Animal models in research

Research workers usually use laboratory animals as models of humans or some other target species. The research involves a long-term objective, such as developing a new drug for diabetics, screening a particular compound for human toxicity, studying a gene or mutation found in both animals and humans or studying a fundamental process such as gene transcription. The short-term objective is to use the animal model in experiments to determine how it responds to the treatments. If it is a faithful model of humans, then humans should respond in the same way. Animals, and other models, are used because the research can not be done on humans for practical or ethical reasons.

Using models is a two-step process:

  • A specific model is chosen because it is believed to be appropriate to the condition being investigated and is thought likely to respond in the same way as humans to the proposed treatment(s) for the character being investigated. This “belief” may be based on specific evidence, or it may be assumed from the biological similarity between animals and humans. It may, of course, be incorrect.
  • Having chosen the model it is essential that any experiments in which it is used are well designed, i.e. are capable of demonstrating a response to any imposed treatment. If the model happens to be insensitive or the experiments are badly designed so that they are incapable of distinguishing between treated and control groups, say as a result of using too few animals, then the model is not appropriate to its purpose.

There are five key features of models used in biomedical research:

  1. There can be substantial asymmetry in the number of similarities and differences between the model and the target. In theory, the model and the target only need to have a single feature in common, but there can be any number of differences. This means that useful models can sometimes be highly abstract, such as a mathematical equation or computer simulation. Moreover, the more fundamental the biological process, the more likely the animal model and humans will respond similarly.
  2. Some differences between the model and the target are necessary, otherwise the animal would not be a model. Differences are as important as similarities as they allow us to do things with the model which would not be possible with a human. Mice are widely used because they are small, prolific, and we can manipulate their genetics in ways not possible with humans. These differences from humans make them more, not less, valuable as models of humans for some applications. But characters like small size may make them unsuitable for other applications, such as heart surgery.
  3. Models are highly specific to a particular study.  Strains of mice and rats which develop cancer, heart disease, diabetes or neurological diseases could be of great interest in the study of these diseases, but these animals would probably be unsuitable for regulatory toxicology where long-lived strains are usually required. Thus, it is impossible to say whether “the rat” is a good or bad model of humans without specifying the context of the proposed study.
  4. Models need to be validated. Research using animal models usually aims to predict a response in humans.  When a new treatment for a particular disease or condition is developed in animals clinical trials will normally show whether or not the model was valid. If not, it may either be because the model was biologically invalid, or because the experiments using the model were badly designed.
  5. Models are subject to improvement through further research. A lot of animal research is aimed at understanding the animal as a potential model for particular human conditions. Models are not just found, they need to be developed, and this requires an understanding of the biology of the species and the effects of various interventions such as inactivating

    Hypothetical example

    Problem: is this new compound a carcinogen?

    1. Choose model systems which, according to current evidence, are the best predictors of human carcinogenesis., e.g. Ames test, cell mutation assay, heterozygous mouse P53+/- knockout, Rat.
    2. Before doing the tests, specify action to be taken according to outcome, e.g if the chemical is non-mutagenic in the Ames test and cell mutation assay and does not cause cancer in mouse P53+/- knockout and the rat we will assume that it is not a human carcinogen.
    3. Do the experiments. These need to be done well otherwise they may give the wrong answer.
    4. In doing the experiments it may be found that one or more are inappropriate, in which case the results would have to be discarded. Otherwise the results will be accepted, and appropriate action will be taken.

      Note that this strategy will only give the right answer if the models are good biological models of humans and the experiments have been well designed and correctly analysed.

      Note also that some people talk about “extrapolating” from animal experiments to humans. However, there is no such extrapolation. Models are used to test hypotheses judged to be relevant to deciding whether or not a compound is a carcinogen.