HI Tom.
While this is truly not in my "wheelhouse" of knowledge, here is what I would suggest:
First, if the testing method or the requirements were determined to be erroneous, I think that is a good thing for you in this situation. Part of your company's or you CMO's investigation should always include whether the correct test and specification requirements were performed and checked against. So that would mean to me that you might be able to reasonably argue that if <3> is the more correct methodology your testing was in error and therefore the results (and the analysis of those results) are in error as well. This is classically one of the examples of a "lab error" that would allow you to invalidate the original result analysis (failing the <905> requirements) and then reanalyze the testing results against <3>.
That said, there needs to also be a root cause and CAPA performed on this issue as well. Why was the incorrect method chosen? What about the testing in <905> is so different from <3> that it would not be reasonable to test against the <905> method and get the same data analysis in this case? Who made the decision to use <905> and not <3> and what qualifications did they have to make that decision or what qualifications were they missing (more importantly in my mind) that should counsel against the presumption that the original test was a valid determinant of your product's stability?
With your question about changing the labeling of the product to storing between 15-25C versus 15-30C, that is a little more difficult. First question to ask is why the testing of this one lot shows 2 OOS results at the 30C temperature that were not noted in other lots? This could be a situation where a particular lot was not properly manufactured or it could mean that your product is not as stable as you originally thought. Then the rest of this analysis is going to depend heavily on how you are able to answer the question about what is so different between <905> and <3> for this product that the <905> result analysis is not valid. Since I am by no means an expert (in case that wasn't already abundantly clear!) in these two methods from USP, I would suggest that someone highly familiar with the methods perform a gap assessment that shows why your testing under <905> should be invalidated and all testing should be analyzed under <3>. I will say, just be careful with that analysis because (while it might not seem obvious here) it could become a double edged sword in that the data from all prior studies would need to be reanalyzed under whichever standard methodology you determine is more applicable...
------------------------------
Victor Mencarelli
Director Regulatory Affairs
United States
------------------------------
Original Message:
Sent: 02-May-2018 11:26
From: Tom Stothoff
Subject: Uniformity in Container USP <3>
Our situation. Topical Ointment product in Phase 2 of development. Packaged in 25g tubes. We learned early on that the product "melts" at the 40C accelerated conditions so we have been testing only at 25C/60%RH and intermediate 30C/75%RH. We recently observed two OOS's at 30C/75%RH for the Content Uniformity Test at 9 and 12 months. Assay results exceeded the high end of the limit (110%) for the sample that came from the bottom of the tube. The thought is the API is migrating downward at the 30C/75% conditions as a result of the ointment being "softer". But our CMO has been testing according to USP <905> Uniformity of Dosage Units and has now realized they should have been testing according to USP <3> Uniformity of Container. I just joined the company so to me it appears applying the <905> criteria was an error from the beginning and USP <3> should have been used all along.
- I think USP <3> is the appropriate test. Agree?
- Our CMO wants to revise the storage condition for clinical supplies from 15-30C to 15-25C as a result of the OOS's, but I'm wondering if there is a way to avoid this. Results would have passed using USP <3> criteria. To date we have acceptable 12M data at 25C/60%RH on several batches. We'd like to assign a 2-year shelf life based on the 12M data. Any thoughts on this strategy or rationale to avoid the label change of the storage condition?
Thanks
Tom
Chicago