FELASA Module 9. Animal Models and Experimental Design

FELASA Module 9: Animal Models and Experimental Design
 When using an animal model you have to consider whether you have a model of
high fidelity or high discrimination
 In a model where young seagulls are responding to a model that looks like their
mother, they are more likely to respond to a sick with a red spot on it than a
realistic model
 The only clue the animals have that something is their mother is the contrast
between the spot and the white beak/stick and this contrast is higher on the
stick
 The lookalike model is a high fidelity model, whereas the stick is a high
discrimination model
 Most animal models are models of high discrimination
 US NIH defines a model as a living organism in which normative biology or
behaviour can be studied, or in which a spontaneous or induced pathological
process can be investigated, and in which the phenomenon in one or more
respects resembles he same phenomenon in humans or other species of animal
 This means we select one phenomenon in the animal which is comparable to
humans that we want to model
 In spontaneous models this phenomenon occurs spontaneously
 In induced models the phenomenon is induced chemically or surgically
 For example by causing an animal to have type 1 diabetes by removing the
pancreas
 A negative model is where the phenomenon never occurs
 For example if you study why a pig never activates coagulation factor VII
when fed a high fat diet
 An orphan model is where a veterinary patient develops a disease which looks
like a human disease
 E.g. when a cat develops obesity and type 2 diabetes
 Models monitor various parameters
 E.g. blood pressure, blood values and weight
 You need to determine the most important parameter (your primary readout)
 Other parameters are secondary readouts- these are good to use for
mechanistic explanations of changes in your primary readout
 Animal models are used to obtain information about disease and its prevention,
diagnosis and treatment
 They are essential for drug discovery and testing medicine
 Using animal models allows us to perform experiments that would be
impractical or prohibited in humans
 If drugs are tested in a poor model they may not have the same result in humans
 Therefore drugs should be tested in multiple high discrimination models for
different aspects of the disease
 Example of a high fidelity vs high discrimination model in human health
 Chimpanzees have a high fidelity to humans but infecting them with HIV isn’t
useful because they don’t develop signs of infection
 It is better to infect cats, which have a low fidelity to humans because they can
be infected with feline immunodeficiency virus disease (from the same family as
HIV) and they will develop similar symptoms
 External validity (how well a model translates to the species it is modelling) is split
into three criteria

,  Construct validity: The extent both human and the model phenomenon studies
can be explained by the same theory
 E.g. the cause of the disease is the same
 Face validity- the similarity of the appearance of the phenomenon between
animals and humans
 E.g. The symptoms of feline immunodeficiency virus are the same as HIV
 Predictive validity- a measure of how much a drug has the same effect in
humans and animal models
 Rats with their pancreas removed have low construct validity as a diabetes model
as humans don’t develop diabetes from a missing pancreas
 The model has low face validity because the rats will develop extra symptoms
 If you want to test the effect of insulin to alleviate diabetes it has high predictive
validity
 For validation of an animal model it is important to consider the cause and effect
relationship between an intervention and an outcome- this is the internal validity
 Internal validity is dependent on experimental design, randomization, proper
controls and the ability to reproduce research
 Reproducibility crisis- up to half of all studies can’t be replicated
 This includes cherry-picking data, publication bias for positive results and
insufficient reporting of experimental settings
 Using animal models allows us to standardize experimental environments which
decreases variance and increases power
 This reduces group size (reduction in the 3Rs
 High variance means high reproducibility but low power and also more animals
needed
 To reduce variation pathogen free mice have been bred
 These mice have little diversity in gut microbiota
 Gut microbiota can play an important role in rodents
 This means that the breeder the animals are purchased from can influence
results
 In an experiment with uncontrolled variation positive results in a heterogenous
animal are more reproducible and have less type I errors (i.e false positives)
 Negative results in a heterogenous group are more likely to be wrong so more
type II errors should be expected
 Controlling variation helps this- this can be done by performing experiments on
smaller groups of mice from multiple vendors
 The more complex a model is the higher the variability and frequency of
unexpected results
 Test tube experiments have complexities in molecular interactions
 Cell cultures experiments have complexities in how molecules interact inside
and between cells
 Ex vivo experiments have complexities in how cell populations interact in an
organ
 In vivo experiments have all the listed complexities in a diverse population
 There are two types of variability
 Technical variability
 Comes from inaccurate instruments/tools, inexperienced experimenters and
poor planning
 Reduced through better methods, instruments, planning, training and
experience
 Reducing technical variability improves the accuracy of an experiment