RESUMO
Scientific and societal interest in the analysis of aggregate toxicity derives from the fact that people are seldom exposed to single chemicals, but rather to multiple agents from different sources and even to mixtures of agents from a single source. Many descriptive terms and mathematical, graphical, and statistical models have been used to evaluate the toxicity of simple mixtures. It is not very easy to distinguish clearly the intrinsic differences, distinctions and limitations of these models when applied to characterizing interactive toxicity. A series of experiments were performed to illustrate model-dependent consistencies and differences in interactive toxicity. Cultured murine renal cortical cells, target cells for metal toxicity, were treated with selected concentrations of one metal or binary mixtures of metals to give conditions of dose-additivity, response additivity, or with only one toxic member of the binary mixture. The cytotoxicity was determined at 24h by lactate dehydrogenase release. The data were analyzed graphically and mathematically by (a) Carter's statistical isobologram, (b) Barton's non-linear, and (c) Kodell and Pounds' linear models to characterize the interaction. These models were compared and contrasted for robustness, and consistency using these common data sets. The models gave generally consistent conclusions, but each model has limitations and strengths for assessing particular mixtures scenarios. This comparison illustrates the complexity of extrapolating conclusions between models, and difficulty of public health assessment from exposures to multiple chemicals in the environment.