We very much appreciate the thoughtful reviews of our paper by Drs. Hoel, Moolgavkar, Portier and Ye. The only point we wish to re-emphasize is in regard to our motivation in developing quantitative simulation models for dioxin and other liver tumor promoters. Our approach was to allow the broad base of biological data to guide development of a comprehensive hypothesis for dioxin's mode of action and then to develop quantitative simulation models to test for model plausibility and to estimate values of critical parameters. We were also especially interested in directly linking regional measures of tissue dose (or regional measures of hepatocyte response) with promotional potency.

The negative selection hypothesis for tumor promotion coupled with the extensive work on mitoinhibition during treatment with multiple promoters caused us to focus on functional differences that might exist in foci from control animals and from promoter treated animals. These considerations formed the biological basis of our two-cell hypothesis. With phenobarbital, the purported role of TGF-ß and the mannose-6-phosphate receptor in the mitoinhibition and promotion provided a sound basis for suggesting general differences in clones in treated versus control rats. However with dioxin, the growth stimulatory and mitosuppressant factors are not known. At this point, the work required to assess whether a two-cell model is really needed for dioxin is not computational, it is experimental. Do clones in control and dioxin treated animals have differences in phenotype related to differential growth advantages? Are there mitosuppressant growth factors or growth factor receptors absent form dioxin treated foci and present in foci from the control animals? One candidate protein factor for further study with dioxin might be tumor necrosis factor- (TNF-). A role TNF- in dioxin toxicity was first proposed by scientists the National Institute of Environmental Health Sciences in the 1980's

Ultimately, our main interest is in the mode of action of tumor promotion, including linkages between tissue dose and tissue responses throughout the liver and the impact of these tissue responses on cell growth characteristics of normal and initiated hepatocytes. The quantitative models discussed in our perspective are critical components of hypothesis formulation, hypothesis clarification and hypothesis testing. As the mode of action of these promoters is clarified by further experimentation and by development of refined simulation models, it will become apparent whether the U-shaped curves are real and whether they should directly influence risk assessments for these promoters.