Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

The rat bone marrow micronucleus test: Statistical considerations on historical negative control data

: Igl, Bernd-Wolfgang; Bitsch, Annette; Bringezu, Frank; Chang, Steffi; Dammann, Martina; Frötschl, Roland; Harm, Volker; Kellner, Rupert; Krzykalla, Volker; Lott, Jasmin; Nern, Marlies; Pfuhler, Stefan; Queisser, Nina; Schulz, Markus; Sutter, Andreas; Vaas, Lea; Vonk, Richardus; Zellner, Dietmar; Ziemann, Christina


Regulatory toxicology and pharmacology 102 (2019), pp.13-22
ISSN: 0273-2300
Journal Article
Fraunhofer ITEM ()

Recent updates of the OECD Guidelines for the Testing of Chemicals (Section 4: Health Effects) on genotoxicity testing emphasize the use of appropriate statistical methods for data analysis and proficiency proof. Updates also concern the mammalian erythrocyte micronucleus test (OECD 474), as the currently most often performed regulatory in vivo test. As the updated guideline gives high importance to adequate statistical assessment of historical negative control data to estimate validity of experiments and judge results, the present study evaluated statistical methodologies for handling of historical negative control data sets, and comes forward with respective proposals and reference data. Therefore, the working group “Statistics” within the German-speaking “Gesellschaft für Umwelt-Mutationsforschung e.V.” (GUM) compiled a data set of 891 negative control rats from valid OECD 474-studies of four laboratories. Based on these data, Analysis-of-Variance (ANOVA) identified “laboratory” and “strain”, but not “gender” as relevant stratification parameters, and argued for approximately normally distributed micronucleus frequencies in polychromatic erythrocytes per animal. This assumption provided the basis for further specifying one-sided parametric tolerance intervals for determination of corresponding upper historical negative control limits. Finally, the stability of such limits was investigated as a function of the number of experiments performed, using a simulation-based statistical strategy.