PRINCIPLES OF TOXICOLOGY

294 CHEMICAL CARCINOGENESIS
• The pharmacokinetics of the chemical are not dose-dependent.
• The dose–response relationship is linear.
• DNA repair is not dependent on dose.
• The response is not dependent on the age of the animal.
• The test dose need not bear a relationship to human exposure.
Following a review of these assumptions, however, the Board of Scientific Counselors within the NTP
concluded that the implicit assumptions underlying linear extrapolation from the MTD do not appear
to be valid for many chemicals, and that both the criteria for selecting doses tested in the chronic
bioassay, as well as the method for extrapolating these results, should be reevaluated. Regarding the
issue of alternative criteria for selecting the highest doses to be tested in a chronic bioassay, the
following criteria recommended earlier by Squire seem to address a number of the issues that are raised
by implicit assumptions associated with the MTD:
1. The MTD should induce no overt toxicity, that is, no appreciable death, organ pathology, organ
dysfunction, or cellular toxicity.
2. The MTD induces no toxic manifestation predicted to shorten lifespan.
3. The MTD does not retard body weight by 10 percent.
4. The MTD is a dose that in two-generational studies is not detrimental to conception, fetal or
neonatal development, and postnatal development or survival.
5. Takes into consideration important metabolic and pharmacokinetic data.
There are a number of attractive features of this proposed definition for the MTD. Because the ultimate
goal of all toxicity testing is to identify all potential hazards we should be guarding against, and to
develop exposure limits that will prevent all toxicities, the criteria listed above allow other toxicological
considerations, under specific circumstances, to set a reasonable upper limit on the doses tested for
carcinogenicity. If high doses produce biochemical changes not seen at lower doses, and if at these
doses the chemical produces other toxicities we have already identified and must prevent by limiting
exposure, then these toxic endpoints may set a reasonable upper limit on the dose range we should
employ to test for other toxicities (e.g., cancer).
While a number of additional arguments and example compounds can be cited in support of
changing the doses tested in chronic animal cancer bioassays, especially if we are going to use the
results in the risk assessment and risk management areas, this particular feature of the test protocol
has placed regulatory agencies on the horns of a dilemma that is difficult to escape. It seems only
logical to attempt to maximize the sensitivity (ability to detect carcinogens) of this test by using the
highest dose possible. Testing the maximal dose helps eliminate the chance of producing false negative
responses. Testing the maximal dose helps ensure the statistical significance of small but important
changes, and helps set a manageable limit on the number of animals that must be tested to be able to
statistically identify a positive response. On the other hand, by maximizing the dose that is tested, we
seem to be incurring a considerable number of positive responses, the results of which, after further
testing for mechanisms at considerable additional expense, do not seem to be relevant, serious human
hazards, at least at the doses to which humans are exposed. The possibility that this may be a substantial
problem with the current testing scheme is indicated by analyses showing some 44 percent of the
positive test results observed in NTP bioassays as positive (carcinogenic) only at the highest dose tested
and not at a dose that is 25–50% of the MTD. Thus, it would appear that for almost one-half of the
chemicals tested thus far, the carcinogenicity of the chemical is strictly a function of the high doses
being tested.