Weighted multiplicity adjustments for animal carcinogenicity tests

Area of Science:

• Toxicology and Pharmacology
• Statistical Analysis in Biological Studies

Background:

• Analyzing animal carcinogenicity data is challenging with sparse data and low background tumor rates.
• The Haseman rule is commonly used to increase statistical power in these scenarios while controlling false positive rates.

Purpose of the Study:

• To evaluate the global error rates of the Haseman rule and its variants in animal carcinogenicity studies.
• To develop and present alternative weighting methods to address the deficiencies of the Haseman rule.

Main Methods:

• Exploration of the performance of the Haseman rule and its variants.
• Development of novel weighting methodologies.
• Application of proposed methods to actual pharmaceutical company carcinogenicity data.

Main Results:

• The study found that the Haseman rule and its variants can result in unacceptably high global error rates.
• Alternative weighting methods were developed and demonstrated to correct these deficiencies.

Conclusions:

• The Haseman rule's limitations necessitate the use of alternative statistical approaches for accurate animal carcinogenicity data analysis.
• The proposed weighting methods offer a more reliable approach for evaluating tumor rates in sparse data settings.