Estimation of benchmark dose ratio distributions for subchronic-to-chronic extrapolation using meta-analysis

Area of Science:

• Environmental Health
• Toxicology
• Risk Assessment

Background:

• The International Programme on Chemical Safety (IPCS) established a probabilistic framework and the Approximate Probabilistic Analysis (APROBA) software for deriving chemical reference values.
• This framework estimates uncertainty and variability, including extrapolation from subchronic to chronic toxicity data, using benchmark doses (BMD).

Purpose of the Study:

• To re-evaluate the subchronic-to-chronic extrapolation distribution using a more extensive dataset.
• To assess the consistency of the IPCS framework and APROBA default parameters with newly analyzed toxicological data.
• To identify potential limitations in APROBA's uncertainty quantification for chemical risk assessment.

Main Methods:

• Estimated the subchronic-to-chronic distribution using a larger dataset encompassing histopathological and clinical endpoints, beyond historical body and organ weight data.
• Compared the derived distribution and uncertainty estimates with the default parameters and assumptions within the APROBA software.
• Evaluated dose-response model fit using multiple indicators beyond standard goodness-of-fit statistics.

Main Results:

• The analysis confirmed that key assumptions of the IPCS method and APROBA's default values are generally consistent with the new data.
• However, predictions for dichotomous response data exhibited greater uncertainty than assumed by APROBA, leading to approximately 25% lower reference values in an example case.
• APROBA's default parameters do not fully capture the uncertainty in predicted chronic BMDs, especially when BMDs fall outside the observed dose range or confidence limits are unquantifiable.

Conclusions:

• While the IPCS framework is largely supported, APROBA's default settings may underestimate uncertainty for certain toxicological endpoints.
• More rigorous evaluation of dose-response model fit is crucial for accurate uncertainty quantification in chemical risk assessment.
• The findings highlight the need for careful application and potential refinement of probabilistic methods in setting chemical safety reference values.