According to current policy, chemicals are evaluated for possible canc
er risk to humans at low dose by testing in bioassays, where high dose
s of the chemical are given to rodents. Thus, risk is extrapolated fro
m high dose in rodents to low dose in humans. The accuracy of these ex
trapolations is generally unverifiable, since data on humans are limit
ed. However, it is feasible to examine the accuracy of extrapolations
from mice to rats. If mice and rats are similar with respect to carcin
ogenesis, this provides some evidence in favor of interspecies extrapo
lations; conversely, if mice and rats are different, this casts doubt
on the validity of extrapolations from mice to humans. One measure of
interspecies agreement is concordance, the percentage of chemicals tha
t are classified the same way as to carcinogenicity in mice and rats.
Observed concordance in National Cancer Institute/National Toxicology
Program bioassays is around 75%, which may seem on the low side-becaus
e mice and rats are closely related species tested under the same expe
rimental conditions. Theoretically, observed concordance could underes
timate true concordance, due to measurement error in the bioassays. Th
us, bias in concordance is of policy interest. Expanding on previous w
ork by Piegorsch et al., we show that the bias in observed concordance
can be either positive or negative: an observed concordance of 75% ca
n arise if the true concordance is anything between 20 and 100%. In pa
rticular, observed concordance can seriously overestimate true concord
ance. A variety of models more or less fit the data, with quite differ
ent implications for bias. Therefore, given our present state of knowl
edge, it seems unlikely that true concordance can be determined from b
ioassay data.