There’s one category of dialkyl nitrosamines that has been explicitly stated as to not be of concern - the EMA have indicated that at least one alpha-hydrogen is required (EMA, https://www.ema.europa.eu/documents/referral/nitrosamines-emea-h-a53-1490-assessment-report_en.pdf ); however at the moment all others should at least initially be considered as part of the Cohort of Concern. Note that not having an alpha hydrogen does not guarantee a negative result - nitrosodiphenylamine (NDPhA) is carc positive and Ames positive in some unusual strains - but does lift the compound from CoC to TTC-controllable levels (NDPhA is the single least potent positive nitrosamine in CPDB)
That said, there is a wide range of chemical structures and potency - 4 orders of magnitude in terms of TD50 values [Thresher et al 2020] - and many structures, like nitrosodiphenylamine does, fall outside the range of TD50s that would indicate the compound as being CoC. Kevin Cross and I recently presented initial SAR work at QSAR2021 - as in fact Rachael Tennant and I did last summer as a webinar - and we continue to work extensively in this area and have a number of upcoming presentations and publications. See my previous posts in the community for more details as to upcoming presentations, and past work is available here: https://www.lhasalimited.org/publications/?custom_in_AdditionalAuthors=4053&orderby=StartDate+desc - we literally have a paper with a title comparable to this thread, for example…
Broadly, the approach that we are taking can be categorised as “mechanism-based read-across” - identify which features that may impact mechanism of action or potency are present in a compound and the mechanisms by which they do so, then use that classification to determine suitable analogues with reliable toxicity data, leveraging the extensive toxicity data available through Vitic or CPDB/LCDB (carcdb.lhasalimited.org) through both data science techniques and our expert understanding as chemical toxicologists. This is the approach permitted under the EMA guidance, but reinforced by requiring consideration of mechanistic similarity rather than simple structural similarity.