Because acute toxicity test data were used to develop and test this approach, abil-
ity of the narcosis model to predict effects from chronic exposures is less certain.
Overall, this source of uncertainty received a rank of high for magnitude of
uncertainty because information relating body burdens of metals to toxic effects
is limited and because the abilities of the various approaches to predict effects
from chronic exposures have not been thoroughly tested. It would be moderately
difficult to reduce this uncertainty given the complexity of interactions between
organisms and environmental media.
Human receptors: Dose-response models for carcinogens and
The dose-response assessment defines the relationship between the dose of a
contaminant and an effect. Potential toxic effects of contaminants are
categorized as either carcinogenic (cancer causing) or noncarcinogenic (acute,
subchronic, or chronic systemic effects). The mechanisms, models, and toxicity
factors derived for each of these categories differ. In both cases, high doses
administered to experimental animals or obtained from exposure estimates in
epidemiological studies are extrapolated to the low exposure levels expected in
the environment. The uncertainty factors applied to toxicity factors range from 10
to 10,000, leading to a magnitude of uncertainty ranking of high. It is possible,
with moderate difficulty given that test subjects are human, to test the accuracy
of individual uncertainty factors.
Potential carcinogens. The toxicity of a potential carcinogen is described by
a cancer slope factor (CSF). The slope factor is based on the assumption that
even a small number of molecular events can lead to changes in a single cell
leading to uncontrolled cellular proliferation. In other words, unlike non-
carcinogens, slope factors are derived by assuming that there is no concentration
or dose threshold below which the carcinogen does not pose any risk. A slope
factor is obtained by fitting a mathematical model to observed tumor-response
data followed by an estimation of the slope out of the range of the data. Differing
extrapolation procedures or models may lead to large differences in projected
risk at low doses. In addition, uncertainty in the carcinogenic mechanism of a
contaminant can make the choice of model more difficult.
Nonhuman animal data are often the only data available for deriving CSFs.
Consequently, the equivalent human dose to the animal dose used in the study
relies on the assumption that different species are equally sensitive to
contaminant effects given the same dose per unit of body surface area absorption.
(The calculation uses body weight raised to the 2/3 power, since surface area has
been shown to be proportional to this quantity.)
The USEPA Carcinogen Risk Assessment Verification Endeavor Work Group
describes the level of confidence that should be attributed to cancer slope
estimates (1998c). There is considerable uncertainty in the toxicity factors used
to estimate human health risks. One method of quantitatively assessing the
uncertainty involves constructing a probability density function by combining all
the available animal and epidemiological data (McKone and Bogen 1992). This
may be appropriate for contaminants where the mechanism of action is well
Chapter 5 Uncertainty in Tier IV Risk Assessments