Novel statistical approach for evaluating flow cytometric in vitro micronucleus data.

Area of Science:

• Toxicology
• Genetics
• Biostatistics

Background:

• Current statistical methods for in vitro micronucleus data analysis are designed for small sample sizes.
• Flow cytometry enables evaluation of significantly larger cell numbers, leading to smaller variance estimates.
• Traditional tests applied to large datasets can yield statistically significant results lacking biological relevance.

Purpose of the Study:

• To investigate alternative statistical methods for analyzing in vitro micronucleus data from large sample sizes.
• To compare the performance of traditional, alternative, and a novel statistical approach (z-score).
• To determine a more biologically meaningful method for assessing genotoxicity from high-throughput screening.

Main Methods:

• Analysis of 95 in vitro micronucleus experiments conducted between 2011-2013.
• Comparison of recommended, widely used, and sample-size independent statistical tests.
• Evaluation of the "z-score" (z-factor) as a metric for identifying 'hits' in high-throughput screening.
• Integration of the z-score with historical vehicle control data to establish biological meaningfulness.

Main Results:

• Traditional and alternative statistical methods resulted in 42-68% of experiments showing statistically significant outcomes (p < 0.05).
• The z-score statistic reduced the number of significant calls to 27%.
• Combining the z-score with biological criteria restored the original positive frequency of 19%.

Conclusions:

• Traditional statistical tests may be overly sensitive for large sample sizes generated by flow cytometry in micronucleus assays.
• The z-score offers a more appropriate and less sensitive tool for analyzing large-scale genotoxicity data.
• A combination of the z-score and biological relevance criteria is recommended for robust analysis of in vitro micronucleus data.