Introduction
The definition of hyperkinetic disorder according to ICD-10 is based upon the simultaneous presence of three main behavioural problems, i.e. attention deficit, overactivity and impulsiveness. They need to be present in more than one situation and to cause impairment in functioning. The problems need also to have been present before the age of 7 years. The DSM-IV category is called attention deficit/hyperactivity disorder (ADHD). The criteria are based upon the same list of behaviours as those that characterize the ICD-10 definition of hyperkinetic disorder.
The 4th edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) describes three subtypes of ADHD, i.e. (1) the predominantly inattentive type, (2) the predominantly hyperactive/impulsive type and (3) the combined type with symptoms of inattention, impulsivity and hyperactivity (American Psychiatric Association
1994). ADHD represents the extremes of normal behaviour in the domains of attention and activity, which makes a clear diagnosis difficult. In addition, several comorbid disorders can be found in children and adolescents with ADHD, including oppositional defiant disorder and conduct disorder (50%), anxiety disorders (25–35%), mood disorders (15%) and learning disabilities (Biederman et al.
1991). In view of the high prevalence of comorbid disorders, clinical, neuropsychological and neuroimaging studies of children and adolescents with ADHD will consist of relatively heterogeneous patient groups. It is therefore important to describe the core disorder of ADHD, e.g. by identifying biological markers, which could improve the diagnosis. This may also help to develop new treatment strategies.
The current pharmacotherapy with psychostimulants goes back to 1937 when Bradley discovered that amphetamines ameliorate disruptive behaviour in children (Bradley
1937). All drugs that are found to be therapeutically effective in ADHD affect central catecholaminergic neurotransmission, namely the dopaminergic and noradrenergic systems.
These findings suggest that dysfunctions of catecholaminergic neurotransmitter systems contribute to the symptoms of ADHD. In addition, patients with frontal brain lesions show some behavioural similarities with ADHD patients (Benton
1991; Heilman et al.
1991; Levin
1938; Mattes
1980) and ADHD has been shown to be highly heritable (Bobb et al.
2005; Fisher et al.
2002; Teicher et al.
2000). These findings suggest that ADHD is based on some specific neurobiological dysfunctions.
Several animal models of ADHD have been suggested and discussed (for review see also Kostrzewa et al.
2008; Russell et al.
2005; Sagvolden et al.
2005; van der Kooij and Glennon
2007). The quality of these animal models depends on their ability to mimic the symptoms and to reflect the neurobiology of ADHD. Most models are solely based on similarities in symptoms. Since our knowledge concerning the neurobiological alterations in ADHD remains sketchy, it is too early to propose valid animal models of ADHD. The present review will come to the conclusion that none of the currently discussed models fulfil all necessary validation criteria.
Dysfunctional systems in ADHD
Mefford and Potter (
1989) postulated a noradrenergic dysfunction of the locus coeruleus (LC) as one of the earliest models of ADHD. This model was supported by findings in monkeys, which showed that the LC is involved in selective processing of sensory stimuli (Aston-Jones et al.
1997), which is partly modulated by alpha
2-autoreceptors (Simson and Weiss
1987). An increase in noradrenaline (NA) suppresses basal firing and enhances responses to stimuli, i.e. an increase in NA leads to more focused behaviour, while a reduction in NA would increase the response to irrelevant stimuli. In addition, adrenaline is known to inhibit the tonic activity of the LC. A deficit in one of the two systems might therefore disrupt stimulus-evoked responding, and this could induce deficits in sustained attention.
In contrast to Mefford and Potter (
1989), Pliszka et al. (
1996) suggested a dysfunction in two neurotransmitter systems. Studies in humans have shown that attention is distributed in a posterior and anterior system (Posner and Petersen
1990). The posterior system includes the superior parietal cortex, the superior colliculus and the pulvinar nucleus. This system receives a dense innervation from the LC (Holets
1990). NA enhances the signal-to-noise ratio and primes, according to Pliszka et al. (
1996), the posterior system to orientate to novel stimuli. Attention then shifts to the anterior system, which is known to control executive functions. It consists of the PFC and the anterior cingulate gyrus. The sensitivity of this system is modulated by DA from the ventral tegmental area (VTA). According to Pliszka et al. (
1996), a noradrenergic dysfunction could inhibit the priming of the posterior system and lead to attention deficits. A loss of DA could induce deficits in the anterior system and impair executive functions.
A third model suggested by Arnsten et al. (
1996) is based on a dysfunction of the alpha
2-autoreceptors in the prefrontal cortex (PFC). The PFC inhibits the processing of irrelevant sensory stimuli through connections with the association cortex (Cavada and Goldman-Rakic
1989) and therefore protects on-going tasks from interference (Alexander et al.
1976; Knight et al.
1989). This function is regulated by the LC, since ascending noradrenergic fibres stimulate postsynaptic alpha
2-adrenoreceptors on the pyramidal cells in the PFC (Aoki et al.
1994) leading to a reduction in spontaneous firing (Hasselmo et al.
1997). Therefore, the activity of the LC primes the PFC to suppress task-irrelevant stimuli and inhibits behaviour. According to Arnsten et al. (
1996), a reduced NA activity causes a partial denervation of the alpha
2 receptors in the PFC, thereby disrupting the inhibitory control of children with ADHD. Based on this model, the central deficit in ADHD is a lack of inhibition induced by a decrease in brain NA.
These models differ but they also have certain points in common, e.g. the central role of the PFC and catecholamine neurotransmitters. The models by Mefford and Potter (
1989), Pliszka et al. (
1996) and Arnsten et al. (
1996) emphasize the role of NA in focusing on relevant stimuli or tasks. Pliszka et al. (
1996) and Arnsten et al. (
1996) suggest that a reduced noradrenergic activity contributes to attention deficits and distractibility. However, these models need to be tested against findings in patients with ADHD.
Animal models in research
Animal models of diseases are supposed to show phenomenological similarities with the modelled disease. In animal models of ADHD, one would expect the three core symptoms of this disorder to be present, i.e. attention deficits, hyperactivity and impulsivity (Rhee et al.
1999). These symptom similarities represent the face validity of the model (Willner
1991). However, as Willner (
1991) has pointed out, face validity also includes a resemblance regarding aetiology, treatment and the physiological basis of the modelled disease. Most of these aspects cannot be used for validation since they are currently objects of research. Face validity is therefore frequently reduced to symptom similarities. Validity based on symptom similarities alone might be misleading, since not every hyperactive rat is a valid model of ADHD. There may be several alternative reasons why a certain behaviour is observed. The presence of a certain disease symptom does not necessarily reflect the presence of the entire disease. Furthermore, similar behavioural expression does not necessarily indicate that this expression has the same biological substrate. This indicates that models based on symptom similarities alone are weak and that other criteria are needed for the validation of an animal model. Willner (
1991) has suggested to check for aspects of construct validity and predictive validity.
Construct validity means that the model conforms to an established or hypothesized pathophysiological basis of the disorder. A disturbance within the fronto-striato-cerebellar system has been postulated in ADHD. An animal showing hyperactivity because of alterations in this system has both construct and face validity. Construct validity is more important than face validity because it is a certain theoretical framework that connects the behavioural symptoms with the modelled disease.
Another criterion used in validating an animal model is predictive validity, which is the ability to predict previously unknown aspects of the genetics, neurobiology and pathophysiology of a disorder or to provide potential new treatments. In practice, drugs with similar effects in human disease and animal model are often used to validate the model.
In summary, the validity of an animal model should not solely be based on behavioural similarities. Both construct and predictive validity have also to be considered. Construct validity depends on the knowledge about the human neurobiology of the modelled disease. Since this knowledge is often limited, construct validity is relatively weak.
Discussion
Although ADHD is a common disorder among children and adolescents, little is known about its neurobiological basis. It has been suggested that disturbances within the fronto-striatal system and altered levels of the neurotransmitters DA and NA are involved in the pathophysiology of ADHD. This is based on several indications. For example, patients with prefrontal lesions show behavioural similarities with ADHD patients (Benton
1991; Heilman et al.
1991; Levin
1938; Mattes
1980) and the right PFC volume is reduced in children with ADHD (Castellanos et al.
1996b; Filipek et al.
1997; Hynd et al.
1990). Both basal ganglia and frontal lobe volumes correlate with impaired attention and inhibition (Casey et al.
1997; Semrud-Clikeman et al.
2000). Finally, methylphenidate increases the reduced blood flow in prefrontal regions of individuals with ADHD (Langleben et al.
2002; Lou et al.
1984,
1989). DA and NA are important neurotransmitters in these brain regions, and a dysfunction of these neurotransmitters appears to be likely. This is also underlined by the fact that treatment with psychostimulants reduces ADHD symptoms.
Different pathophysiological mechanisms have been suggested on the basis of altered dopaminergic and noradrenergic neurotransmission. For example, Arnsten et al. (
1996) have suggested a sole reduction in noradrenergic function while Pliszka et al. (
1996) postulated a combination of dopaminergic hypofunctioning and noradrenergic dysfunctioning as the basis of the core symptoms of ADHD. Studies investigating neurotransmitter levels in patients revealed conflicting results. Some studies found indications of an altered activity in catecholaminergic metabolites (Oades et al.
1998; Shaywitz et al.
1977; Shekim et al.
1977,
1979,
1983,
1987), while others found no differences (Rapoport et al.
1978; Shetty and Chase
1976; Wender et al.
1971). However, it has to be considered that the patients investigated in these studies differed with regard to comorbidity, medication and other relevant factors. Catecholamine metabolites may not reflect the neurochemical status of patients with ADHD since neither plasma nor urinary levels of HVA and MHPG correlate with hyperactivity or predict the response to stimulant treatment (Castellanos et al.
1994,
1996a).
Given the high heritability of ADHD (Gilger et al.
1992; Rhee et al.
1999), the investigation of genes involved in catecholamine functioning is another research strategy. Recent studies have suggested that both the DAT gene and the D
4 receptor gene are associated with ADHD (Bobb et al.
2005; DiMaio et al.
2003). However, several studies were not able to establish any association (Bakker et al.
2005; Frank et al.
2004; Langley et al.
2004; Mill et al.
2004).
Although the available data clearly indicate that dopaminergic, noradrenergic and probably serotonergic activities within the fronto-striatal system play an important role in ADHD, there is no prevailing concept of the neurobiology of ADHD. This might reflect the heterogeneous nature of ADHD, and it may not be reasonable to expect a unique biological profile in ADHD.
Further knowledge about the neurobiology of ADHD may be provided by animal models. However, these models will provide reasonable conclusions only if certain validation criteria are fulfilled.
Several animal models of ADHD have been proposed (for review see also Kostrzewa et al.
2008; Russell et al.
2005; Sagvolden et al.
2005; van der Kooij and Glennon
2007), and most of these models were initially based on the presence of hyperactivity. However, face validity of an animal model of ADHD should also include impulsive behaviour and attention deficits. In addition, deficits in learning or executive functions might also be indicators of face validity. Investigations concerning impulsivity and attention deficits are still missing for some of the ADHD models including the acallosal mouse, the neonatal 6-OHDA lesion model, the neonatal hypoxia model, the NHE rat, the DAT-KO mouse and the developmental cerebellar stunting model. Further research is therefore needed in order to validate these models regarding face validity. Most of these models have predictive validity since treatment with psychostimulants reduces hyperactivity. It has sometimes been argued that good predictive validity is given only if both amphetamine and methylphenidate are effective in these models. However, the fact that there are responders and non-responders to methylphenidate among patients with ADHD suggests different types of pathophysiology in ADHD. Differential response to amphetamine and methylphenidate in animal models might therefore reflect different pathophysiological mechanisms. Data concerning predictive validity are not available for the acallosal mouse and the NHE rat, while the treatment of the developmental cerebellar stunting rat with amphetamine leads to an increase in hyperactivity.
With regard to construct validity, alterations in dopaminergic or noradrenergic activities have been reported for all models except the acallosal mouse and developmental cerebellar stunting. The validity of these two models of ADHD is therefore limited. However, the developmental cerebellar stunting rat might have some potential as an animal model, since human studies have suggested a role of the cerebellum in ADHD. Not all models fulfil therefore the criteria necessary. The SHR is the best studied animal model with regard to validity. However, the hypertension in this rat and the use of the WKY rat as control in most studies put in question the use of SHRs as an animal model of ADHD.
Even if all criteria are fulfilled, the models show differences. For example, the SHR shows an impaired DA release, and both neonatal 6-OHDA-lesion rat and coloboma mouse have a decreased DA transmission while the DAT-KO mouse shows an increased DA transmission. Nevertheless, all these animals present with symptoms of ADHD, namely hyperactivity. Both increased and decreased dopaminergic activity can therefore lead to ADHD-like symptoms. This suggests that a dysbalance between presynaptic and postsynaptic activities might be important. There is a similar problem with noradrenergic activity in these models. Pliszka et al. (
1996) and Arnsten et al. (
1996) have postulated a decreased noradrenergic function in ADHD. Depletion of NA in neonatal rats by administering 6-OHDA in combination with a selective DAT-inhibitor (Teicher et al.
1986) has been shown to induce motor hyperactivity (Raskin et al.
1983), learning deficits (Roberts et al.
1976) and attention deficits (Carli et al.
1983). The main source of central NA is the LC, which innervates the entire cerebral cortex, various subcortical areas, cerebellum and spinal cord. The LC has been found to play an important role in attention, arousal, orientation and vigilance (Solanto
1998) since its neurons selectively respond to target stimuli. Tonic LC activity corresponds with the arousal state, and both very low and very high LC activities are associated with impaired vigilance (Arnsten
1997; Aston-Jones et al.
1994). However, the above-mentioned animal models show either unaltered noradrenergic functioning or an increase in NA functions, while none of the models show a decrease in noradrenergic activity.
Based on the SHR model, one might conclude that increased noradrenergic activity and decreased dopaminergic activity represent the characteristic dysbalance of catecholamines in ADHD. However, there are indications that the opposite might also be true, i.e. both increased and decreased dopaminergic activity can lead to ADHD-like symptoms. The same appears to hold true for noradrenergic activity. Furthermore, the increase in noradrenergic activity in the SHR is closely connected to hypertension, which is one of the most confounding factors in this animal model.
The question therefore arises, which model best represents the nature of ADHD. So far, studies with patients have only shown that the structural alterations in the fronto-striatal-cerebellar system, functional alterations in catecholaminergic systems and genes coding for the DAT and the D4-receptor are associated with ADHD. In regard to construct validity in ADHD animal models, this means that every animal with alterations in these systems has some construct validity for ADHD. Therefore, it is the combination of face, predictive and construct validity that makes an animal model more or less valid. This illustrates the basic problem in validating animal models: the more is known about the biology of a disease the more conclusive is the comparison between animal model and modelled disease. However, it is the lack of information that makes it necessary to develop a model in order to learn more about the biology of the modelled disease.