Module 3 - Strategic case studies in practice

However, the alkylating potential alone does not only determine the mutagenic and therefore carcinogenic potency. N- Nitrosamines with more bulky side chains (e.g. NPYR, NPIP) do have higher TD 50 values as compared to those with small side chains. They are nevertheless high potential cohort of concern carcinogens with a TD 50 below 1.5 mg/kg/day.

2.4.2.2. DNA repair processes

Among the genotoxic N -nitrosamines, different nitrosamines (or more specifically their reactive metabolites) can generate different types of DNA lesions. For instance, NDMA has so far been linked to different alkyl-group types of guanine adducts, adenine adducts, and thymine adducts whereas tobacco-related N -nitrosamines such as NNN and NNK are primarily linked to so called bulky DNA adducts on the different nucleosides. At the same time, there is only a limited capacity of DNA repair systems for different DNA lesions. This would imply that that to some degree, N -nitrosamine exposure from pharmaceutical impurities is integrated into the more general N -nitrosamine exposure as long as there is some commonality in the type of DNA lesions. The full DNA lesion profiles of potent nitrosamines such as NDMA and NDEA remain unclear but it is well established that they generate pro-mutagenic 0 6 -alkylguanine adducts (i.e. the modification of guanine through the addition of small alkyl-groups such as e.g. -CH 3 , -C 2 H 5 ) which are most commonly repaired via dealkylation by DNA alkyl transferases (AGT, also known as methyl-guanine-methyl- transferase, MGMT). Other nitrosamines such as NDBA, NDELA, NMEA, NNK and NNN are also known to generate 0 6 -alkylguanosine adducts (Dennehy & Loeppky 2005, Coulter et al 2007, Kotandeniya et al 2013, von Hofe & Kleihues 1986, Bonfanti M et al. 1990). The activity of this enzyme is considered to have a high variability within humans. In a small study by Lees et al. 2002, the interindividual variability of MGMT activity in human colon mucosa was 6- to 7-fold. In a larger cohort this variability can be expected to be even greater. In rodent cell lines (e.g. rat liver), MGMT repair enzymes have been reported to be inducible in contrast to human cell lines. This may indicate potential differences in detoxification between rats and humans (Fritz et al. 1991, Grombacher & Kaina 1995, Fritz & Kaina 1992) Animal data generated in lifetime bioassays are the most reliable source to conclude on the carcinogenicity of chemicals and human relevance. Reliable human data are currently not available for most chemicals and are also lacking for N- nitrosamines. The characterization and ranking of the carcinogenic potency based on animal data is difficult and only possible in a limited fashion. There are several aspects to be considered. One major factor is that it is experimentally practically impossible to reduce the exposure or dose beyond a certain level as one has to separate increasingly small chemical carcinogenesis effects from the background levels of target organ relevant neoplasms. In other words, to detect increasingly subtle toxicological neoplasms against the background levels of neoplasms in a given organ as the exposure decreases, the size of the experimental groups needs to increase substantially. The largest mammalian chemical carcinogenesis study so far used ~24,000 animals 10 – far more than all nitrosamine carcinogenicity studies reported so far - and only reached a sensitivity of 1 in 100 cases for liver and bladder neoplasms. This has to be considered in relation to the theoretical 1 in 100 000 excess risk for oncogenesis that serves as reference for establishing levels of mutagenic impurities that are expected to pose negligible carcinogenic risk according to ICH M7(R1), meaning that all estimates of human risk from animal data are based on a theoretical linear extrapolation (generally considered the most conservative extrapolation approach) from a higher animal dose to a 2.4.3. N- Nitrosamine carcinogenicity in animals

10 Bruce R.D. et al (1981), Fund. Appl. Toxicol, 1 (1981), pp. 67-80.

EMA/369136/2020

Page 37/90