The characteristics of outbred stocks are not widely understood. If they were, these stocks would not be so widely used! Someone using an outbred stock generally knows nothing about the genetic characteristics of individual animals, what genes they carry or how heterozygous they are. Background data on characteristics will be unreliable because they can change rapidly. The increased variability means either that on average more animals will be needed or the experiment will be less able to detect treatment effects than if an isogenic strain had been used. There are not even any methods of quality control to ensure that the correct stock is being used. Some of these characteristics are discussed below:
The extent of the genetic variation depends on the previous history of the colony.
It can range from almost zero, when a closed colony has been maintained with small numbers of animals for many generations or has experienced some genetic bottleneck (such as being reduced to small numbers in one generation), to something like a wild colony where about 30% of loci may be heterozygous. The extent of the genetic variation in any given colony is unknown unless it has been specifically tested. Thus somebody who really needs a genetically heterogeneous stock is advised to test available stocks at a number of marker loci or synthesise one by crossing several inbred strains.
Each individual is genetically unique
In a stock segregating genetically at many loci each individual will be genetically unique, and the genotype of an individual at any given locus will be unknown unless that individual has been specifically genotyped. This means that an animal assigned to a control group will be different from one assigned to a treatment group. Any differences observed between them may be due to the effect of the treatment or it may be because they are genetically different. In order to take this into account the samples sizes need to be increased, or if not the experiment will be less powerful than if the animals had been genetically identical.
Another consequence is that anyone wanting to know, for example, the major histocompatibility type of an animal or whether it carries a gene for retinal degeneration will have to test that individual. In contrast, most inbred strains have already been typed at these two loci, with all individuals being genetically identical.
Finally, when daughter colonies are set up the breeding stock will only carry a sample of the genes present in the parental colony. The size of the breeding nucleus will determine whether it is a good sample. It is usually recommended that new colonies are set up with at least 25 breeding pairs.
Outbred stocks are phenotypically more variable than isogenic strains
However, two members of an outbred stock taken at random are likely to be much more similar than two individuals from different inbred strains. So if the aim is to test a treatment on a genetically heterogeneous population, the best strategy is to use several inbred strains in a factorial experimental design. A set of four or five strains should cover a wide range of phenotypes. Anyone wanting to find a really extreme phenotype will want to test as many strains as possible. This can be done using very few animals per strain. A preliminary screen might use only one animal of each strain in each treatment group. The factorial design has the major advantage that the genetic variation is under the control of the investigator and does not contribute to "noise" which can obscure treatment effects. Moreover, by using several strains the investigator will find out the extent to which the observed characters and the response are under genetic control. This is never clear when using an outbred stock.
The characteristics of the colony can change quite quickly
This can happen in several ways not found with inbred stains. Genetic contamination due to the mixing of two colonies is much more difficult to detect in an outbred stock than in an inbred strain. As many genetic loci are segregating it is necessary to monitor not only those loci which are genetically fixed (as with inbred strains), but also loci which are segregating. In this case the frequency of each gene (or more correctly, allele) at a sample of marker loci needs to be determined. This is likely to require large sample sizes.
Some level of inbreeding is unavoidable, although it will be negligible in large colonies. Where it does occur to any appreciable extent it will change gene frequency as alleles become fixed, and this will inevitably lead to genetic drift. This does not occur in inbred strains which are already fully inbred.
Selective breeding is totally ineffective in altering the characteristics of inbred strains, but can dramatically alter the characteristics of outbred stocks as shown by the many examples where this method has been used to develop new or improved animal models. For example, most outbred stocks weight more than inbred strains as a result of generations of the elimination of smaller animals born in the breeding colonies. Similar selection occurs with inbred strains, but as these are genetically fixed, there is no genetic change in them. Natural selection will also alter characteristics as the stock responds to any environmental changes which might occur. Thus, the presence of a sub-clinical infection will mean that resistant animals will tend to leave more offspring, thereby altering strain characteristics.
Both inbred strains and outbred stocks can change as a result of new mutations. However, in inbred strains they will either be eliminated as a result of further inbreeding (with a probability of 0.75) or fixed in the colony (probability 0.25). With outbred stocks new mutations may segregate within the colony for many generations before they are discovered.
Stocks with the same names from different breeders will be genetically different
Daughter colonies are inevitably slightly different from the parental colony, and once set up will start to change as a result of all the forces discussed above. So stocks with the same names from different breeders or even from different animal rooms will be genetically different. The extent of these differences will depend on all of the above factors which are likely to alter stock characteristics. One consequence is that published historical data on the characteristics of a particular stock may be entirely misleading.
Use of an outbred stock gives no assurance that it will be susceptible to the treatment
Scientists often reject isogenic strains on the grounds that a single strain may be resistant to the experimental treatment, such as a potentially toxic chemical. However, there are many examples where outbred stocks are resistant and inbred strains are susceptible. For example, Sprague-Dawley rats were totally resistant to diethylstilbestrol in a study in which ACI inbred rats were highly susceptible (Shellabarger et al J.Nat.Can.Inst. 61:1505, 1978), and the LD50 for TCDD is nearly 1000 time higher in Han:Wistar stock of rats than in other rat strains or stocks (Pohjanvirta et al 1988, Tox. Appl. Pharm. 29:131).
Thus, the use of an outbred stock does not solve the problem of resistance to the test chemical (or other treatment). The only way to overcome this problem is to use several strains in the factorial experimental design, as explained in the page on multi-strain experiments.
The genetic variation in an outbred colony of rats or mice is not comparable with that found in the human population
The human population is very large and partly sub-divided as a result of geographical and ethnic factors. It is subject to a wide range of environments which have shaped human evolution. Thus genes such as those causing sickle cell anaemia & the thalasemias have been maintained in some populations as a defence against malaria, but not in others where malaria is no problem.
In contrast most outbred stocks of mice and rats have been maintained for many generations as relatively small closed colonies in a protected environment, often with less than 100 breeding individuals. They have been selected for tameness, fast growth rate, breeding performance and ability to thrive in a laboratory environment. As a result they have lost some of the genetic variation and their phenotypes have drifted so that different stocks have different characteristics. Thus they are likely to be much less variable than human populations. Moreover, most experiments involve quite small sample sizes so that much of the genetic variation present in any colony is absent from the particular sample of animals which are used.
Moreover, in a genetically heterogeneous stock many genotypes can result in the same phenotype, and it is the phenotype which is important in research. Alleles often have a dominant/recessive mode of inheritance so that animals of genotype AA and Aa will have the same phenotype although they have different genotypes. Epistasis is also important. Many characters have a polygenic mode of inheritance in which the genotype at several loci determines the phenotype. Thus “plus” alleles (say for susceptibility) at some loci can be cancelled out by “minus” alleles at other loci, so that two animals may have very different genotypes, but be phenotypically similar.
Outbred stocks are a worse model of humans than are inbred strains of the same species because they lack relative sensitivity to experimental treatments
A good model will be like the thing being modelled (the target) with respect to the characters of interest, but must also be unlike the model in other respects so that it can be used to generate information which is not possible using the target. Thus a good toxicological model of humans will respond to toxic agents in a similar manner to humans, but it must be small, easy to handle and economical to maintain. It must also be ethically more acceptable to use it than to use humans. The availability of isogenic strains of mice and rats which are more sensitive (because they are more uniform) in their response to experimental treatments when properly used means that these provide a better, more powerful, model capable of detecting effects which may not be detected using outbred stocks. Thus, the suggestion that outbred animals are a better model of humans simply because humans are outbred is about as logical as the claim that because humans have no tail, so we should use animals with no tail for our research.