for example,
MDD of Ampicillin Trihydrate is 12 g (as Ampicillin) (Equivalent to 13.85 g of Ampicillin trihydrate).
Acceptable Limit for NDEA in Ampicillin trihydrate: 0.0265 µg/day ÷ 13.85 g/day = 0.0019 µg/g.
Case as mentioned above will be challenging for analytical scientists to achieve 0.0002µg/g as LOQ Value.
How can we develop an analytical method with minimum errors/ artifact results.
Ampicillin is produced by the acylation of 6-aminopenicillanic acid with D-(-)-α-phenylglycine by either microbiological or chemical synthesis (Ivashkiv, 1973).
Where on earth can NDEA come from?
Is it only a speculative example?
E.g. use of diethylamine to obtain ampicillin diethylammonium salt as intermediate when used as late stage purification strategy for ampicillin based on solubility considerations.
A high daily doses and analytical limitation argument could be used to try to justify no specification setting in case of consistent no detection above the lowest achievable LOQ between 0.2 and 2 ppb (in combination with LTL arguments if preferred, see Bercu 2021 on NDEA). Method guidance for NDEA testing can be derived from Ph. Eur. chapter 2.5.42.
If the identified scenario is reagent-based (most likely NDEA prepresence in DEA used and not formation in process), reagent testing can be considered instead, if it is sufficiently straightforward to design a limit and testing on the reagent. Assessment strategies on where to test and how to control typically take into account that P. Gehlert and W. Rolle have described that nitrogen dioxide and diethylamine forms NDEA very fast at 25 °C (Experientia. 1977 May 15;33(5):579-81. doi: 10.1007/BF01946507). In addition, contact of DEA (residue) with activated carbon in the process and the fact that diethylamine participating in salt formation is not available for nitrosation can be used in the evaluation.
Depending on your territory of focus provisions for this in applicable guidance can be available.
If API testing (if any) is used to also analytically show API to reagent (or intermediate) mass balancing is possible (to the degree until the LOD/LOQ technical limitations start) or that there is analytical verification for the identified likelihood of a reagent origin, possibly higher LOD/LOQ is usable in case of relevant detection in API or intermediate. Specification design on DEA should not be impossible if the faith and purging is clear and showable with technical arguments and API testing with higher LOD/LOQ and can add further weight to the use of high daily doses and analytical limitation arguments (as these are then used to shift the control rather than to remove it).
Usage of alternative amines or building in a simple NDEA removal step on DEA in API production could be an alternative pathway, allowing to step away from NDEA in DEA testing controls to manufacturing controls with time if the as is NDEA risk is real.
So as a starting point, when the NDEA risk is real, the best technically achievable method development plan commensurate to the risk is identified typically I would say.