182 • Alternatives to Animal Use in Research, Testing, and Education 
cies that extrapolation to humans is only rough. 
Third, the LD 50 of a given substance varies signifi- 
cantly from laboratory to laboratory, and even in 
the same laboratory. 
Various regulatory classification schemes make 
distinctions between levels of toxicity ("highly toxic” 
versus "toxic,” versus "moderately toxic,” versus 
“nontoxic"). The LD 50 for two neighboring levels 
typically differs by a factor of 4 to 10. Yet, the 
reproducibility of test results does not justify even 
these distinctions . A recent study, though not nec- 
essarily typical, indicates the magnitude of the 
problem. A series of LD S0 tests were performed 
in 60 European laboratories for five substances 
on one species . The LD S0 for one substance ranged 
from 46 mg/kg body weight to 522 mg/kg, possi- 
bly ranging over three toxicity levels in some clas- 
sification schemes. Although the variations were 
not this large for the four other chemicals tested, 
the smallest variation was 350 to 1,280 mg/kg. Each 
test was done with 50 or more animals so that the 
results would be precise (61). 
Using Fewer Animals 
The standard LD 50 requires at least three groups 
of 10 animals or more each. An alternative proce- 
dure for determining the Approximate Lethal Dose 
(ALD) was developed as early as the 1940s (29), 
in which individual animals are administered doses 
that increase by 50 percent over the previous dose. 
Depending on the initial dose level, the total num- 
ber of animals needed is usually 4 to 10. Because 
the test substance might not be cleared between 
doses or because there may be cumulative effects, 
the ALD can be lower than the LD S0 , perhaps by 
70 percent, though more typically by less than 20 
percent (29). 
Many other acute toxicity tests that require fewer 
animals than the LD 50 have been developed (14, 
17,33,61,69,71,94,105,107).Most require that the 
doses increase sequentially, thereby allowing the 
experiment to stop when a certain limit is reached. 
Thus, fewer animals die in the conduct of a test, 
but its duration could increase from 2 weeks to 
a month or more. Although many investigators 
are moving to less precise LD 50 tests, no generally 
accepted alternative seems to have emerged. 
The Limit Test and 
Other Refinements 
If a substance is not lethal at high doses, its pre- 
cise LD 50 is not very important. In the limit test 
(80), a small number of animals is given a single 
oral dose, e.g., 5 g/kg body weight. If no animals 
die and no major ill effects occur, no further test- 
ing is needed. However, this limit is so high that 
this approach may have little practical value in re- 
ducing animal use (24). 
Rather than determining the dose that is lethal, 
studies can also be done to detect toxic effects at 
doses that are not lethal. As with the LD 50 , increas- 
ing doses can be administered to a small number 
of animals, perhaps stopping when some limit is 
reached. This approach can be further refined so 
that animals that are in distress could be sacrificed 
without affecting the outcome of the test (14). 
In Vitro and Nonanimal Methods 
Cell toxicity— changes in cell function or death 
of cells— can sometimes be used to detect acute 
toxicity. However, cell toxicity cannot be expected 
to function as a replacement for the LD S0 because 
lethality can occur by so many mechanisms that 
are supercellular. Cell toxicity is particularly use- 
ful in comparing members of chemical families, 
such as alcohols and alkaloids (79). 
At present, mathematical modeling has limita- 
tions, although it may have some utility in range- 
finding and in screening substances for testing 
(109). Modeling of acute toxicity fails to meet one 
of the criteria suggested by a working party on 
quantitative structure activity relationships, namely 
that the mechanism by which the response occurs 
should involve a common rate -determining step 
(88). Nonetheless, in a large study involving thou- 
sands of substances, a computer program was de- 
veloped that predicted LD 50 values within a fac- 
tor of 2.5 for 50 percent of the substances and 
within a factor of 6 for 80 percent. Considering 
