Cleaning Verification – Investigating Unknown Peaks in Chromatography
Introduction
This procedure provides recommendations for conducting an investigation into unknown chromatographic peaks or spots/streaks observed during cleaning verification. In this procedure “peak” refers to a spot or streak on a TLC plate as well as a peak in a HPLC, GC, or other chromatogram.
Some unknown peaks may warrant further investigation when observed during verification testing for product and/or cleaning agent residues. The decision to investigate is based on a risk assessment (1, 2) that includes the size of the peak relative to the Residue Acceptable Limit (RAL) and history of the process. A step-wise, methodical investigation should follow any decision to investigate the unknown peak.
Recommendation & Rationale
Because unknown peaks observed in chromatograms for cleaning verification samples cannot be assumed to provide an equivalent response to a known standard, specific compound, Relative Response Factors (RRF) are typically not assigned to unknown peaks. Based on the approach used at a number of GMP sites, the following considerations should be made to ensure investigations take place for potential contaminants that may negatively impact the next product manufactured in that equipment:
Investigation Considerations
Because unknown peaks observed in chromatograms for cleaning verification samples cannot be assumed to provide an equivalent response to a known standard, specific compound, Relative Response Factors (RRF) are typically not assigned to unknown peaks. Based on the approach used at a number of GMP sites, the following considerations should be made to ensure investigations take place for potential contaminants that may negatively impact the next product manufactured in that equipment:
– Analytical test methods can only accurately detect residues for which they have been developed and validated. Unknown peaks cannot be assumed to have the same Relative Response Factor (RRF) as the residue of interest, because the unknown has not been identified and a RRF for the unknown has not been determined. Therefore, an accurate quantitation of an unknown peak is not possible.
– Comparison of an unknown peak to the RAL determines if an investigation is initiated. There are multiple Safety Factors typically applied in the RAL calculations. Therefore, the risk of having an unknown peak present that is significant to toxicity or dose limits is relatively low considering the conservative approach chosen to calculate the RAL.
– It is very unlikely that there is an additive or cumulative pharmacological effect of the individual peaks. The primary residue being removed during cleaning is that for which we may expect potential pharmacological impact, and upon which the RAL is based.
– The investigation of the sum of unknown peaks is technically challenging and impractical in many cases. It would require an evaluation of all peaks that make-up the sum, regardless of their relative size. Alternatively, the largest peaks that make up the sum could be investigated, but this becomes the same approach as investigation of each peak that exceeds the RAL. Such an effort to sum all peaks and investigate each provides little value considering the Safety Factors already built into the RAL.
– This approach of not summing peaks has been based on common benchmarking within many API site.
Confirmation of Results
Investigative evaluation of the equipment and unknown contaminant prior to re-clean should first include confirmation of the analytical result. If the analytical result is confirmed, the investigation should then include a review of cleaning procedures, visual inspection of the equipment and reporting of the failure and trending of past results. Once the initial investigative evaluation has been conducted the equipment is typically released back to production by the site quality team, upon successful completion of the equipment re-cleaning. This may be done prior to the completion (e.g. approval) of the formal investigation report.
Limits
For worst-case limits calculations, the unknown peak investigation should initially be based upon 100% of the worst-case (e.g. lowest) RAL of the target residue. If the limits are then subsequently recalculated based on the specific product A to product B changeover (vs. worst-case limit), the new limit could release the equipment for subsequent use to produce product B. This approach is based on risk and the multiple safety factors already included within the calculated matrix.
Response Factor
If an unknown peak is subsequently identified and a corresponding RRF established, the requirements of the investigation and the cleaning limit for that peak may be impacted.
Trends
Trending of unknown peaks investigated should include past incidents (i.e. history) and be included within periodic review for cleaning procedures.
Investigative Steps and Possible Causes
The following list is intended to provide a practical sequence of investigative steps to evaluate unknown peaks:
Artifacts
Confirm the unknowns by re-analyzing the sample solution using the same instrument and the same sample vial (if applicable). Also analyze the same sample solution from the original sample submitted from the plant, which has been loaded into a new vial (if applicable). If available, prepare and inject a duplicate swab or rinse sample solution submitted from the same location. Blank swab sample extracts may also be injected. If a pattern emerges suggesting the unknown is an artifact of the analysis, unconfirmed, and/or due to the swabs themselves, a re-clean of the equipment may not be required. If the unknown is considered to be an artifact, it does not require additional investigation of the cleaning procedure.
Solvent/Cleaning Agents
Compare the peak to the solvents and/or cleaning agents from the manufacturing process and/or cleaning process to determine if the unknown peak is a result of solvent or cleaning agent interference. Cleaning evaluation studies would have hopefully uncovered this interaction. For example, toluene and dimethylformamide are solvents that have commonly been observed to cause unknown peaks in chromatography. If an investigation into an unknown peak determines that it is due to an interfering solvent, then future observations of this peak would not require repeated identification of that peak.
Degradation
The peak may also be the result of product degradation or reaction with the cleaning agent. For example, a material containing a carboxylic acid can react with methanol during cleaning to produce an ester. If an investigation into an unknown peak determines that it is due to product degradation, then future observations of this peak would not require repeated identification of that peak. . If the degradation product is likely to be generated during the cleaning procedure, then the analytical method should be updated to include identification of the additional peak.
Sampling
Determine if the unknown peak is due to sample preparation, such as residues from lab gloves, sample containers or caps.
Previous Product
If practical based on the test method used, compare the peak to the previous one or two products produced in the same equipment prior to the product being cleaned out. If the unknown peak is determined to be due to a previously produced product, the analytical method specific to that previously produced product should be used to determine an accurate result for that product. If the equipment has been determined to contain the previous product at a level exceeding the respective RAL, the equipment may need to be re-cleaned for the previous product using an appropriate cleaning procedure and limits for that previous product. In addition, the cleaning procedure for the previous product should be re-assessed.
Additional Inspection and Sampling
If the investigation recommends additional inspection and sampling (e.g. after the unknown peak is confirmed, but not due to solvent interference) then the following actions are suggested:
Request the operations group to perform another visual inspection of the equipment for residues and describe any that are found. Compare results against the physical characteristics of manufacturing process raw materials, by-products, or indications of equipment failure (lubricant leaks, etc).
If deemed necessary, and it is safe to access the location of the residue, sample the unknown residue and attempt to identify it through other analytical means such as mass spectroscopy.