In vivo comet assay of acrylonitrile, 9-aminoacridine hydrochloride monohydrate and ethanol in rats

Re-analysis results using medians of the data from the JaCVAM-organized international validation study of the in vivo rat alkaline comet assay

“…Among genotoxic non-carcinogens, 9-aminoacridine hydrochloride monohydrate (9-AA) was called as positive in liver using the median based approach in both the laboratory [7] and our analyses while it was negative using the means approach [2]. The other genotoxic carcinogens and genotoxic non-carcinogens gave the same calls by both approaches.…”

“…Data re-analysis was conducted by two types of investigations, i.e., one used each laboratory analysis report [4][5][6][7][8][9][10][11][12][13][14][15][16] and the other was our unified analysis. For both analyses, based on TG 489 [3], the median % tail DNA for each slide was determined, the mean of the 2 median values was calculated for each animal, and then the mean of the individual animal means was determined to give a group mean.…”

“…In laboratory analysis, statistical significance between a negative control group and chemical treatment groups except for the positive control group was analyzed with a one-way analysis of variance (ANOVA) using log-transformed values (3 labs; Lab B: one-sided Simple meanapproach Significant 18 3 3 24 Significant in either statistic 0 5 3 8 Non-significant 2 4 47 53 Number of data 20 12 53 85 Significant: statistically significant in Dunnett's and linear trend test, non-significant: no statistically significant in both statistical analysis methods, significant in either statistic: statistically significant in either but not both of the two statistical analysis methods. and p < 0.03 [4], Lab D: p < 0.05 [6], Lab G: p < 0.001 [9]) followed by a multiple-comparison such as Dunnett's test (3 labs; Lab C: p < 0.05 [5], Lab L: two-sided and p < 0.05 [14], Lab M: two-sided and p < 0.05 [15]), a multiple-comparison with Dunnett's test for nonlog-transformation values (5 labs; Lab E: two-sided and p < 0.05 [7], Lab F: two-sided and p<0.05 [8], Lab I: two-sided and p < 0.05 [11], Lab J: p < 0.05 [12], Lab N: two-sided and p < 0.05 [16]), or a post-hoc right one-sided pairwise comparison using log-transformed values (1 lab; Lab K: one-sided and p < 0.05 [9]). One laboratory (Lab H) did not use statistics because the values in all treatment groups were lower than the respective control values [10].…”

“…One laboratory (Lab H) did not use statistics because the values in all treatment groups were lower than the respective control values [10]. Dose-dependency was also examined in 8 of 12 laboratories with a statistical analysis method, such as a linear trend test [4,7,8,[11][12][13][14]16]. …”

“…Here, we refer to the former method of analysis as the simple mean approach and the latter as the median based approach. Due to TG 489, the individual laboratory reports of the main validation study (Phase 4-2) from most of the participating laboratories were revised and described using their own statistical approach based on using medians (and means) [4][5][6][7][8][9][10][11][12][13][14][15][16]. Also, we separately re-analyzed the Phase 4-2 data based on the median based approach in the same manner as the original statistical analysis of the JaCVAM-organized validation study as reported by Uno et al [2] in order to compare the sensitivity/specificity using the simple mean approach with the median based approach.…”