Dose-Response Modeling with Summary Data from Developmental Toxicity Studies.

Area of Science:

• Developmental Toxicology
• Risk Assessment
• Statistical Modeling

Background:

• Dose-response analysis of binary developmental data typically requires individual fetus or litter-specific data.
• Risk assessors often lack access to detailed litter-specific data, relying on dose-group summaries.
• Absence of litter-specific data leads to incorrect variance estimation, known as overdispersion or the 'litter effect'.

Purpose of the Study:

• To develop a method for analyzing fetal malformation summary data when litter-specific information is unavailable.
• To adjust summary data using historical design effects for accurate dose-response analysis.
• To compare benchmark doses derived from adjusted summary data with those from traditional litter-level analyses.

Main Methods:

• Utilized historical data to estimate the design effect for adjusting summary data on fetal malformations.
• Applied statistical models designed for binomial responses to the adjusted summary data.
• Compared the benchmark doses obtained from the adjusted summary data analysis with those from nested dichotomous models using litter-level data.

Main Results:

• Summary data on fetal malformations can be satisfactorily adjusted using estimated design effects from historical data.
• Analysis of adjusted summary data yielded benchmark doses comparable to those derived from litter-level data.
• The proposed method provides a viable alternative for dose-response analysis when detailed litter data are absent.

Conclusions:

• Adjusting summary data with historical design effects enables accurate dose-response analysis in developmental toxicity studies.
• This approach mitigates issues arising from missing litter-specific data, improving risk assessment.
• The methodology offers a practical solution for utilizing available data to determine benchmark doses effectively.