Hair dyes are easily accessible and represent a big part of the cosmetic industry. Given their frequent use, hair dyes as well as other cosmetic products must be safe. Specifically, cosmetic products are not allowed to contain substances that are carcinogenic, mutagenic or toxic to reproduction. In this thesis, we focused on carcinogenic and mutagenic aspect of twelve most frequently represented substances used in oxidative hair dyes. The selection of the latter was conducted from a review of 30 different oxidative hair dyes. Since 2013, cosmetic products and their ingredients are not allowed to be tested on animals, therefore we used in silico methods to predict their carcinogenic and mutagenic potential. These methods are based on physico-chemical properties of certain chemical and its structural alerts. Furthermore, they are also based on the following presumption; those substances, which are structurally related also share the same mechanism of action. In our research, we used the following programmes: OncoLogic, Toxtree, T.E.S.T. and Derek Nexus. We compared the generated predictions with results from databases and data obtained from in vitro and in vivo studies. While T.E.S.T. only predicts mutagenicity and OncoLogic only predicts carcinogenicity, other two programmes are capable of predicting mutagenicity as well as carcinogenicity. Only a few predictions in terms of carcinogenicity potential were correct, because in most other cases the prediction differed from the data from literature. When assessing the carcinogenic potential, programmes Derek Nexus and Toxtree possess the highest predictive strength, because they predicted the least false positive results, as opposed to OncoLogic, which failed to generate correct predictions. Most of the mutagenicity results from in vitro and in vivo studies are contradictory. Ames test in bacteria is most frequently used test for predicting in vitro mutagenicity. By comparing results from the Ames test we found out that Derek Nexus and Toxtree were able to generate the most correct predictions, when compared with T.E.S.T. Overall; the highest predictive strength can evidently be assigned to Derek Nexus and Toxtree. However, one has to bear in mind that this decision is based only on a few examples of correct predictions for carcinogenicity potential and some correct predictions for mutagenicity potential. In silico and in vitro methods represent a good as well as the only alternative to in vivo studies, but their predictive strength is still rather low, which we can see from comparing the results. In the future more accurate QSAR models should be developed for appropriate evaluation of cosmetic ingredients.
|