Broad white matter impairment in multiple system atrophy

Abstract Multiple system atrophy (MSA) is a rare neurodegenerative disorder characterized by the widespread aberrant accumulation of α‐synuclein (α‐syn). MSA differs from other synucleinopathies such as Parkinson's disease (PD) in that α‐syn accumulates primarily in oligodendrocytes, the only source of white matter myelination in the brain. Previous MSA imaging studies have uncovered focal differences in white matter. Here, we sought to build on this work by taking a global perspective on whole brain white matter. In order to do this, in vivo structural imaging and diffusion magnetic resonance imaging were acquired on 26 MSA patients, 26 healthy controls, and 23 PD patients. A refined whole brain approach encompassing the major fiber tracts and the superficial white matter located at the boundary of the cortical mantle was applied. The primary observation was that MSA but not PD patients had whole brain deep and superficial white matter diffusivity abnormalities (p < .001). In addition, in MSA patients, these abnormalities were associated with motor (Unified MSA Rating Scale, Part II) and cognitive functions (Mini‐Mental State Examination). The pervasive whole brain abnormalities we observe suggest that there is widespread white matter damage in MSA patients which mirrors the widespread aggregation of α‐syn in oligodendrocytes. Importantly, whole brain white matter abnormalities were associated with clinical symptoms, suggesting that white matter impairment may be more central to MSA than previously thought.

Moreover, beyond α-syn, the relocation of p25α from the myelin sheath to the cytoplasm is probably an early event in the disease process (Song et al., 2007). This relocation leads to significant changes in oligodendroglial morphology and a reduction in smaller diameter myelinated axons. White matter degeneration may thus be a central feature of this disease, in addition to neuronal loss. Diffusion magnetic resonance imaging (MRI) studies have previously reported white matter abnormalities in MSA. For example, regions such as the cerebellar peduncles, pons, and corpus callosum have been found to be particularly vulnerable in MSA, consistent with the clinical manifestation of the disease (see Table S1, Hara et al., 2018;Worker et al., 2014;Zanigni et al., 2017).
We aimed to take a step back and instead of focusing on focal abnormalities we sought to take a global perspective to investigate if there are whole brain white matter abnormalities in MSA. In order to do so, in vivo structural MRI and diffusion tensor imaging (DTI) were performed on MSA patients (n = 26), Parkinson's disease (PD) patients (n = 23), and controls (n = 26). Data were examined using a comprehensive, spatially refined whole brain approach encompassing both the superficial white matter adjacent to the cortical gray matter, and the large deep white matter fiber tracts. We chose to investigate superficial and deep white matter because both have been shown to be vulnerable in neurodegenerative diseases (Phillips, Joshi, Piras, et al., 2016;Phillips, Joshi, Squitieri, et al., 2016b;Rulseh et al., 2016) and have unique properties (see Table S2).
We hypothesized that there would be broad whole brain white matter microstructure abnormalities in MSA patients, but not PD patients, compared to controls. It was also hypothesized that broad whole brain white microstructure abnormalities in MSA patients would be associated with clinical measures.

| Standard protocol approvals, patient consents, and data availability
The study was approved by the local ethics committee according to French public health legislation (ID RCB 2012-A01252-41). All participants provided written informed consent. Assessments were carried out at the Toulouse Clinical Investigation Center and the Toulouse NeuroImaging Center. Study data are currently not available in a public repository.

| Subjects
Patients diagnosed with MSA and PD (n = 30 per group) were consecutively recruited via the outpatient clinics of the MSA French National Reference Center and the PD Expert Center of Toulouse. Diagnoses were performed by movement disorder specialists (A. P.-L. T. and O. R.). As described elsewhere (Barbagallo et al., 2016), inclusion criteria for patients were: (a) diagnosis of MSA or PD as per international diagnostic criteria (Gilman et al., 2008); (b) Hoehn and Yahr scores less than 4 on treatment; (c) no history of neurological or psychiatric diseases other than MSA or PD, (d) no evidence of a significant cognitive deficit (Mini-Mental State Examination [MMSE] score > 24); (e) no prior exposure to deep brain stimulation; (f) no evidence of movement artifacts, vascular brain lesions, or brain tumor.
Moreover, 27 sex-and age-matched healthy controls were recruited via the Toulouse Clinical Investigation Center. Overall, 12 subjects (4 MSA, 7 PD, and 1 healthy control) were excluded from analyses due to movement artifacts or severe atrophy as agreed between two of the authors (O. P. and P. P.) after visual inspection before data analysis was performed. Data from the remaining 26 MSA patients, 23 PD patients, and 26 healthy controls were analyzed for the current study.
Within the MSA group, 15 were diagnosed with the MSA subtype MSA-P and 11 with MSA-C due to exhibiting predominantly parkinsonian versus cerebellar symptoms, respectively.

| Motor and cognitive assessments
Motor disabilities in MSA and PD patients were examined using the Motor Examination scores of the Unified MSA Rating Scale, Part II (UMSARS-II) and the Unified Parkinson's Disease Rating Scale, Part III (UPDRS-III), the reference scales to assess motor function in MSA and PD, respectively (Fahn, Elton,, & Members of the UPDRS Development Committee, 1987;Wenning et al., 2004). Cognition was assessed in all subjects using the MMSE (Folstein, Folstein, & McHugh, 1975).

| Superficial white matter
In brief, T1-weighted images were processed using BrainSuite's cortical surface extraction pipeline (http://brainsuite.org/processing/ surfaceextraction/, v16), which produces surface models of the cerebral cortex from T1 MRI (Shattuck & Leahy, 2002). Next, the surfaces for each subject were registered to a reference atlas surface using BrainSuite's surface/volume registration software (SVReg; http:// brainsuite.org/processing/svreg/) (Joshi & Shattuck, 2012;Joshi, Shattuck, Thompson, & Leahy, 2004. Outputs from SVReg were inspected to ensure proper segmentation and surface/volume registration. As an additional quality control for the diffusion data, we used Region of interest labels from the SVReg output (specifically, the caudate, thalamus, putamen and cerebellum) were then applied to the mean diffusivity image. Finally, to allow cross-subject sampling of anatomically comparable superficial white matter mean diffusivity, diffusivity was sampled along each vertex of the white matter surface that had been mapped to the atlas reference via SVReg. In order to extract the whole brain superficial white matter mean diffusivity for each subject, the diffusivity value at each vertex was then extracted and the mean value from across these vertices was calculated.

| Deep white matter
FMRIB Software Library (FSL-TBSS) was run (Smith et al., 2006). In short, data were subjected to eddy current and motion correction, followed by brain extraction. Dtifit was used to generate mean diffusivity, fractional anisotropy (FA), radial diffusivity, and axial diffusivity.
A study-specific mean FA skeleton representing the centers of all white matter tracts was generated and used to transform all subjects to MNI152 space. The white matter skeleton was thresholded at 0.2.
The mean FA skeleton was then binarized and applied to each subject's mean diffusivity image, which allowed the extraction of the whole brain mean diffusivity value.

| Selection of diffusion parameter
The parameter chosen to be the focus of this study was mean diffusivity, which is the mean of the three eigenvalues and corresponds to the molecular diffusion rate (lower values mean low diffusivity) (Soares, Marques, Alves, & Sousa, 2013). Mean diffusivity was chosen rather than other diffusion metrics such as FA and axial/radial diffusivity for a number of reasons. One, to reduce the number of tests.
Two, mean diffusivity has been demonstrated to be a sensitive biomarker in both the deep white matter and in the superficial white matter Narr et al., 2009;Phillips, Joshi, Squitieri, et al., 2016b). Three, the difference between axial and radial diffusivity values is considerably smaller in the superficial white matter than in the deep white matter (Phillips et al., 2013;Phillips, Joshi, Piras, et al., 2016;Phillips, Joshi, Squitieri, et al., 2016b). This reduces the usefulness of individually investigating these measures and FA.
Mean diffusivity is on average lower in the deep white which reflects the restricted diffusion and generally greater coherence of white matter connections in a particular voxel while the mean diffusivity is on average higher in the superficial white matter which reflects the greater diffusion and generally increased tissue complexity within a particular voxel. The mean diffusivity within the superficial white matter is higher because it contains cortico-cortical short-range connections, long-range axonal projections, as well as a high proportion of "interstitial neurons." However, volumetrically, the superficial white matter is largely composed of myelin and oligodendrocytes Reveley et al., 2015;Suárez-Solá et al., 2009

| Correlations between whole brain diffusivity and clinical measures
To investigate whether superficial and deep white matter changes in MSA patients were related to UMSARS-II and MMSE, we ran a partial correlation analysis within SPSS between these measures and whole brain superficial and deep white matter mean diffusivity with sex and age as covariates. Based on robust a priori hypotheses about directionality of these correlations (given the prerequisite of previously established group differences), correlations were performed one tailed. Levels of significance were set at alpha = 0.05.

| Exploratory vertex-and voxel-based analysis
For increased spatial resolution, we applied an exploratory vertex In order to examine the specific location of white matter changes in the brain exploratory vertex and voxel-wise analyses were conducted. They revealed that mean diffusivity levels were increased in the MSA group throughout the brain, as can be observed in Figure 2 for the superficial white matter and Figure

| White matter differences in PD versus controls
In contrast to the marked increase in whole brain white matter mean diffusivity observed in MSA patients compared to controls, this difference was not observed for PD patients (Figure 1). Vertex-and voxelbased analyses exploring further potential regional effects confirmed this finding (results not shown).

| Relationships between white matter mean diffusivity and clinical variables in MSA
Correlation analyses explored whether the increased whole brain white matter mean diffusivity observed in MSA patients was associated with their clinical profile. As shown in Figure 4, whole brain white matter mean diffusivity was significantly correlated with both motor and cognitive functions. More specifically, scores on the UMSARS motor subscale were positively correlated with whole brain deep and superficial white matter mean diffusivity (r = .53, p < .001 and r = .39, p = .03, respectively). The MMSE scores were negatively correlated with whole brain superficial

| Widespread white matter abnormalities in MSA
Our results suggest that there are pervasive white matter abnormalities in MSA. First, large whole brain differences in mean diffusivity were observed between MSA patients and controls in both deep matter and superficial white matter ( Figure 1). These measures were calculated using two different methodological approaches and these dual observations both confirmed the widespread distribution of the white matter abnormalities over the brain. At the vertex and voxel levels, abnormalities were particularly prominent in white matter in the primary motor cortex, cerebellar peduncles, cerebellum, and the frontal lobe, which are among the brain regions known to be GCI rich in MSA (Papp et al., 1989). However, abnormalities were not limited to those regions, just as neuropathological studies have reported GCIs beyond those regions as well (Papp et al., 1989). Second, whole brain diffusiv-  shown to be associated with cognitive function in Alzheimer's disease (Phillips et al., 2013) Huntington's disease (Phillips, Joshi, Squitieri, et al., 2016a) and anti-N-methyl-D-aspartate receptor encephalitis (Phillips et al., 2018), which indicates that it may be a sensitive marker of broad nonspecific cognitive dysfunction.
The above group differences and correlations converge to suggest that abnormal mean diffusivity in the superficial and deep white matter are equally pathognomic to MSA. We were interested in both types of white matter tissue for the sake of providing a spatially comprehensive framework, but also because of their distinct cellular speci- white matter across neuropathologies but also in the healthy brain.

| In vivo MRI in MSA: Interpretation of results and prior literature
The results provide support for our primary hypothesis that MSA would be characterized by widespread white matter abnormalities.
in vivo MRI is a limited modality at uncovering the precise nature of individual cellular processes compared to intensive lab benchwork, such as immunochemistry, cell-based or postmortem cell staining which has been used to identify the distribution of GCIs in the brain (Papp et al., 1989;Papp & Lantos, 1994). However, DTI is a practical methodology to achieve a broad in vivo overview of the white matter.
This quality of MRI makes it ideally suited to investigate our hypothesis since white matter is made up primarily of oligodendrocytes, and white matter microstructure can be assessed based on directionality of the overall movement of water molecules as measured by diffusion MRI. Moreover, ex vivo DTI studies coupled with histological assessments have described some of the processes which may underlie changes in DTI parameters, including demyelination (Yano et al., 2017).
White matter microstructure abnormalities have been previously documented in MSA using DTI (Table S1). Most of these prior investigations focused their investigations on predefined brain regions, although more recent work has also reported widespread abnormalities using TBSS (Hara et al., 2018;Rulseh et al., 2016). This research has been helpful in identifying particularly vulnerable regions such as the cerebellar peduncles, pons, and corpus callosum, consistent with the clinical manifestation of MSA (Hara et al., 2018;Worker et al., 2014;Zanigni et al., 2017). Though the previous focal approaches are clearly informative and have like our own study demonstrated abnormalities in white matter in MSA (Table S1), the novelty of our findings lies in the whole brain topography of these white matter abnormalities, which appear more pervasive than previously thought. It is possible that white matter abnormalities in MSA are initially highly localized in vulnerable brain regions and then expand as the disease evolves. An alternative explanation is that the white matter abnormalities are persistent from the beginning, and that these become increasingly severe as the disease evolves. With our current design, we are not able to determine the origins of this pervasiveness.

| Specificity of white matter damage: MSA versus PD
Our second hypothesis was that widespread white matter abnormalities would be specific to MSA. The underlying rationale was that MSA is unique in that the accumulation of α-syn occurs in oligodendrocytes, whereas in PD, α-syn accumulates in neurons. As expected, no global white matter damage was observed in PD compared to controls. These findings are in line with recent well-powered DTI studies suggesting that the white matter is largely intact in PD, although focal abnormalities, particularly in the substantia nigra, have been consis-  (Bassil et al., 2017;Nykjaer et al., 2017;Yazawa et al., 2005), which would be expected to compromise the integrity of white matter microstructure. Changes in oligodendroglial morphology and axonal myelin have been observed even before GCIs are developed (Song et al., 2007). Moreover, the widespread nature of our diffusion results largely mirror the distribution of GCIs found in postmortem studies, where GCIs have been found in the deep white matter fibers of the putamen, the internal capsule, external capsule, corpus callosum, anterior commissure, corticopontine tract, pyramidal tract, cerebellar white matter, and the superficial white matter (neocortex cortical layers five and six) and the frontal lobe (primary motor, premotor areas, and supplementary motor cortical area) (Papp et al., 1989;Papp & Lantos, 1994).
For example, transgenic mice that overexpress wild-type human α-syn specifically in oligodendrocytes develop GCI-like α-syn deposits and exhibit loss of oligodendrocytes and neurons (Yazawa et al., 2005).
However, it has been suggested that there is widespread oligodendrocyte dysfunction early or even initially during MSA pathogenesis, prior to cell loss . Recent neuropathological evidence collected in MSA patients comparable to our own in terms of age and disease duration shows largely preserved oligodendrocyte numbers in the presence of significant microgliosis, astrocytosis and neuronal loss, both sub-cortically (Salvesen et al., 2015a) and in the neocortex (Salvesen et al., 2015b). The present investigation examined the white matter supporting the deep subcortical and cortical gray matter where these widespread neuronal, astrocytes and microglia abnormalities were described.
There are other neurobiological events co-occurring alongside the putative oligodendrocyte dysfunction which may further contribute to white matter abnormalities in MSA patients. First, astrocytic and microglial activation has been previously reported in MSA (Vieira, Radford, Chung, Guillemin, & Pountney, 2015).

| Limitations
The first limitation is the relatively reduced sample size, which results from the challenge of recruiting patients with this rare disease.  (Phillips et al., 2013). A further limitation is that with our current design we cannot determine to what extent the pervasive abnormalities we observe in MSA patients are focal to begin with and progressively expand throughout the brain.
Future work needs to address these questions using early and late stage MSA patients, and reliable markers of disease progression.

| CONCLUSIONS
We find strong evidence for pervasive white matter abnormalities in MSA, but not PD, suggesting that white matter dysfunction may be more central to MSA than previously thought. The fact that white matter abnormalities were specific to MSA raises the possibility that these are related to the presence of GCIs, which are pathognomic to the disease. The link between white matter integrity in MSA and oligodendrocyte (dys)function has been hitherto largely neglected in clinical research. We discuss some of the mechanisms which likely underlie the observed white matter abnormalities in MSA in the light of existing neuropathological and experimental evidence. Further studies are needed to definitely confirm that white matter changes are driven by GCIs and their downstream effects and examine potential relationships with concomitant neurodegeneration.

ACKNOWLEDGMENTS
We thank the patients for participating in the study, the Centre