Module 3 - Strategic case studies in practice

It is recommended that for NMEA, NNN, NMA, NDPA and any other N- nitrosamine with no or non- reliable toxicity studies which may be identified in the future in pharmaceuticals, a similar approach based on SAR considerations using the AI of the closest related N- nitrosamine for which a robust AI could be calculated should be chosen. In conclusion, CHMP considers that the primary attention with respect to risk by exposure of patients should be paid to the highly carcinogenic N- nitrosamines such as NMPEA, NDEA, NDMA, NMEA, NNK, NNN, NMOR, NMA, NDPA, NDBA, NPYR, MNNG, NMBA, NPIP and closely related molecules for which no data are available. In addition, in vitro assays in bacteria like the Ames assay or tests in mammalian cells cannot be used as a quantitative surrogate for carcinogenic potency. They only might serve as a qualitative read out for a mutagenic potential. As outlined in the CHMP Art 31 referral on sartans the generic and internationally agreed methodology for calculation of excess risk is the linear extrapolation using the TD 50 calculated of animal cancer studies as the point of departure as described in ICH M7(R1). The linear extrapolation framework is a conservative/precautionary regulatory risk assessment approach for genotoxic carcinogens that stipulates that i) there are no toxicological thresholds (no ‘dose’ is safe, which is debated with regard to biological plausibility), ii) the exposure-outcome relationship must always be monotonic (also subject to debate but a precautionary and pragmatic premise), iii) any other biological variables must always be insignificant in relation to the exposure (which is very questionable as one reaches very low or high exposure levels) and iv) that the ‘risk-per-unit-dose’ is always constant (also debatable at very low or high exposures). These premises create a theoretical conservative framework where risk in relation to exposure is considered additive. The methodologies to calculate the excess risk for humans and guiding decision making on immediate market actions in case of nitrosamine contaminations has followed so far the ICH M7(R1) approach for defining an AI. The AI in the context of ICH M7 is defined as an intake level that poses negligible cancer risk, or for serious/life-threatening indications where risk and benefit are appropriately balanced. The approach recommended in ICH M7(R1) is to use the TD 50 , as the point of departure for the calculation of excess cancer risk and calculating the dose associated with a theoretical excess cancer risk of 1:100,000 as the AI from which the limit is calculated based on the maximum daily dose of the medicinal product. A well acknowledged and accepted source for TD 50 values from cancer studies is the CPDB. The TD 50 calculated in the CPDB provides a robust reference value as long as the studies are well described and are multiple dose group studies with a minimum of 3 dose groups and 50 animals per dose per sex. The extrapolation to the excess risk level for cancer is performed by linear back extrapolation to the dose theoretically causing a 1:100,000 risk by dividing the TD 50 by 50,000 (50% or 0.5 x 100,000). For a person with a bodyweight of 50 kg the AI level is then calculated as AI = 50 x (TD50/50,000). The ad-hoc expert group expressed preference for the BMDL 10 model to define usable point of departure metrics, stating however that in certain cases the TD 50 model could be used. This approach is also harmonised across PROAST and BMDS software for quantal cancer bioassay data. The model averaging approach was also considered more suitable and easier and could overcome considerations with model selection. The BMDL is calculated as lower confidence limit (usually 90%) of a dose corresponding to a defined increase of a toxicological effect compared to controls. This increase is called benchmark response (BMR) or critical effect size (CES e.g. 5 or 10%). The corresponding dose is called Benchmark dose (BMD). Major limitations were identified by the Ad-hoc expert group with the 2.4.5. Generic methodology to calculate excess risk for humans

EMA/369136/2020

Page 43/90