Benchmark dose (BMD) modeling: current practice, issues, and challenges.

Area of Science:

• Toxicology and Risk Assessment
• Biostatistics and Computational Biology

Background:

• Benchmark dose (BMD) modeling is the current standard for establishing points of departure in risk assessment.
• Key benefits include comprehensive data utilization, enhanced consistency, and improved statistical uncertainty accounting.
• Discrepancies persist between regulatory bodies like the US Environmental Protection Agency (US EPA) and the European Food Safety Authority (EFSA) regarding specific modeling aspects.

Purpose of the Study:

• To highlight ongoing disagreements in benchmark dose (BMD) modeling methodologies.
• To identify areas requiring further research for harmonizing risk assessment practices.
• To emphasize the importance of model confidence over selecting the lowest benchmark dose lower confidence limit (BMDL).

Main Methods:

• Discussion of differences in benchmark response (BMR) selection for continuous data.
• Analysis of the implications of using unrestricted models and various mathematical models.
• Exploration of model averaging as a potential solution to single-model selection challenges.

Main Results:

• Divergences in regulatory recommendations (US EPA vs. EFSA) lead to variations in benchmark dose lower confidence limit (BMDL) values.
• Current practices may not adequately account for model confidence or biological relevance in all cases.
• Model averaging shows promise but requires integration of biological considerations.

Conclusions:

• Further research is needed to refine benchmark dose (BMD) modeling, particularly concerning benchmark response (BMR) and uncertainty factor (UF) interplay.
• Incorporating biological data into model averaging weights is essential for robust risk assessment.
• Addressing these methodological differences will facilitate harmonization and advance the field of risk assessment.