Are you comparing to the qualification threshold calculated via ICH Q3A/B?
Because a higher specification than the qualification threshold based on a scientific justification is possible under ICH Q3A/B (often accepted outside of the context of nitrosamines) and “reducing to the safe level of 1500 ng/day” should still be arguable considering the data (negative in vivo TGR assay) supports at least this? From the perspective that a non mutagenic substance should not be controlled below the ICH M7 TTC, due to an atypically low API MDD (and considering that such low API MDD is probably quite diluted in the dosage form, the technicality of measuring lower levels than 1500 ng/day can play as well (if root-cause wise DP testing is relevant), the limits should remain practical). As it is a combination of safe and attainable or detectable levels, an in globo approach departing from the screening data could make sense.
Can you share the calculation you are using (e.g. what is MDD of medicine and API)?
Often a qualification threshold is higher than 1.5 µg/day, but also for the other thresholds I wasn’t expecting below 1.5 µg/day easily unless for very low API doses (µg’s per day).
But maybe you are looking at much less than 1.5 mg API per day and the reporting threshold?
1.5 mg API/day: 1500 ng/day ICH reporting threshold is still higher than the 150 ng/day based on 10% of 1500 ng/day (E.g. CPCA/EMA guidance)).
And for 1.5 mg API per day you would also get qualification threshold 15 µg/day/identification threshold 5 µg/day.
Maybe you can find back in ICH Q3A/B discussions from the design phase what the lowest API MDD in mind was.
It is conceptually interesting though if acceptance of ICH Q3A/B based on negative in vivo TGR means you can fully leave the framework of nitrosamine guidance behind and focus on what ICH Q3A/B only says for scientific justifications, for omission of reporting,… and engage with (national) CA on that. For higher API daily doses the decision making suggested in ICH Q3A/B guidance would not often conflict with nitrosamine guidance principles (considering also the high limits compared to the typical detections), but I hadn’t thought about very low API MDDs (do many exist below 1.5 mg/day?) and their possible formulation in diluted medicines. Those might get both lower benefit of ICH Q3A/B principles and technical challenges for reaching a low LOQ.
See also the Burns et. al. comments on "attainable levels of control/detection’:
The assessment of the essential medicines list identified no nitrosamines within category 1 to be formed from secondary amines in the API structures themselves, however 3 structures were identified in category 2, including N -nitrososertraline and N -nitrosoepinephrine. In the case of N -nitrosoepinephrine a 100 ng AI in the drug product (a 0.1 mg/ml solution of API) correlates to a concentration of just 1 ppb. At present the CPCA framework does not incorporate the less than lifetime approach, however this scenario, though certainly an extreme case, clearly indicates the need to consider impact of attainable levels of control/detection for a 1 in 100,000 lifetime risk of cancer versus the acute treatment benefits.
Burns, M. J., Ponting, D., Foster, R. S., Thornton, B. P., Romero, N. E., Smith, G. F., Ashworth, I. W., Teasdale, A., Simon, S., Schlingemann, J. (2023). Revisiting the Landscape of Potential Small and Drug Substance Related Nitrosamines in Pharmaceuticals. Journal of Pharmaceutical Sciences, Redirecting.
I’m wondering now if that is one of the reasons not all regulators find it easy to integrate ICH Q3A/B in their nitrosamine guidance. Maybe it is not because regulators don’t want to endorse the compounds not being mutagenic or not in the CoC, maybe it is because of the system of a variable limit (leading to internitrosamine differences (value-wise) in the approach due to different MDD (range) of the APIs) and loosing oversight on AI values)?
I’ve seen also cases where the readacross AI was higher than an ICH Q3A/B AI would suggest, but there is no clarity yet if scientific justification would allow to go back to the readacross AI.
Just like with the EAT and readacross, bringing examples in practice and discussing with regulators will clarify a lot. Possibly some more industry position papers to come on this (although a big focus seems to be the positive in vivo TGR now).
Considering the cost and limited capacity for in vivo TGR studies scientific advice ahead of studies is probably preferred.