Determining organ weight toxicity with Bayesian causal models: Improving on the analysis of relative organ weights

Area of Science:

• Toxicology
• Pharmacology
• Biostatistics

Background:

• Organ weight changes are key indicators in animal toxicity testing for new chemical entities.
• Traditional relative organ weight (organ-to-body weight ratio) analysis inadequately controls for body weight variations.
• This limitation can lead to misinterpretation of chemically induced organ damage and flawed safety conclusions.

Purpose of the Study:

• To address the limitations of relative organ weight analysis in toxicity studies.
• To introduce a more robust statistical framework for interpreting organ weight changes.
• To enhance the accuracy and ethical considerations of animal toxicity testing.

Main Methods:

• Proposed a causal modeling framework as an alternative to traditional statistical methods.
• Utilized a Bayesian approach within causal modeling to handle unequal variances and multiple testing.
• Demonstrated the simplicity and interpretability of the proposed causal modeling analysis using freely available software.

Main Results:

• Causal modeling provides a direct assessment of scientific interest regarding organ weight changes.
• The Bayesian approach within causal modeling allows for modeling unequal variances and controlling for multiple testing.
• This framework facilitates clearer interpretation of results, directly providing evidence of safety.

Conclusions:

• Causal modeling offers a superior method for analyzing organ weight data in toxicity studies compared to relative organ weights.
• The proposed Bayesian causal modeling framework is interpretable, statistically sound, and ethically advantageous.
• Adoption of this method can lead to more accurate safety assessments, reducing wasted resources and improving animal welfare.