Animal Models: Trouble with Translation

“The best laid schemes of mice and men often go awry,”
Robert Burns, To a Mouse
Image by Sabor7.
In a previous blog, we discussed the important factors you should consider when you pick an animal model for your disease study. Now, before you dive in and hope to create your own bench-to-bedside cure, there are some difficulties you should be aware of with translational models. These translational models are employed in animal systems to treat a disease or understand a mechanism. The goal, then, is to find something that will translate to humans and possibly cure the disease. One of the biggest challenges in inducing a disease or genetic manipulation in an animal is that this can bring out an unexpected phenotype or affect pathways that may not be relevant to humans or human diseases. Some people have stated mice are the healthiest animals in the world since we’ve cured just about every disease in them. But, it has been very difficult to get similar treatment results in humans. For example, there are dozens, possibly hundreds of cures for Type 1 Diabetes in mice, but none have been effective for humans.

Image from The Simpsons, Fox Television.
Failure with clinical drug trials is not uncommon, as noted by Janet Woodcock, deputy commissioner of the FDA, “Nearly half of all molecular entities that come into development fail.” So, why the high failure rate? There can be several factors that can make translational studies difficult. The first factor was already mentioned: the animal model. Studies that don’t make use of the proper animal or lack standardized methods can present biased or incorrect conclusions. Michael Festering, a former animal scientist at the UK Medical Research Council, stated that roughly half of the papers based on animal models contain errors. “Whether those are serious enough that the conclusions are invalid is debatable.”
Furthermore, many papers don’t report whether the tests were blind or randomized, creating the potential for bias. For example, imagine you are conducting a drug study on rats. When you reach to pull one out of the test cage, the lethargic rats are the easiest to catch, while the active ones scurry away. Whether you realize it or not, you’ve immediately created a bias. In this instance, a good experimental approach would be to randomly assign the rats into different study groups (control and experimental) ahead of time based on their tag ID numbers. Some labs will hide negative results or eliminate outliers from their data set, which is more commonplace than you might think.

This is why methodology for these animal studies should be standardized to avoid vague and irreproducible results. In fact, the ability to reproduce results is critical for confirming findings, which led to BioLegend’s support for the Reproducibility Initiative.

Comic by Mike Baldwin.

Image from GRIMM.
Picking the right animal can still be a difficult task. Ethically and financially, dogs, cats, and non-human primates may not be the easiest things to get a hold of. So, many labs are left choosing mice, which are easily manipulated in terms of breeding and genetic alterations. But, mice are evolutionarily farther from humans than monkeys, and mechanisms discovered in mice may not carry over to humans. In mice, amyotrophic lateral sclerosis can occur in SOD-1 knockouts or by TDP-43 protein-associated mechanisms. Research related to the SOD-1 knockout in humans has yielded no results, and new findings are now supporting the idea of the TDP-43 protein being more heavily involved1. Aside from the animal, the disease itself can make translational research difficult. For example, HIV’s high mutation rate, multiple subtypes, latency, and attack of immune cells has made an effective vaccine elusive for decades2.
Thus, it is important that you understand your model. And in some cases, you may want to consider whether an animal model is necessary. Alternatives, as mentioned in this paper, are becoming available so you can fulfill the 4 R’s of research (Replacement, Reduction, Refinement, and Responsibility) when it comes to using animals.
The ENCODE (Encyclopedia of DNA Elements) project was recently established by the NIH and Human Genome Research Institute. ENCODE is analyzing and comparing mouse and human genomes on a broad scale. This program will allow scientists to see whether DNA sequences are conserved between mice and humans, and also where they might diverge. With this understanding, researchers can better understand the limitations of their mouse model.
ENCODE can help you prepare.
It’s clear that translational studies have their own set of difficulties. But, as researchers, you can help to alleviate some of this by adhering to standardized practices, reporting all of your findings, and considering the appropriateness of your animal model. What else do you think you can do to help? Let us know at
Additional References:
  1. The trouble with animal models
  2. How close are we to an AIDS Vaccine?

Just when I thought I got the hang of science...
Contributed by Ken Lau, PhD.
Insert Note Here
Save Close Clear
Lab Timer
Login / Register
Remember me
Forgot your password? Reset password?
Create an Account