Subdivided Historical Data to Assess Replicability of the Rat Embryo-Fetal Developmental Toxicity Study

Area of Science:

• Toxicology
• Developmental Biology
• Scientific Reproducibility

Background:

• Scientific reliability hinges on replicability, ensuring consistent experimental outcomes.
• Embryo-fetal developmental toxicity (EFDT) studies can assess replicability via in vitro models, targeted in vivo studies, or second species studies.
• This research examines the replicability of whole-animal studies using historical rat data.

Purpose of the Study:

• To evaluate the within-study replicability of whole-animal embryo-fetal developmental toxicity (EFDT) studies.
• To assess the consistency of results from historical rat data in EFDT studies.
• To establish benchmarks for acceptable variability in EFDT data.

Main Methods:

• Downloaded data for two endpoints from five full studies on the National Toxicology Program (NTP) website.
• Divided each study group into two replicate sets to assess within-study replicability.
• Utilized summary statistics, scatter plots, modified Levene's test, and Cohen's d for analysis.

Main Results:

• Replicate means deviated minimally from original studies (0.4%-3.7%) and between replicates (≤7%).
• Coefficients of variation (CV%) remained consistently low across subgroups (<10% in most cases).
• Variance testing indicated significant differences in two studies, with one showing opposite fetal weight effects in a subgroup.

Conclusions:

• Observed 5%-7% differences between idealized replicates may define the lower bound for acceptable variability in EFDT data.
• This study provides a foundation for enhanced replicability evaluations in EFDT research.
• Findings may inform future regulatory guidelines for EFDT study design and data interpretation.