In-vivo microstructural white matter changes in early spinocerebellar ataxia 2

Abstract

Objective

White matter (WM) integrity of Spinocerebellar ataxia 2 (SCA2) is poorly understood, more so in the early stages of SCA2. In this study, we evaluated the microstructural integrity of the WM tracts with an emphasis on the nature of in-vivo pathological involvement in early SCA2.

Materials and methods

We evaluated the MRI images of 26 genetically proven SCA2 patients with disease duration less than 5 years and 24 age- and gender-matched healthy controls using tract-based spatial statistics (TBSS) to identify the WM tract changes and their clinico-genetic correlates (age at onset, duration of disease, ataxia severity, and CAG repeat length) using standard methodology.

Results

The mean age at onset and duration of disease was 28.7 ± 8.51 years and 3.5 ± 0.69 months, respectively. The mean CAG repeat length was 42.5 ± 4.6 and ataxia score was 16.1 ± 4.9. Altered DTI scalars signifying degeneration was present in bilateral anterior thalamic radiation (ATR), corticospinal tract (CST), inferior fronto-occipital fasciculus (IFOF), superior and inferior longitudinal fasciculus (SLF and ILF), uncinate fasciculus (UF), cingulum, corpus callosum (CC), forceps major, and forceps minor (corrected p<0.05). DTI scalars representing demyelination was seen in the superior cerebellar peduncle (SCP) and cerebellar WM. There was a significant correlation of SARA score with axial diffusivity of the bilateral cingulum, ATR, CST, forceps minor, IFOF, ILF, SLF, and SCP on the right side (corrected p<0.05).

Conclusion

Extensive WM involvement is present in early SCA2. The DTI scalars indicate degeneration and demyelination and may have clinical implications.

1. Introduction

Spinocerebellar ataxia 2 (SCA2) is a neurodegenerative disorder caused by nucleotide repeat expansion in the ATXN2 (12q24.1) gene [1]. Clinically, it manifests with a plethora of manifestations such as cerebellar symptoms, extrapyramidal involvement, cognitive dysfunction, motor-neuron involvement, pyramidal signs, and autonomic dysfunction [2]. Prevalence-wise, it is the second-most common subtype of SCA in the world [3]. However, the pace of progress in research into the disease is slow. In the recent years, research has started focussing on neuroprotection and disease modification in pre-symptomatic and early symptomatic patients of SCA2. Many therapeutic strategies are presently in different stages of preclinical testing.

Understanding the pathology of any disease has significant relevance in the field of neurology. It can help in identifying biomarkers, understanding clinical manifestations, and also in formulating therapeutic strategies. However, for many rare diseases such as SCA2, acquiring brain samples for pathological study is difficult, especially in the pre-symptomatic stage and in early stage of disease. While cell and animal models are helpful, these techniques are time consuming, labour intensive, and technically difficult. Another viable strategy is through the analysis of magnetic resonance imaging (MRI). MRI has emerged as an important tool for clinical research in recent years due to its ability to detect the in-vivo structural changes of the brain. Use of advanced neuroimaging analysis on MRI allows one to quantitatively assess the microstructural damage even in the apparently normal appearing brain regions.

Pathological studies in SCA2 have reported a distributed pattern of neuron loss, gliosis and myelin loss [4,5]. However, histopathological studies are scarce and their observation represents the late neuropathological changes of SCA2. Neuroimaging studies have consistently reported grey matter (GM) atrophy of cerebellum, hindbrain, parahippocampal gyrus, insula, thalamus, precuneus, and fronto-temporo-parietal cortices [6-8]. Compared to GM changes, white matter (WM) changes has only been seldom studied systematically. In one such study, Mascalchi et al evaluated ten patients with SCA2 and identified WM changes involving the brain stem, cerebellar peduncle, corpus callosum, internal capsule, and WM of the frontal cortex, sensorimotor cortex, thalamus, and cerebellum [9]. Another study by Hernandez-Castillo et al on 14 subjects also reported a similar distribution of involvement [10]. However, the cohorts valuated had variable and longer duration of disease and small sample sizes. Furthermore, they described the WM changes using limited DTI parameters and did not provide any information on the nature of neuropathological changes. Nonetheless, the presence of WM microstructural abnormalities in SCA2 suggests that in addition to characteristic cerebellar GM involvement, WM degeneration and demyelination may also be important pathophysiological features. This is especially pertinent since myelin loss is reported in pathological studies, albeit in later stages of disease. Understanding the WM involvement and the in-vivo neuropathology would help us better understand the disease.

In this study, we evaluated the WM changes in genetically proven SCA2 using DTI-Tract based spatial statistics (DTI-TBSS). To identify the early WM microstructural changes, we recruited patients with duration of disease less than 5 years from the onset of ataxic symptoms. In addition to identifying the WM tracts with microstructural changes, we also used the DTI matrix variables to delineate the in-vivo pathology of individual WM tracts.

2. Materials and methods

This study was performed prospectively with a cross-sectional design in the Department of Neurology at the National Institute of Mental Health and Neurosciences (NIMHANS), India after appropriate scientific and ethical approval. The inclusion criteria for this study included (a) genetically confirmed SCA2, (b) age more than 18 years, (c) duration of disease < 5 years from onset of ataxic symptoms, (d) patients capable of independent gait (without any need of support). The exclusion criteria were (a) contraindication for MRI scan, (b) uncooperative patients, (c) age> 65 years, (d) presence of comorbidities affecting brain (previous stroke, tumours, head trauma, etc.). All subjects provided informed consent to participate in the study. Thirty-three subjects with genetically proven SCA2 and 30 age and gender-matched healthy controls were recruited using convenient sampling.

Clinical details

The demographic and clinical details of patients were recorded systematically. The ‘age at onset (AAO)’ was defined as the age at which the first symptom was noticed by the patient or relative. The ‘duration of disease’ was defined as the number of years from the AAO to the time of MRI scan. The SARA score was used as the measure of severity of ataxia.

MRI acquisition

The MRI data were acquired in Siemens SKYRA 3 Tesla MRI system. Routine conventional T2 weighted images and FLAIR images were acquired. DTI was performed by using single-shot spin-echo, echo-planar sequences in axial sections with the following settings: TR= 8400 ms, TE= 91 ms, FOV= 240×240 mm, section thickness= 2 mm, voxel size= 2×2×2 mm, and no intersection gap. DTI was performed along 64 directions with a b value = 1000 s/mm2 and NEX-2. In addition, the images without diffusion weighting were acquired corresponding to b = 0 s/mm2.

Diffusion tensor imaging-Tract Based Spatial Statistics

The Diffusion Toolbox software tool in the Functional MRI of the Brain (FMRIB) Software Library (FSL; http://www.fmrib.ox.ac.uk/fsl/fdt/index.html) was used for calculating the DTI indices including fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD). The standard procedure of voxel wise cross-subject analysis of DTI data was performed using Tract-Based Spatial Statistics (TBSS) following the preprocessing steps [11].

Preprocessing of DTI data in the raw format included eddy current correction which included motion correction to correct for distortions due to gradient directions that were applied. This was followed by another common preprocessing step to align all the images with each other before estimating diffusion-related measures. The pre-alignment was both to correct for head motion during the session and to reduce the effects of gradient coil eddy currents. FLIRT was applied for full affine alignment of each image to the non-diffusion weighting image. FA values were calculated following this step. Finally, Brain Extraction Tool (BET) was applied to the non-diffusion-weighted image, to exclude non-brain voxels for further consideration.

Voxel-wise statistical analysis of FA was performed using TBSS. First, the individual FA images from both the control and patient groups were non-linearly aligned to the predefined FSL FMRIB58 FA map using a resolution of 1mm in the standard MNI152 space. Subsequently, a mean FA image was created and thinned to create a mean FA skeleton which represented the centers of all tracts common to the group (http://fsl.fmrib.ox.ac.uk/fsl/tbss). The aligned FA image was projected on the skeleton. Finally, voxel-wise statistical analyses were performed across the subjects for each point on the common skeleton. A permutation test was performed on the final skeleton to detect white matter differences between the patient and the control groups using the FSL/RANDOMISE tool (5000 permutations) corrected for multiple comparisons and threshold-free cluster enhancement. Subsequently, the average FA value was analyzed using two-sample two-tailed t test, with age and TIV included in the model as nuisance covariates, with p<0.05 as the confidence threshold after correction for multiple comparison. The same procedure was applied on the MD, AD, and RD maps. The results were obtained by projecting the mean skeleton of patients> controls on the anatomical template image built-in in FSL, and labels were made using John Hopkins University (JHU) white matter tractography atlas.

Clinico-genetic and neuroimaging correlation

Correlations were assessed between the DTI indices (mean FA, MD, AD and RD values) of WM tracts and clinico-genetic variables such as SARA score, AAO, duration of disease, and CAG repeat number. A threshold of p<0.05 after multiple comparison was considered significant.

Statistical analysis

Data was analysed using R software. For continuous variables, data was expressed using mean and standard deviations and comparison between groups was performed using independent t-test (after establishing normality of data). Comparison of frequencies (eg. gender distribution) was performed using chi-square test. A threshold of p-value < 0.05 after correction for multiple comparison was considered as significant.

3. Results

Six patients with SCA2 and five healthy controls were excluded from the analysis due to the presence of excessive movement artifacts. Furthermore, one patient and two healthy controls were removed from the analysis due to the presence of gliosis (SCA2), large arachnoid cyst (HC), and widespread WM changes (HC) in MRI. DTI-TBSS analysis was performed on the MRI images of 26 patients (Male: Female = 14:12) with SCA2 and 24 healthy controls (Male: Female = 14:10).

Demographic and clinical variables

There was no significant difference in the mean age among patients with SCA2 and HC (33.0 ± 10.02 vs 34.1 ± 11.88 years; corrected p = 0.45). In patients with SCA2, the mean AAO and duration of disease was 28.7 ± 8.51 years and 3.5 ± 0.69 years, respectively. The mean CAG repeats of the abnormal allele in patients with SCA2 was 42.5 ± 4.6 (range: 34 – 61, median: 41). The mean SARA score was significantly higher in SCA2 than HC (SARA: 16.1 ± 4.9 vs 0.1 ± 0.1; corrected p value = 0.001) (table 1).

Table 1. Demographic and clinical characteristics of Spinocerebellar ataxia 2 and healthy controls.

Demographic parameters	Patient	Control	P value
Number	26	24	-
Gender distribution (Male:Women)	14:12	14:10	0.65
Mean age (in years)	33.0 ± 10.02	34.1 ± 11.88	0.45
Mean age at onset (in years)	28.7 ± 8.51	-	-
Mean duration of disease (in years)	3.5 ± 0.69	-	-
Mean CAG† repeats	42.5 ± 4.6	23.4 ± 2.3	-
Mean SARA ‡ score	16.1 ± 4.9	0.1 ± 0.1	-

^†CAG: Cytosine-Adenine-Guanine Trinucleotide repeat

^‡Scale for assessment and rating of ataxia. Mean CAG and SARA scores were not compared between patients and controls.

Microstructural white matter changes in SCA2

WM changes were characterized on the basis of changes in the DTI scalars such as Fractional anisotropy (FA), Mean diffusivity (MD), Axial diffusivity (AD), and Radial diffusivity (RD).

On comparison of patients and controls, extensive WM changes were detected in the DTI scalars of different tracts.

Decreased FA was obtained bilaterally in the anterior thalamic radiation (ATR), corticospinal tract (CST), inferior fronto-occipital fasciculus (IFOF), inferior longitudinal fasciculus (ILF), superior longitudinal fasciculus (SLF), uncinate fasciculus, superior cerebellar peduncle (SCP), cingulum, corpus callosum, forceps major, forceps minor, and cerebellar WM (FWE corrected p<0.05).

Increase in MD was seen in the ATR, CST, IFOF, ILF, SLF, forceps major and minor, uncinate fasciculus, SCP, and cerebellar WM on both sides of the brain (FWE corrected p<0.05).

Decrease in AD was seen in the bilateral IFOF, ILF, SLF, right ATR and left forceps major and minor (FWE corrected p<0.05).

Increase in RD was found in the ATR, CST, forceps major and minor, IFOF, ILF, SLF, uncinate fasciculus, SCP, and cerebellar WM (FWE corrected p<0.05). The WM tracts with significantly altered diffusion parameters is represented in figure 1.

Figure 1.

White matter tracts in spinocerebellar ataxia 2 with altered DTI scalars. The white matter skeleton is represented by green colour. Tracts with decrease in FA and AD are represented by red and yellow colours and increase in MD and RD are represented by blue and light blue colours.

Decreased FA was present in the bilateral anterior thalamic radiation (ATR), corticospinal tract (CST), inferior fronto-occipital fasciculus (IFOF), inferior longitudinal fasciculus (ILF), superior longitudinal fasciculus (SLF), uncinate fasciculus, superior cerebellar peduncle (SCP), cingulum, corpus callosum, forceps major, forceps minor, and cerebellar WM.

Increase in MD was seen in the ATR, CST, IFOF, ILF, SLF, forceps major and minor, uncinate fasciculus, SCP, and cerebellar WM on both sides of the brain.

Decrease in AD was seen in the bilateral IFOF, ILF, SLF, right ATR and left forceps major and minor.

Degeneration and demyelination changes

Two distinct patterns of involvement of WM fiber was observed in our patients. A decrease in FA and AD with an increase in MD and RD, a pattern consistent with axonal degeneration, was obtained in the ILF, SLF, IFOF, uncinate tract, thalamic radiation, forceps major, and forceps minor. In contrast, a combination of normal AD, decrease in FA, and increase in MD and RD, suggestive of demyelination, was seen in the SCP and cerebellar WM. The cingulum and the corpus callosum had decrease in FA without any significant changes in other DTI measures.

Clinico-genetic correlates of white matter changes in SCA2

SARA score inversely correlated with AD in the bilateral cingulum (cingulate gyrus), right ATR, right corticospinal tract, right forceps minor, right IFOF, right ILF, right SLF and right SCP (corrected p<0.05). There was no significant correlation of SARA score with the FA, MD or RD in other WM tracts. There were no significant correlations between CAG repeat length, AAO and duration of disease with DTI scalars of any WM tracts.

4. Discussion

The DTI-TBSS is a whole brain-based approach without any a priori hypothesis, performed on diffusion MRI sequences to identify the global WM changes [12]. Hence it is an appropriate technique to study rare disorders such as SCA2. This study identified extensive WM changes in the supratentorial and infratentorial tracts, especially involving the association fibers of the brain. Furthermore, the DTI scalars revealed the presence of demyelination changes in the SCP and cerebellar WM whereas the other WM tracts had changes suggestive of neurodegeneration.

The association fibers connect different cortical areas that are discrete and spatially distributed in the same hemisphere. On the basis of length of fibers, they are classified as short and long association fibers. While short association fibers connect WM in adjacent gyri, long association fibers interconnect distant areas of the hemisphere [13]. In this study, we identified the long association fibers (SLF, ILF, IFOF, uncinate, and cingulum) to be particularly affected in SCA2. In addition to the association fibers, projection fibers such as the ATR, which provides projections from the anterior and midline thalamic nuclei to the frontal lobe and the corticospinal tract was also involved. Furthermore, commissural fibers such as forceps major, forceps minor, and the corpus callosum, which connects similar areas of both hemispheres of the brain were also affected. In their structurally and functionally intact state, the WM fibers connect distant parts of the brain and integrate different regions of the brain to work together as a functional network to subserves a host of neurobehavioral functions. They also expand the operational capacity of neurons by enabling rapid and efficient transfer of information that complements the information processing of the neurons. Hence, the disruption of these fibers may result in neurological impairment due to the ‘disconnection’ of these networks [13].

Only few previous studies have systematically studied the WM changes using TBSS in SCA2. Studies by Mascalchi et al and Hernandez-Castillo et al identified WM changes limited to the brain stem, cerebellar peduncle, corpus callosum, internal capsule, and WM of the frontal cortex, sensorimotor cortex, thalamus, and cerebellum [9,10]. In contrast to their observations, this study revealed more widespread WM microstructural changes despite shorter duration of disease. Whether the difference in observations could be attributed to differences in cohort size, demography, or MRI field strengths, the results of all studies point to significant involvement of WM even in the early disease. Recent studies on presymptomatic subjects have also revealed significant WM microstructural changes [14].

In addition to identifying the WM tracts involved, we also used the changes in the DTI metrics (FA, MD, AD, and RD) to speculate the nature of in-vivo microstructural changes. Whereas, FA is a sensitive, non-specific summary measure of microstructural integrity, MD is an inverse measure of the membrane density and is highly sensitive to changes in cellularity, edema, and necrosis. The AD represents the diffusion of water along the principal axis of diffusion and tends to be variable in WM pathologies. For example, AD tend to decrease in axonal injury whereas AD remains unchanged in demyelination. RD represents water diffusion in the perpendicular direction to the principal direction and increases in demyelination. The permutations and combinations of FA, MD, AD and RD values can hence provide valuable insight into the in-vivo structural changes in the WM tracts [15-17]. Our results indicate a predominant degenerative pattern in the supratentorial brain and demyelinating pattern in the infratentorial structures.

The DTI metrics suggesting axonal degeneration in supratentorial and infratentorial brain structures have previously been reported by Mascalchi and colleagues [9]. However, a demyelinating pattern of brain involvement in DTI studies is a novel finding, especially in the early disease. The most likely explanation for the differential involvement of supratentorial and infratentorial brain may be the greater progression of the disease in the infratentorial structures with secondary or mixed demyelination. A previous longitudinal study by Mascalchi and colleagues in a cohort of SCA2 with varying duration of disease (mean value: 12.8±7.3 years) has reported accelerated volume loss and microstructural changes in the midbrain, basis pontis, MCP, SCP, posterior medulla, and cerebellum as compared to the supratentorial regions [18,19]. Furthermore, pre-symptomatic SCA2 carriers are known to have volume loss in the infratentorial structures, although a comparison with supratentorial structures was not performed [20]. Nonetheless, these results imply a differential rates of neurodegeneration in the supratentorial and infratentorial brain. However, whether these advanced pathological changes may occur within five years of symptomatic disease can only be conclusively proven with neuropathological studies. Another interesting but controversial explanation may be the presence of demyelination in the early pathological process of symptomatic SCA2. In the recent years, few studies have provided direct and indirect evidence of demyelination in SCA2. These include histopathological studies and case reports reporting demyelination changes [21-24]. A recent study has demonstrated impaired ceramide-sphingomyelin metabolism in mouse models and human cerebellar tissue in SCA2 challenging the traditional view of secondary myelin involvement [25]. This metabolomic study of spino-cerebellar tissue revealed consistent downregulation of multiple proteins involved in ceramide-sphingomyelin metabolism suggesting broad disturbance of lipid homeostasis in large cerebellar neurons, oligodendrocytes, and other glial cells causing myelin instability and demyelination [25]. If true, this observation may render immunomodulation using synthetic sphingosine analogue such as Fingolimod relevant in SCA2 as disease modifying agents [25]. However, these findings have to be confirmed with well-designed studies preferably with neuropathological confirmation.

Another observation from this study is the correlation between SARA score and AD changes of the cingulum, ATR, CST, forceps minor, IFOF, ILF, SLF, and SCP, predominantly on the right side. The SARA scale is mainly aimed at measuring the degree of cerebellar involvement. Hence, the supratentorial correlations and its laterality are difficult to explain and may be spurious in nature. However, it is known that damage to the input and output pathways such as the spinocerebellar pathways, cortico-ponto-cerebellar pathway, cerebello-thalamic inputs, and cerebellar peduncles may also contribute to ataxia and incoordination [26]. Previous studies have also reported similar supratentorial correlations with ataxia severity [10,27,28]. Hence, these correlations obtained should be interpreted with caution.

Another significant observation from our study was the absence of correlation of duration of disease with WM diffusion changes. Previous DTI studies in SCA2 have also not observed such a correlation [10,27]. This is probably because the ‘duration of disease’ is an ambiguous concept. It is well known that neurodegeneration precedes the clinical symptoms and that onset of ataxia may not be a reliable indicator of onset of neurodegeneration [29,30]. Recollection bias and inaccurate reporting may also cause spurious ‘duration of disease’ estimates. A non-linear progression of WM changes may also underlie the lack of statistical correlation.

There are few limitations of this study. Despite a larger cohort size compared to other studies using TBSS, our sample size is still small. Also, the lack of histopathological correlate in this study also limits the generalizability. Nevertheless, converging proof from multimodal studies including histopathology, animal models, and metabolomics all point towards demyelination as a definite pathophysiological mechanism operating in SCA2. This opens up immunomodulation as a viable option for disease modification. Future studies on larger cohorts with longitudinal designs starting in the pre-symptomatic stage and in different stages in the natural history of disease may provide stronger evidence.

5. Conclusion

This study explored the WM involvement in the early stages of SCA2 and identified widespread WM changes affecting most tracts, especially the long association tracts. The in-vivo pathological changes in SCA2 included degenerative and demyelination changes. Converging evidence from different studies suggests that demyelination could be an early pathological phenomenon in SCA2 and can be potentially modified using immunomodulation. However, studies in larger cohorts are necessary to validate our observations.

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.