Harmonization of AI for NDSRIs with insufficient carcinogenic data

For nitrosamine impurities without sufficient carcinogenic data to derive a specific acceptable limit, EMA, Health Canada and Anvisa provide a class specific TTC for nitrosamine of 18 ng/day as default option while FDA applies 26.5 ng/day.
The most conservative limit for nitrosamine impurities published by MFDS(South Korea) so far is 26.5 ng/day. We have almost finalize a guideline for industry on controlling nitrosamine impurities and will share within this community once it is completed.

Lately EMA updated their list of AI which includes the limits of N-nitrosorasagiline and N-nitrosodabigatran (18 ng/day). Considering the global harmonization of nitrosamine control, I was wondering how FDA regulates such NDSRIs and how pharmaceutical companies in US determine the limit of those NDSRIs with insufficient carc data - whether to apply “18 ng/day” same as EMA’s approach or “26.5 ng/day”.
It would be really helpful if someone can provide any information on this issue.

Thank you!

1 Like

@heajin nice question…
Same thing i also observed…
But however whenever there is no carcinogenicity data, need to use default AI limit as per EMA & USFDA respectively, as per guidance…


AI for NDSRIs with insufficient carcinogenic data can be derived using read across (i.e., structural analogue, if robust good quality data is available) approach.

So below are two scenarios which would be accepted by regulators:

  1. Use nitrosamine class specific AI value provided by regulator
  2. Propose new AI using read across (i.e., structural analogue, if robust good quality data is available) approach with robust and detailed justification support

How can anyone use read-across approach when the nitrosamine in question is an API-nitrosamine… of course there are no similar nitrosamine compounds with carc data available. So basically all API-nitrosamines are always going to be limited to default 18 ng/day. Or ?

1 Like

@Filip it does depend on the existence or not of a structure to do the read-across. In the case of N-Nitrosodabigatran it is used the default value 18 ng/day, but in the case of N-Nitrosovarenicline it is used read-across N-nitroso-1,2,3,6-tetrahydropyridine as the point of departure. So, It depends on the API structure


As @Naiffer_Host already mentioned, it all depends on structure type of NDSRI. It is not always the case that NDSRI must use class-specific nitrosamine limit. Robust literature search is very crucial to identify suitable analogue with robust well quality carcinogenicity data, even for NDSRI. With this one may support higher AI limit than default class-specific nitrosamine limit. An expertise in read across methods application would be very critical skill while justifying the use of analogue.