Drawbacks of in vitro models are that they have been developed mainly for screening purposes selleck products by the pharmaceutical industry and are not validated for certain categories of industrial chemicals. Therefore, training with the latter compounds and taking into account uncertainty is needed. This methodology allows for the determination of human pharmacokinetics of test compounds administered at doses much lower than the

expected pharmacologically effective or toxic levels (FDA, 2008). Microdosing has been used as part of human drug clinical testing to evaluate drug ADME (Coecke et al., 2005b) but has not been widely accepted for testing chemicals. This is not used universally and is done on a case-by-case basis. This technology, once installed is cost-effective to study new chemical entities and has the advantage of requiring only very low doses of radiolabelled compounds. One limitation to this technology is that the dose has to be lower than 100 μg, thus if this is significantly different from the therapeutic dose and the pharmacokinetics profile is different, then the low dose pharmacokinetics data may have decreased relevance compared to the toxic/effective concentration. Another disadvantage of this method

is that humans are purposely FG 4592 exposed to radiation for biomedical research and its use should therefore be justified (as recommended by the International Commission of Radiation Protection in Publication 62 (ICRP, 1991)). There are radiation dose constraints for volunteers under different conditions and these are discussed in the recommendations from the ICRP

(ICRP, 2007). In order to refine and improve existing in vivo study types, as well as reduce the number of animals used, for chemical testing, it was recommended to increase information gained from one study by incorporating Lck additional endpoints into the study, e.g. using peripheral blood for metabolomics and the micronucleus (MN) test. It is noted that inclusion of more endpoints, e.g. kinetics, may be difficult to implement for small animals, e.g. mice. In addition, inclusion of positive controls for each endpoint may mean extra animals are needed, although, for some endpoints which have sufficient historical data, such as the in vivo MN test, additional positive controls are not an absolute requirement. The different industry sectors have generated a vast amount of data using similar models; however, the sharing of this data across sectors has not been as fast flowing. The workshop recommended the sharing of in vivo data, coordination and information exchange between research projects and sectors. Companies should be encouraged to share in-house additional data from long-term studies so that in vivo studies are not unnecessarily duplicated and in silico/in vitro methods can be validated.