Another important component in the analysis of assays for carcinogenic potential, both in vivo and in vitro, is that of the dose-response to a particular test agent. Just as with radiation carcino- genesis (Chapter 3), the effectiveness of the induction of neoplasia by an agent is dependent on the dose of that agent administered to the test animal. The dose-response curves for the stage of initiation in rodent liver neoplasia were already noted in Chapter 7, as well as the dose-response of promoting agents in the mouse epidermal model of multistage carcinogenesis. The differ- ences in the shapes of these two curves were discussed in that chapter and are considered again below.
Other factors may also influence a dose-response curve, such as the toxicity of the agent, the bioavailability of the agent, and the metabolic or pharmacokinetic characteristics of the agent within the living organism. A classic dose-response of a complete carcinogen and some of its ramifications are seen in Figure 13.12 (Druckrey et al., 1963). In this figure, curve 1 shows the relationship between the daily dose administered and the median total dose of animals develop- ing carcinoma. Thus, the left ordinate indicates the sum of all doses administered up to a 50% tumor incidence, therefore relating the total dose to the tumor incidence. In this way the straight- line relationship, if extrapolated, would proceed through the origin. Curve 2 relates the daily dose of carcinogen to the median induction time of the appearance of the first neoplasm. While extrapolation of curve 1 through the origin indicates that there is no dose at which some inci- dence of neoplasms is not apparent, it should be noted that if the daily dose is less than 0.1 mg/kg, no experimental data points are available. Furthermore, extrapolation of curve 2 to this low-dose region indicates that at doses lower than 0.1 mg/kg, the rats used in this experiment and whose lifetime is approximately 1000 days will not live sufficiently long for carcinomas to appear. In the assay depicted in Figure 13.12, the animals were administered a carcinogenic agent at a constant rate in the diet. Littlefield and Gaylor (1985) noted that with another com- plete carcinogen, 2-acetylaminofluorene, both the dose rate and the total dose administered are important in the final analysis. These workers demonstrated that when the total doses of this test agent were similar, the higher dose rates with shorter time periods induced a higher prevalence of neoplasms.
The use of the maximal tolerated dose (MTD) has been criticized because of the toxicity it induces, paralleled by an increase in cell proliferation in a number of susceptible organs (Ames
Figure 13.12 Dose-response relation seen in the chronic feeding of 4-dimethylaminostilbene to rats. 1. Relation between the daily dose and the median total dose for animals that developed carcinoma. 2. Rela- tion between daily dose and median induction time. The abscissa shows the daily dose; the ordinate on the left is the total dose administered, and on the right is the time from the beginning of the experiment. All scales are logarithmic. (Modified from Druckrey et al., 1963.)
and Gold, 1990). At the other end of the dose-response curve, the problem of the presence or absence of a threshold (no-effect level) of a carcinogenic agent is still hotly debated. It has al- ready been noted that, on theoretical grounds, agents capable of damaging DNA might not be expected to exhibit a threshold, whereas those exerting their effects through complicated recep- tor-mediated pathways, as with promoting agents, would be expected to exhibit a threshold of their effects (Aldridge, 1986). In addition, as briefly pointed out in the previous chapter (Chapter 12, Figure 12.5), carcinogenic agents, whether DNA-damaging or not, may exhibit a lower than control effect at very low doses (Kitchin et al., 1994; Teeguarden et al., 1998). Despite these observations, as well as the data depicted in Figure 13.12 and our knowledge of the mechanism of action of promoting agents, regulatory agencies have in general not seen fit to alter the philos- ophy that carcinogenic agents do not have threshold dose levels. As outlined below, this philoso- phy has permeated much of the regulatory action taken to date with respect to agents shown to be carcinogenic in any form of life.
In most epidemiological studies it has not been possible to determine the doses of the agents to which humans have been exposed, and only occasional, rather crude retrospective dose-response curves are available. However, a summary was made by the Meselson Committee (1975) of dose levels of several known human carcinogens that appear to be carcinogenic for certain human populations. These estimated levels were compared with levels of these agents known to produce neoplasms in animals (Table 13.7). They show that the cumulative doses re- quired per unit of body weight for tumor induction in the human and in experimental animals are of the same order of magnitude for four of the six agents. However, a more detailed comparison would require a correction for the short observation time in many of the studies on humans (as in the case of diethylstilbestrol and vinyl chloride), since many cancers in humans do not appear for 20 to 30 years after exposure. In addition, both vinyl chloride and diethylstilbestrol cause a very rare neoplasm in humans that is not usually seen in experimental animals. Thus, the effec- tive doses of several agents known to be carcinogenic for humans and rodents are not markedly dissimilar in the two species. If this conclusion can be extended to other chemical carcinogens in the human environment, then both the qualitative and quantitative extrapolations of such find- ings in the animal to the human situation have some degree of validity.