Risk Assessment - Toxicology of Metals - page 280

in this handbook). Data on acute and chronic toxicity
in experimental animals provide useful information
on potential effects in humans. To provide such use-
ful information, animal studies have to be well per-
formed. Test guidelines are available for the various
specifi c animal bioassays for short-term, subchronic,
and chronic toxicity, for developmental and repro-
ductive toxicity, immunotoxicity, and carcinogenicity
(see www.oecd ). Available human and animal data for
metals on some of these specifi c effects are provided
in separate chapters of this handbook, for immuno-
toxicity in Chapter 11, for reproductive and develop-
mental effects in Chapter 12, and for carcinogenicity in
Chapters 5 and 10. Animal studies published in recent
years often fulfi l the OECD requirements or similar
guidelines, but older data may not. Positive fi ndings
may still be useful even if the size of groups do not
meet the requirements of the guidelines. The absence
of observed effects, however, cannot be assessed from
such incomplete data.
In addition to the categories of bioassays mentioned,
specifi c tests may be required for effects like skin and
eye irritation. Traditionally, such irritation effects have
been assayed in whole animals, but recent achieve-
ments concerning alternatives to animal testing have
made it possible to test some of these irritation effects
in alternative systems (see http://ecvam.jrc.it ). Particu-
larly for effects like irritation, implying some degree of
animal suffering, it is important from an ethical point
of view to use nonanimal assays as much as possible if
such systems can provide the same degree of safety for
humans in the fi nal risk assessment.
When human data are entirely or almost entirely
lacking, hazard identifi cation is based on animal and
in vitro data.
Only when the nonhuman data are of good qual-
ity, can a reasonable degree of safety for humans be
achieved. Even limited human data complementing
good nonhuman data improves the risk assessment
considerably both from a qualitative (hazard identi-
fi cation) and a quantitative (dose-response analysis)
point of view (see Section 4).
In vitro data on cells and molecular systems provide
very useful information when mechanism of action for
toxic metals can be described. In situations in which
such mechanisms can be shown to be valid at dose
levels occurring in human tissues to realistic expo-
sure levels for humans and animals, such in vitro data
are very valuable for extrapolation of observations in
animals to humans (see Chapter 6).
The role of metals in carcinogenesis was discussed
in an international meeting in 1980 (Belman and Nor-
dberg, 1981). No fi rm conclusions in regard to the
use of so-called short-term tests were reached at that

time, although lifetime studies on animals were con-
sidered predictive of probable carcinogenicity in man,
especially when the routes of exposure and the sites
of the tumors are the same. These are the same evalua-
tion principles as used by IARC in their documents on
carcinogenic risks to humans (IARC, 2006a,b). There
are several examples of metals/metalloids (arsenic,
beryllium, cadmium, chromium, and nickel) that have
proven carcinogenic in both humans and animals.
At the international meeting referred to previously
(Belman and Nordberg, 1981), some chronological
observations were made on the carcinogenicity of
metallic compounds as judged by case reports, ani-
mal experiments, and epidemiological studies. This
evaluation has been updated by us, including all metal
compounds classifi ed by IARC as human carcinogens
or possible human carcinogens (IARC, 2006a,b) and is
shown in Figure 1.
The reason that caution should be exercised
when extrapolating data from animals to humans
is not only because of the possible differences in
metabolism related to species differences. Exposure
situations will also differ. Often exposure in animal
experiments is higher and more acute than exposure
in real life. On the other hand, if only single substance
exposures occur, this situation excludes several pos-
sibilities for interactions. Furthermore, the number
of animals exposed is usually small compared with
the number of people that may be exposed and does
not include the very young or very old and other risk
groups, such as people with defi cient or suboptimal
nutritional intakes and people who already suffer
from diseases.
As mentioned (Section 3.2), the IARC classifi cation
system for carcinogenicity in humans uses these princi-
ples for classifi cation into Group 1. Experimental data
in animals and in vitro systems are the main basis for
classifi cation of exposures into Group 2, although epi-
demiological data are also required for classifi cation
in Group 2A. As stated in the preamble of the IARC
monographs (IARC, 2006b).

3.3.1 IARC Group 2

This category includes agents for which, at one
extreme, the degree of evidence of carcinogenicity in
humans is almost suffi cient , as well as those for which,
at the other extreme, there are no human data but for
which there is evidence of carcinogenicity in experi-
mental animals. Agents are assigned to either Group
2A ( probably carcinogenic to humans ) or Group 2B ( possi-
bly carcinogenic to humans ) on the basis of epidemiologi-
cal and experimental evidence of carcinogenicity and
mechanistic and other relevant data. The terms probably