A Survival-Adjusted Quantal-Response Test for Analysis of Tumor Incidence Rates in Animal Carcinogenicity Studies Shyamal D. Peddada and Grace E. Kissling Biostatistics Branch, National Institute of Environmental Health Sciences, National Institutes of Health, Department of Health and Human Services, Research Triangle Park, North Carolina, USA Abstract In rodent cancer bioassays, groups of animals are exposed to different doses of a chemical of interest and followed for tumor occurrence. The resulting tumor rates are commonly analyzed using a survival-adjusted Cochran-Armitage (CA) trend test. The CA trend test has reasonable power when the tumor-response curve is linear in dose, but it may be underpowered for a nonlinear response. An alternative survival-adjusted test procedure based on isotonic regression methodology has previously been proposed. Although this alternative procedure performs well when the tumor response is nonlinear in dose, it has less power than the CA trend test when the response is linear in dose. Here, we introduce a new survival-adjusted test procedure that makes use of both the CA trend test and the isotonic regression-based trend test. Using a broad range of experimental conditions typical of National Toxicology Program (NTP) bioassays, we conducted extensive computer simulations to compare the false-positive error rate and power of the proposed procedure with the survival-adjusted CA trend test. The new procedure competes well with the survival-adjusted CA trend test when observed tumor rates are linear in dose and performs substantially better when observed tumor rates are nonlinear in dose. Further, the proposed trend test almost always has a smaller false-positive rate than does the survival-adjusted CA trend test. We also developed an order-restricted inference-based procedure for performing multiple pairwise comparisons between each of the dose groups and the control group. The trend test and the multiple pairwise comparisons test are demonstrated using an example from a study conducted by the NTP. Key words: cancer bioassay, Cochran-Armitage trend test, multiple pairwise comparisons, National Toxicology Program, order-restricted inference, poly-3 trend test. Environ Health Perspect 114:537-541 (2006) . doi:10.1289/ehp.8590 available via http://dx.doi.org/ [Online 10 November 2005] Address correspondence to S.D. Peddada, Biostatistics Branch, MD A3-03, National Institute of Environmental Health Sciences, Research Triangle Park, NC 27709 USA. Telephone: (919) 541-1122. Fax: (919) 541-4311. E-mail: peddada@niehs.nih.gov We thank G. Dinse and B. Gladen for their helpful comments. We also thank the reviewers and the editor for several useful comments that have improved the article. This research was supported by the Intramural Research Program of the National Institute of Environmental Health Sciences, National Institutes of Health. The authors declare they have no competing financial interests. Received 16 August 2005 ; accepted 9 November 2005. The full version of this article is available for free in HTML or PDF formats. |