Ch. 8— Alternative to Animal Use in Testing • 187 
sex-linked recessive lethal assay (18). Treated males 
are mated with untreated females, and the progeny 
are mated to each other. The number and charac- 
teristics of the male progeny are evaluated to de- 
termine if lethal mutations (that is, mutations that 
prevent viability) have occurred. 
Other tests involving fruit flies also exist or are 
likely to be developed. End points that can be meas- 
ured include the loss, gain, or breakage of chro- 
mosomes detected by examining germ cells. With 
the availability of mutant strains, the measurement 
of reverse mutations can be a valuable tool. Eye 
color is a popular method of following genetic ef- 
fects in the fruit fly (18). 
CARCINOGENICITY 
Many assays meant to replace carcinogenicity 
testing are designed to detect the initiation of can- 
cer rather than the formation of tumors. First, de- 
tecting initiation is faster and easier than detect- 
ing cancer. Second, although not all initiation leads 
to cancer, certain kinds are considered reliable 
surrogates for the disease. 
A major problem with evaluating the predictive- 
ness of alternatives to whole animals for carcinoge- 
nicity testing is that very few human carcinogens 
have been positively identified. Most substances 
treated as human carcinogens, although docu- 
mented to be known animal carcinogens , must be 
viewed as probable or suspected human carcino- 
gens. The development of alternatives is somewhat 
hampered by a lack of epidemiologic data on hu- 
mans. 
Various molecular and physicochemical prop- 
erties of substances have been correlated to car- 
cinogenicity. Some structure -activity models de- 
veloped for families of chemicals have predicted 
the carcinogenic properties for 75 to 97 percent 
of them. The chemicals modeled include polycyclic 
aromatic hydrocarbons (123), nitrosamines (89,99, 
121), and aromatic amines (124). 
The Ames Test 
Because mutation is often the first step in car- 
cinogenesis, the Ames test has been suggested as 
a possible screen or replacement for carcinoge- 
nicity testing. It has been evaluated for this pur- 
pose, both alone and as one in a battery of tests. 
Alone, it is less predictive than whole-animal tests. 
In a battery, it has been shown to be about as pre- 
dictive as animal testing for certain families of 
chemicals and substantially less predictive for 
others for the substances tested. Table 8-1 shows 
the predictiveness of mouse and rat bioassays and 
the Ames test for some known human carcinogens . 
The Ames test has been performed thousands 
of times in over 2,000 laboratories throughout the 
world and has provided results on over 1,000 
chemical substances since it was developed less 
than two decades ago. Portions of this large body 
of analytical data have been reviewed in over a 
dozen evaluation studies with the intent of deter- 
mining the test’s ability to predict carcinogenicity 
(6,19,66). These evaluations show that the percent- 
age of human carcinogens that are also mutagens 
(mutagenic carcinogens) ranges from 50 to 93 per- 
cent and is most likely about 80 percent (48). About 
20 percent of the human carcinogens were not 
mutagens (nonmutagenic carcinogens) in the Ames 
test, and it is believed that cancer associated with 
these carcinogens is initiated by a mechanism other 
than mutation. 
A critical analysis of several studies (19) identi- 
fied several sources of variation. These include 
methods of chemical selection, sample coding, use 
of a high proportion of chemicals known to work 
well or poorly with Ames testing, and differences 
in metabolic activation during the test procedure . 
The conclusion was that a reasonably careful ap- 
plication of the Ames technique to a nonbiased 
group of chemicals would be expected to yield a 
predictive accuracy of approximately 80 percent 
for mouse and rat carcinogens. 
The Ames test tends to be positive for a large 
proportion (about 40 percent) of substances that 
have not been identified as carcinogens in rodent 
bioassays. It should be noted, however, that these 
