Among humans, some respond badly to stress; others don’t. We attribute that to differences in personality. Yet when we study animals, we expect them all to respond in the same way: to show the same biochemical and behavioural changes. In fact, of course, there is “mousality”. Animals vary in sensitivity and resilience just as we do.

As an example, when mice are exposed to the smell of a cat, they almost all show an acute anxiety-like response. In certain animals, but by no means all, the anxiety becomes chronic and looks a bit like post traumatic stress disorder. So if we want to model PTSD, it is animals with a persisting stress response which we should select.

Subgroups as a useful tool

To some researchers animal heterogeneity is an unwanted obstacle. But if we look at the different subgroups, instead of concentrating on mean data for the group as a whole, we have valuable opportunities to increase our understanding of mechanisms underlying sensitivity and resilience, and response or non-response to medication.

Work on a potential new antidepressant illustrates how animal variability gives us insight into dose-response. At a standard dose, some animals show a behavioural response; others don’t. Increasing the dose increases the overall effect. OK. But by looking at individual animals we can see that what the higher dose has done is to increase the number of responders. It has not increased the strength of the effect in those who were already responders at the lower dose.

To give another example, animals that are dominant in a food competition test respond less to antidepressants than animals that are submissive. Can we use that difference to investigate possible genetic or molecular mechanisms that might underlie this different response to medication?, Professor Einat asked.

This theme was taken up by Catherine Belzung of INSERM, Tours, France.

Dissecting non-response

We have known since the results of STAR*D that around a third of patients with MDD respond well to antidepressant therapy while around two-thirds have partial responses or do not respond at all. A large array of factors seems to have some value in predicting likelihood of response. Variation in level of stress hormones is one. Can this be modelled in mice?, Professor Belzung and colleagues wondered.

The answer is that it can. We do see variability in stress hormones in mice, and that variability does seem to influence behavioural responses to antidepressant medication. In an attempt to understand a possible underlying mechanism, the French group is now looking at Ki67, a marker of cell proliferation, in the hippocampus of their responding and non-responding animals.

It would be risky to say that animal non-response translates directly into human non-response. Face validity and predictive validity are not the same. But intra-group variability can be an opportunity to develop animal models of human treatment resistance and, potentially, to develop new therapeutic options.

In discussion, Professor Einat was asked about the role of the most obvious of individual differences: gender. Following the prevailing view that female mice were likely to be inherently more variable, because of hormonal fluctuations, he had for many years used only male animals in his studies. But he has been impressed by a recent review suggesting that gender in fact makes no real difference to variability of response. And he is now proposing to study female animals as well.