A validation maintenance strategy used by some companies is to perform a single run “confirmatory” protocol (also called a “verification” protocol) on a regular basis (such as annually) on a drug product that has been qualified (or validated). This confirmatory protocol is just one part of what the company might do as part of “Stage 3” of lifecycle validation.
If a company makes multiple products and validates them individually (that is, not using a product grouping approach), then a confirmatory run is done on each product on an annual basis. If for production scheduling reasons one drug product is not to be manufactured during a given year, then the annual confirmatory run can be skipped for that year (for that one drug product). There is no reason to force scheduling of manufacture of that drug product. However, if a year is skipped because of no manufacture in that year, then it is prudent to make sure that a confirmatory run be performed on the first manufacture of that drug product in a subsequent year. If that is the approach utilized for confirmatory runs where drug products are validated individually, then it is helpful to include some verbiage in an appropriate document to clearly state that approach. An example of such wording might be “If a drug product is not manufactured in a given year, then no such confirmatory run is required. In such as case a confirmatory run should be conducted for the first manufacture of that drug product in a subsequent year.” Let me clarify that the approach in this paragraph is a possible approach if drug products are validated individually.
Now we’ll turn to the situation of where cleaning validation is done in a product grouping approach. In a product grouping approach, a worst case product (“most difficult to clean” product) is established, and a protocol of three runs is performed on that worst case product that then applies to all products in the group. If a yearly confirmatory run is to be done, it is only done on that worst case product. If that worst case product is manufactured every year, life is straightforward – the annual confirmatory run is always on that worst case product. A successful confirmatory run on that worst case product should provide supporting evidence that the cleaning process for all products in the group is in a “state of control”.
But, what happens if there is no scheduled manufacture of that worst case product in a given year. Certainly one option (albeit an unattractive one) is to force manufacture of that one product so that a confirmatory protocol can be performed. Another option is to perform a confirmatory run on a different product where routine manufacture is scheduled for that year. That different product may be either an equivalent “worst-case” product or may be a “second” worst case product (second in terms of easier to clean as compared to the worst case, but more difficult to clean than all the other products in the group that are scheduled to be manufactured in that year).
What do I mean by “equivalent” worst case product? This would apply in situations where, by whatever criteria is used for determining the worst case product, there is more than one product that is the worst case. This may happen is cases where I am using a “point” system to evaluate a variety of parameters relevant to “difficulty of cleaning”, and there are two (or more) products that have an equal number of maximum points (assuming higher numbers are a worst case). In such a case, it makes no sense to require a validation protocol of three runs on both of the worst case products. My general recommendation in such a case is to select one of the “maximum points” products on a different criterion, such as frequency of manufacture (to enable me to get all three runs done in a more timely manner) or capabilities of the analytical method to measure residue values. It may also happen in situations when I am using a “point” system, but where I form “bands” to categorize the difficulty of cleaning. For example, if the maximum points is “100”, I may have bands of “91-100”, “81-90”, “71-80” and “≤70”. (Note that these ranges in the bands will generally be dependent on the various parameters and the type/nature of products manufactured.) An analogy is the Occupational Exposure Bands (OEBs) that are used by some companies to rank the relative safety of actives for EHS purposes. The idea is that with one product rated 88 and a different product rated 84, there may be no practical difference in the difficulty of cleaning. So, if I have two or more products identified as equivalent worst cases, I chose one to use for my initial validation of three runs. As much as possible (based on scheduling) I continue to do my annual confirmatory run on that same product. If that product is not manufactured in a given year, I then choose any “equivalent” product for my annual confirmatory run.
What do I mean by “second worst case” product? I will illustrate this by using a “point system”. Let’s say I don’t use “bands”, but rank the products strictly on the total points for each product in the group. For example, here is a possible ranking:
Product A: 92 points
Product B: 88 points
Product C: 81 points
Product D: 62 points
My initial validation of three runs is done on Product A. In subsequent years my annual confirmatory run is also done on Product A. But, in a given year Product A is not scheduled to be made. In that case my annual confirmatory run is done on the second (or next) most difficult-to-clean product, which is Product B. Why is this appropriate? The purpose of the annual confirmatory run is to provide evidence that my cleaning processes are in a state of control. By performing a confirmatory run on Product B (the worst case for the products made in that year), I am providing evidence that the cleaning processes for Product B, Product C and Product D are covered. Yes, it does not strictly apply to Product A, but Product A is not manufactured during that year. However, the first time I make Product A again, I should perform a confirmatory protocol run and continue to use Product A in years when manufacture of it is scheduled.
In the case of using either an “equivalent” or a “second” worst case product for an annual confirmatory run, the residue limit that is used for that run should be the same residue limit used in my original validation protocol for the worst case product. Note that I am assuming here that the approach for product grouping is selecting the most “difficult-to-clean” product, but for the (numerical) residue limit using the lowest limit of any product in the group. Some might argue that for a yearly confirmatory protocol in cases where the worst case product is not being manufactured, I should only use the lowest limit for any product excluding the limit of that worst-case product. However, the correct approach is not to modify the limit in such a manner. For example, using the example of the four products A, B, C and D, if A was the worst case and was used for my original validation, but because of scheduling I am doing my annual confirmatory run on B, my residue limit would not change. If my original validation for A had a limit of X mcg/swab (which is the lowest limit calculated among the four different products), then by the logic of my grouping approach when I do a confirmatory run on B the residue values should also be less than X mcg/swab. In other words, if A should be cleaned to below X mcg/swab, then B (which is easier to clean) should also be cleaned to below that same limit.
For those of you not familiar with the product grouping approach of the “most difficult to clean” product with limits set at the lowest limit of any product in the group, please see the May 2017 Cleaning Memo. Furthermore, this current Cleaning Memo does not cover all options that might be considered for an annual confirmatory run, nor does it address all the issues that should be considered as part of validation maintenance.