Introduction

The spinocerebellar ataxias (SCAs) are a clinically and genetically heterogeneous group of autosomal dominantly inherited progressive ataxia disorders. Up to now, almost 30 different gene loci have been found. In 19 SCAs the affected genes and causative mutations have been identified. The most common SCAs, which together account for more than half of all affected families, are SCA1, SCA2, SCA3 and SCA6 [1, 2]. Each of these disorders is caused by a translated CAG repeat expansion mutation [3–7]. Expanded CAG repeats are dynamic mutations with variable length that may change during transmission. Patients with longer repeats have an earlier disease onset, a more severe phenotype and faster disease progression than patients with shorter repeats [8].

The clinical phenotype of the common SCAs, SCA1, SCA2, SCA3 and SCA6 is characterized by prominent cerebellar ataxia. Phenotypical differences between these disorders mainly depend on the occurrence of additional non-cerebellar symptoms. Thus, SCA1 patients frequently have pyramidal tract signs, whereas peripheral neuropathy is often present in SCA2 and SCA3. Basal ganglia symptoms are more frequent in SCA2 and SCA3 than in SCA1, while brainstem oculomotor signs including saccadic slowing and ophthalmoparesis are encountered in all three genotypes. However, saccadic slowing is most prominent in SCA2 [9–15]. Other than in SCA1, SCA2 and SCA3, non-cerebellar symptoms in SCA6 are rare and usually mild [16, 17].

It is generally assumed that these disorders are uniform with respect to the type and severity of cerebellar ataxia symptoms. However, a proof for this assumption is lacking. This is mainly due to the fact that validated clinical instruments that allow to separately assess ataxia symptoms in a simple and reliable fashion have only recently been developed [18]. The Scale for the Assessment and Rating of Ataxia (SARA) is a novel clinical rating scale that is based on a semi-quantitative assessment of ataxia on an impairment level. SARA underwent a rigorous validation procedure involving three clinical trials in large groups of SCA, non-SCA ataxia patients and controls [18, 19]. Cerebellar oculomotor symptoms which are part of the spectrum of cerebellar ataxia are not considered in SARA because the first SARA validation trial indicated that the results of oculomotor examination were determined by a different underlying factor than the remaining ataxia symptoms [18]. The Inventory of Non-Ataxia Symptoms (INAS) was devised to assess the presence or absence of clinical signs that may occur in SCAs, but are not covered by SARA. INAS consists of 30 items, among which items 15 (broken up smooth pursuit), 17 (downbeat nystagmus), 18 (gaze-evoked nystagmus on horizontal testing), 23 (hypometric saccades) and 24 (hypermetric saccades) are related to cerebellar oculomotor signs [20].

To define and compare the spectrum of ataxia in the common SCAs, SCA1, SCA2, SCA3 and SCA6, we analysed the results of the individual SARA items, of SARA subscores related to posture and gait, speech and limb kinetic function as well as the INAS items 15, 17, 18, 23 and 24 obtained at the baseline visit of the EUROSCA natural history study. The EUROSCA natural history study is an ongoing multicentric cohort study of 526 patients with either SCA1, SCA2, SCA3 or SCA6 [20]. To better understand the anatomical basis of the ataxia symptoms, we correlated the severity of ataxia symptoms with volumetric measures of the brainstem and cerebellum obtained in a magnetic resonance imaging (MRI) substudy of the EUROSCA natural history study [21].

Patients and Methods

The study was performed at 17 European centres which together form the EUROSCA clinical group. Patients were eligible when they had progressive, otherwise unexplained ataxia and a positive molecular genetic test for either SCA1, SCA2, SCA3 or SCA6. Cases were ascertained with the help of an electronic patient registry that contains data of all SCA patients that have been in contact with one of the study centres. Patients were consecutively recruited within a predetermined period between July 2005 and August 2006. The study population consisted of 526 patients (SCA1 117, SCA2 163, SCA3 139, SCA6 107) [20]. The subgroup which underwent MRI consisted of 82 patients (SCA1 48, SCA3 24, SCA6 10) [21]. Inclusion into the MRI substudy depended on the informed consent and the availability of MRI scanners and scanning time of the participating centres. Demographic, genetic and clinical data of the study population are given in Table 1.

Table 1 Demographic, genetic and clinical data of the study population
Full size table

Assessments were done according to a written study protocol. There were maximally three investigators at each centre. All investigators were experienced in the use of the applied scales.

To measure ataxia symptoms, we used the SARA that has been recently developed and validated. SARA has eight items that yield a sum score of 0 (no ataxia) to 40 (most severe ataxia): 1 = gait (score 0–8), 2 = stance (score 0–6), 3 = sitting (score 0–4), 4 = speech disturbance (score 0–6), 5 = finger chase (score 0–4), 6 = nose-finger test (score 0–4), 7 = fast alternating hand movements (score 0–4) and 8 = heel–shin slide (score 0–4). Items 5–8 are rated independently for both sides, and the arithmetic mean of both sides is included in the SARA sum score. In addition, we formed subscores related to posture and gait, speech and limb kinetic function. The posture and gait subscore was formed by adding the scores of items 1–3, the limb kinetic function subscore by adding the scores of items 5–8, while the speech subscore is identical with item 4 [18, 22].

Cerebellar oculomotor signs were assessed with the INAS. INAS consists of 30 items, each of which is related to one of the following 16 symptoms or syndromes: areflexia, hyperreflexia, extensor plantar response, spasticity, paresis, amyotrophy, fasciculations, myoclonus, rigidity, chorea, dystonia, resting tremor, sensory symptoms, oculomotor signs, urinary dysfunction and cognitive impairment. For the purpose of this study, item 15 (broken up smooth pursuit), item 17 (downbeat nystagmus), item 18 (gaze-evoked nystagmus on horizontal testing), item 23 (hypometric saccades) and item 24 (hypermetric saccades) were considered. In these items, only the presence or absence of the respective symptom is determined. Hypometric (item 23) and hypermetric saccades (item 24) were summarized as dysmetric saccades, if one of these symptoms occurred. Reliability of INAS ratings was tested in two large multi-centre trials. For most INAS items, it was found to be acceptable [20] (own unpublished data).

The MRI studies were performed, as previously described [21]. In brief, patients were scanned on 1.5 T scanners (Siemens Sonata, Siemens Vision, Philips, General Electrics) in eight participating imaging centres (Bonn, Brussels, Essen, Milan, Nijmegen, Paris, Santander, Tübingen and Warsaw). The imaging protocol comprised a 3D T1-weighted anatomical dataset measured with a standard head coil. For each scanner model, a custom-tailored imaging protocol was used to make datasets as similar as possible without sacrificing image quality. Image protocols were chosen to optimize image contrast and signal-to-noise ratio for each scanner. According to the different manufacturer-specific conditions, the following settings were used: Siemens: MPRAGE, TRπ 1,200–1,500; Philips: FFE, TRπ 6–15 and General Electrics: TRπ 2,200–3,000. The following settings were identical on all scanners: TE 2–7, TI 450–700, matrix 256 × 256, field of view 230 mm, slice thickness 1 mm and voxel size of 1 × 1 × 1 mm. The resulting datasets were converted to ANALYZE format.

All MRI datasets were analysed by one investigator blinded to the diagnosis. MRI-based volumetry was implemented by using the method previously described and validated for its reproducibility [23]. Volumetry consisted of semiautomatic presegmentation based on predefined anatomical landmarks and automated region-growing segmentation. Brain structures that could not be distinguished by the region-growing algorithm were delineated manually. Calculation of the volumes was achieved by multiplying the number of voxels per regions of interest and the voxel size. The cranial border of the brainstem was defined by a plane through the mammillary body and posterior commissure shifted caudally by 1/3 the distance to the cranial notch of the pons. A parallel plane to the latter intersecting with the posterior rim of the foramen magnum represented the caudal border. The posterior border of the brainstem was defined by a plane through the obex and posterior commissure shifted posteriorly to include the inferior colliculus. Paravermal sagittal planes separated the cerebellar vermis from the hemispheres. The total intracranial volume (TICV) was estimated by tracing the dura mater using a 1-in-10 sampling strategy [24]. To adjust for individual differences of head size, TICV was used for the normalization of localized brain volumes.

Statistical analyses were performed with SAS 8 software (SAS Institute, Cary, NC, USA) and SPSS 17.0 for Windows (SPSS Inc., 2008). Test results were considered significant at the 0.05 level. Patient clinical characteristics were compared among SCA groups with an analysis of variance for continuous variables and with a [chi]2 test for categorical variables. Descriptive results are given as mean ± standard deviation or as median and lower (Q1) and upper quartile (Q3). For comparison of the SARA, we used Kruskal–Wallis one-way analysis of variance with Bonferroni correction. Wilcoxon–Mann–Whitney test on each pair of groups was used to qualify the differences. For the comparison of INAS, we used Pearson’s chi-square test with Bonferroni correction.

To explore the association of regional brain volumes with SARA scores, we calculated Spearman’s rank correlation coefficient. As this part of the study had an exploratory character, we did not perform a Bonferroni correction.

The procedures followed were in accordance with the ethical standards of the ethics committees of the contributing centres and with the Helsinki Declaration 1975, as revised in 2000 (World Medical Association Declaration of Helsinki 2000). Informed and written consent was obtained from all study participants.

Results

Ataxia Symptoms

The four genotypes were comparable in terms of disease duration (Table 1). There were no differences between the SARA sum scores in SCA1 (15.6 ± 9.1), SCA2 (15.8 ± 8.0), SCA3 (15.1 ± 8.6) and SCA6 (15.0 ± 6.7). Similarly, the posture and gait subscore (items 1–3), the speech subscore (item 4) and the limb kinetic subscore (items 5–8) did not differ between the genotypes (Table 2). At the level of the single items, we found differences for item 3 (sitting, p = 0.0050), item 5 (finger chase, p = 0.0025) and item 6 (nose–finger test, p < 0.0001), whereas the scores of the remaining items were not different (Table 3). Further analyses showed that the differences in item 3 (sitting) were due to a better performance of SCA6 patients compared to all other genotypes. In item 5 (finger chase), SCA3 patients performed better than the other genotypes, while item 6 (nose–finger test) was less impaired in SCA3 and SCA6 compared to SCA1 and SCA2. In a subgroup of SCA1, SCA3 and SCA6 patients who underwent MRI, sum scores, subscores and single item scores did not differ between the genotypes with the exception of the speech subscore which was lower in SCA3 compared to SCA1 and SCA6 (Tables 2 and 3).

Table 2 Scores of SARA subscores
Full size table
Table 3 Scores of SARA items
Full size table

Cerebellar Oculomotor Deficits

The prevalence of cerebellar oculomotor signs assessed in this study differed between the genotypes (Table 4). Post hoc analyses showed that INAS item 15 (broken up smooth pursuit) was more frequent in SCA3 and SCA6 compared to SCA1 and SCA2. Downbeat nystagmus (item 17) was encountered in 20.6% of all SCA6 patients, but only rarely in the other genotypes. Horizontal gaze-evoked nystagmus (item 18) was frequently found in SCA3 and SCA6. Prevalence was lower in SCA1 and even lower in SCA2. Dysmetric saccades (item 23 and/or 24) were most frequent in SCA6 followed by SCA1 and SCA3, while dysmetric saccades were rare in SCA2.

Table 4 Prevalence of cerebellar oculomotor signs
Full size table

Correlation of Ataxia Symptoms with Volumetric Measures

In a subgroup of SCA1, SCA3 and SCA6 patients, we correlated clinical signs with the volumes of the brainstem, cerebellar vermis and cerebellar hemispheres. Representative MRI scans are shown in Fig. 1. SCA1 patients with higher posture and gait subscores had smaller volumes of the brainstem (r = −0.28, p < 0.05). In SCA3, we found inverse relationships between the posture and gait subscore and the brainstem volume (r = −0.60, p < 0.01) as well as the cerebellar hemispheric volume (r = −0.36, p < 0.05). In addition, the limb kinetic subscore was inversely correlated with the brainstem (r = −0.65, p < 0.01) and cerebellar hemispheric volumes (r = −0.35, p < 0.05). There were no significant correlations between SARA subscores and brainstem and cerebellar volumes in SCA6. Correlation of the single item scores with the volumes of the brainstem, cerebellar vermis and cerebellar hemispheres yielded largely congruent results (data not shown). Beyond this, we found an inverse correlation between gait ataxia (item 1) and cerebellar volume in SCA6 (r = −0.56, p < 0.05; Table 5).

Fig. 1
figure 1

Representative MRI scans (left and middle: axial T2 TSE, right: sagittal 3D T1) of a 45-year-old male SCA1 patient (disease duration 6 years), a 38-year-old male SCA3 patient (disease duration 15 years) and a 61-year-old male SCA6 patient (disease duration 5 years)

Full size image
Table 5 Correlation of brainstem and cerebellar volumes with SARA subscores
Full size table

Discussion

We have used two novel clinical instruments, SARA and INAS, to assess and compare the spectrum of ataxia and cerebellar oculomotor symptoms in a large cohort of SCA1, SCA2, SCA3 and SCA6 patients. SARA has eight items which are related to gait, stance, trunk stability during sitting, speech and coordination of arm and leg movements [18]. When we used three subscores that were related to posture and gait, speech and limb kinetic function, we failed to find differences between SCA1, SCA2, SCA3 and SCA6. Comparison of the scores of the individual SARA items revealed differences only in item 3 which is related to sitting and items 5 and 6 which are related to control of arm movements. Sitting was more stable in SCA6 than in the other genotypes. In addition, SCA3 and to a lesser degree SCA6 patients had less ataxia of arm movements than SCA1 and SCA2 patients. Similarly, comparison of the scores of the subgroups that underwent MRI did not reveal relevant differences between the genotypes.

INAS was designed to assess non-ataxia and oculomotor symptoms in SCA patients. For the purpose of this study, we selected those INAS items that are related to cerebellar oculomotor functions. In contrast to previous electrooculographic studies, the present study completely relied on clinical observation and lacked quantification. On the other hand, our clinical approach allowed enrolling a large number of SCA patients that by far exceeds that of all earlier eye movement studies in SCAs [25–28]. Whereas the phenotype of SCA1, SCA2, SCA3 and SCA6 patients was quite similar with respect to ataxia of gait, stance and limb movements, there were marked differences of the frequency of cerebellar oculomotor deficits. Pursuit deficits, horizontal gaze-evoked nystagmus and saccadic dysmetria were most frequently encountered in SCA6 followed by SCA3, whereas these abnormalities were less frequent in SCA1 and SCA2. These findings are in good accordance with previous studies that observed pursuit deficits, horizontal gaze-evoked nystagmus and saccadic dysmetria in patients with all four genotypes, but noted a low frequency of horizontal gaze-evoked nystagmus in SCA2 [26, 27, 29]. Downbeat nystagmus was encountered almost exclusively in SCA6, a finding that confirms previous reports on downbeat nystagmus in SCAs. However, the prevalence of downbeat nystagmus in SCA6 was lower than previously reported, a discrepancy that is probably due to lower sensitivity of clinical testing [25, 26, 28].

Although the cerebellar cortex has a remarkably homogeneous microanatomical structure, there is evidence for subdivision of the cerebellar cortex into functionally distinct compartments that emerge as a consequence of specific afferent and efferent connections. Vestibular afferents strongly project to the flocculonodular lobe, spinal afferents to anterior and posterior parts of the vermis and the paravermal hemispheric regions of the anterior lobe and cortical afferents via the pontine nuclei to the cerebellar hemispheres. With respect to efferents, the cerebellar cortex is divided into three sagittal zones: the vermis projecting via the fastigial nucleus to brainstem and spinal targets, an intermediate zone projecting via the interposed nucleus to the red nucleus and thalamus and a lateral zone projecting via the dentate nucleus to the thalamus. In addition, the flocculonodular lobe projects to the vestibular nuclei [30–32]. As a consequence of these anatomical connections, lesions of the flocculonodular lobe and caudal vermis typically result in truncal and gait ataxia, lesions of the rostral vermis in gait ataxia and hemispheric lesions in limb ataxia. Similar symptoms may occur as a result of lesions in different anatomical structures and tracts that project to the cerebellar compartments or are targets of them. For example, ataxia of limb movements may not only occur as the consequence of cerebellar hemispheric damage but also of pontine and thalamic lesions [33–35]. For the understanding of clinical–pathological correlations in degenerative ataxias, it is therefore appropriate to conceive the anatomy of the cerebellum and its connections in terms of vestibulocerebellar, spinocerebellar and pontocerebellar circuits rather than cerebellar cortical divisions [31].

The flocculus and posterior vermis have a key role in the cerebellar control of eye movements. Bilateral lesions of the flocculus result in downbeat nystagmus. The flocculus is also part of the neural integrator network that is the anatomical substrate of the ability to keep eyes in an eccentric position. Consequently, floccular lesions results in a drift towards the primary position which is compensated by corrective saccades towards the eccentric position resulting in gaze-evoked nystagmus. Impaired control of saccadic eye movements by the posterior vermis results in saccadic hypermetria. Cerebellar control of smooth pursuit involves the flocculus and posterior vermis [36].

In the light of this knowledge, the SARA data suggest that vestibulocerebellar, spinocerebellar and pontocerebellar circuits are involved in SCA1, SCA2, SCA3 and SCA6 to almost the same degree. The clinical differences that we found between the genotypes were minor and detected only at the single item level, whereas the more robust analysis using SARA subscores did not confirm any differences. The large concordance of the spectrum of ataxia symptoms between SCA1, SCA2, SCA3 and SCA6 is surprising, as the cerebellum and its connections are differentially affected in SCA1, SCA2, SCA3 and SCA6. In SCA6, there is an almost exclusive degeneration of the cerebellar cortex [37], whereas SCA1, SCA2 and SCA3 are characterized by a more complex pattern of degeneration. SCA1 and SCA2 are prototypes of olivopontocerebellar atrophy with degeneration of the inferior olives, pontine nuclei and cerebellar cortex [38, 39]. In SCA3, there is prominent degeneration of the spinocerebellar tracts, pontine nuclei and dentate nuclei, whereas the cerebellar cortex is relatively spared [12]. The different degeneration patterns occurring in the various SCAs have been reproduced in MRI studies including a recent multicentric morphometric study performed in the present patient cohort [21, 40].

To shed light on the seeming paradox between the uniform spectrum of ataxia and the differential involvement of cerebellar and brainstem structures in the SCAs, we correlated the SARA subscores with brainstem and cerebellar volumes as measured by MRI volumetry in a subset of SCA1, SCA3 and SCA6 patients. This part of our study has an exploratory character because the number of patients, in particular SCA6 patients, for whom MRI data were available, was small. Moreover, the morphometric analysis was restricted to the brainstem, cerebellar vermis and cerebellar hemispheres, whereas functionally important structures such as the flocculonodular lobe or the dentate nuclei were not segmented and measured. MRI data from SCA2 patients were not available, but the degeneration pattern of SCA2 closely matches that of SCA1. Despite these limitations, the results give important clues to the understanding of the pathoanatomical basis of ataxia. Instead of finding a relation between ataxia symptoms and volume loss of certain anatomical structures across all three genotypes, we rather found patterns of correlations that were specific for each disorder. In SCA1, posture and gait abnormalities correlated with brainstem atrophy. In SCA3, posture and gait abnormalities and limb kinetic dysfunction correlated with both, brainstem and cerebellar atrophy. In the SCA6 MRI subgroup that comprised only ten patients, we found weak evidence for a relation between gait ataxia and cerebellar atrophy. A recent study of 17 patients with a cerebellar cortical degeneration, among them 11 with SCA6, reported robust correlations between the SARA posture and gait subscore and atrophy of the medial and intermediate cerebellum and between the SARA limb kinetic subscore and atrophy of the cerebellar hemispheres [22].

The present data together with knowledge from earlier neuropathological and clinical studies support the view that vestibulocerebellar, spinocerebellar and pontocerebellar circuits in SCA1, SCA2, SCA3 and SCA6 are functionally impaired to almost the same degree, but at different anatomical levels. In SCA1 and possibly SCA2, brainstem degeneration appears to be most critical for the development of ataxia, whereas cerebellar degeneration plays the major role in SCA6. The anatomical structures responsible for ataxia in SCA3 appear to be more distributed between brainstem and cerebellum.

Whereas the clinical spectrum of ataxia of gait, stance and limb movements was widely uniform among the four genotypes, examination of cerebellar oculomotor function revealed major differences suggesting that the oculomotor regions of the cerebellum, in particular the flocculus, are most heavily affected in SCA6 followed by SCA3, whereas the oculomotor cerebellum appears to be only mildly affected in SCA1 and SCA2. This assumption appears reasonable to explain the differences between SCA6 and SCA3, since involvement of the cerebellar cortex is much milder in SCA3 than in SCA6. However, cerebellar cortical degeneration is severe in SCA1 and SCA2, and there is no indication that oculomotor parts of the cerebellum are spared in these disorders [29]. The seemingly low prevalence of “cerebellar” oculomotor disturbances in SCA1 and SCA2 is rather due to prominent pontine involvement resulting in saccadic slowing. As a consequence, corrective saccades are also slow and hard to discern in clinical testing [29]. Similarly, saccadic hypermetria is unlikely to occur in patients with slow saccades. Thus, the defective saccadic system occludes cerebellar deficits of pursuit, gaze holding and saccadic accuracy in SCA1 and SCA2.

Conclusion

Our data suggest that vestibulocerebellar, spinocerebellar and pontocerebellar circuits in SCA1, SCA2, SCA3 and SCA6 are functionally impaired to almost the same degree, but at different anatomical levels. The seemingly low prevalence of cerebellar oculomotor deficits in SCA1 and SCA2 is most probably related to the defective saccadic system in these disorders.