Inclusion of biological outlier values was found to bias the results of rat uterotrophic assays towards false negatives, i.e., not identify uterotrophic effects in treated populations. The present investigation was conducted to identify the background variability in the rat uterotrophic assay and to evaluate the need to exclude biological outlier values in untreated control groups. The Styrene Steering Committee (SSC) of the European Chemical Industry Council (CEFIC) co-sponsored this work with Argus Research Laboratories (Argus). The rat uterotrophic response assay originally was used as a pharmacology screen to identify estrogenic agents. Classically, 5 to 10 immature female rats (18 to 22 days of age) are administered an agent for three or four days. At sacrifice on the following day (21 to 26 days of age), the uterus is removed, weighed and a uterine weight/terminal body weight ratio calculated. This in vivo assay has been adapted for use in identifying the potential estrogenicity of chemicals, generally using 10 immature female rats per group, more closely controlling the ages, and adding one or more positive control groups to demonstrate sensitivity and response of the test system. Statistically significant increases in the positive control group means for absolute and relative uterine weights, as compared with the untreated (or vehicle-treated) means, is generally interpreted as identifying a sensitive test system. The untreated (and/or vehicle-treated) control group is then compared with the various test groups, and statistically significant increases in the mean absolute and relative uterine weights are identified as evidence of estrogenicity of the agent. Although not fully described previously, the inherent biological variability existing in both untreated and treated animals, can confound interpretation of the data, especially when numbers are relatively small. Our laboratories have identified that under controlled GLP-compliant conditions, some Wistar rats [randomly assigned (weight-ordered) to groups of ten at 22 +/- 1 days of age, and sacrificed when 26 +/- 1 days of age] in untreated control groups have high relative uterine weights that skew data distributions such that statistically significant differences are not present between untreated control and positive control groups. Based on these observations, further evaluations of untreated control and positive control (DES-DP, 2.5 micrograms/kg, b.i.d.) populations of three rat strains [Wistar--Chbb:THOM-SPF, Wistar--Crl:(WI)BR and Sprague-Dawley--Crl:CD(SD)IBS BR VAF/Plus "International Genetic Standard"] were made to define when such normal findings should be considered biological outliers, and whether outlier values should be excluded from analyses. Our data indicate that body weight is not always predictive of uterine weight, that relative uterine weight outlier values occur in each of these rat strains, and that statistically significant differences exist between groups of untreated control animals when outlier values are included in analyses. Of 98, 60 and 60 untreated control rats in the three respective strains, 11 (11.2%), 16 (26.7%) and 15 (25.0%) had relative uterine weights > or = 0.150%, and 5 (5.1%), 4 (6.7%) and 9 (15.0%) of these rats had relative uterine weights > or = 0.200%, values within the positive control range. All positive control rats attained relative uterine weights > or = 0.100%. Of 50, 60 and 60 positive control rats in the three respective rat strains, 27 (54%), 47 (78.3%) and 36 (60%) had relative uterine weights > or = 0.200%, 9 (18%), 2 (3.3%) and 7 (11.7%) had relative uterine weights > or = 0.300% and 5 (10%), 1 (1.7%) and 3 (5%) had relative uterine weights > or = 0.400%. The incidences of relative uterine weights > or = 0.300% in the positive control group may indicate the presence of high responders. Histological evaluations of uteri of positive control rats and untreated control rats with relative uterine weights > or = 0.
Read full abstract