Specifically for nitrosamines and their by regulators and scientists reviewed key mechanism of concern: The Thresher studies showed that the sensitivity for nitrosamines is 93%, showing good correlation between the Ames test result and the carcinogenicity of the nitrosamine. As such, the Ames test as described in the OECD 471 guideline can be used as an indicator of the carcinogenic potential of a nitrosamine, and can be an element to an overall assessment of the potency of a nitrosamine. To derisk mutagenicity/carcinogenicity, it is typically not the only element in the weight of evidence, and todays regulator guidances remain conservative from a precautionary perspective on what the EAT can support limit-wise.
Redirecting.
In addition to this, the photo-activation mechanism linked to NMOR has been shown to be inducable in the Ames test when using irradiation to activate NMOR, this is the key concept of the paper you have shared. So what works to induce the NMOR mechanism should also be helpful to show if the same is possible on other substrates or not and can help to show that the current literature is correct in identifying alpha-hydroxylation related photo-induced effects (especially when not alpha-hydroxylatable substrates are evaluated) and further help to establish a possible link between +S9 Ames + and -S9/photo-induced Ames +. This is truly a stress test, making abstraction if in vivo photo-induction is relevant based on the ADME and exposure route of the specific NDSRI.
Overall, there seems to be agreement in the scientific community that even if it could be (in general terms) a false positive, a positive in the Ames test should be endorsed from a precautionary principle.
In case you would only like to rely on in vivo carcinogenicity data or in vivo mutagenicity data in combination with newer data on the correlation between in vivo mutagenicity and in vivo carcinogenicity for nitrosamines:
Then I find it strange your papers focus on the CPCA for the nitrosamine AI. CPCA is a model designed for regulatory decision making on nitrosamine AIs (built on small nitrosamine data), building in a high degree of precaution to compensate for scientific uncertainty and does minimize the risk for unacceptable intakes, but does not claim to accurately predict NDSRI AIs (which can be an important element for correlation building with clinical observations). During SOT 2024 FDA scientists have also presented new data to support this.
For example your reporting on “Nitroso-sertaline is as toxic as NDMA/NNK” is not precise (which is not problematic for supplying safe medicines, but might be problematic for correlation building with clinical observations): it being treated by the model CPCA in the NDMA/NNK box does not prove the real AI is as low as the one of NDMA/NNK. The EMA reported EAT negativity on nitroso-sertraline stresses this further (cf. shift from 100 ng/day to 1500 ng/day guided acceptable intake in EU). If you would not support this test because the EAT protocol does not involve photo-activation: if you don’t agree that photo-activation is another route to similar activity and that building protective strategies around alpha-hydroxylation risks is not sufficient, it is strange your papers refer to CPCA AIs as well, as these are built on this mechanistic consideration.
Correlation building/bias risk for NDSRIs is probably influenced by the lack of studying the real AI (based on literature review or going further than only the FDA list and also looking to HC, TGA, EMA lists) and the lack of knowledge of contamination levels. Abstraction of this probably requires big studies like the one on ranitidine (allowing to approach general causality in a multifactorial context). And just like the discussion on mutagenicity and carcinogenicity correlation, there might be further discussion on the correlation between photo-induced mechanisms and enzymatic mechanisms. From an environmental exposure scenario it is easy to assume photo-induction though, but in vivo there is also the systemic availability element.
Again, CPCA is a model and not a primary source of nitrosamine acceptable intakes.
Full elimination of exposure of humans to nitrosamines is an ideal scenario you propose but probably not realistic, giving the multitude on exposure routes and the fact that they can also be formed endogenously. (See for example also the recent review of EFSA for nitrosamines via food and the work the EMA has commissioned on studying endogenous nitrosation). So it does remain all about the dose and evaluating the acceptable daily intake and building in scientifically sound protective factors around that intake.
On the urge for elimination:
Where further efforts are done usually this is called “ALARP”, going as low as reasonably/technically possible. Today we can measure ppb levels of nitrosamines, but not ppt or ppq levels, so proving “endless elimination” is not possible (and thus not formalisable in guidance) even if such a production process exists, the question what is the acceptable intake cannot be avoided. Again, I believe in toxicology it is all about the dose and determining the safe dose.
A fair balance has to be struck between protecting the patient and realising access to medicines (risk/benefit). When asking something that cannot be measured (and is not needed to be measured that low from a safety perspective), the access to medicines might be unnecessarily threatened by making getting authorisation for the medicine impossible. Surely diagnostic limitations based on the state-of-the-art are not strange to clinical practice as well and recognisible.
Investigating the principles of the LCR under ICH M7 and the design of CPCA or nitrosamine guidance will probably offer some more perspective as well.
For some background see also: Review of NDSRIs in Pharmaceutical Drugs -Pub