gms | German Medical Science

Background: The in vivo single cell gel electrophoresis assay (“comet assay”) is a widespread and sensitive method to detect direct DNA damage in various tissues. Several regulatory frameworks include the comet assay as a standard genotoxicity assay for the prediction of potential human risks. However, due to the complexity of the method, the diversity of different variables (e.g. organs), and, in addition, numerous methodological steps with potential impact on the test outcome, there is, currently, no universal statistical strategy available to handle comet assay data in vivo. Nevertheless, the respective OECD Guideline No. 489 (2016) (“In vivo mammalian alkaline comet assay”) gives respective recommendations, for example, the computation of the median primary endpoint (%tail intensity) per slide followed by taking the arithmetic mean of all medians per animal.

Methods: In general, there is a variety of different well-known strategies for analyzing comet data (see e.g. [1], [2], [3], [4]). Moreover, based on experimental data, different simulation settings were carried out to analyze the effect of different slide summaries, for example the median, arithmetic mean, geometric mean, on the final test decision. Initial examinations are based on a simple linear model, which was extended to a multivariate linear model involving additional endpoints (tail moment, tail length) and corresponding correlation structures. In addition, we will embed historical control information into the statistical strategy using a multivariate Bayesian model. A nonparametric modelling approach will also be investigated and compared from a frequentist (bootstrap) and from a Bayesian point of view.

Results: It turned out, that the distribution of e.g. %tail intensities can be extremely skewed and contaminated with “extreme” values, and that different statistical strategies may differ to a large extend. For example, the choice of the slide summary can strongly effect the interpretation of the test results and can vary even between positive or negative.

Conclusion: Statistical methods matter; this is particularly true for the analysis of comet assay data in vivo. The statistical strategy incl. the summary of data on the cell level has an immense impact on the final test decision and must be chosen with extreme care.

The authors declare that they have no competing interests.

The authors declare that an ethics committee vote is not required.