The Control of The Consumer Risk in the Ames Assay
- PDF / 266,024 Bytes
- 5 Pages / 504 x 720 pts Page_size
- 74 Downloads / 187 Views
Drug Infonnation Journal, Vol. 31, pp. 363-367, 1997 Printed in the USA. All rights reserved.
THE CONTROL OF THE CONSUMER RISK IN THE AMES ASSAY MARKUSNEUH~USER Solvay Pharmaceuticals, Hannover, Germany
LUDWIGA. HOTHORN Universitat Hannover, LG Bioinformatik, Hannover, Germany
Preclinical tests in genetic toxicology represent safety studies. Therefore, the primary concern in the evaluation of Ames test data is the control of the consumer risk, that is, the risk of erroneously concluding safety. Hence, an equivalence test procedure is adequate. This approach is presented beside the two-fold rule and classical tests for differences with respect to an order restricted alternative. Key Words:Ames assay; Consumer risk; Difference approach; Equivalence test procedures for ordered many-to-one designs
INTRODUCTION THE AMES (1) ASSAY IS a short-term test in genetic toxicology (2) most often used worldwide. This assay is based on the detection of mutated histidine-dependent cell strains of Salmonella typhimurium. The mutated bacteria can mutate back to the wild-type, particularly if they are exposed to mutagenic compounds. The endpoint for evaluation is the number of revertant colonies on replicated plates. This endpoint is a count. There are some difficulties in the statistical evaluation of the Ames assay: As the distribution of the discrete endpoint is not clear, different distributions are used: the Poisson distribution and two special cases of the generalized Poisson distribution according to Consul and Jain (3), the negative binomial distribution is an extension to the Poisson distribution as well (4).Furthermore, the normal distribution is used after a suitable transformation, Small and unbalanced sample sizes are used according to guidelines, for example, three per dose group and five for the negative control group are common sample sizes. With the small sample sizes, the variance may be reduced by ties. As a consequence, this can lead to a false positive result, @Thevariances are frequently nonhomogeneous because of small sample sizes and ties. Moreover, the variance can increase with increasing effects, and
Presented at the DIA Workshop “Statistical Methodology in Non-Clinical and Toxicological Studies,” March 25-27, 1996, Bruges, Belgium. Reprint address: Markus Neuhauser, Solvay Pharmaceuticals, Hans-Bockler-Allee 20, D-30173 Hannover, Germany.
363 Downloaded from dij.sagepub.com at FLORIDA ATLANTIC UNIV on May 22, 2015
Markus Neuhauser and Ludwig A. Hothorn
364
There is the possibility of a downturn at high doses. This so-called muta-tox problem should be taken into account. There are different statistical approaches: One can ignore the muta-tox problem; this strategy is frequent in praxi. Another pragmatic method is to reject high doses before testing (5). A transformation can moderate the muta-tox problem. Moore and Felton (6) propose a weighted regression. Another method is stepwise testing. There are step-up and step-down procedures (7,8). Chen (9) proposes trend tests for umbrella alternatives. On
Data Loading...
