The cerebellum in dystonia – Help or hindrance?

Abstract

Dystonia has historically been considered a disorder of the basal ganglia. This review aims to critically examine the evidence for a role of the cerebellum in the pathophysiology of dystonia. We compare and attempt to link the information available from both clinical and experimental studies; work detailing cerebellar connectivity in primates; data that suggests a role for the cerebellum in the genesis of dystonia in murine models; clinical observation in humans with structural lesions and heredodegenerative disorders of the cerebellum; and imaging studies of patients with dystonia. The typical electrophysiological findings in dystonia are the converse to those found in cerebellar lesions. However, certain subtypes of dystonia mirror cerebellar patterns of increased cortical inhibition. Furthermore, altered cerebellar function can be demonstrated in adult onset focal dystonia with impaired cerebellar inhibition of motor cortex and abnormal eyeblink classical conditioning. We propose that abnormal, likely compensatory activity of the cerebellum is an important factor within pathophysiological models of dystonia. Work in this exciting area has only just begun but it is likely that the cerebellum will have a key place within future models of dystonia.

Introduction

Dystonia has long been considered to be a manifestation of basal ganglia dysfunction, similar to other movement disorders. However, there is accumulating evidence from a wide variety of sources that the cerebellum may have a role to play in the pathophysiology of dystonia. Here we review this evidence, and demonstrate how the intimate structural and functional connections between cerebellum and basal ganglia appear to be involved in patients with dystonia.

The cerebellum and the basal ganglia receive input from multiple cortical areas and have been traditionally been thought to modulate motor control via distinct thalamic nuclei that project to the primary motor cortex (Alexander et al., 1986). However studies using viral tracers in primates reveal the macro-architecture of an increasing number of cerebellar and basal ganglia projections. Multisynaptic circuits link the cerebellum and basal ganglia with the primary motor cortex, supplementary motor area (SMA), pre-SMA, oculomotor, prefrontal, and posterior parietal cortex (Lynch et al., 1994, Middleton and Strick, 2001, Akkal et al., 2007, Prevosto et al., 2010). Many cortical areas project topographically to specific cerebellar and basal ganglia territories that reciprocally innervate these same cortical areas (Bostan and Strick, 2010). There is also a substantial direct communication between the basal ganglia and the cerebellum: a disynaptic projection linking the dentate nucleus (output stage of cerebellar processing) to the striatum (input stage of basal ganglia processing) (Hoshi et al., 2005), and a forward connection from the subthalamic nucleus of the basal ganglia to the cerebellar cortex (Bostan et al., 2010). The reciprocal communication between these two major subcortical structures suggests that they directly modulate each other. Brainstem nuclei provide another junction for the cerebellum and basal ganglia to interact, for example in cats the red nucleus receives input from both the basal ganglia and cerebellar nuclei which then project directly to motor nuclei (Pong et al., 2008). These neuronal circuits provide an anatomical substrate for the cerebellum and basal ganglia to have wide ranging functions in motor and non-motor domains. Dysfunction in either structure could induce either compensatory activity or disruption in the other.

To date, there is a paucity of information regarding the neuropathology of dystonia, and there has not been specific exploration of cerebellar pathology or cerebellar-basal ganglia projections in brains of dystonia patients. Reported autopsy studies in sporadic primary dystonia and DYT 1 dystonia have not shown an overt neurodegenerative process or clear patterns of cell loss (Holton et al., 2008, Standaert, 2011).

Pharmacological and mutant mouse models of dystonia provide further data supporting a role of the cerebellum in the genesis of dystonia (Neychev et al., 2008, Ledoux, 2011). For example, tottering mice mutants exhibit paroxysmal dystonia due to a point mutation in a gene that codes for a calcium channel (Wakamori et al., 1998). Clinically and electrophysiologically these episodes have characteristics similar to human dystonia (Jinnah et al., 2005). Surgical removal of the cerebellum abolishes dystonic attacks in these mice (Neychev et al., 2008). Elimination of dystonic movements following cerebellectomy has also been found in other murine models of dystonia (LeDoux et al., 1993, Devanagondi et al., 2007). Similarly, dystonia is abolished if the tottering mouse is bred with an additional genetic mutation that causes Purkinje cell degeneration (Campbell et al., 1999). In a pharmacological mouse model for dystonia, microinjection of low doses of kainic acid into the cerebellar vermis of mice generates dystonia of a severity proportional to kainite dose (Pizoli et al., 2002). Microdialysis of the striatum reveals dystonic attacks to be associated with reductions in striatal dopamine in both tottering mice and the kainic acid pharmacological model, which suggests that that cerebellar activity can directly influence the dynamics of striatal dopamine (Neychev et al., 2008). Recently, a pharmacological model of rapid onset dystonia parkinsonism (DYT12) has been created (Calderon et al., 2011) by selective blockade of the sodium–potassium ATPase pump (which is mutated in the disorder). Both cerebellar and basal ganglia blockade of the sodium pump were needed to cause dystonic symptoms, and lesioning of cerebellar output nuclei or the disynaptic cerebellar-basal ganglia link caused significant resolution of symptoms (Calderon et al., 2011).

There are a substantial number of case reports linking dystonia to structural lesions of the cerebellum. However, there is considerable heterogeneity amongst cases with variable lesion location, aetiology and extent, type of dystonia produced, time interval between initial insult to onset of dystonia, and quality of clinical data. It has long been recognised in both adult and paediatric neurology that posterior fossa tumours can present with cervical dystonia (Grey, 1916, Extremera et al., 2008). A review of 25 cases of secondary cervical dystonia with a range of aetiologies in adults revealed that structural lesions of the brainstem and cerebellum were the most frequent cause of cervical dystonia (44%), with basal ganglia lesions accounting for less (24%) of cases (LeDoux and Brady, 2003). In two cases of cerebellopontine angle tumours, the cervical dystonia improved following successful removal of the tumour (Krauss et al., 1997). Focal limb dystonia has also been associated with cerebellar lesions. In an intriguing case, successful treatment of an isolated tuberculoma of the left cerebellar hemisphere led to parallel resolution of left arm dystonia (Alarcon et al., 2001). Other cases document the emergence of late-onset oromandibular dystonia after bilateral cerebellar infarction, blepharospasm/torticollis after bilateral cerebellar infarction, and left hemidystonia following ipsilateral vertebral artery occlusion (Rumbach et al., 1995, O’Rourke et al., 2006, Waln and LeDoux, 2010).

Patients with genetic degenerative cerebellar disorders (for example spinocerebellar ataxia type 3, SCA3) commonly demonstrate dystonia as part of their clinical phenotype, and sometimes dystonia may be the predominant presentation (Münchau et al., 1999). The neurodegeneration in such patients is widespread however, and the dystonia is usually assumed to be the result of basal ganglia rather than cerebellar degeneration. Pathological studies have indeed confirmed involvement of other motor system structures such as the pallidum and substantia nigra (Van Gaalen et al., 2011). We have reported two separate series of patients with a syndrome of cervical dystonia and mild cerebellar ataxia (DYTCA) of undetermined etiology (Kuoppamaki et al., 2003, van de Warrenburg et al., 2007). Dystonia is the more prominent and disabling symptom in this disorder, with the cerebellar ataxia being relatively mild and slowly progressive. Imaging findings vary between patients from cerebellar and brainstem atrophy to normality. Twelve further patients have been described in a separate series; although in these patients marked cerebellar atrophy (albeit with mild cerebellar signs) was the norm (LeBer et al., 2006). We have speculated whether the cerebellar pathology in our DYTCA patients contributes to (or perhaps is even wholly responsible for) the development of their dystonia. The electrophysiological studies that lend some support for this hypothesis are detailed below.

It is worth noting that in primary dystonia there is an absence of clear cerebellar signs on clinical examination, even when there is neurophysiological evidence to support cerebellar dysfunction (Teo et al., 2009). This might be taken to support a more compensatory role for the cerebellum in some forms of primary dystonia, and at the very least tells us that the role of the cerebellum in dystonia is more complicated than simply a loss or gain of cerebellar function.

In vivo functional and structural imaging studies in dystonia can broadly be divided into studies of (i) grey and white matter structure and integrity (ii) neurotransmitters and (iii) brain metabolism at rest and during learning and motor tasks. The hereditary dystonias are particularly interesting as the incomplete penetrance of clinical manifestation in patients with mutations of DYT 1 or DYT 6 allow one to make distinctions between patterns of abnormality related to genotype and phenotype.

Voxel-based morphometry applied to high resolution MRI has demonstrated subtle changes in grey matter, with increases in putamal, internal globus pallidus and prefrontal cortex as a common pattern across different types of primary dystonia (Egger et al., 2007). Both increases and decreases of cerebellar grey matter volume have been found with this technique in different types of dystonia and thus further studies are required to elucidate the significance of these observed changes (Draganski et al., 2006, Delmaire et al., 2007, Obermann et al., 2007). Diffusion tensor imaging (DTI-MRI) can be used to assess microstructural white matter integrity with fractional anisotropy (FA) as a measure of axonal coherence (Sen and Basser, 2005). Axonal integrity is reduced in the subgyral white matter of the sensorimotor area in both manifesting and non-manifesting DYT1 carriers, with additional FA reductions in the dorsal pons at its juncture with the superior cerebellar peduncle in manifesting subjects (Carbon et al., 2004, Carbon et al., 2008b) Additionally, DTI-MRI combined with probabilistic tractography techniques in DYT1 and DYT6 have demonstrated reduced connectivity of the cerebello-thalamic pathway near the dentate nucleus (Argyelan et al., 2009). This was most pronounced in clinically affected mutation carriers compared with clinically unaffected mutation carriers. DTI-MRI in non-hereditary primary dystonias has also demonstrated white matter integrity abnormalities but as yet cerebellar connectivity has not been specifically studied (Colosimo et al., 2005, Bonilha et al., 2007, Fabbrini et al., 2008, Delmaire et al., 2009).

As dystonia has traditionally been conceptualised as a basal ganglia disorder, abnormalities of dopaminergic neurotransmission have been investigated using radioligand binding. Decreased striatal D2 radioligand uptake has been demonstrated in various forms of primary dystonia including hand, cervical and cranial dystonia and patients with DYT1, DYT6 and DYT11 mutations regardless of clincial manifestation (Perlmutter et al., 1997, Naumann et al., 1998, Beukers et al., 2009, Carbon et al., 2009). However, a recent study suggests that D3 rather D2 receptor affinity is reduced in focal dystonia, thus further work with increasingly specific radioligands is needed to investigate whether this pattern is seen across other types of primary dystonia (Karimi et al., 2011). This work provides an interesting link with work presented above in animal models that cerebellar activity may directly modulate striatal dopamine (see above).

Alterations in regional brain function at rest can be measured using positron emission tomography (PET) with selective radioligands. Patients with sporadic and genetic forms of dystonia demonstrate relative increases in regional metabolic activity in the posterior putamen/globus pallidus, supplementary motor area (SMA) and cerebellum (Eidelberg et al., 1995, Eidelberg et al., 1998, Niethammer et al., 2011). Elevated network activity persists during sleep in manifesting DYT1 carriers and is also present in non-manifesting DYT1 carriers (Eidelberg et al., 1998). Contrasting findings with regard to cerebellar metabolism have been found in DYT6 carriers, but all clinically affected DYT1 and DYT6 patients show relative metabolic increases in the pre-SMA and parietal association regions (Carbon et al., 2010).

Sequence learning is a task that requires cerebellar processing (Molinari et al., 2008). Sequence learning ability and task-related brain activation is abnormal in non-manifesting carriers of the DTY1 deletion (Ghilardi et al., 2003). Sequence learning in conjuction with an equiperformance study design in this patient group resulted in overactivation of the lateral cerebellum, perhaps as a compensation for lack of recruitment of pre-frontal regions in order to achieve normal motor performance (Carbon et al., 2008a).

A normal motor-related activation pattern (NMRP) has been proposed by combining PET activation data with multivariate network modelling in control subjects (Niethammer et al., 2011). The NMRP is characterised by contributions from the cortico-striato-pallidal-thalamocortical and cerebello-thalamo-cortical motor circuits. Groups of dystonic patients that have been studied to date (sporadic cervical dystonia and manifesting DYT1 and DYT6) demonstrate increased activity of the NMRP (Niethammer et al., 2011). Furthermore, the increased activity of the NMRP correlated with severity of dystonia and also microstructural changes observed by fractional anisotropy in cerebellar outflow as described above (Niethammer et al., 2011).

A large number of fMRI-BOLD studies have been conducted in patients with idiopathic primary dystonia and cerebellar abnormalities have repeatedly been described. Patients with musician’s dystonia and focal hand dystonia show abnormal cerebellar activation during different tapping tasks (Kadota et al., 2010, Wu et al., 2010) and abnormal cerebellar activation is observed during writing in writer’s cramp (Preibisch et al., 2001, Hu et al., 2006). Task-related activity in the cerebellar nuclei, posterior vermis, right paramedian cerebellar hemisphere and dorsal pons was however inversely related with the severity of hand dystonia and proposed to reflect secondary compensatory reorganization. Abnormal cerebellar activation in spasmodic dysphonia (Simoyan and Ludlow, 2010) during voice production and essential blepharospasm during eyeblinking (Baker et al., 2003) is also reported. Patients with cervical dystonia show BOLD signal increase in a number of brain regions including the cerebellum during passive movement (Obermann et al., 2010).

Electrophysiological studies in dystonia have revealed a distorted balance between the excitatory and inhibitory circuitry of the sensori-motor system at various levels and there are abnormal responses to protocols inducing plasticity-like effects (Deuschl et al., 1992, Chen et al., 1995, Edwards et al., 2003, Quartarone and Pisani, 2011). Investigation into the role of the cerebellum in these observed changes in dystonia is at an early stage. Firstly we compare the contrasting neurophysiological profiles seen in dystonia and cerebellar disorders. We then discuss eyeblink classical conditioning (EBCC), a paradigm that is heavily cerebellar dependent that is abnormal in dystonia. Finally we summarise recent work examining cerebellar inhibition in dystonia and areas for future investigation.

Studies investigating cortical excitability profile in primary dystonia using transcranial magnetic stimulation (TMS) have reported reduced cortical inhibition, observable as a relatively greater increase in motor evoked potentials (MEPs) with increasing stimulus intensities (Ikoma et al., 1996), shortening of the cortical silent period (CSP) (Chen et al., 1997) and reductions in short intracortical inhibition (SICI) (Ridding et al., 1995, Chen et al., 1997, Siebner et al., 1999). A more limited number of studies have assessed the cortical excitability profile of patients with cerebellar lesions. The balance between cortical excitatory and inhibitory circuitry appears disturbed, but the shift is opposite to that seen in primary dystonia, with increases in motor cortical thresholds (Schwenkreis et al., 2002), abnormal prolongation of the CSP (Wessel et al., 1996, Oechsner and Zangemeister, 1999, Tamburin et al., 2004), reduced intracortical facilitation (ICF) (Liepert et al., 1998, Liepert et al., 2004) and an increase in SICI (Liepert et al., 2004). We have compared measures of cortical excitability in five patients with DYTCA and found SICI to be increased (Talelli et al., 2011), in contrast to typical primary dystonia, but similar to patients with cerebellar lesions (Liepert et al., 2004). Of interest, in myoclonus dystonia (DYT 11) the expected decrease in SICI seen in primary dystonia is absent (Meunier et al., 2008). Thus, cortical excitability profiles in dystonia are different with certain types (DYTCA, myoclonus dystonia) having patterns that more closely resemble patients with cerebellar disorders rather than typical primary dystonia. More work is clearly needed in this area, including testing response to plasticity protocols in these disorders to see if they are different from typical primary dystonia. Little is known about the role of the cerebellum in response to plasticity protocols, but preliminary work suggests that in cerebellar degeneration electrophysiological responses to plasticity inducing paradigms are normal (Teo et al., 2008).

Perhaps the most compelling electrophysilogical evidence for cerebellar involvement in dystonia is seen when studying EBCC. This is a paradigm of associative motor learning in which paired presentation of a conditioned (CS) and unconditioned stimulus (US) leads to the production of a conditioned eyeblink response (CR). Eyeblink classical conditioning has extensively been studied in humans and animals and is critically dependent on the cerebellum (Gerwig et al., 2007). Patients with Parkinson’s disease perform as well as healthy controls on EBCC (Sommer et al., 1999), indicating that basal ganglia dysfunction does not necessarily impact significantly on this learning paradigm. In contrast, patients with adult onset focal dystonia have abnormal eyeblink classical conditioning (Teo et al., 2009). Previous studies in animals and humans has revealed a cerebellar circuitry underlying EBCC in which the cerebellar cortical Purkinje cell (PC) receives convergent afferent information about the CS and US via two separate pathways with an additional potential convergence upon the underlying interpositus nucleus (IN) (Yeo and Hesslow, 1998). It is noteworthy that structural imaging studies in dystonia identify grey matter abnormalities in the area of the cerebellar cortex that is involved in this circuit (Draganski et al., 2006).

Purkinje cells in the cerebellar cortex have an inhibitory connection with the underlying dendate nucleus, which in turn displays a disynaptic excitatory connection through the ventral thalamus to the contralateral M1 (Middleton and Strick, 1998). Paired pulse transcranial magnetic stimulation (TMS) protocols can be used to study this pathway. In healthy subjects, a conditioning pulse delivered over the cerebellar cortex 5–7 ms prior to a test pulse over the contralateral M1 will result in reduction of the MEP amplitude relative to a test pulse given alone over this cortical area: “cerebellar brain inhibition” (CBI). This inhibitory effect is thought to arise from activation of Purkinje cells that will consequently inhibit the dentate nucleus and thus reduce the disynaptic excitatory drive from cerebellum to motor cortex (Saito et al., 1995, Ugawa et al., 1995). In eight patients with idiopathic focal limb dystonia, a cerebellar conditioning pulse had no effect on the test pulse MEP amplitude, SICI or ICF (Brighina et al., 2009). The authors hypothesized that the reduced cerebellar modulation of motor cortex excitability could arise through hyperactive purkinje cells in dystonia, that may be compensating for basal ganglia dysfunction. Another possibility is a reduced integrity of the cerebello-thalamo-cortical pathway in idiopathic primary dystonia, as has been described for hereditary primary dystonia (Argyelan et al., 2009).

It is possible that aberrant CBI in dystonia might interact with another phenomenon commonly reported in dystonia: abnormal surround inhibition. This is a muscle-specific modification of the excitability of the corticospinal pathway where just prior to and in the early phase of movement, muscles not involved in the planned movement but surrounding the active muscles show a decreased excitability due to active inhibition (Sohn and Hallett, 2004b). Surround inhibition (SI) has repeatedly been reported to be disrupted in patients with primary dystonia and could account for the overflow of muscle activation seen in this movement disorder (Sohn and Hallett, 2004a, Beck et al., 2009). The mechanism through which this inhibition is regulated has repeatedly been investigated but remains unknown and we have recently explored in healthy subjects whether there is a relationship between SI and CBI (Kassavetis et al., 2011). We did not observe a muscle specific modulation of CBI in parallel with SI, as CBI was reduced in both active and surround muscles at the onset of movement. However, the cerebellum has been proposed to be involved in the movement initiation processes and the observed change of the cerebellar inhibitory drive to the motor cortex at onset of movement is consistent with this. Although yet to be explored, it is possible that the abnormal CBI known to be present in dystonia could interfere with this process.

Regional cerebral blood flow in the ipsilateral cerebellum is negatively correlated with reaction time (Horwitz et al., 2000) and an increased reaction time is observed in patients with cerebellar dysfunction (Grill et al., 1997) as well as a decreased premovement corticospinal excitability (Battaglia et al., 2006). It is intriguing that this increase in reaction time also applies to patients with primary dystonia (Jahanshahi et al., 2001) as well as the lack of MEP facilitation normally present before movement (Gilio et al., 2003). This altered release of motor programs in dystonia has been attributed to basal ganglia dysfunction, but a role for the cerebellum in these abnormalities is also conceivable.