The Effect of Historical Data-Based Informative Prior on Benchmark Dose Estimation of Toxicogenomics

Area of Science:

• Environmental toxicology
• Genomics
• Computational toxicology

Background:

• High-throughput toxicogenomics is crucial for assessing environmental chemical toxicity.
• Limited data in dose-response analyses leads to uncertainties in parameter and benchmark dose (BMD) estimation.
• Bayesian methods integrating historical data via prior distributions offer a potential solution but are understudied.

Purpose of the Study:

• To evaluate the effectiveness of informative priors in genomic dose-response modeling and BMD estimation.
• To identify plausible informative priors for gene and pathway-level BMD estimates.
• To assess the impact of time-specific informative priors on BMD estimation.

Main Methods:

• Derived a general informative prior and eight time-specific informative priors for seven continuous dose-response models.
• Utilized Bayesian methods to integrate historical data.
• Conducted real data-based simulations to evaluate BMD estimation with and without informative priors.

Main Results:

• Derived informative priors showed sensitivity to the specific datasets used for elicitation.
• Time-specific informative priors improved or maintained BMD estimation accuracy compared to noninformative priors.
• Informative priors significantly decreased uncertainty and slightly enhanced correlation with apical endpoint-derived points of departure.

Conclusions:

• Historical data-based informative priors offer significant benefits for BMD estimation in toxicogenomics.
• The study systematically examined the effects of informative priors, highlighting their utility in advancing toxicogenomic practices.
• Plausible informative priors enhance the reliability and reduce uncertainty in toxicity assessments.