This is a continuation of a discussion of products with highly hazardous actives, and deals with the question of routine monitoring for highly hazardous products. Hopefully (if you follow my Cleaning Memos) you know that I generally define a “highly hazardous product” as a drug product with an active that has a significant toxicity concern apart from the therapeutic effect. Others define it as a drug product containing an active with a PDE/ADE of less than 10 micrograms/day. Whichever definition you use (and there is significant overlap), I think you know what I am referring to.
By “routine monitoring” I generally mean testing that is routinely done after cleaning validation is done (that is, after completion of the validation protocols performed as part of Stage 2 qualification process in a lifecycle validation approach). By “routinely”, I mean testing done either on every occurrence of the cleaning process or on some reduced schedule such as monthly, quarterly, or based on a frequency such as “every X batches”. The testing that might be done includes visual examination, final rinse sample testing, and swab sample testing. If visual examination is done for routine monitoring, there may not be the same extensive equipment disassembly that is done during the validation protocols. If swab sampling is done, there may be a reduced number of swab sample locations (selecting for example, the “worst cases” of what was done in the validation protocols). Finally, the testing that is done may be for the drug active, for the cleaning agent, for bioburden, for endotoxin, or for visual assessment. Note that I have included visual assessment as part of both sampling and analytical methods because it really entails aspects of both.
Let me also clarify that by routine monitoring, I am not referring to the practice of repeating one run of a validation protocol on a yearly (or other frequency) basis. However, it is possible to include that practice as part of the Stage 3 “continued process verification” (also called “ongoing process verification” or “validation maintenance”) that is a part of a lifecycle validation approach. I’m sure that you will see that some of the things discussed for routine monitoring also apply to that yearly verification protocol or yearly confirmatory protocol. However, the routine monitoring as I usually present it is not done as part of a separate protocol, but is apart of cleaning procedure and/or batch record of that cleaning process.
Okay, I have tried to set the stage. Now let’s talk about how this applies to highly hazardous products. As discussed in the February Cleaning Memo, those highly hazardous products are a greater patient safety risk (and consequently a greater business risk), so I want to pay more attention to those products. I certainly should pay more attention to the design/development of the cleaning procedure (lifecycle validation Stage 1) and to the design/execution of the validation protocol (lifecycle validation Stage 2). Our focus here is what I should do for routine monitoring (part of lifecycle validation Stage 3). Certainly whatever is done for routine monitoring for non-highly hazardous products should be done also for highly hazardous products. But, what more should I do?
In most cases, I probably should not have to do anything additional for visual examination, for cleaning agent testing, for bioburden testing and for endotoxin testing. However, it would certainly reduce risk if I were to do some level of additional testing for the highly hazardous active itself. What are the possibilities?
One is to routinely test a final rinse water (or solvent) sample for the active. Rinse sampling (if it encompasses the entire equipment sampled) can be a valuable technique for getting an overall picture of any problem that might occur with a cleaning process. Typically, that analysis for highly hazardous products should be a specific analytical procedure, such as HPLC, that tells me exactly how much active was present in the rinse sample. I can compare that to both my protocol acceptance limit as well as to trend charts and/or to action/alert levels (the latter having been established based on process capability analysis of rinse sample testing).
It probably is not good enough to try to just analyze the final rinse for TOC; TOC can certainly tell you how much carbon is present, but it cannot tell you whether that carbon is from the active or from excipients or from processing aids or from cleaning agents. Furthermore the practical LOD/LOQ for TOC samples is probably not low enough (in most cases for highly hazardous actives) to provide a confirmation that a limit is being met for a cleaning validation sample (where you are subtracting out a blank value to determine what could be present due to residues on surfaces). That said, it may be possible for routine monitoring to use just a “straight” UV (ultraviolet) method on the rinse sample. You might wonder why I suggest that UV, also a “non-specific” method, might be useful, whereas I think TOC is probably not useful.
The difference is this. For UV I can generally measure my active at a much lower concentration as compared to measurement by TOC. Secondly, there are likely to be fewer interferences for the UV measurement as compared to interferences for a TOC measurement. For TOC most carbon containing species will interfere and contribute positively to the TOC value. For UV, species (either with or without carbon) will only interfere if they have significant absorption at a wavelength close to the λmax of the highly hazardous active being analyzed. Even if there is some positive interference, you can still show acceptable levels of the highly hazardous active if, based on the total response at the λmax of the highly hazardous active, the assumed value of that highly hazardous active is below the acceptance value (or within a trend line). The value of a straight UV method as compared to a HPLC method may be the time required for analysis, and therefore the time required for release of the equipment for subsequent manufacture. Note that even with HPLC analysis only, the equipment may be released at risk pending completion of that analysis. However, rather than doing that, I would recommend you perform a UV measurement, and then release it at risk pending a subsequent confirming HPLC analysis.
Enough for rinse sample testing. What about swab sample testing? First, the discussion about selection of an analytical method to use for rinse samples also applies to swab samples (although for UV analysis, I will also have to consider possible interferences from the swab itself in addition to interferences from the excipients, processing aids, and cleaning agents). Furthermore, as suggested earlier it may be possible to reduce the number of swab samples by only sampling the “worst cases” of the worst cases previously chosen for my validation protocol. In some cases, the determination of the “worst of the worst” may be easy. For example, suppose I review the results of the twenty-seven sampling locations in a cleaning validation protocol run, and all samples except for three locations gave results below my LOD or below my LOQ. And for those three other locations I consistently (in all three protocol runs) obtained results that were acceptable (below my limit), but they were a quantified number above the LOQ. This type of data suggests (but is not an “ironclad” determination) that if I had problem with my cleaning process, it would show up in those three locations. However, selecting those three locations for swab locations for routine monitoring would be a reasonable approach. For those of you who might wonder at my selection of twenty-seven swab locations, I just used that number to see if you were really reading and paying attention. I realize that number is certainly on the high side and a little “over the top”, but it was just an example.
Another possibility for choosing the “worst of the worst” is to look at my initial rationale for selecting swab locations. I may have selected some locations based on complex geometry and other locations based on representative materials of construction. Particularly if all my protocol data is consistent (that is, I cannot separate out those locations with consistent higher values from those with consistent lower values), looking at the initial criterion for selection of swab sampling locations may be helpful. Let emphasize, however, that it is not necessary to reduce the number of swab sampling location. However, if a sound risk assessment is done, in light of the need to be more efficient in manufacturing of drug products (without sacrificing quality) the reduction of swab sampling locations for routine monitoring for highly hazardous products should be considered.
Let me close by stating that I am not trying to present a “cookie-cutter” approach to routine monitoring for highly hazardous products. What I would like you to do is understand your manufacturing and cleaning processes, and based on a scientifically and logically based risk assessment, decide what is appropriate and applicable for the specific situation in your company.