Confirmatory testing - on how many batches?

Confirmatory Testing & Analytical Challenges
1

Dear Community,

I have a question regarding the expectations for nitrosamine confirmatory testing following a recent LoD from the authorities.

We applied for skip testing based on consistently low nitrosamine results (all below 30% of the AI). The authorities acknowledged this approach and requested a more detailed nitrosamine risk assessment, which unfortunately had not been included in the original submission.

In addition, they commented as follows:

“Moreover, the number of tested batches should be justified, since according to the Q&A document ‘for marketed products, test results from 10% of annual batches, or 3 per year, whichever is highest, should be submitted. If fewer than 3 batches are manufactured annually, then test results from all manufactured batches should be submitted.’ Further data should be submitted, if relevant.”

I am uncertain how this statement from EMA Q&A (Question 8) should be interpreted.

Specifically, does the requirement for testing 10% of annual batches (or a minimum of 3 batches per year) refer to:

  1. The number of batches that should be included in the initial confirmatory testing package, or
  2. The minimum number of batches that should continue to be tested during routine monitoring / skip-testing activities?

To illustrate my concern, let’s assume a marketed product with a newly identified NDSRI. A suitable analytical method is developed and validated, and the product is manufactured at a rate of approximately 200 batches per year, with a shelf life of 5 years.

Would this imply that confirmatory testing data would ultimately need to cover: 20 batches per year (10% of annual production), and all shelf-life years for which batches remain on the market, resulting in approximately 100 batches of confirmatory testing data (20 from each year, so 20 x 5)?

Even with a 3-year shelf life, this could still mean around 80 batches, considering that batches manufactured in previous years remain within expiry and available on the market.

This interpretation seems quite extensive, so I would be interested to hear how others have approached this requirement in practice.

How do you interpret this expectation, and what has been your experience with confirmatory testing for marketed products with NDSRIs?

Thank you in advance for your insights.

1 Like
2

The request from authorities is regarding the number of batches tested during confirmatory testing. Your example of 5 years, 200 batches per year is correct, a total of approximately 100 batches should be tested according to the guidance. While this is probably an overkill, it is the current guidance (probably due to the initial uncertainty around nitrosamines). Also, consider that the guidance even says that if you have more than one potential source of risk (i.e, two API sources) the efforts should be proportionate. To be fair, we have applied in the past some level of matrixing, but we tend to keep to the 10% of annual batches tested.

If you are in June, for example, I would not consider 4 full years for selecting batches. It would be just whatever batches are not expired.

4 Likes
3

Hi! It’s so “good” that many of us start having actual practical issues with authorities with reference to interpretation of the GLs… perhaps this means that some “grey zones” will be explained in a better way in the revised document.
apart from this…
specifically for your current request…
if I were you, I would try first option 1. (10% of annual batches as the number of batches that should be included in the initial confirmatory testing package).
please note that number of batches set as a frequency for skip testing is a GMP issue - not driven by ICH guideline.
whats more, they do expect see data during shelf life. so you could test any retain batches in the middle or near end of shelf life. I think that 3 batches could suffice their request.
I am not very sure if different strengths of product, what authorities expect and that’s one of my questions in another open thread.
and of course, please make sure to present a robust root cause assessment.

kind regards.

2 Likes
4

Thank you for your answer!
Yes, this definitely seems and feels like an overkill…
In case representative batches are selected from release, mid-SL and EoSL, three from all time periods, and all results are below LOQ (10% of the AI) or even below LOD, would you still go the extra mile to test the 20% of the annual batches?

5

Yes, that was our interpretation as well. We initially followed the classic 3-batch approach, selecting batches representing the beginning, middle, and end of shelf life. In a simple scenario with a single product matrix, this would already amount to 9 tested batches.

However, what I struggle with is how to reconcile this approach with the wording in the EMA Q&A stating that test results from 10% of annual batches (or at least 3 batches per year, whichever is higher) should be submitted.

For example, if 10% of the annual production corresponds to 10 batches, this already exceeds the traditional 3-batch confirmatory testing approach. Furthermore, for products with longer shelf lives, the number of potentially relevant batches increases significantly because multiple production years remain simultaneously on the market. Following that logic, the number of batches could easily double or triple.

I also very much agree with your point regarding the use of retain samples for confirmatory testing. However, I have another concern. What would be the expectation in a situation where retain samples stored under room temperature remain consistently below the LOD throughout the entire 24-month shelf life, but stability studies at intermediate condition show an increase over time, for example reaching 11-30% of the AI at 12 months?

The guidance frequently refers to testing retain samples and does not appear to specifically discuss the use of stability batches for confirmatory testing. In practice, however, there are cases where storage conditions and aging clearly influence nitrosamine levels. In such situations, relying exclusively on retain samples may not fully reflect the highest nitrosamine levels that could occur during shelf life.

Has anyone received feedback from authorities on how stability data should be integrated into the confirmatory testing strategy when there is evidence of nitrosamine formation during storage?

6

Please also check EMA Q&A 14 for new and ongoing MAAs. Perhaps this helps to define your control strategy and batch selection

I would say that this is related to Zone you want to register your product. In any way if results reach up to 30% AI, and you are about to include nitrosamine in specs (skip test) this would not change your strategy

Retain samples usually represent the worst case scenario - assessors loooove to assess the “worst case scenario” hahaha

2 Likes
7

I’m just enough of a statistician to get myself into trouble, but I would strongly consider a Statistical Power Analysis to determine the number of samples needed for confirmatory testing. If you were to bring that data and use a conservative approach, I think there would be a real chance of acceptance.

1 Like