Last month we talked about what to do if residue limits change. That was a prospective (going forward) issue. This month we’ll discuss a retrospective issue. Specifically, what could (or should) be done if a serious issue comes up and (while you believe that previously manufactured product probably is not cross-contaminated), there are issues that have arisen such that your company or (more likely) a regulatory agency asks you to do a retrospective evaluation to determine whether or not any previously manufactured product that has been released is, in fact, contaminated due to a cleaning issue. There are really two avenues that might be pursued, and most likely both avenues will be followed.
One avenue is to evaluate all data related to GMP manufacturing (including cleaning) procedures. In other words, since not manufacturing a product under GMPs is problematic, my company should evaluate past practices (cleaning procedures and validation of that cleaning). That avenue will not be the focus of this Cleaning Memo. The second avenue (the focus of this Cleaning Memo) involves actually testing released product for API (drug active) residues of previously manufactured products. While it is possible that the contaminating residue could be a microbiological residue or cleaning agent residue, it is not likely that those would be the major risk or major concern. If the residue of concern is not the active, but rather a degradant, a similar approach to what will be presented could be used. In fact, the recent experience with nitrosamines (such as nitrosodimethylamine, or NDMA) provides a partial blueprint for how to address potential residues of prior drug actives.
So, let’s assume our approach will be to analyze past manufactured product for residues that could be suggestive of cross-contamination. In other words I will be looking for residues of a different active (one that shouldn’t be there at an unacceptable level). The questions to ask include the following.
For Question #1, the focus should be on batches that are still within their expiry period. If the suspected route of potential contamination is known, then that may narrow down the batches based on a given time frame or a given equipment or equipment train. For example, if there could have been an issue with the quality of a cleaning agent used, I may limit my evaluation to the times when equipment was cleaned with that suspect cleaning agent. If I had seen visually clean issues with certain equipment (or certain types of equipment), then I might focus my efforts on products manufactured on that equipment. If the suspected contamination is from the equipment surfaces of the immediately prior manufactured product, then in campaign manufacture the batch most likely to have higher residues is the first batch of a campaign. It should be clear, however, that this requires an investigation and risk evaluation as to the extent of products and product batches to be tested. A related issue is whether I should (or can) only test finished product and not different physical forms (for example, tablets versus powder) at an earlier stage of manufacture.
For Question #2, the answer is typically related to the suspected cause. For example, if it is likely that any potential contamination could be due to transfer from equipment surfaces that would preferentially transfer to the first part of a next manufactured batch, then perhaps the focus should be on the initial portions of that next manufactured product. Caution should be used here. For example, if my suspected source of potential contamination is a powder blender, then (unless I have retains of the beginning, middle, and end of the blend itself), I may be limited in what I could test. Realize that if the suspected contamination is from the blender equipment itself, it is likely that any transferred residue would be evenly dispersed throughout the blend, and therefore testing finished product (tablets) may not show significant differences. However, if the suspected source is from an indirect product surface via an airborne route, then trying to determine which part of the batch should be tested may be problematic; I can’t be certain whether the transfer happened at the beginning of the batch, in the middle, or at the end.
For Question #3, if the suspected contamination is from the equipment surfaces of the immediately prior manufactured product, then it makes sense to test for active (API) of the immediately prior product. In contrast, if potential contamination is from a visually soiled “‘indirect product contact surface”, the transfer may not necessarily be to the next batch, but to several batches later (which is a good reason to double down on a visual assessment at the end of cleaning; see the September 2020 Cleaning Memo).
For Question #4, the first thing is that it is not a requirement that residue levels be non-detectable. The whole reason for doing limit calculations is to determine a maximum acceptable value for residues in the finished product (even though in validation, we measure residues on the equipment surfaces). So being below the limit of Quantitation (LOD) is not a requirement (unless the acceptance limit is precisely at the LOD). The limit value that is used for this analysis is preferably the L1 limit, or the limit in the next product (see the Cleaning Memo of September 2012 if you are not familiar with this terminology). There are two possible options here. One is to select the L1 acceptance limits used in the validation protocol for that product. That value for the protocol may actually be more stringent than the value that would be directly applicable if a calculation were made considering only the two products involved. For example, if I make four (4) products on the same equipment, and am calculating the limit for the active of A, I do a calculation based on that active getting into each of B, C, and D, and use the lowest limit for my protocol for the cleaning of A. If that lowest limit is based on the transfer of A into D, but if the actual product suspected of having that residue is C, then the L1 value for transfer of the active of A into C may be higher. Note that in this latter case, if the residue value is higher than protocol value (based on the A to D calculation), while it may mean that the actual hazard due to the A to C situation is acceptable, it may be problematic in that it suggests that the cleaning validation may not be in a state of control.
For question #5, the key for the analytical method is first determining how to analyze at the L1 limit in an actual next product. Specifically, do the active and/or the excipient in that next product interfere with an established analytical procedure? Assuming that an analytical procedure developed for cleaning validation purposes of A has dealt with possible interferences from the excipients in A, the question is whether the active and excipients in B (the tested product) will also interfere. In addition to this interference concern, it should also be realized that the L1 limit is likely below the L4b or L4c limits typically encountered in a cleaning validation protocol. A final analytical issue is that of sample preparation so that analysis is possible. For example, prior to analysis the tablets are typically crushed to a fine powder so that any actives being targeted, including (for example) the active of the prior product, will be accurately measured in the tablets. Care must be exercised so that all lab vessels used for this procedure are scrupulously clean (to avoid contamination in the sample preparation itself).
The purpose of this Cleaning Memo is not to be a turn-by-turn roadmap to get you through this process. Rather, it is designed to be a travel advisory so that your trip is successful. That said, hopefully if you have a robust cleaning validation program, you will never have to take this particular trip. It is not a pleasant journey.