Validation of the statistical parameters and model selection criteria of the benchmark dose methods for the evaluation of various endpoints in repeated-dose toxicity studies

抄録

The benchmark dose (BMD) approach is one of the important techniques in dose-response assessment for the risk assessment of chemicals and adapted by various international organizations. We investigated the appropriateness of the statistical parameters and model selection criteria for BMD lower bound (BMDL) estimation by BMD software (BMDS) (developed by the US Environmental Protection Agency) and PROAST (developed by the National Institute for Public Health and the Environment of the Netherlands). Publicly available repeated-dose toxicity study data (226 dichotomous datasets and 151 continuous datasets) were used for the investigation. Our findings were applied to establish BMD technical guidance for BMDS for the evaluation of various endpoints in repeated-dose toxicity studies. Under the Japan Chemical Substance Control Law (CSCL), the DRA-BMDS guidance (i.e., Division of Risk Assessment-BMDS guidance) is used for the evaluation of a “Priority Assessment Chemical Substance.” Namely, selecting of an extra risk of 10% (dichotomous data) or a level change of 1SD (continuous data) as a default benchmark response. Running all the models without or with parameter constraints. Selecting the model that calculated the lowest BMDL but excluding the one that estimated a BMD/BMDL ratio ≥ 10 or lowest dose/BMDL ratio ≥ 10. We believe that the DRA-BMDS guidance can assist risk assessors in the selection of the BMD model.