Preclinical Candidate Evaluation Department, Supernus Pharmaceuticals Inc. , Rockville, MD , USA.
Toxicol Mech Methods. 2014 May;24(4):284-93. doi: 10.3109/15376516.2014.881946. Epub 2014 Jan 27.
Caramel coloring chemicals, 2-methylimidazole (2-MI) and 4-methylimidazole (4-MI) have been used extensively in coloring soft drinks. The health effects of these chemicals have been concerned in the recent years.
In the present investigation, 2- and 4-MI were subjected to three commonly used structure-activity relationship (SAR) software to understand the utility of such software as a method of alternatives to animal testing in predicting potential genotoxicity and tumorigenicity.
Three SAR software: Osiris, ToxTree and DEREK, were used. Published procedures and/or manuals of respective software were utilized to generate data outputs and the data were evaluated in comparison with available toxicological data on 2- and 4-MI.
The results show that these software predicted genotoxic activity in comparison with published genotoxicity for 2- and 4-MI. However, only one of three software used (Osiris) predicted imidazole ring in 4-MI to be tumorigenic; other software predicted them to be negative.
Based on the weight of evidence of SAR results observed in this study and the genotoxicity and tumorigenicity reported using actual in vitro and in vivo animal testing in literature, it was concluded that the models used are useful for routine screening of chemicals; however, for better prediction, additional models may be employed. Software's ability to predict health effects depends on the type of structural alerts used as knowledgebase in developing such software.
Three computational software used in this study predicted genotoxic activity of 2- and 4-MI, but did not predict tumorigenicity conclusively when compared to literature reported animal data. Additional mechanistic non-clinical studies may be conducted to better understand reported tumorigenicity.
焦糖色素化学物质 2-甲基咪唑（2-MI）和 4-甲基咪唑（4-MI）已广泛用于饮料着色。近年来，人们对这些化学物质的健康影响表示关注。
在本研究中，将 2-MI 和 4-MI 进行了三种常用的结构-活性关系（SAR）软件分析，以了解此类软件作为替代动物测试的方法在预测潜在遗传毒性和致癌性方面的应用。
使用了三种 SAR 软件：Osiris、ToxTree 和 DEREK。使用了各自软件的已发布程序和/或手册来生成数据输出，并将数据与 2-MI 和 4-MI 的可用毒理学数据进行了比较评估。
结果表明，与 2-MI 和 4-MI 的已发表遗传毒性数据相比，这些软件预测了遗传毒性活性。然而，只有三种软件中的一种（Osiris）预测了 4-MI 中的咪唑环具有致癌性；其他软件预测为阴性。
根据本研究中观察到的 SAR 结果的证据权重，以及文献中实际体外和体内动物测试报告的遗传毒性和致癌性，得出结论认为，所使用的模型可用于常规筛选化学物质；但是，为了更好的预测，可以使用其他模型。软件预测健康影响的能力取决于用于开发此类软件的知识库中的结构警报类型。
本研究中使用的三种计算软件预测了 2-MI 和 4-MI 的遗传毒性活性，但与文献报道的动物数据相比，未能明确预测致癌性。可能需要进行额外的机制性非临床研究，以更好地了解报告的致癌性。
