A bootstrap analysis of four in vitro short-term test performances

Area of Science:

• Toxicology
• Genetics
• Statistical Modeling

Background:

• In vitro short-term tests are crucial for predicting chemical carcinogenicity.
• Understanding the association between in vitro assays and rodent carcinogenicity is vital for risk assessment.
• Existing data from the U.S. National Toxicology Program (NTP) provides a valuable resource for this analysis.

Purpose of the Study:

• To estimate confidence intervals for association measures between four in vitro tests and rodent carcinogenicity.
• To evaluate the relationships among the in vitro tests themselves.
• To assess the sensitivity, specificity, and accuracy of these tests in identifying chemical carcinogens.

Main Methods:

• Utilized the bootstrap statistical technique to derive variability intervals for association measures.
• Applied multivariate statistical methods in conjunction with bootstrap.
• Analyzed data from Salmonella, mouse lymphoma L5178Y cell mutation, chromosomal aberrations, and sister-chromatid exchanges assays from the NTP database.

Main Results:

• The bootstrap technique provided robust confidence intervals for association measures.
• Multivariate analysis combined with bootstrap demonstrated the reliability of NTP data.
• The study offers insights into the performance and interrelationships of the in vitro mutagenicity assays.

Conclusions:

• The in vitro tests analyzed, when evaluated using bootstrap and multivariate methods, show robust associations with rodent carcinogenicity.
• The findings support the reliability of the NTP database for assessing mutagenicity and carcinogenicity relationships.
• This approach enhances the understanding of in vitro assay performance in predicting chemical carcinogenicity.