Testing a New Chromatography Column for Cleaning Effectiveness

Cleaning validation is a critical consideration in the pharmaceutical industry. Inadequate cleaning can result in contamination of drug products with bacteria, endotoxins, active pharmaceuticals from previous batch runs, and cleaning solution residues. Such contaminants must be reduced to safe levels, both for regulatory approval and to ensure patient safety.

Regulatory agencies worldwide require validation of pharmaceutical cleaning processes. None, however, offer a specific formulation for validation. Cleaning processes vary according to the nature of the drug being produced, the type of equipment being used, and whether equipment is dedicated or multi-use, among other variables. Thus, a one-size-fits-all validation strategy is impossible. A validated pharmaceutical cleaning process must use principles that are scientifically sound and reproducible. It is the pharmaceutical company's responsibility to set acceptance criteria and to explain the scientific basis for those limits to the regulatory authorities.

Companies must have written standard operating procedures (SOPs) in place that detail the cleaning processes used for individual pieces of equipment or systems. Validation protocols also must describe each operator's responsibilities, and include details of acceptance criteria and timelines or circumstances for revalidation. Protocols should contain sampling procedures, analytical methods and their sensitivities, and descriptions of methods and materials, including cleaning agents used in all processes. Instructions should be included for disassembly and reassembly of equipment when necessary, removal of previous batch identification, protection of equipment from contamination before use, and inspection of equipment immediately before use, as well as maximum time lapse permitted between the end of a production process and the start of cleaning procedure.¹

In addition, pharmaceutical companies must provide the regulatory authorities with a validation report, approved by management, which demonstrates the effectiveness of the cleaning process. The report must include data indicating that residues specific to the drug product and process in question are reduced to acceptable levels by the cleaning process.

Such regulatory scrutiny of cleaning processes means that pharmaceutical companies must carefully consider the cleanability of new equipment and to the systems already in place for cleaning existing equipment.

Construction material should be a primary consideration in selecting equipment and establishing cleaning protocols. Equipment should not be reactive, additive, or adsorptive with the process materials that contact them. Materials such as seals, valves, gaskets, hoses, or tank surfaces that come in contact with drug products cannot cause contamination. Possible interactions of equipment materials with cleaning products must also be examined.

Other things to consider for equipment design include difficult- to-clean areas such as piping, connections, and valves. Surfaces should be smooth and the system should have no dead legs where sediment can settle. Piping and instrumentation diagrams can identify problem areas.

Depending on the drug product being manufactured, different cleaning methods and analyses may be required. First one must identify the substances to be removed. The chemical and physical properties of the residues determine the best method to remove these from equipment surfaces. Some characteristics to consider include solubility, hydrophobicity, and reactivity. In addition, removal of the cleaning agent will have to be demonstrated, so its composition must be known.

CASE STUDY: A CHROMATOGRAPHY COLUMN

Cleanability — Considerations for New Equipment

The following example details cleaning protocols for a chromatography column. While some specifics may vary, the procedures will apply to cleaning most pharmaceutical equipment.

Chromatography columns frequently are located at the end of the process train, where purity is essential. Columns have certain intrinsic cleaning challenges, making them an exceptional test case for cleaning effectiveness. Packed beds influence hydraulic dynamics within a column, making it difficult to achieve the high linear velocities necessary for clean-in-place (CIP) procedures. Intraparticulate accessibility varies, creating sampling difficulties because the effectiveness of CIP is based on exposure. Ligands may be sensitive to cleaning agents and resin affinities may bind biomolecules that should be removed. O-rings, seals, and bed supports can be difficult to clean thoroughly and ancillary equipment such as pumps, valves, hoses, and safety devices must be sanitized in isolation. T-intersections and dead-legs in connecting tubing also may require special attention.

Despite the difficulties, CIP procedures are desirable to minimize cleaning-associated downtime, reduce material expense, reduce operator exposure, and avoid repacking. Also, standardized CIP procedures prevent repeated tests and judgment errors. Cleaning out of place (manually) requires additional equipment and time, resulting in efficacy variance. Disassembly is labor intensive and excludes resin from concurrent sanitization.

The column tested in this study was built to minimize inherent cleanability problems. It includes a new adjuster arrangement instead of a conventional O-ring and a wedge-shaped seat seal designed to maximize flow in the seal area. Other design enhancements result in improved flow rate and eliminated settling.

Figure 1. Washout of phenol red from a chromatography column.

Test plan. Following the design and prototype manufacture, the column's cleanability was tested at a small scale using a variety of tools and challenges. The tests are described in detail. This plan can be applied to maximize equipment performance and to validate cleanability.

• Dye testing was administered to visually inspect flow coverage.
• Computational flow design helped identify problem areas and model potential design solutions.
• Riboflavin clearance testing and visual inspection with UV excitation helped evaluate the cleaning strength needed and to assess the ability to remove strongly adhesive contaminants.
• Endotoxin and bacteria assays were conducted to test inactivation of pyrogens and removal of microbiological organisms.

Dye removal test. To determine the overall flow coverage in the system and how well-swept the wetted parts were, packed resin was slurried in a dye–salt solution, and then washed out with water. The washout was monitored with conductivity and the bed dissected for traces of dye. The derived peak is indicative of pack symmetry. The test is inexpensive, simple to perform, and non-hazardous. It should be noted that its qualitative nature might irreversibly bind media and identify only gross flow problems.

Figure 2. The riboflavin residue detected on a screw showed a design weakness, which was corrected

Dye testing was performed with phenol red. Figure 1 shows that no visible dye remained in the column; the column passed the test.

Riboflavin Clearance Testing. Rib-oflavin clearance testing was conducted to test relative cleanability and identify problem areas in the system. Accessible wetted surfaces were spray-painted with riboflavin ethanol (EtOH) solution; 100 ppm riboflavin was recirculated throughout the system. The system was then flushed with a cleaning solution. Removal was monitored with inline UV measurements. All wetted surfaces were examined with UV and the system was disassembled to inspect seal interfaces. The industry's standard acceptance criteria require that no riboflavin be present under visual or UV excitation; UV traces return to baseline. Level of detection (LOD) is 500 ppb for absorbance, –5 ppb for fluorescence. This test is very sensitive, relatively inexpensive, and identifies specific problem areas. It is also non-hazardous and an industry standard. However, it is a comparative, indirect test because riboflavin is not a natural contaminant.

Figure 2 shows that the riboflavin test revealed a weakness in the original design of the column, a dead space where the bed supports were attached to the column. The edge supports were redesigned and the problem resolved.

Endotoxin spiking. To determine the capacity to inactivate or remove endotoxins from a column in situ, the column was spiked with endotoxins and subsequently cleaned. Columns are acrylic 450 mm, glass 100 mm; and the medium used was Matrex Cellufine CG700. To simulate a worst-case scenario, the test used a higher concentration of endotoxins than would ever be encountered in production and more volume than the column could hold. The system was spiked with >1.0 x 10⁴ EU/mL (670 ng/mL) solution. A typical cleaning procedure followed, after which elluent samples were analyzed for endotoxins with a kinetic assay (limulus amebocyte lysate, LAL). Industry criteria are 5 log reduction in concentration (endotoxin units per millimeter). The United States Pharmacopoeia (USP) recommends a <0.25>2 This test is a quantitative, realistic challenge, but relatively expensive. It also exposes a column to material that could contaminate the entire system; thus, endotoxin spiking is prohibited.

The static inactivation trials, performed with no solution replenishment, and gentle shaking at room temperature with this buffer, showed that 1.0 M offered significant enhancement over 0.5 M NaOH. Positive product controls (PPCs) demonstrated negligible enzymatic inhibition. The study of concentration-dependent inactivation kinetics indicate that only 1.0 M NaOH achieves inactivation <>

Figure 3. Endotoxin removal from chromatography columns.

Lipopolysaccharide (LPS) spiking results. All final rinse values were >0.24 EU/mL. A trial with a 450-mm column filled with reverse osmosis (RO) H₂O to a bed height of 5 cm resulted in a log removal value (LRV) of 7.2, with final traces below LAL LOD. LRVs of 5.8 and 5.9 were obtained with the 100 mm and 450 mm columns when packed with Matrex Cellufine GC700 media (nominal 17 cm packed bed). The tests demonstrated scalable cleanability with these columns and this methodology. Media have a negative impact on inactivation kinetics due to mass transfer and the shielding effect. However, this test proves that not only the column itself, but also the Matrex Cellufine CG700 resin can be cleaned on different scales. Dynamic, realistic cleaning is much more effective at inactivation than merely static exposure alone. For all trials, feed water LPS content was below the LOD for the LAL assay (0.100 EU/mL.) This is further evidence that negligible, if any, levels of contaminants were held up in the column and would be released through diffusion (Figure 3).

Microbial challenge. Microbial challenges were developed to characterize the susceptibility of various microorganisms to treatment conditions, to develop a cleaning procedure that can effectively sanitize the column, and to determine the ability of the column to be sanitized after severe bacterial and fungal contamination. Typical acceptance criteria include a 5 log reduction in microorganism concentration in the effluent (CLU/mL), insignificant levels of growth in swab, media, and component samples, and 10 CFU per 100 mL WFI.² Specific biologics may contain customized information on the appropriate USP monograph. Microbial challenge tests are quantitative, realistic challenge scenarios, but they are very expensive and preliminary assay development is required.

The USP compendium recommends specific microorganisms for use in microbial challenge testing.

Static exposure to 0.5M NaOH showed mixed results. Gram-negative bacteria are easy to kill. Bacillus subtilis spores are unaffected by NaOH. Staphylococcus aureus, Candida albicans, and A. niger are initially inhibited by 0.5M NaOH, but stabilize. Heat treatment at 50°C compared with ambient trials showed increased inhibition of C. albicans than with 0.5M NaOH. At similar inoculum concentrations, all of the controls exhibited < ± 3.0% change when heat was combined with NaOH treatment. Destruction of B. subtilis spores was increased, but S. aureus showed enhanced resistance at 50°C, possibly a stress response.

Reducing S. aureus concentrations from 10^7.5 to 10^5.5 leads to increased inactivation kinetics. Protective effects of S. aureus glycocalyx and culture-derived flocculation were examined and were not a factor.

Kinetic reaction studies were conducted. Gram-negative bacteria were killed rapidly with 0.5 M NaOH. Spores cannot be effectively cleaned with sodium hydroxide. Elevated temperatures (50°C) do not offer a clear cleaning advantage for the trialed organisms. Static application of 0.5 M NaOH may not provide adequate removal of non-gram negative species and is very time dependent.

Figure 4. Challenge bacteria concentration after various testing stages.

The process was tested at scale up using a Vantage L (32 x 250 mm) column and Matrex celluline CG700 chromatography media with a packed bed height of 5.0 cm. Cleaning procedures included forward and reverse flow for 4 hours with both 0.5M and 1.0M NaOH used. Four samples were taken during the water rinse phase. Long static soak proved insufficient; dynamic flow with direction reversal is beneficial to the effectiveness of the sanitization process. 1.0 M NaOH was required to sanitize the Vantage L column and spores.

The final experimentally derived cleaning SOP was as follows:

• One column volume (CV) (plus pipe hold-up volume) of water is circulated at 100 cm/hr in reverse flow;
• Two CV of 1.0M NaOH is circulated at 100 cm/hr in reverse flow;
• One CV of 1.0M NaOH is pumped through the column at the same linear velocity in forward flow;
• Two CV of 1.0M NaOH (total) is recirculated for two hours at 100 cm/hr.;
• Neutralization: Water is flushed through the column at 100 cm/hr until the effluent pH is <>
• Total microbial exposure time to a pH of 12–14 in less than 3 hr.

Microbial challenge scale-up. To demonstrate that the column design is readily cleanable, the column was challenged with an extremely concentrated microbial cocktail, subsequently cleaned, and validated, as follows.

• Column: Acrylic 450 mm glass 100 mm QuikScale column
• Medium: Matrex Cellufine CG700 resin
• Cleaning agent: 1.0 M NaOH
• Microbes: E. coli, S. aureus, P. aeruginosa, C. albicans
• Initial inoculum concentration: 2.68 x 10⁸ culture (static phase), (-1 x 10¹² total cells loaded)
• Breakthrough concentration: 960 x 10⁷ clm/mt — stationary
• Procedure:

• Inoculate column with high concentrations of micro-organisms.
• Incubate column for 18 hr.
• Perform sanitization procedure.
• Following neutralization, analyze column with ortho-gonal methods.

• Enumeration: spread plating
• Acceptance criteria: (USP26-NF21S1 monograph <1231>). Water for Pharmaceutical Purposes states not more than 10 CFU per 100 mL WFI may be present. Very low levels of post cleaning microbial presence are necessary to prove that the column can be readily sanitized.

DETECTION METHODS

• Swab locations (according to Le Blanc):³

• Calcium alginate in trypticase soy broth (TSB), rolling biodirectional > 1 cm²

• Turbidity of rinse
• Membrane filtration of rinse and plating of membrane
• Desorbtion of cells from quantified amount of chromatography media
• Immersion of column seals and gaskets in liquid broth

Figure 5. The cleaning efficacy of the column was determined through surface swabs on different column wetted parts, then analyzed for microbial growth.

Results. No growth of the spiked organism present in the original cocktail was observed in the rinse samples (LOD = 1 CFU/100 mL). All swabs and chromatography media samples (pour plating and liquid desorption) were negative for growth. Application of the developed cleaning protocol significantly reduced populations of microorganisms. Rigorous and redundant examination procedures provided several metrics that can be used to support the claim of sanitary design (Figures 4 and 5).

CONCLUSIONS OF COLUMN CLEANABILITY CASE STUDY

• 1.0 M NaOH is a potent anti-microbial agent.
• Efficacy is greatly enhanced by dynamic rinsing and recirculation.
• Contact time >2 hr is recommended for maximum sanitization.
• Sanitization procedures can be effectively conducted at room temperatures.
• Gram negative bacteria are killed rapidly.
• All cleaning should be preceded by a 1 CV flush with sterile buffer.
• Pour-plating is inferior to a discrete desorption step because of the high turbidity of solidified agar on chromatography media plate, which makes it difficult to resolve colonies.
• Background flora necessitate controlled environment for challenge studies.
• These results showed that the column met the metrics of cleanability described.

The columns proved to be cleanable with a variety of challenge materials in extremely contaminated conditions. The investigation allowed design changes to be made that not only improved the system, but also provided quantifiable evidence that the final product could be cleaned sufficiently in typical user scenarios.

Because each cleaning process is different and presents unique challenges, cleaning protocols and agents must be selected on the nature of the contaminants, media, equipment, and the level of sanitization required. Ultimately, the pharmaceutical company is responsible for validating its own process. However, purchasing equipment that has been tested with rigorous scaled-down cleaning trials helps the company develop and test its own processes with confidence.

ACKNOWLEDGMENTS

The author thanks Aaron Noyes, formerly of Millipore, for his contribution to this research.

Chris Antoniou is group R&D manager and Hilary Carter is development engineer, both in the BioPharmaceutical Division, Millipore Corporation, 80 Ashby Road, Bedford, MA 01730, Tel: 781.533.2650, Fax: 781.533.3134,

REFERENCES

1. Lindsay, J. Cleaning and Cleaning Validation: A Biotechnology Perspective, PDA, 1996.

2. U.S. Pharmacopeia, 26 / National Formulatory 21, S1

3. LeBlanc, DA. Sampling, Analyzing, and Removing Surface Residues Found in Pharmaceutical Manufacturing Equipment. Microcontamination 1993 May: 37-40.