Martin Heather L, Svendsen Claus, Lister Lindsay J, Gomez-Eyles Jose L, Spurgeon David J
School of Biosciences, University of Birmingham, Edgbaston, Birmingham, B15 2TT, United Kingdom.
Environ Toxicol Chem. 2009 Jan;28(1):97-104. doi: 10.1897/07-215.1.
Ecological risk assessments must increasingly consider the effects of chemical mixtures on the environment as anthropogenic pollution continues to grow in complexity. Yet testing every possible mixture combination is impractical and unfeasible; thus, there is an urgent need for models that can accurately predict mixture toxicity from single-compound data. Currently, two models are frequently used to predict mixture toxicity from single-compound data: Concentration addition and independent action (IA). The accuracy of the predictions generated by these models is currently debated and needs to be resolved before their use in risk assessments can be fully justified. The present study addresses this issue by determining whether the IA model adequately described the toxicity of binary mixtures of five pesticides and other environmental contaminants (cadmium, chlorpyrifos, diuron, nickel, and prochloraz) each with dissimilar modes of action on the reproduction of the nematode Caenorhabditis elegans. In three out of 10 cases, the IA model failed to describe mixture toxicity adequately with significant or antagonism being observed. In a further three cases, there was an indication of synergy, antagonism, and effect-level-dependent deviations, respectively, but these were not statistically significant. The extent of the significant deviations that were found varied, but all were such that the predicted percentage effect seen on reproductive output would have been wrong by 18 to 35% (i.e., the effect concentration expected to cause a 50% effect led to an 85% effect). The presence of such a high number and variety of deviations has important implications for the use of existing mixture toxicity models for risk assessments, especially where all or part of the deviation is synergistic.
随着人为污染的复杂性不断增加，生态风险评估必须越来越多地考虑化学混合物对环境的影响。然而，测试每一种可能的混合物组合是不切实际且不可行的；因此，迫切需要能够根据单一化合物数据准确预测混合物毒性的模型。目前，有两种模型经常用于根据单一化合物数据预测混合物毒性：浓度相加模型和独立作用模型（IA）。这些模型所产生预测的准确性目前存在争议，在其用于风险评估能够完全合理之前，需要得到解决。本研究通过确定IA模型是否能够充分描述五种农药与其他环境污染物（镉、毒死蜱、敌草隆、镍和咪鲜胺）的二元混合物的毒性来解决这一问题，这些物质对线虫秀丽隐杆线虫的繁殖均具有不同的作用模式。在10个案例中的3个案例中，IA模型未能充分描述混合物毒性，观察到显著或拮抗作用。在另外3个案例中，分别有协同、拮抗和效应水平依赖性偏差迹象，但这些均无统计学意义。所发现的显著偏差程度各不相同，但所有偏差都使得对生殖产出所观察到的预测效应百分比出现了18%至35%的错误（即，预期导致50%效应的效应浓度却导致了85%的效应）。如此大量且多样的偏差的存在对于将现有的混合物毒性模型用于风险评估具有重要意义，尤其是在所有或部分偏差具有协同性的情况下。
