OR WAIT 15 SECS
An adulteration limit of 100 ?g/25cm? (4 ?g/cm?) was proposed for pilot-plant facilities. The dynamic changes in equipment, formulation, and residue determination made implementation of a constantly changing, calculated adulteration limit impractical. A single adulteration limit was simpler to communicate and document, making compliance achievable. The limit would be used only after it was determined to be lower than a health-based evaluation and a visual-cleanliness assessment.
A well run cleaning-validation program requires a significant amount of planning and resources. Planning takes into account the size, configuration, and complexity of the manufacturing equipment; the physical properties of the soils encountered, which directly affect the cleaning procedure used; the detergent to clean the equipment; the type of swab or rinse sampling to capture residue levels; the analytical test methods to determine residue levels; the periodic monitoring of the system to ensure ongoing compliance; change control to address new residues and equipment; appropriate documentation; and training for personnel cleaning equipment and performing validation challenges. Required resources include equipment downtime for the validation, raw materials for the formulations tested, analytical instrumentation for analysis, detergents for cleaning, solvents for testing, and personnel to perform the validation.
The entire cleaning-validation program and its execution depend directly on the acceptable residue limit (ARL) for the formulation residue. The active pharmaceutical ingredient (API) residue is monitored because it is the most phamacologically active component of the formulation. The ARL established for a program must be scientifically justified on the basis of the needs and capabilities of the manufacturing facility. In its Guide to Inspection of Validation of Cleaning Processes (1), the US Food and Drug Administration stated that residue limits should be logical, practical, achievable, and verifiable. The agency did not intend to set acceptance limits for validating cleaning processes. One of the examples cited as a possible residue limit used in industry, however, was 10 ppm of carryover into the subsequent batch manufactured.
The purpose of cleaning validation is to prevent the cross-contamination of a drug formulation. The primary concerns of cross-contamination are an evaluation of the therapeutic-dose carryover or the toxicity of the potential contaminant. Any contaminant is undesirable, but one with a pharmacologic or toxicological effect is far more serious. The Food, Drug, and Cosmetic Act (2) states: "A drug ... shall be deemed to be adulterated if it consists in whole or in part of any filthy, putrid, or decomposed substance," that is, something that will cause an adverse pharmacological effect. This definition supports the pharmacological and toxicological concerns.
If, however, the amount of a residue is below the level at which it would have an unintended, adverse health effect, what is the allowable level from an adulteration standpoint? Should a health-based limit be the only limit for equipment cleaning? Should the analytical-method detection limit be the standard (3)? Or, if an intermediate adulteration limit exists, what rationale determines that level? In the past, analytical methodology had limited sensitivity, making health-based and adulteration-based contaminations essentially equivalent. The increased sensitivity of analytical methodology seemingly has created a contradictory situation. Modern analytical detection limits are far lower than the pharmacological levels of drugs, thus creating a divergence between the previously equivalent health-based and adulteration definitions.
A logical residue limit would be one that demonstrates no pharmacological or toxicological effect, without regard to analytical sensitivity. This would indicate that residue levels can be as high or as low as the health-based limit will allow. This approach, however, would allow equipment to be visually dirty for relatively safe residues, which indicates an inadequate cleaning procedure. A logical residue limit also should leave the equipment visually clean. Setting residue limits any lower than the health-based and visible levels does not appear to be necessary or logical and adds no additional value to the cleaning process.
Several cleaning-validation programs use the dual health-based–adulteration-based criterion (4–6). Evaluations based on toxicity, tablet weight, number of doses administered, swab recovery, swab area or equipment surface area, and batch size determined the ARLs from both an adulteration-based and health-based perspective. The lower of the two limits was the designated ARL for the formulation and the equipment.
The cleaning-validation program in our pilot-plant facility calculated ARLs for both the health-based and adulteration-based criteria and used the lower of the two (7). The dynamic nature of the pilot plant and the drug-development process necessitated regular re-evaluation of the ARL for each API. Each API required a health-based evaluation, but the majority of development compounds had such low toxicity that their adulteration limits were lower. A proposed alternative to the constant re-evaluation of adulteration limits was to use a constant adulteration limit of 100 μg/25 cm2 swab or 4 μg/cm2 . This technique would streamline the cleaning process and minimize potential errors without compromising quality or safety.
Health-based risk assessment
To implement an adulteration limit, it must be lower than the associated health-based ARL. The adulteration limit would be lower at the level at which compounds are not likely to be potent, highly toxic, or carcinogenic. A corresponding allowable daily intake (ADI) for this category of material is 100 μg/day (0.1 mg/day) (8).
The calculation for the health-based ARL with an ADI of 0.1 mg/day includes:
in which ADI is the allowable daily intake for the compound, SSA is the shared product-contact surface areas of the manufacturing equipment train, and recovery is the percentage of spiked material recovered for assay.
Table I shows the health-based ARLs with associated parameters for an ADI of 0.1 mg/day, tablet weight from 0.1–1.5 g and maximum daily dose from 1–10 tablets. The number of tablets per batch ranged from 266–10,000 (small) to 3600–160,000 (medium) to 46,666–1,000,000 (large), based on batch sizes of 0.4–1 kg (small), 5.4–16 kg (medium), and 70–100 kg (large), respectively. It was assumed that the next batch was manufactured in equipment with the same SSA.
Table I: Calculated health-based acceptable residue limit (Î¼g/swab).
The shaded portions of the small, medium, and large batch ranges in Table I fell below the proposed limit of 100 μg/swab (4 μg/cm2 ). The low end of each range assumed the smallest batch size, the largest tablet weight (1500 mg), and the highest number of tablets dosed (10).
In the pilot plants, about 5% of manufactured batches fell below the 100 μg/swab (4 μg/cm2 ) limit. The batches that fell below the 4 μg/cm2 limit were typically for clinical studies with large tablet sizes (>1000 mg), multiple-tablet doses (>6 tablets/dose) or small batch sizes (<500 g). Small batch sizes generally are for first-in-man studies or preclinical and Phase I studies. These clinical-trial programs are dosed on small populations to establish dose levels.
The risk of falling below the 4 μg/cm2 limit for compounds with ADIs >100 μg/day was small, based on the site data. Also, of the 1225 swab samples taken in support of cleaning validation, none failed the ARL for the compound tested, greater than 98% of the swabs were below 1 μg/cm2 , and more than 99.5% were below 4 μg/cm2 , further reducing potential risk.
Finally, a continuing program for monitoring cleaning effectiveness using visible residue limits (VRL) was conducted. Of the manufactured batches that fell below the 100 μg/swab (4 μg/cm2 ) limit, the highest VRL was 1.23 μg/cm2 , which is well below the health-based limit. Therefore, for compounds with ADIs >100 μg/day, it is extremely unlikely that the adulteration limit selected will be greater than the health-based limit.
The adulteration limit was used when it was lower than the health-based limit. For development compounds in the pilot plant (7), the adulteration limit originally was calculated using the following equation:
in which UAL was the upper acceptance limit; SSA the shared surface area, MBS the minimum batch size for the equipment train, and recovery the fraction of spiked material recovered for assay. The UAL of 10 μg/g (10 ppm) cited by FDA (1) was used in various cleaning-validation programs (3, 9, 10). The SSA was the combined product-contact surface areas of the manufacturing equipment train. The MBS provided the most conservative limit because any residue would be most concentrated in the subsequent batch. The swab area of 25 cm2 was used widely (3, 7, 11, 12) in industry.
Calculation of an allowable adulteration level was a logical cleaning limit for a pilot plant but in the long term proved to be impractical. The number of factors that went into the pilot-plant production schedule, the drug-development formulations, and the residue determination made an adulteration assessment a constantly changing number.
Number of pilot-plant programs. The pilot plant manufactured varied formulations of numerous compounds. The programs in the pilot plant increased with new compounds and decreased as programs ceased development or were transferred to commercial manufacturing. Schedulers, formulators, equipment cleaners, analytical chemists, and quality personnel were involved. The number of programs and personnel along with the associated ARL calculations, documentation, and communication made it difficult to maintain a consistent, compliant program.
Number of new compounds. The number of new compounds entering the pilot plant was significantly greater than the number of new compounds entering a commercial manufacturing facility. These new programs had to be included in the overall cleaning-assessment program. Validating a new compound required significant analytical method development and validation. In addition, the small, early-phase manufacturing equipment was in great demand, and extended downtime for cleaning validation support was problematic.
Subsequent product. The manufacturing schedule in the pilot plant was variable. Equipment was scheduled for use several weeks in advance, but other programs sometimes took priority. Even knowing the subsequent product was no guarantee that a particular formulation was the same as the previous one manufactured for the product. Calculating an ARL based on the subsequent product manufactured in the equipment was problematic.
Current equipment train versus subsequent equipment train. The equipment train was the order in which equipment was used to manufacture a formulation. Blending, granulation, roller compaction, drying, and tablet pressing were examples of unit operations that together manufactured a clinical formulation. The ARL calculation for the equipment train assumed that the same train was used for the subsequent product. This almost never was the case in a pilot plant, which made the value of the ARL limited. An alternative ARL calculation considered each individual piece of equipment without regard to the manufacturing train. This type of ARL consideration became exceedingly cumbersome without adding increased value to the ARL process.
Formulation changes. The formulation for each research compound evolved during development. A dry-filled capsule for a Phase I compound became a film-coated tablet in Phase II. Refinements in formulation composition also were common. Excipient levels changed to optimize the physical properties of the formulation. Formulation changes also resulted from scale-up issues. Formulation modifications often changed the ARL for the subject compound.
Establishing effective dose. The effective dose of an API was unknown when clinical trials began. Dose levels could cover several orders of magnitude for early clinical trials. The results of the ongoing trials determined the dose for the next phase of testing and the eventual market-dose level. Formulations of early-development compounds had different dose levels and batch sizes. The amount of API in a formulation had a direct effect on the ability to clean the equipment. Therefore, cleaning could be assessed after every batch, based on the API factors involved.
Scale-up. Each clinical trial required a larger batch size than the previous trial, which necessitated equipment with a larger capacity. The physical interactions of the formulation components for larger batches often resulted in changes to the formulation or required a different type of manufacturing equipment. Each time the batch size or the equipment train changed, the ARL was reassessed.
Effect of small batch size or unit operations. In the pilot-plant environment, initial batch sizes were very small, often on the order of several hundred grams. Calculating an ARL based on the smallest batch size reduced the cleaning limit to a low level and potentially affected a Phase II compound manufactured just before a Phase I compound. Similarly, calculating an ARL for each individual piece of equipment in the manufacturing train resulted in very low cleaning limits for small surface-area equipment.
Rather than calculate the ARL for every clinical batch manufactured, certain assumptions were made to generalize the equation. The most conservative assumption used the minimum batch size for the equipment. This assumption made the generalized ARL lower than a specifically calculated ARL under most circumstances.
The adulteration limit calculation was the following:
Table II shows the product-contact surface areas for a typical equipment train for Phase I (small), Phase II (medium), and Phase III (large) formulations. Table III shows the range of ARLs for a constant UAL of 10 μg/g, a swab area of 25 cm2 , and a recovery of 100%. The calculated adulteration limit varied from 17 to 216 μg/swab for the same compound, depending on the manufacturing train. The adulteration limit varied from batch to batch for the same compound, making reassessment a routine occurrence.
Table II: Product-contact surface areas (SSA) for typical equipment trains by phase.
The variable adulteration limit also brought into question the value of the calculated limit to the overall cleaning program. For small batches, the limit was far below the 10 μg/g level. For larger batches, the calculated adulteration limit was greater than the visually clean level, thus making it obsolete.
Table II: Product-contact surface areas (SSA) for typical equipment trains by phase.
Analytical limits. High-performance liquid chromatography and total organic carbon methods were used most frequently. Each analytical test method had very low detection limits, either in the ppm or ppb range. Analytical limits were lower than the calculated health-based and adulteration-based limits. On the basis of instrumental capabilities, the use of analytical limits was considered for the adulteration limit.
Using 0.1% of the subsequent API as the adulteration limit was not appropriate. The 0.1% limit, determined during release testing, was intended for qualifying impurities that were associated with the manufacturing process or related compounds, and not for extraneous impurities caused by cross-contamination. Acceptance limits should reflect the capability of the cleaning processes (13).
Visible residue limits
The determination and use of VRLs demonstrated that the vast majority of formulations and APIs had VRLs lower than 100 μg/25 cm2 swab (4 μg/cm2 ) (14, 15). Of the 54 formulations evaluated to date, all were well below 100 μg/25 cm2 swab. Of the 102 APIs, excipients, and detergents evaluated, only five excipients and one API had VRLs greater than 100 μg/25 cm2 swab. Limited applications of VRLs have saved resources without sacrificing quality (16, 17).
Swab area. An often cited adulteration limit was 10 ppm or 100 μg/swab, using a swab area of 25 cm2 for cleaning validation. This amount was a feasible limit. The swab area was not as important as the scientifically justified limit but was a practical compromise to obtain a representative residue sample against the occasional need to swab smaller pieces of equipment.
The factors that affect the adulteration calculation made it an impractical situation for a pilot-plant application. A constantly changing adulteration limit caused documentation problems and made compliance difficult to enact and enforce.
An alternative single adulteration limit was proposed for compounds with ADIs >100 μg/day. An adulteration limit of 100 μg/25 cm2 swab (4 μg/cm2 ) was satisfactory as long as the equipment was visually clean. This limit ensured that there were no toxicity cross-contamination problems and that the equipment was visually clean. A single, scientifically determined adulteration limit is logical, practical, achievable, and verifiable, making it a justifiable adulteration limit for a pilot-plant facility.
Richard J. Forsyth* is an associate director in global clinical GMP quality with Merck & Co., Inc., WP53C-307, West Point, PA 19486, tel. 215.652.7462, fax 215.652.7106, email@example.comAlain Leblanc is a facility manager at Merck Frosst Canada's Center for Therapeutic Research. Mark Voaden is head of service for validation and compliance for facilities with Merck, Sharp & Dohme in the United Kingdom.
*To whom all correspondence should be addressed.
Submitted: June 20, 2006. Accepted: Aug. 24, 2006.
Keywords: adulteration limit, cleaning validation, compliance, pilot plant.
1. US Food and Drug Administration, Guide to Inspection of Validation of Cleaning Processes, (Rockville, MD, Office of Regulatory Affairs, 1993).
2. Food, Drug and Cosmetics Act, Chapter V, Section 501(a)(1), 1938.
3. FDA, CDER Human Drug CGMP Notes, 9 (2) (Rockville, MD, Division of Manufacturing and Product Quality, Office of Compliance, Center for Drug Evaluation and Research, 2001).
4. G.L. Fourman and M.V. Mullen, "Determining Cleaning Validation Acceptance Limits for Pharmaceutical Manufacturing Operations," Pharm. Technol. 17 (4), 54–60 (1993).
5. A.O. Zeller, "Cleaning Validation and Residue Limits: A Contribution to Current Discussions," Pharm. Technol. 17 (10), 70–80 (1993).
6. K.M. Jenkins and A.J. Vanderwielen, "Cleaning Validation: An Overall Perspective," Pharm. Technol. 18 (4), 60–73 (1994).
7. R.J. Forsyth and D. Haynes, "Cleaning Validation in a Pharmaceutical Research Facility," Pharm. Technol. 22 (9), 104–112 (1998).
8. D.G. Dolan et al., "Application of the Threshold of Toxicological Concern Concept to Pharmaceutical Manufacturing Operations," Regul. Toxicol. Pharmacol. 43 (3), 1–9 (2005).
9. R. Baffi et al., "A Total Organic Carbon Analysis Method for Validating Cleaning Between Products in Biopharmaceutical Manufacturing," J. Parenter. Sci. Technol. 45 (1), 13–19 (1991).
10. R.J. Romanach et al., "Combining Efforts to Clean Equipment in Active Pharmaceutical Ingredient Facilities," Pharm. Technol. 23 (1), 46–58 (1999).
11. M.A. Stege et al., "Total Organic Carbon Analysis of Swab Samples for the Cleaning Validation of Bioprocess Fermentation Equipment," BioPharm 9 (4), 42–45 (1996).
12. D.A. LeBlanc, "Establishing Scientifically Justified Acceptance Criteria for Cleaning Validation of Finished Drug Products," Pharm. Technol. 22 (10), 136–148 (1998).
13. FDA, CDER Human Drug CGMP Notes, 6 (2), 5–6 (Rockville, MD, Division of Manufacturing and Product Quality, Office of Compliance, Center for Drug Evaluation and Research, 1998).
14. R.J. Forsyth, V. Van Nostrand, and G. Martin, "Visible Residue Limit for Cleaning Validation and its Potential Application in a Pharmaceutical Research Facility," Pharm. Technol. 28 (10), 58–72 (2004).
15. R.J. Forsyth and V. Van Nostrand, "Application of Visible Residue Limit for Cleaning Validation in a Pharmaceutical Manufacturing Facility," Pharm. Technol. 29 (10), 152–161 (2005).
16. R.J. Forsyth and V. Van Nostrand, "Using Visible Residue Limits for Introducing New Compounds into a Pharmaceutical Research Facility," Pharm. Technol. 29 (4), 134–140 (2005).
17. R.J. Forsyth, J. Hartman, and V. Van Nostrand, "Risk-Management Assessment of Visible-Residue Limits in Cleaning Validation," Pharm. Technol. 30 (9), 104–114 (2006).