Is a theoretical evaluation of NDSRIs acceptable for a drug product when the impurity is not listed on any health authority website, but its acceptable intake (AI) has been established via the Carcinogenic Potency Categorization Approach (CPCA)?
Furthermore, the risk assessment indicates no risk from structural attributes, Route of Synthesis (ROS) or excipients, and the NDSRI is only cited in the DMF, with testing showing non-detectable (ND) levels in the API.
Hi Reshma,
An NDSRI mostly shares structural similarity with the drug substance.
If the NDSRI been cited in the DMF, then the description on the origin of the NDRSI would surely be part of the NRER document. If not request the drug substance manufacturer on the origin.
Testing and concluding as non-detectable levels would not be sufficient to conclude zero risk, the suitability of the analytical method to detect the NDSRI at less than 10% of the AI to be verified. Appropriate method with LOD & LOQ as defined authorities to be verified.
Thank you so much for the response!
Adding up, if the method validation concluded with NDSRI content as ND/BDL observation after the LOD and LOQ were established in Drug Formulation, still required testing…
Test for three consecutive commercial batches and if found less than 10% of AI limit then regular testing not required. If the values are more than 10% and within 30% skip is recommended. If found above 30% of the limit routine testing is recommended. Again if it was observed at BDL / ND at the initial stage, the potential for the NDSRI to spike on stability also to be monitored. It all depends on how good the assessment that is made on the drug product.
A lot of great advice here. I would come back to where the advice started. There may be a good reason the impurity is not listed. It is possible that with more information, you understand that this impurity is not possible given the API synthetic route, etc. I have seen this be the case and the API manufacturer test for it anyway.
EMA has the best guidance on lot selection for confirmatory testing. As I recall, it is 10% of the annual batches manufactured or 3 batches, whichever is higher. The wording gets a little vague, but they must be spread out over the shelf life of the product. I think the best interpretation or at least best practice would be that you are testing this 10% of batches or 3 batches for every year of shelf life. So, a total of at least 9 batches if the product has a shelf life of 3 years.
Thanks, Dr Brown, for the clean-up.
Please correct me if I’m wrong, but my understanding is that the number of batches selected for testing should be commensurate with the level of risk identified through the assessment. If the assessment indicates a minimal risk, I understand that initially performing confirmatory testing on three batches would be sufficient.
The 10% concept would then apply where the source of the risk has been identified and the impurity levels are demonstrated to be consistent across the batches.
Also the 10% of batches can be done for an already established / commercialised product. If its a new launch, when three batches are available the testing is done to understand the risk..
Just Mr. Brown. Appreciate the conversation. There are definitely a lot of grey areas in the implementation of the guidelines. You’re completely right about the path forward for new products. Here, that grey area is centered around the assessment of “no risk” and what a reviewer would accept. FDA’s guidance simply says testing at least 3 batches throughout the product shelf life is necessary when a risk is identified. However, that does not mean that they would accept just 3 batches, especially when that means that only one sample close to expiry would be tested. I’ve definitely seen FDA approve worse strategies.
However, at the end of the day, do what would make you feel comfortable that the issue is settled. Don’t just look at this as a way to check a box. If FDA comes back with questions, you can always test more.
Thank you so much for the lovely conversation and the insightful suggestions. I truly appreciate your time and guidance.
Hi all,
In principle, I agree with what is written.
I would like to highlight that, according to the EMA, the LOQ should be set at 10% of the AI limit. Therefore, acceptance criteria should require results to be below the LOQ (i.e., below 10% of the AI limit).
Getting back in the initial query (theoretical assessment), I agree that in case NDSRI is API related (NO-API) actual results are commonly requested. But in case it is a nitrosamine/NDSRI for example from an intermediate (early in synthesis) then perhaps theoretical assessment could be accepted (as per ICH M7). I think that other colleagues have already established experiences with such cases as posted elsewhere.
Moreover, regarding the number of batches, the assumption that the number of batches is directly correlated with risk is a reasonable one. However, this relationship is not explicitly defined in the relevant guidelines.
From our experience, health authorities typically request data from both fresh batches and batches tested up to the end of their shelf life, with all results being <10% of the AI limit if omission is sought. For example, in a recent submission, we presented NDSRI data generated at the end of shelf life, with all results reported as ND. Nevertheless, the health authorities came back and requested additional data from newly manufactured batches.
kindly
A lot of questions. The database is just an example based on what the FDA or other agencies have approved. So, even if the information is not in the database, you need to calculate per CPCA or do a surrogate study or in vitro Ames Test and propose a limit. FDA says they want it in a PAS, even if the value is per CPCA, but I know in some cases, they have allowed CBE 0. The issue of “no risk” is that you have to demonstrate it by testing three lots. If the amount is persistently below 10% of the CPCA based limit, you can justify not testing unless you make sweeping changes somewhere.
I am always glad to hearing of your opinion Aloka
This is exactly what I have understood by continuous screening/ reading of Guideline -also taking into account available information in ICH M7. Do you mean 3 lots (end of shelflife or irrespective)?
Yet, again based on our experience, this is not enough for many assessors who keep requesting 3 “fresh” and further results during shelf life (to ensure that impurity is consistently below 10% till end of shelflife), in order to demonstrate omission.
Eleni, ideally three lots at different stages of shelf life. So, one fresh lot, one in the middle and one at the end of shelf life is possibly okay. However, I would use judgement based on what I see in the fresh lot.
Thank you Aloka I will keep it in mind. Perhaps this would be clearer upon revision of ICH M7