Ch. 7— The Use of Animals in Testing • 155 
refers to tests with exposure for at least 1 year 
or most of the lifetime of the test species. Sub- 
chronic usually refers to tests of intermediate dura- 
tion— 3 to 6 months. Short-term repeated-dose tox- 
icity tests last from 2 to 4 weeks. 
Some have suggested that there is little to be 
gained by exposures of more than 6 months dura- 
tion for chronic toxicity testing (18,34). One com- 
mentator has argued that studies of 3 to 6 months 
are easier to interpret because the complicating 
effects of aging are avoided (44,45). Another finds 
longer tests necessary for detecting effects that 
occur only late in life or for which cumulative tox- 
icity is an important consequence (42). 
Throughout repeated-dose testing, animals 
would be observed for general appearance, res- 
piratory problems, central and peripheral nerv- 
ous system function, coordination, and behavioral 
changes. During and following the course of ex- 
posure, observations are made of hematology 
(hematocrit, white cell count, platelet count, clot- 
ting factors), ophthalmology, electrolyte balance, 
carbohydrate metabolism, liver and kidney func- 
tion (as determined from concentrations of cer- 
tain substances in the blood), body weight, and 
the appearance of lesions . After the animals have 
been sacrificed, observations are made of body 
surfaces, orifices, cavities, and organs. Microscopic 
examinations are made of selected tissues and or- 
gans, of gross lesions, and of organs that changed 
in size. One technique used in repeated -dose tox- 
icity testing to determine whether the toxic effects 
are reversible is to give a satellite group the high- 
est dose of the test substance and then give the 
animals time to recover before sacrificing them. 
Carcinogenici ty 
Cancer is a major human health concern, strik- 
ing one out of four and killing one out of five Ameri- 
cans (53). Consequently, carcinogenicity is an im- 
portant animal test. Detecting human carcinogens 
presents special problems because a latency period 
of 20 years or more can occur. Animal testing, par- 
ticularly in rodents, is useful because the latency 
period for tumor formation is much shorter (1 to 
2 years for rodents), thus allowing potential hu- 
man carcinogens to be detected during testing and 
before use, at which point they could become ma- 
jor public health problems. It is also much easier 
to control the animal environment than the hu- 
man environment, and therefore to investigate 
causal relationships. 
Although many human carcinogens were dis- 
covered without animal testing, several have been 
identified by first using such tests, e.g., DES, vinyl 
chloride, and bis (chloro -methyl) ether (55). Ani- 
mal use has its limitations; many substances cause 
cancer only in certain species . The known human 
carcinogens benzene and arsenic have never 
proved to be animal carcinogens. Hundreds of sub- 
stances have been identified as carcinogens in tests 
with one or more animal species but not in hu- 
mans, in part because of insufficient human epi- 
demiologic data and in part because some of them 
undoubtedly do not cause cancer in humans (41). 
Nonetheless, the use of animals in testing for car- 
cinogenicity is widely endorsed (55). 
Carcinogenicity testing is more costly and re- 
quires far more animals than other tests . Chronic 
toxicity testing may use about 160 rats and 32 dogs, 
whereas carcinogenicity testing would use about 
400 rats and 400 mice. (In order to economize, car- 
cinogenicity testing and chronic toxicity testing 
are often combined.) Cancer is easy to detect if 
tumors are visible, but it can only be detected in 
its early stages by microscopic examination of mul- 
tiple samples of 30 or more tissues and organs that 
may appear normal. Typically, 500,000 data points 
must be analyzed (41). 
These large numbers of animals and multiple 
data points are needed for statistical reasons . Can- 
cer has a high background incidence and large var- 
iations from animal to animal, making it difficult 
to establish that cancer was caused by the test sub- 
stance. The higher the incidence of spontaneous 
cancers, the more difficult it is to establish a link 
between cancer and the test substance. For ex- 
ample, if the background rate of cancer is 10 per- 
cent and the common criterion for statistical sig- 
nificance of 0.05 is used, the number of animals 
required to detect carcinogenicity in 90 percent 
of the tests is as shown in table 7-1. As can be seen, 
if a test substance causes cancer in 80 percent of 
the animals, 48 animals are needed to demonstrate 
carcinogenicity. If the incidence is only 15 percent, 
over 3,000 animals are needed. It has been sug- 
gested that the background incidence could be re- 
