Early red nucleus atrophy in relapse‐onset multiple sclerosis

Abstract No study has investigated red nucleus (RN) atrophy in multiple sclerosis (MS) despite cerebellum and its connections are elective sites of MS‐related pathology. In this study, we explore RN atrophy in early MS phases and its association with cerebellar damage (focal lesions and atrophy) and physical disability. Thirty‐seven relapse‐onset MS (RMS) patients having mean age of 35.6 ± 8.5 (18–56) years and mean disease duration of 1.1 ± 1.5 (0–5) years, and 36 age‐ and sex‐matched healthy controls (HC) were studied. Cerebellar and RN lesions and volumes were analyzed on 3 T‐MRI images. RMS did not differ from HC in cerebellar lobe volumes but significantly differed in both right (107.84 ± 13.95 mm3 vs. 99.37 ± 11.53 mm3, p = .019) and left (109.71 ± 14.94 mm3 vs. 100.47 ± 15.78 mm3, p = .020) RN volumes. Cerebellar white matter lesion volume (WMLV) inversely correlated with both right and left RN volumes (r = −.333, p = .004 and r = −.298, p = .010, respectively), while no correlation was detected between RN volumes and mean cortical thickness, cerebellar gray matter lesion volume, and supratentorial WMLV (right RN: r = −.147, p = .216; left RN: r = −.153, p = .196). Right, but not left, RN volume inversely correlated with midbrain WMLV (r = −.310, p = .008), while no correlation was observed between whole brainstem WMLV and either RN volumes (right RN: r = −.164, p = .164; left RN: r = −.64, p = .588). Finally, left RN volume correlated with vermis VIIb (r = .297, p = .011) and right interposed nucleus (r = .249, p = .034) volumes. We observed RN atrophy in early RMS, likely resulting from anterograde axonal degeneration starting in cerebellar and midbrain WML. RN atrophy seems a promising marker of neurodegeneration and/or cerebellar damage in RMS.

images. RMS did not differ from HC in cerebellar lobe volumes but significantly differed in both right (107.84 ± 13.95 mm 3 vs. 99.37 ± 11.53 mm 3 , p = .019) and left The red nucleus (RN), a large neuronal structure located in the most rostral part of ventral midbrain, owes its name to its high iron content, which makes it clearly identifiable both in fresh tissue sections and in T2-weighted magnetic resonance imaging (MRI) sequences. According to its cytoarchitecture, the RN is divided in a rostral parvocellular part (pRN), accounting for up to 80% of the nucleus volume and receiving inputs primarily from the nucleus dentatus, and a caudal magnocellular part (mRN, 20%) receiving inputs primarily from nucleus interpositus and nucleus emboliform (Massion, 1988;Onodera & Hicks, 2009;ten Donkelaar, 1988). mRN and pRN have different functional roles. While mRN is involved in intra-and inter-limb coordination (Lavoie & Drew, 2002) and probably plays a role in compensating lesions of the corticospinal tract (Belhaj-Saif & Cheney, 2000;Siegel, Fink, Strittmatter, & Cafferty, 2015), pRN has been suggested to be involved in more complex functional networks, including those integrating cognitive-motor functions such as motor learning, error encoding, timing, and control of the ongoing movement (Habas, Guillevin, & Abanou, 2010;Lang et al., 2017;Reid et al., 2009). These observations further point out the association of specific cerebellar lobules with cognitive and affective functions (Koziol et al., 2014;Lazzarotto et al., 2020;Schmahmann & Sherman, 1997). Finally, worth of interest is also the possible role of cerebellar and RN damage in the early development of disabling multiple sclerosis (MS) symptoms such as fatigue and depression (Lazzarotto et al., 2020).
The cerebellum and the brainstem are major sites of pathology in MS and their involvement, although frequently asymptomatic, is known to be associated with a more severe prognosis (Damasceno, Von Glehn, Brandao, Damasceno, & Cendes, 2013;Tintore et al., 2010). Lesions damaging RN structure and connections may cause severe clinical picture, such as the rubral tremor (Koch, Mostert, Heersema, & De Keyser, 2007). Due to its major connections with the cerebellum, RN atrophy can result from trans-synaptic degeneration following cerebellar damage. Indeed, in a preclinical study in mutant mice, the loss of Purkinje cells was followed by the progressive atrophy of deep cerebellar nuclei due to anterograde trans-synaptic degeneration (Triarhou, Norton, & Ghetti, 1987).
Given these premises, we aimed to explore whether RN atrophy can be detected in the early disease phases, and whether this is associated with cerebral or cerebellar MRI metrics of white matter (WM) or gray matter (GM) damage, and early physical disability in relapse-onset MS (RMS) patients.
All patients that were diagnosed to have MS in the period March 2014 to August 2018, were selected according to the following criteria: (a) age range, 18-65 years; (b) clinical disease duration <5 years; (c) no history/evidence of neurologic or psychiatric disorders other than MS; (d) no history of alcohol or drug abuse; (e) good quality of MRI (i.e., we excluded those with movement artifacts).
Each participant gave written informed consent and the study was approved by the local Ethics Committee, according to the IV revision of Declaration of Helsinki.

| MRI data processing
Cerebellum and brainstem were identified and isolated automatically, and a mask was automatically generated by calculating the maximum of the probability of each voxel belonging to one of these structures. Each obtained mask was visually inspected. Next, the isolated cerebellum was aligned to the SUIT atlas template via an affine transformation (for the linear part of the normalization) and a nonlinear transformation using the diffeomorphic anatomical registration using Exponentiated Lie algebra (Avants et al., 2011). The individual cerebellum was therefore resliced in the atlas space, modulating in order to grant volume preservation (D'Ambrosio et al., 2017). Finally, by applying an inverse transformation matrix derived from the previous co-registration step, the SUIT atlas was aligned to the native subject space, and lobular volumes were calculated.
Two experienced investigators (MM, DP), blinded to clinical data, visually inspected all RN. Only in three cases the pipeline was re-run due to inconsistent segmentation.

| STATISTICAL ANALYSIS
Statistical analysis was performed using SPSS Statistics 23 (SPSS Chicago, Ilinois). A Shaphiro-Wilk test and visual analysis of histograms were used to test the normality of the data.
Student's t test was used to test group differences in term of age, χ 2 test was used to test possible differences in terms of sex.
Group differences in terms of cerebellar and RN volumes were tested via multivariate Generalized Linear Model (GLM), including ICV as a covariate of no interest to account for head size. Correlations between MRI metrics and clinical disability were explored with Spearman's rank correlation coefficient.
All p values were two-sided and considered statistically significant when p < .05. Since our study is exploratory, we did not adjust for multiple comparisons.

| DISCUSSION
Looking for early markers of neurodegeneration in RMS, we explored RN changes in a very early disease phase. Since the cerebellum and its connections are major sites of MS-related pathology (Parmar et al., 2018) and a pre-clinical study in Purkinje cell degeneration mutant mice disclosed trans-synaptic degeneration of deep cerebellar nuclei (Triarhou et al., 1987), our working hypothesis was that early RN atrophy could be the result of trans-synaptic anterograde axonal neurodegeneration starting in cerebellar WM and GM lesions.
We observed that RN volume in RMS, having very short disease duration (1.1 ± 1.5 years), was significantly lower compared with matched HC and correlated with cerebellar and midbrain WMLV, but not with supratentorial WMLV. Moreover, RN volume directly associated with the volume of some cerebellar lobules, but not with cerebellar GMLV. These observations suggest that RN atrophy could, at least partly, result from an axonal degeneration starting in cerebellum and midbrain WM lesions.
Our findings are particularly interesting on the light of previous observations in other neurological diseases. Indeed, RN volume was increased in patients with Parkinson's disease and significantly correlated with disease duration and severity, suggesting a compensatory activation of this nucleus (Colpan & Slavin, 2010), probably aimed at compensating the impaired function of the extra-pyramidal network.
Moreover, in nonhuman primates (Philippens, Wubben, Franke, Hofman, & Langermans, 2019) larger RN volume inversely associated with the severity of extra-pyramidal symptoms. Finally, a damaged striato-thalamo-cortical pathway was found to result in a compensatory increase in RN activation (Philippens et al., 2019). All these observations suggest a compensatory function of RN in the presence of an extra-pyramidal disorder.
The RN seems to play a compensatory role even in patients with severe corticospinal tract injury following ischaemic stroke. Indeed, elevated levels of neuronal activity were observed in the RN of the affected hemisphere, especially in the early phases following stroke, suggesting that RN could work as a compensatory structure to support the residual motor functions (Yeo & Jang, 2010). Microstructural RN changes, reflecting plastic and functional remodeling and associated with better motor function recovery, have been confirmed in other studies (Takenobu et al., 2014).
The above-summarized findings in Parkinson's disease and stroke strongly differ from our observations. Indeed, while direct pyramidal and extra-pyramidal damages are followed by a compensatory reaction of the RN, cerebellar damage associates with its atrophy. Thus, the initial RN atrophy observed in early RMS phases, may be considered a potential MRI marker of neurodegeneration following cerebellar damage and merits to be further investigated as negative prognostic factor. Indeed, the early NR atrophy might impair its potential functional compensatory role in case of corticospinal damage, that is, one of the major determinants of disability in MS.
Our data are in line with the increasing evidence of atrophy of deep gray matter nuclei in early RMS stages (Eshaghi et al., 2018), which indicates that this MRI metric may have a prognostic value since it associates with an increased risk of EDSS progression (Eshaghi et al., 2018) and cognitive decline in a relative shorter time (Riccitelli et al., 2011;Schoonheim et al., 2015). Interestingly, we also found an inverse association between RN atrophy and the cerebellar FS score of EDSS, suggesting that NR atrophy may be a specific marker of cerebellar damage rather than a generic marker of diffuse neurodegeneration.
The initial and significant, although slight correlations between RN atrophy and some cerebellar lobule atrophy indicates that the loss of volumes of these structures may proceed in parallel, probably as a Health, during the conduct of the study.

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