You are here:
A Four-Step and Four-Criteria Approach for Evaluating Evidence of Dose Addition in Chemical Mixture Toxicity
Citation:
Hertzberg, R. C., R. Lyles, L. Haber, J. E. SIMMONS, J. E. SIMMONS, M. Kohrman-Vincent, D. W. HERR, AND V. C. MOSER. A Four-Step and Four-Criteria Approach for Evaluating Evidence of Dose Addition in Chemical Mixture Toxicity. Presented at International Toxicology of Mixtures Conference, Arlington, VA, October 21 - 23, 2011.
Impact/Purpose:
We propose a four step evaluation of dose additivity based on a stringent interpretation of the U.S. EPA definition of dose addition
Description:
Dose addition is the most frequently-used component-based approach for predicting dose response for a mixture of toxicologically-similar chemicals and for statistical evaluation of whether the mixture response is consistent with dose additivity and therefore predictable from the dose-response curves of the components. Toxicologic similarity as interpreted by the U.S. EPA (1986; 2000) goes much further than merely having the same major effects in the same target organs. Its first assumption is that similar chemicals can act as toxicologic clones, so that the dose-response curves of two similar chemicals are congruent once their doses are scaled by their relative potencies (scaled equivalence is called toxicologic proportionality). We propose a four step evaluation of dose additivity based on a stringent interpretation of the U.S. EPA definition of dose addition. The four steps are: 1) use nonlinear regression to fit all candidate dose-response models to the component data; 2) pick a preferred model based on fit, model simplicity and toxicological knowledge (e.g., is there a plateau response, a saturable process); 3) fit the dose additive version of the preferred model (the Combined Prediction Model) to the merged single chemical data; and 4) evaluate four comparisons regarding support for dose addition. Those comparisons include: a) the prediction model compared with the component data, to decide toxicologic proportionality of components; b) the fit of the mixture model (the preferred component model formula fitted to the mixture data); c) the fit of the prediction model to the mixture data; and d) the difference between the prediction model and the mixture model at all tested doses. With binary and 7-component mixtures of carbamate pesticides, these four pieces of evidence give different indicators of deviation from dose addition, posing different questions for further toxicologic investigation. This is an abstract of a presentation and does not reflect EPA policy.