Apart from mutations with obvious phenotypic effects, it is not normally possible to detect genetic variation using a single outbred stock or inbred strain. Hence the need to do multi-strain experiments. These can usually be done without increasing the total number of animals using a factorial experimental design.
Multi-strain experiments:
Can be used to find a sensitive or resistant strain for further study. Fewer animals will be needed for many experiments if the strain is more sensitive
They can be used as a first step in identifying gene loci controlling strain differences. Once differences are detected, the genes responsible can be mapped and identified.
Can be used in toxicological and pharmacological screening to minimise the chance that effects are missed due to genetic resistance of either outbred stocks or inbred strains.
Can be used to optimise routine pharmaceutical screening experiments. Use of more sensitive strains may mean that sample size can be reduced (see Shaw R, Festing MF, Peers I, Furlong L. 2002. Use of factorial designs to optimize animal experiments and reduce animal use. ILAR J 43:223-232.)
Multi-strain experiments do not need to use more animals
These experiments use a factorial experimental design in which treatment differences are assessed across several strains in a single experiment. The “group” is a sample of animals which are assigned to a treatment. They do not all need to be alike provided it is possible to match the genotypes of the treated and control groups.
This may be seen clearly in Example 1, which involves two treatments, eight isogenic strains but only sixteen animals. Group size is eight, but from eight different strains. This experiment is similar to a human study involving monozygous twins. The results will be analysed using a paired Student’s t-test.
Examples of multi-strain designs:
Principles: this explains how several strains can be used in one experiment without using lots of animals
Example 1 Is a comparison of two experiments involving 16 mice, using eight inbred strains or one outbred stock. The treatment means are exactly the same in the two experiments, but whereas the differences are statistically highly significant when using isogenic strains (using a paired t-test), the same differences are not statistically significant in the outbred stock where a two-sample t-test has to be used.
Example 2 shows the results of a real experiment to study the effect of the anti-oxidant BHA on the activity of a liver enzyme EROD, and to show whether the response was under genetic control. It involved sixteen mice of four isogenic strains, using a randomised block experimental design.
Example 3 Examples 1 and 2 were concerned with quantitative outcomes. This example explains why a multi-strain experiment is more powerful when the outcome is qualitative, such as presence/absence of a tumour.
Example 4 Here the haematological response to chloramphenicol in a multi-strain experiment involving four inbred strains is compared with with a similar experiment using a single outbred stock of mice. Both used the same total number of animals. The multi-strain experiment was substantially more powerful than the one using a single outbred stock.