Patients with MSA-C

Patients with SCA

In all types of SCAs studied, the NAA/Cr in the cerebellar hemispheres and vermis and NAA/Cho in the cerebellar hemispheres were significantly reduced (p < 0.05) at both assessments (Table 2).

In patients with SCA2, the Cho/Cr in the cerebellar hemispheres was significantly reduced (p < 0.05) at both MRS assessments (Table 2) and in the vermis only at the second MRS assessment, as compared with the healthy controls.

In patients with SCA3, NAA/Cho in the vermis appeared consistently reduced and Cho/Cr in the cerebellar hemisphere became significantly reduced than healthy controls at 2^nd MRS assessment (p < 0.05). (Table 2) In patients with SCA6, NAA/Cho appeared significantly reduced in the vermis at initial assessment. (p = 0.024) (Table 2).

Longitudinal changes in the MRS metabolites

There was no significant difference in the MRS metabolites between initial and follow up assessments in the study groups (Table 2), except for an increase of Cho/Cr with time in the cerebellar hemispheres in patients with SCA2 (p = 0.031) (Table 2). Furthermore, there was no difference in the monthly changes of metabolites between different groups of patients (Table 5 and Figure 2).

	Cerebellar hemispheres			Vermis
	NAA/Cr	Cho/Cr	NAA/Cho	NAA/Cr	Cho/Cr	NAA/Cho
p #	0.471	0.168	0.261	0.693	0.269	0.273
SCA2	0.09 ± 0.17	0.13 ± 0.14	−0.08 ± 0.28	−0.25 ± 0.53	−0.18 ± 0.23	−0.08 ± 0.68
SCA3	−0.04 ± 0.22	−0.04 ± 0.26	0.01 ± 0.61	−0.07 ± 0.20	0.05 ± 0.21	−0.22 ± 0.37
SCA6	0.06 ± 0.22	−0.11 ± 0.16	0.27 ± 0.44	−0.18 ± 0.13	−0.16 ± 0.16	0.00 ± 0.45
MSA-C	0.14 ± 0.59	0.02 ± 0.37	0.33 ± 1.14	0.03 ± 0.32	−0.14 ± 0.36	0.51 ± 0.86

SCA2: Spinocerebellar ataxia type 2; SCA3: Spinocerebellar ataxia type 3; SCA6: Spinocerebellar ataxia type 6; MSA-C: Multiple system atrophy- cerebellar type.

^╥Data are presented as 100X MRS ratios.

^#Kruskal-Wallis test.

The correlation between the changes of MRS metabolites and SARA scores

The changes of NAA/Cr (R = −0.332, P < 0.001 in the cerebellar hemispheres; R = −0.521, P < 0.001 in the vermis) and Cho/Cr (R = −0.380; P < 0.001 in the cerebellar hemispheres; R = −0.529, P < 0.001 in the vermis) were inversely correlated with the changes of SARA scores (Figure 3).

MSA-C

In general, patients with MSA-C tended to have a worse neuronal integrity and less production of cell membranes than those with SCA3 or SCA6 at the first MRS assessment (Tables 2 & 3). With time, the differences remained similar at follow-ups (Table 4), suggesting a persistent and more severe disruption in the neuronal and membrane integrity in MSA-C. MSA-C is a rapidly progressive neurodegenerative disease manifesting ataxia and autonomic dysfunction [23]. Our previous cross-sectional study in patients with MSA-C also corroborated these observations [18].

MSA-C and SCA2 are two relentlessly and rapidly progressive diseases manifesting ataxia with neuronal drop-outs and impairment of membrane synthesis. Lactate peak and myo-inositol were previously proposed to be indicators to differentiate SCA2 from MSA-C [14]–[16]. In this study, we found that, at the early stage of the diseases, the MRS features were similar between MSA-C and SCA2. As disease progressed, the different speed of changes in Cho/Cr and NAA/Cr in the cerebellar hemispheres might help differentiate MSA-C from SCA2 (Table 4).

SCA

In line with previous reports [15],[16], patients with SCA2 have a more severe impairment in neuronal integrity, given the lower NAA/Cr in the cerebellar hemispheres than those with SCA3 or SCA6 at the start and at the follow-up assessment, even being of similar clinical severity (Tables 2, 3 & 4). The SARA scores increased significantly in SCA2 and SCA3 but not in SCA6 during the period of follow up (Tables 1 and 6). Although there was little difference in the monthly changes of SARA scores between subtypes of SCAs, the change was still higher in SCA2 (Table 6). Pathological studies in the past have revealed, in SCA2, a marked loss of Purkinje cells in the cerebellar cortex and a loss of myelinated fibers in the cerebellar peduncles and white matter, sparing the dentate nucleus, and the neuronal degeneration preceded the onset of clinical symptoms [24] .While a prominent spinopontine degeneration with a relative sparing of the olivocerebellar regions in SCA3 [25] and a pure cerebellar degeneration with a loss of Purkinje cells in the cerebellar cortex in SCA6 [26],[27] were observed. The synaptic loss in the cerebellum and brainstem is most severe in SCA2 [28],[29]. The MRS changes observed by us are corroborated by the pathological changes.

The disease progression reflected by the monthly changes of SARA scores was most severe in SCA2, MSA-C, followed by SCA3 and SCA 6. Similar results have been documented in our earlier report [30].

Interval changes

Progressive reduction in the cerebellar volume, correlating with disease duration and age, has been documented in patients with MSA-C [31], SCA2 [32], SCA3 [33] and SCA6 [34]. In SCA2, the presence of pre-symptomatic region-specific atrophy was observed and a possible developmental component to the pathomechanisms was once proposed [32]. This is consistent with the early MRS changes in SCA2 when the clinical severity represented by SARA scores is still low and disease duration remains short.

The limitations of this study are, first, the sizes of the patients were relatively small, especially in SCA2 and SCA6. The small patient numbers could be the reason contributing to the lack of significance in disease duration between groups. We tried to minimize these heterogeneities by recruiting only patients who had their first MRS examination at a disease duration less than 12 months. Furthermore, the MRS metabolites from both cerebellar hemispheres were averaged for the ease of analyses, thus the data of MRS from the cerebellar hemispheres should actually be twice as many as those from the vermis. Hence, the reason that some differences of MRS in patient groups were observed only in the cerebellar hemispheres, but not in the vermis, could likely be attributed to the limited patient numbers. Third, we only focused on the metabolites in the cerebellar hemispheres and vermis. Fourth, we only correlated the metabolites on MRS with total SARA scores. The rates of progression in different compartments of SARA might be different [30]. Further studies with MRS in a larger cohort, addressing also the changes of metabolites in the other regions of the brain while correlating the changes between MRS metabolites and different compartments of SARA, would be desirable to help unravel the pathophysiology of the SCAs and MSA-C.

Patients and controls

This study was approved by the Institutional Review Board of Taipei Veterans General Hospital, Taipei, Taiwan. Patients who had undergone SARA and MR spectroscopy at least twice over a period of 10 years were enlisted. The interval between evaluations with SARA and the MR spectroscopy must be approximately within 3 months. From March 2004 to January 2014, a total of 47 patients with either one of the SCAs or MSA-C were studied sequentially in the afore-mentioned manner. Among them, 38 patients had their first MRS assessment with a disease duration less than 12 months and the follow-up MRS after one year. Among them, 5 were molecularly identified to be SCA2, 18 SCA3, and 3 SCA6. Twelve patients met the criteria of MSA-C. Forty-four healthy individuals without any history of neurological diseases served as controls. The MRS features in the basal ganglion were validated according to the normal data proposed by Brain Ross and Else Rubaek Danielsen [35]. The demographic features of the study subjects are listed in Table 1.

Images and spectroscopic acquisition

Brain MRI and MRS were performed using a 1.5-T system (Signa EXCITE, GE Medical Systems, Milwaukee, WI). The MRI protocol consisted of an axial T1-weighted three-dimensional fast-spoiled gradient recalled acquisition in steady state images (TR 8.58 msec, TE 3.62 msec, inversion time [TI] 400 msec, voxel resolution 0.75X0.75X1.5 mm³) and an axial T2 fast spin-echo sequence [TR 4000 msec, TE 256.5 msec, voxel resolution 348X512].

After MR imaging, proton MRS was recorded in the cerebellar hemispheres and vermis by using single-voxel stimulated echo acquisition mode sequence (3000/15/13.7/96 [TR/TE/mixing time/excitations], spectral width = 2500 Hz, number of points = 2048, voxel resolution = 2 cm × 2 cm × 2 cm). The voxel of interest (VOI) in each subject was placed in a uniform manner by the same investigator (JFL). Care was taken to avoid cerebrospinal fluid space within the VOIs. The peak areas for N-acetyl aspartate (NAA) at 2.02 parts per million (ppm), Creatine (Cr) at 3.03 ppm, and Choline (Cho) at 3.22 ppm were measured using the Functool provided by the MR company (GE XVi, Milwaukee, WI). Peak integral values were expressed relative to the Cr peak. Metabolite intensity ratios were automatically calculated at the end of each single voxel acquisition including NAA/Cr and Cho/Cr. The NAA/Cho ratio was also calculated for comparison. MRS results with full width at half maximum (FWHM) > 6 Hz were disqualified from the MRS analyses to ensure high quality.

Statistical analyses

Comparisons of the age at examination between the patient groups and healthy controls and comparison of the disease duration and the SARA scores at first and second evaluations between the patient groups were performed using nonparametric analyses (Kruskal-Wallis H test) given that some of the sample sizes were small and the assumption of a Gaussian distribution was not appropriate. Similarly, given the non-Gaussian distribution of the MRS parameters, comparison of the metabolites on MRS (NAA/Cr, Cho/Cr and NAA/Cho in the cerebellar hemispheres or vermis) between different groups of patients and healthy controls, between patients of disparate types of SCA and MSA-C, between patients with different types of SCA at initial and follow-up MRS assessments, and comparison of the monthly changes of metabolites on MRS and SARA scores were all performed using the non-parametric Mann–Whitney U-test. The interval changes of metabolites on MRS and SARA scores were evaluated by pair-t test. The correlation between metabolites on MRS and SARA scores was assessed with Pearson’s correlation coefficient. Differences were considered significant at p < 0.05. Data are given as mean ± standard deviation.