks and fiberglass wash tubs) used to clean these closures. The Food and Drug Administration (FDA) and International Conference on Harmonization
(ICH) have published guidelines for validating cleaning processes. (2-6) Currently, rinse water and swab testing are the most common approaches used for cleaning validation and verification. This report describes a strategy to meet the challenge of verifying the cleanliness of vial stoppers, and the associated washing equipment, by demonstrating the post-wash removal of the cleaning agent polysorbate 20
(PS20). The common sampling practice for stainless steel and other surfaces that can be reached by hand is to swab representative areas that have been cleaned, extract the swab, and then analyze the extract to demonstrate the effectiveness of the cleaning procedure. Vial rubber stoppers present a unique challenge since they cannot be swabbed like other surfaces due to their size, shape, and surface characteristics. The extraction procedure and the solvent must be compatible with both the sample and the TOC assay. Therefore, various approaches were evaluated during development of the testing procedure. The safety acceptance criterion established for PS20 was <5[micro]g/[cm.sup.2]. Therefore, based on guidance from toxicology considerations and assessment of the compatibility of PS20 with process, product, and manufacturing components, the analytical test was required to provide a detection range that contained the acceptance limit of 125 [micro]g of PS20 per swab taken from a 25 [cm.sup.2] surface and 25 [micro]g of PS20 per stopper surface of 5 [cm.sup.2].
The surfaces of interest consisted of halobutyl and butyl elastomer rubber from the stoppers, fiberglass reinforced polyester from the washing tubs, and stainless steel from the tanks. Total organic carbon (TOC) was selected as the analytical technique for this study because it has been successfully used throughout the past decade to analyze aqueous-based cleaning samples for the presence of organics and was expected to provide adequate sensitivity. (7-9) A procedure for the extraction and determination of PS20 in the low parts-per-million (ppm) range on the surfaces of stoppers and cleaning equipment (stainless steel and fiberglass-reinforced polyester (FRP)) was developed and validated.
EXPERIMENTAL
Materials
The experimental setup utilized a sucrose standard (42.1% carbon by mass), the system suitability check reagent, parabenzoquinone (1, 4-benzoquinone, PBQ) (66.7% carbon by mass), and purified water. All glassware was rinsed with dilute aqua regia (1 part concentrated HN[O.sub.3]: 3 parts concentrated HCI), followed by a rinse with purified water. The sampling glass bottles (250 mL) with Teflon[R] lined closures were large enough to accommodate the stoppers. The bottles were also used to extract the samples and eliminated the need to transfer the stoppers. Texwipe Alpha[R] Swab TX761 swabs (ITW Texwipe, Kernersville, NC) were used to swab surfaces.
Equipment
A Shimadzu Model TOC-5000 TOC instrument with an ASI-5000 auto-sampler (Shimadzu Scientific Instruments, Columbia, MD) equipped with 4-mL vials was used for all data generation. An Elga Purelab Ultra-purified water system (Vivendi Water Systems, Bucks, UK) and laboratory bench wrist shaker were also employed for this study.
Method
* Establishment of the TOC Assay Range
As described in the introduction, the procedure was designed to provide an assay range that contained the acceptance limit of 5 [micro]g/[cm.sup.2], equivalent to 125 [micro]g of PS20 per swab for an area of 25 [cm.sup.2] surface and 25 [micro]g of PS20 per stopper with a surface area of 5 [cm.sup.2]. The Shimadzu 5000 TOC response is typically linear over a wide range, such as from 0.1 to 100 ppm C or greater. To enable the sample preparation procedure to extract a measurable amount of residual PS20 within the range of the instrument sensitivity (0.1 ppm C is an approximate limit of detection) and to increase the sampling cross-section of the stoppers' batches, multiple stoppers (four or ten) were combined as one composite sample based on the stopper size. See Figure 1. For example, a set of ten stoppers with an area of 5 [cm.sup.2] each, and a total area of 50 [cm.sup.2] extracted with 100 mL will have an assay limit concentration of 2.5 ppm C. Similarly, a swab used to !
cover an area of 25 [cm.sup.2], which is extracted with a 50 mL, volume will have an assay limit concentration of 2.5 ppm C (see Figure 2). It was also discovered that the sparging of PS20 solutions above 10 ppm resulted in inaccurate readings due to sample foaming resulting in sample loss. Therefore, method linearity was evaluated from 0.25 ppm to 5.00 ppm PS20, and the selected sample working range was from 0 to 5 ppm.
* Swab and Stopper Sample Preparation
Swab sampling took place using the following procedure: The fabric head of the swab was wetted with water (for swab spikes and FRP) or 0.012 N HC1 (for stainless steel); any excess liquid was removed by gently shaking the swab. A surface area of 25 [cm.sup.2] was swabbed, after which the swab stem was cut and each swab head was dropped into a glass bottle with a cap. Each swab sample was extracted with 50 mL of purified water by shaking for 10 minutes using a wrist action shaker. A 5 mL aliquot was immediately transferred into a sampling vial without transferring the swab head. Aliquot samples were then analyzed by TOC.
Stopper sets were extracted with 100 mL of purified water by shaking for 10 minutes using a wrist action shaker. Similarly, blank stoppers were spiked with purified water and extracted to generate blank samples. An aliquot of each extract was immediately transferred into a sampling vial. To minimize the effects of carbon leaching from the stoppers, the stoppers were not transferred with the testing aliquot. The aliquots were then assayed by TOC.
* Instrument Calibration and Controls
The TOC instrument was calibrated using a series of sucrose calibration standards, and the system was checked on a daily basis with the PBQ system suitability solution and a sucrose standard check covering a calibration range of 0.1 ppm to 50 ppm C. The instrument demonstrated a linear response to sucrose and PBQ, with correlation coefficients better than 0.995 for these compounds over this range.
RESULTS AND DISCUSSION
Method Development and Validation
* TOC Analysis
TOC analysis is a non-specific method theoretically capable of quantitation of the carbon content of any aqueous sample. The TOC-5000 instrument measures carbon through oxidation of organic carbon to carbon dioxide (C[O.sub.2]), which is subsequently detected by a C[O.sub.2]-selective detector in an effluent stream. The instrument is designed to measure TOC directly as "non-purgable organic carbon" after total inorganic carbon (carbonates and dissolved carbon dioxide) is purged from the sample matrix prior to analysis by helium sparging. Because TOC analysis is a non-specific analytical technique and measures the total carbon from all compounds in a sample, sample stability will not be an issue unless microbial contamination occurs before sample analysis.
* Linearity
Polysorbate 20 linearity was evaluated by analyzing five levels of PS20 (1.00, 1.75, 2.50, 3.75, and 5.00 ppm) in triplicate. Purified water served as the blank and diluent. The working range covered 40% to 200% of the target level of 2.5 ppm (as indicated in the section labeled, "Establishment of the TOC Assay Range," under the 'Method' heading). The percent carbon load for PS20 is 56.7% (see Figure 3 for calculations). The instrument demonstrated a linear response to PS20 and the fitted curve displayed a coefficient of determination of 0.9955 (see Figure 4). The slope value of 0.5164 compared favorably with the theoretical carbon fraction of 0.567 for PS20. Recoveries calculated from measured TOC vs. theoretical ranged from 80-100% for this data set (see Figure 5).
[FIGURE 4 OMITTED]
* Accuracy and Precision
Aqueous PS20 stock solutions were prepared at levels of 1000 and 5000 ppm. Surfaces were spiked with purified water for surface blanks, and with PS20 stock solutions corresponding to levels of 50, 125, and 250 [micro]g per swab (resulting in 40%, 100%, and 200% of the target concentration in the extract). A set of swabs was also directly spiked with aliquots of PS20 stock solutions at the same levels. The 125 [micro]g/swab recovery was determined separately by two analysts for the evaluation of intermediate precision. The surfaces were spiked with aliquots of stock solutions and were allowed to air-dry completely prior to swabbing. Six replicates were prepared at each level. To examine the efficiency of the swabbing procedure, a second moist swab was also collected to test for the presence of residues not removed by the first swab. No significant amount of PS20 was found on the second swabs under these conditions (data not shown).
Sets of ten stoppers were used for the 5 [cm.sup.2] stopper sizes (5 [cm.sup.2] x 10 = 50 [cm.sup.2] total area) and stopper sets of four stoppers were used for the 13 [cm.sup.2] sizes (13 [cm.sup.2] x 4 = 52 [cm.sup.2] total area), as indicated in the area calculations of Figure 2. Stoppers sets (n=10) were spiked with a 500 ppm PS20 stock solution to achieve final PS20 concentrations of 10, 25, and 50 [micro]g per stopper (5 [cm.sup.2]), equivalent to 2, 5, and 10 [micro]g /[cm.sup.2]. Stopper sets (n=4) were spiked with a 500 ppm stock solutions of PS20 for a final PS20 concentration of 25, 62.5, and 125 [micro]g per stopper (13 [cm.sup.2]), equivalent to approximately 2, 5, and 10 [micro]g /[cm.sup.2] at the target level. The 62.5 [micro]g/stopper recovery was determined separately by two analysts for the evaluation of intermediate precision. Each stopper was directly spiked with aliquots of 500 ppm PS20 stock solution and the liquid was allowed to evaporate completely!
prior to extraction. Six replicate sets of spiked stoppers were prepared at each level.
Accuracy was determined via the calculation of recoveries across the range studied for each surface. The swabs directly spiked with PS20 demonstrated recoveries ranging from 65-75% across levels (see Figure 6). Swab recoveries from surfaces ranged from a low of 56.6% for fiberglass-reinforced polyester surfaces spiked at the low level to a high of 78.1% for stainless steel surfaces spiked at the high level (see Figures 7 and 8). Accuracy was also evaluated through the calculation of individual stopper recoveries across levels (see Figure 9). Recoveries ranged from a low of 64.1% to a high of 112%. The precision for spiked swabs was calculated using the six replicates prepared at each level, and ranged from 8.0% to 10.8% across all levels. The precision for the swabs taken from the stainless steel and fiberglass-reinforced polyester surfaces ranged from 4.4% to 14.5%. The precision of the recoveries of PS20 from the various stopper materials ranged from 3.5% to 19.4% RSD.
Assay precision repeatability was determined for recoveries from surfaces by preparing a set of six replicate samples by two analysts for each surface and for Stopper 1. Intermediate precision was demonstrated by comparing the data from each analyst for swab spike recovery performed at 125 [micro]g/swab; the pooled RSD (n=12) obtained was 8.9% with an average recovery of 64%. Similarly, the intermediate precision was evaluated for stopper spike recoveries for each analyst at 62.5 [micro]g/stopper; the pooled RSD obtained was 13% with an average recovery of 95%.
* Limit of Detection and Limit of Quantitation
Limit of detection (DL) and limit of quantitation (QL) for the Shimadzu instrument were calculated using six low level replicates of PS20 solution at the 1.75-ppm level with the following equations, where S is the slope of the calibration curve:
DL = Std. conc. x [%RSD/33%] = 3[delta]/S
QL = Std. conc. x [%RSD/10%] = 10[delta]/S
See Figure 10 for the results of this determination. Based on these results, the limit of detection for the Shimadzu instrument was set at 2.5 ppm of PS20.
Optimization of Stopper Extraction
Carbon contamination from stopper handling, packaging components, leaching of organics, and stopper batch/lot variability were anticipated to contribute significantly to the TOC signal, and therefore, a background correction was implemented.
In order to obtain a uniform background control for each stopper type, a cleaning procedure with 0.01% (w/w) PS20 was designed. A set of 100 5 [cm.sup.2] stoppers was cleaned by soaking in the 0.01% (w/w) PS20 solution at 80[degrees]C for 15 minutes, and then rinsed with at least 1 liter of purified water to ensure that no residual PS20 or particulates were present. The cleaned blank stoppers were air dried, and stored for later use in a sealed glass container. Sets of uncleaned stoppers that were spiked with PS20 had replicate preparation variability from 17% to 57% RSD, whereas clean stoppers that were spiked in a similar manner had an RSD that ranged below 10% for replicate preparations. These results suggested that the uncleaned stoppers contained varying levels of organic contaminants that were removed through the cleaning process.
The stoppers were extracted by placing them in an appropriate container and accurately adding 100 mL of purified water. The sample size consisted of ten stoppers for the 5 [cm.sup.2] stoppers and four for the 13 [cm.sup.2] stoppers. The sample was then shaken with a mechanical wrist shaker for 10 minutes, after which an aliquot was immediately removed for testing.
In order to establish extraction time for sample preparation, stoppers were soaked in the extraction solution for a period of time that would represent a practical length of time required by the analyst to perform the preparation steps. TOC responses indicated that stoppers can be in extraction solution for up to 60 minutes with no noticeable background increase due to leaching of organic constituents from the stoppers (Figure 11). The variability observed between measurements across the hold times fell within the expected precision of the TOC responses for the low levels of carbon (below the quantitation limit) in these samples.
Following the cleaning, a study was done to determine the volume and duration of rinsing needed to remove PS20 from the stoppers. The stoppers were washed with the 0.01% PS20 solution and rinsed four times in fresh volumes of 200 mL of purified water for 10 minutes. An aliquot was taken from each rinse and assayed for TOC. A final water sample was collected after stoppers were soaked in the rinse for 24 hours to evaluate for leaching of organics from the stoppers. Figure 12 presents the levels of PS20 plotted against stages of rinsing. The first 200 mL rinse removed most of the cleaning agent added; the highest level of carbon detected was 1.75 ppm C versus the approximate 50 ppm C originally present in the 0.01% PS20. The TOC values for the water rinses stabilized after subsequent 200 mL rinses at an average response of 0.5 ppm C, corresponding to the background measurement for clean stoppers.
Soaking the stoppers in water for an extended time of 24 hrs was also evaluated, and the TOC response within the stoppers was observed to increase up to 1.45 ppm C (Figure 12). This increase was likely due to stopper leaching.
CONCLUSION
Laboratory procedures for the extraction and determination of the residual cleaning agent PS20 on swabs and vial stoppers by TOC analysis were developed and validated. These procedures were designed to provide an assay range that contained the acceptance limit of 5 [micro]g of PS20 per [cm.sup.2] surface area (equivalent to a prepared sample concentration of 2.5 ppm) for both stoppers and processing equipment surfaces. Surface recoveries for PS20 ranging from 56% to 92% were obtained for stoppers, stainless steel, and FRP. The response of the TOC instrument for PS20 was determined to be linear over the range of 1-5 ppm, with limits of detection and quantitation corresponding to 0.7 and 2.2 ppm, respectively, based on signal-to-noise estimation.
ACKNOWLEDGEMENT
The authors wish to acknowledge Dr. Ed Tidswell for supplying stoppers and the FRP wash tub and for his contributions to the establishment of the surface acceptance limits for PS20.
[FIGURE 12 OMITTED]
REFERENCES
1. FDA. Current Good Manufacturing Practices for Finished Pharmaceuticals
. 2. FDA. Guide to Inspections of Validation of Cleaning Processes
. The Food and Drug Administration. www.fda.gov/ora/inspectref/igs/valid.html.3. FDA. Guide to Inspections of Bulk Pharmaceutical Chemicals
. The Food and Drug Administration. www.fda.gov/ora/inspectref/igs/bulk.html.4. BioPharm May 2002 Info #50 Cleaning Validation
5. FDA. Cleaning Procedures and Protocols in Focus.
6. Amer, G. and Deshmane, P. "Ensuring Successful Validation: The Logical Steps to Efficient Cleaning Procedures." BioPharm Vol. 14, No. 3. 2001. pp. 26-32.
7. Baffi, R., Dolch, G., Garnick, R., Huang, Y., Mar, B., Matsuhiro, D., Niepelt, B, Parra, C., and Stephan, M. "A Total Organic Carbon Analysis Method for Validating Cleaning Between Products in Biopharmaceutical Manufacturing." J. Parent. Sci. Technol. Vol. 45, No. 1, 1991, pp. 9-13.
8. Strege, M., Stinger, T., Farrell, B., and Lagu, A. "Total Organic Carbon Analysis of Swab Samples for the Cleaning Validation of Bioprocess Fermentation Equipment." BioPharm Vol. 9, No. 4, 1996, pp. 25-42.
9. Wallace, B., Stevens, R., and Purcell, M. "Implementing Total Organic Carbon Analysis for Cleaning Validation." Pharm. Tech. Aseptic Processing, May 2004, pp. 40-44.
ABOUT THE AUTHORS
Nieves Juarbe is an Assistant Senior Analytical Chemist and member of the Manufacturing Science and Technology Analytical Sciences laboratory at Eli Lilly and Company. She has an MS degree in Chemistry from the University of Georgia at Athens.
Mark Strege is corresponding author for this article; he is a Senior Research Scientist and a team leader within the Manufacturing Science and Technology Analytical Sciences laboratory at Eli Lilly and Company. Mark has an MS degree in Analytical Chemistry from Purdue University. He can be reached by email at strege_mark_a@lilly.com or by telephone at (317) 276-9116.
Article Acronym Listing
<pre>
C[O.sub.2] Carbon Dioxide
DL Limit of Detection
FDA Food and Drug Administration
FRP Fiberglass Reinforced Polyester
GMP Good Manufacturing Practice
ICH International Conference on Harmonization
PBQ 1, 4-Benzoquinone
PS20 Polysorbate 20
QL Limit of Quantitation
REF Recovery Efficiency Fraction
RSD Relative Standard Deviation
TIC Total Inorganic Carbon
TOC Total Organic Carbon
Figure 1 Stoppers and Surfaces
Sample Size
Vendor (stoppers or
Item (item code) Description swabs)
Stopper 1 West Pharma, Butyl / 13.16 [cm.sup.2] 4
ws-450
Stopper 2 West Pharma, Butyl / 5.48 [cm.sup.2] 10
ws-491
Stopper 3 Hospira, Halobutyl / 4.55 [cm.sup.2] 10
88-1839
Stopper 4 West Pharma, Halobutyl / 4.58 [cm.sup.2] 10
ws-808
Stopper 5 Hospira, Halobutyl / 4.55 [cm.sup.2] 10
88-1856
Stopper 6 West Pharma, Halobutyl / 5.48 [cm.sup.2] 10
ws-375
Stainless Steel N/A 5x5 cm coupons created 1
in-house
Fiberglass- N/A washing tub 1
Reinforced
Polyester
(FRP)
Figure 2 Surface and Stopper Area Calculations
Description Extraction
Material Estimated Surface Area Volume
Swab from 25 [cm.sup.2] 50 mL
Stainless
Steel Surface
Swab from FRP 25 [cm.sup.2] 50 mL
Surface
Stopper 1 13 [cm.sup.2] x 4 stoppers = 52 [cm.sup.2] 100 mL
Stopper 2 5 [cm.sup.2] x 10 stoppers = 50 [cm.sup.2] 100 mL
Stopper 3 5 [cm.sup.2] x 10 stoppers = 50 [cm.sup.2] 100 mL
Stopper 4 5 [cm.sup.2] x 10 stoppers = 50 [cm.sup.2] 100 mL
Stopper 5 5 [cm.sup.2] x 10 stoppers = 50 [cm.sup.2] 100 mL
Stopper 6 5 [cm.sup.2] x 10 stoppers = 50 [cm.sup.2] 100 mL
Figure 3 Conversion Calculations
Carbon load calculation for Polysorbate 20 FW = 1228
Chemical formula: [C.sub.58] [H.sub.114] [O.sub.26]
% Carbon load = (58 x 12/1228) x 100% = 56.7%
Figure 5 Polysorbate 20 Linearity and Recovery
TOC
PS20 Actual Prep Average Response Theoretical
Conc Conc TOC Blank PS20 as
(ppm) (ppm) Response Adjusted Carbon % Recovery
Blank Diluent 0 0.323 0 0
1.000 1.009 0.777 0.453 0.567 80
1.750 1.816 1.236 0.913 1.017 90
2.500 2.422 1.478 1.1547 1.356 85
3.750 4.036 2.385 2.062 2.260 91
5.000 5.033 3.009 2.686 2.819 95
Key for Results:
1. TOC response blank adjusted = result (ppm C) - mean blank signal
(ppm C)
2. Calculated ppm C PS20 = (corrected ppb C)/(0.567)
Figure 6 Spiked Swab Recoveries
Target PS 20 Spike Actual PS 20 Spike Mean % Recovery
Level (mcg/swab) Level (mcg/swab) (n=6) % RSD
50 50 71.3 8.6
125 126 66.3 8.0
125 126 62.5 9.5
250 252 74.1 10.8
Figure 7 Stainless Steel Surface Recoveries
Target PS 20 Spike Actual PS 20 Spike Mean % Recovery
Level (mcg/swab) Level (mcg/swab) (n=6) % RSD
50 54 56.7 6.3
125 135 66.1 6.7
250 271 78.1 7.9
Figure 8 Fiberglass-Reinforced Polyester Surface Recoveries
Target PS 20 Spike Actual PS 20 Spike Mean % Recovery
Level (mcg/swab) Level (mcg/swab) (n=6) % RSD
50 54 56.6 9.3
125 135 60.2 4.4
250 271 60.6 14.5
Figure 9 Spiked Stopper Recoveries
Target PS 20 Spike Actual PS 20 Spike Mean % Recovery
Level (mcg/swab) Level (mcg/swab) (n=6) % RSD
Stopper 1
25.0 27 97.2 12.5
62.5 68 101.8 6.1
62.5 63 89.0 9.8
125.0 135 97.5 16.1
Stopper 2
10 10 64.1 19.4
25 25 66.3 6.8
50 50 67.8 4.9
Stopper 3
10 10 84.7 8.8
25 25 80.0 3.5
50 50 73.6 9.8
Stopper 4
10 11 112.1 8.9
25 27 107.1 7.3
50 54 91.8 15.8
Stopper 5
10 10 100.0 16.5
25 25 86.4 4.6
50 50 78.4 4.6
Stopper 6
10 10 82.4 7.7
25 25 72.3 7.2
50 50 72.4 10.5
Figure 10 Average Limit of Detection (DL) and Limit of Quantitation (QL)
PS20 DL and QL Estimates
DL 3 x sigma/slope 0.372 ppm C 0.66 ppm PS20
QL 10 x sigma/slope 1.242 ppm C 2.21 ppm PS20
Figure 11 Stopper Extraction plus Holding Time
TOC Response, ppm C
ITEM + 0 min. hold + 30 min. hold 60 min. hold
Stopper 1, n=2 0.4 0.3 0.4
Stopper 2, n=2 0.5 0.4 0.5
Stopper 3, n=2 0.8 0.7 0.7
Stopper 4, n=2 0.3 0.5 0.4
Stopper 5, n=2 0.8 0.7 0.9
Stopper 6, n=2 0.4 0.4 0.4
Control Water 0.3 0.2 0.3 </pre>
InfoTrac OneFile (R)
No comments:
Post a Comment