Benchmark Dose Analysis of Ames Test Data: Identifying Appropriate Benchmark Response Values and Modeling Approaches

As interest in the quantitative interpretation of genotoxicity assay data continues to grow, the Ames bacterial mutagenicity assay offers a valuable opportunity due to its long-standing use and extensive dose-response data sets. A key challenge for the quantitative use of this data is determining how best to apply benchmark dose (BMD) analysis to Ames data to support a quantitative risk assessment. This requires careful selection of both an appropriate benchmark response (BMR) and the most suitable modeling approach. Using a database of Ames bacterial mutagenicity test data from regulatory submissions, we evaluate a range of BMRs and several benchmark modeling approaches. Our evaluation indicates that a 50% increase in revertants over the background (BMR50) is a justifiable, interpretable candidate BMR for Ames studies in the pesticides regulatory context. Furthermore, in the filtered Ames-positive data set, the Bayesian implementations (BMABMDR Laplace approximation and Bridge sampling) yielded BMDLs that met the EFSA criteria for use as a reference point for a larger proportion of data sets and had narrower confidence intervals than the PROAST implementation. Together, these findings provide a practical framework for deriving Ames-based potency estimates to support chemical prioritization, read-across, and quantitative weight-of-evidence assessment.