Source: Mouse Genome Informatics email list.
We recently found a treatment that protected B6 (inbred) mice against hypoxia. When we tested the same treatment with ICR (outbred) mice, however, we obtained largely scattered results. Some ICR mice were already relatively resistant to hypoxia without the treatment, while others were still vulnerable even with the treatment. With my limited knowledge in genetics, I understand that the ICR results are more or less expected because of diversity in gene functions in individual outbred mice (could anyone expand on this topic in terms of mouse genetics?). But don't we human beings resemble ICR more than B6? Does the thirst for "clinical relevance" diminish the significance of our finding in B6? Any advice would be appreciated.
I can reply to your question from the perspective of someone using animal models in drug development.
You've fallen into what has been a long-running debate in drug development circles. An individual investigator's preference likely depends on what point in the drug development process they are involved, from target validation to preclincal disease models to toxicology. The further downstream, the more interest there often is in outbred animals.
You are correct in saying that one of the perceived advantages of using outbred strains is that they supposedly more accurately mimic what one would find in humans. From the perspective of drug development, this would make it more likely that one could find evidence of genotype-dependent toxicity (a common reason for a drug to fail, so something one would want to find evidence of as soon as possible), as well as facilitate the identification of biomarkers that could prove useful or pharmacogenomics or to initiate a hunt for new targets through mapping QTLs. Incidentally, such heterogeneity in humans is likely explains at least in part why most drugs only work in ~30% of patients, which is why there is so much interest in identifying markers that can place patients into responding and non-responding groups.
The disadvantage of outbred strains, as you have discovered, it that one's results may be much more irreproducible. And depending on the strength of the effect, you may miss it entirely (of course, the same can happen if you choose the wrong inbred strain for your studies). Increased numbers of animal are often required to overcome this.
An additional factoid to consider is that it has generally been under-appreciated just how inbred most outbred strains actually are. I recently heard Howard Jacob (Medical College of Wisconsin) give a talk in which he showed that in Sprague Dawley rats ~one-third of the genome was fully inbred, and the average number of alleles for most loci was only 2(whereas in a panel of 48 different inbred strains, the total available diversity averaged around 6 alleles/locus). Based on this, he has come up with a combinatorial intercrossing strategy between 4 strains that results in greater diversity.
As someone who works in target validation (using animal models to show that a particular drug target is likely to affect a disease of interest, so as to justify further downstream development), it is my bias that inbred models are the best way to go. The mouse is clearly a reasonable, but far from perfect model of human disease. As they say, it is easy to cure a mouse of cancer, but a lot of those therapies don't end up working in humans. Given this, I'd much rather be able to rapidly obtain reproducible results in the mouse, and then go on and do the real experiment in humans. Focusing on genetic diversity in a mouse model of disease may be putting the cart before the horse in many cases.
Bottom line: if we had your results, we would ignore the outbred model and use the positive results in the inbred model to justify further drug development work. I'm certain most companies would do the same.
-Kevin P. Foley (Millennium Pharmaceuticals, Inc.)
Banbury Conference on Genetic Background in Mice (1997). Mutant mice and neuroscience: recommendations concerning genetic background. Neuron 19, 755-759.