able from other sources, such as (1) structure-activity relationships relating the toxicity of a chemical to other members of its chemical family; (2) the results of in vitro tests of the chemical; and (3) human epidemiological observations about the effect of the chemical.

To make their evaluations of risk, assessors seek accurate mechanism-based empirical data—that is, data based on a solid understanding of the mechanism of a chemical’s toxicity, as determined by developmental toxicologists. Such data are sparse. Several uncertainties limit the estimation of a chemical’s potential for developmental toxicity. Animal bioassays, principally using mammals, currently are considered to provide the most reliable data for extrapolating toxicant effects to humans. Because these bioassays are expensive and time consuming, only a small fraction of the compounds in commerce and in the environment have been fully evaluated for their toxicity potential in animals. The many attempts to devise simpler, less costly test systems involving tissue explants, cell cultures, or purified biological molecules have so far proved to be of only limited value in predicting the actions of compounds on human embryonic and fetal development. Among the reasons for poor predictability are the inherent differences between the simple test systems and humans regarding the uptake, distribution, metabolism, and excretion of chemicals and the lack of understanding about the basic mechanisms of development.

In the absence of accurate mechanism-based empirical data, risk assessors often make four kinds of default assumptions when recommending the acceptable levels of exposure of humans to an environmental agent. First, they assume that animal test results are relevant for humans. Unless there is contradictory evidence, humans are assumed to be the most susceptible mammals, and a factor of 10 below the maximum no-effect exposure level in the animal’s development serves as a basis for setting the acceptable human exposure level. Second, a further 10-fold reduction is introduced to take into account the possibility that the animal’s developmental response, which frequently was obtained at subchronic exposure to the chemical, might not reflect human responses at prolonged (chronic) exposures. Third, a 10-fold reduction is introduced to cover the possibility that susceptibility varies among human individuals, some being inherently more sensitive to the chemical. Fourth, a 10-fold reduction is sometimes introduced if the toxicity database for a chemical is incomplete. Because of the susceptibility of developing systems, an additional child specific factor (usually a 10-fold reduction) is sometimes applied. Although many risk assessors would prefer to use mechanism-based empirical data instead of those defaults (up to a 10,000-fold compounded reduction in acceptable human exposure beyond that given by the animal test) to improve their risk assessments for environmental agents, the test data for the assessors’ use often are sparse because of limited resources and are of unknown applicability to humans because of a lack of understanding of basic mechanisms of developmental toxicity and of differences in humans and animals.

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement