One particularly difficult set of clinical information for animal models to incorporate is that psychological treatments, such as Cognitive Behavioural Therapy (CBT), Interpersonal Therapy (IPT) and cognitive behaviouranalysis system of psychotherapy (CBASP) are effective treatments for depression (Hirschfeld et al., 2002; Elkin et al., 1989). The underlying mechanisms of action of such treatments are very difficult to explain in neurobiological terms and, perhaps, can never be tested in animal models. Of course, some advances have been made in attempting to understand how such treatment approaches may work by applying brain imaging technologies to intervention studies. CBT is able to affect the activation of key brain areas, and has been shown to modify the appearance of (presumably) dysfunctional neuronal activity in phobic anxiety (Paquette et al., 2003). However, both CBT and treatment with a selective serotonin inhibitor (citalopram) have been reported to elicit essentially indistinguishable changes in regional cerebral blood flow during a provocation test in a small cohort of patients with social anxiety (Furmark et al., 2002). How should this be interpreted? Does CBT share ‘pharmacological specificity’ with citalopram in this disorder? More probably, the changes described in both groups reflect functional deactivation of overactive ‘defence’ circuitry by different mechanisms. In depression, using the technique of Positron emission Tomography (PET) brain imaging, the early suggestions are that psychotherapy (Interpersonal Therapy) may not share therapeutic ‘mechanisms’ with venlafaxine (Martin et al., 2001), but that it may share them with paroxetine (Brody et al., 2001). Either these studies illustrate interesting potential differences in the therapeutic mechanisms by which different drugs and psychotherapy exert their effects, or they provide a perfect illustration of the inconsistency and variability of clinical neuroscience. In both studies, the main declared limitation was small sample size (15 ‘v’ 13, Martin et al.; 12 ‘v’ 12, Brody et al.).
How might modelling advance
Epidemiology suggests that the development of depressive disorders is dictated by a broad range of genetic and environmental influences that interact to confer vulnerability. In turn, psychosocial stressors can evoke depressive syndromes in such vulnerable individuals (Brown et al., 1986; Billings et al., 1983). However, the current conceptualisation of depression almost certainly includes a cluster of common clinical features that are generated by different pathophysiological processes. Hence, largely because of this antecedent complexity, the pathophysiology of depression remains elusive despite many decades of study. Many of the key hypotheses that can be generated from existing clinical information are simply untestable in humans. Experimental animals represent the best approach for some of these studies. Unfortunately, as outlined above, there are significant problems with the state of our knowledge from clinical studies. Psychiatric research has been plagued by small scale, short-term studies with weak methods, poor analyses and overinterpretation. As a result, we have failed to generate sufficiently reliable and accurate information about even common mental disorders, such as depression, to facilitate the required laboratory research. The following represent our suggestions for priorities for the advancement of valid animal models of depression:
1. Improved clinical science (i) We urgently require larger scale studies and better study designs. We must promote studies that will definitively address questions of treatment efficacy, predictors of response, and outcomes in different subtypes of depression. Such studies ought to include populations that are representative of those who actually present to health services for treatment. In many respects, and in light of the knowledge of psychiatric neuroscience that has been acquired, we should now set about trying to answer the same questions that were inadequately addressed many years ago. For example, in which ways do those patients with acute and chronic depressive episodes differ? (ii) At the same time, we also need further clinical research to look more closely at specific symptoms and specific symptom clusters within the syndrome of depression, in terms of antidepressant response. For example, what happens to the appetite and eating behaviour of depressed patients and how is it affected by antidepressant treatment? There are many beliefs around this topic, but we still do not know. This will not only advance our understanding of the neurobiology of certain aspects of depression, but also enable more accurate modelling of such symptoms in animals.
2. Basic laboratory science (i) Animal studies must focus on modelling very specific clinical features, mindful of any controversy around the presence, or significance, of the phenomenon under scrutiny. The laboratory rat or mouse will never exhibit depression under any circumstances. In other words, animal models must focus on construct validity and reliability. Face validity is probably an unrealistic goal. Predictive validity may be helpful, but we need to have better information on differential responses to treatment in humans. (ii) Models must pay greater attention to the basal state of targeted neural systems. Administration of antidepressant medication, or electroconvulsive stimulation, to non-depressed humans almost certainly does not elicit the same neural changes as when applied to someone with depression. There is evidence that antidepressants may have different effects on affective functions if the basal state is affected by drug withdrawal, for example (Harrison et al., 2001). Therefore, we should ensure that the basal state of laboratory animals undergoing trials of putative antidepressant agents closely mirrors what is known about the neural changes that occur in depressed humans.
3. Integration between clinical research and basic science (i) Clinicians and non-clinical neuroscientists need to become more involved in the collaborative development both of laboratory and clinical research programmes. Very few clinicians work on animal models of psychiatric disorder. Indeed, relatively few psychiatrists, clinical psychologists or mental health nurses have an interest in any form of research. The research base in psychiatry is shallow, available funding is a fraction of that devoted to other health areas, securing patient participation is difficult, and funding schemes that permit studies of the necessary longitudinal nature and time scale are virtually non-existent. Even fewer mental health professionals have the expertise, training or interest to pursue basic laboratory science and to assist in the development of valid models. Similarly, few basic scientists contribute to clinical research programmes. This polarisation of research activities was brought into sharp focus by a comment made to one of the authors (KM) by a former leading researcher in animal models of depression who confessed that, despite having written about the topic for over 15 years, he had never actually met or spoken to someone with depression.