[1] 42w This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as

[1] 244w Background:The immunological pathophysiologies of chronic inflammatory demyelinating polyneuropathy (CIDP) and multifocal motor neuropathy (MMN) differ considerably, but neither has been elucidated completely. Quantitative magnetic resonance imaging (MRI) techniques as diffusion tensor imaging, T2 mapping, and fat fraction analysis may indicate in vivo pathophysiological changes in nerve architecture. Our study aims to systematically study nerve architecture of the brachial plexus in patients with CIDP, MMN, motor neuron disease (MND) and healthy controls using these quantitative MRI techniques. Methods:We enrolled patients with CIDP (n = 47), MMN (n = 29), MND (n = 40) and healthy controls (n = 10). All patients underwent MRI of the brachial plexus and we obtained diffusion parameters, T2 relaxation times and fat fraction using an automated processing pipeline. We compared these parameters between groups using a univariate general linear model. Results:Fractional anisotropy was lower in patients with CIDP compared to healthy controls (p < 0.001), patients with MND (p = 0.010) and MMN (p < 0.001). Radial diffusivity was higher in patients with CIDP compared to healthy controls (p = 0.015) and patients with MND (p = 0.001) and MMN (p < 0.001). T2 relaxation time was elevated in patients with CIDP compared to patients with MND (p = 0.023). Fat fraction was lower in patients with CIDP and MMN compared to patients with MND (both p < 0.001). Conclusions:Our results show that quantitative MRI parameters differ between CIDP, MMN and MND, which may reflect differences in underlying pathophysiological mechanisms.

[1] 63w Inflammation of peripheral nerves is the underlying disease mechanism that causes muscle weakness and sensory deficits in chronic inflammatory polyneuropathies, including multifocal motor neuropathy (MMN) and chronic inflammatory demyelinating polyneuropathy (CIDP). MMN is characterized by asymmetric weakness without sensory deficits that dominates in the arms, while CIDP may cause pure motor, pure sensory, or mixed deficits that are most pronounced in the legs.

[2] 123w Nerve conduction studies may show conduction blocks and nerve imaging studies have revealed multifocal thickening of nerves in both CIDP and MMN. 1,2 The immunological pathophysiologies of CIDP and MMN differ considerably, but have not been elucidated completely. Autopsy studies, sural nerve biopsy, immunostaining with patient sera in vitro and animal models have provided insight in underlying immunological mechanims. [3][4][5][6][7][8][9] There is an obvious need for additional tools to study the condition of peripheral nerves in vivo to further dissect the underlying pathophysiological mechanisms. Quantitative magnetic resonance imaging (MRI) may bridge this gap: diffusion tensor imaging (DTI) and measurements of T2 relaxation times and fat fraction may indicate specific pathophysiological changes in the myelin sheath and axon in patients with CIDP and MMN.

[3] 144w DTI is an MRI technique that provides quantitative parameters as fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD) and radial diffusivity (RD). These parameters give insight in the microstructural integrity of (nervous) tissue and seem to correlate with histological findings. [10][11][12][13] Previous DTI studies evaluated peripheral nerves such as the tibial, sciatic and median nerve of patients with CIDP or MMN and healthy controls. [14][15][16][17][18][19] The brachial plexus was analyzed in a recent exploratory pilot study in a small cohort of patients and showed different FA values between patients with CIDP and MMN. 20 Studies using other quantitative MRI techniques, e.g. T2 mapping or fat fraction analysis, documented an increase in T2 relaxation time in the brachial and lumbosacral plexus and in the tibial nerve in small cohorts of patients with CIDP. [21][22][23] Complete and systematic studies of the brachial plexus are lacking.

[4] 60w We therefore performed a detailed and systematic quantitative MRI study in a large cohort of patients with CIDP, MMN, motor neuron disease (MND) and healthy controls, to compare diffusion parameters, T2 relaxation times and fat fraction of the brachial plexus. The aim of this study was to interpret these results in light of underlying pathophysiological mechanisms of CIDP and MMN.

[5] 9w This article is protected by copyright. All rights reserved

[1] 45w We processed all DTI data semi-automatically, using a two-step custom-build processing pipeline based on the diffusion toolbox ExploreDTI which allows visualization of the spinal nerve roots, segmented tract analysis and extraction of diffusion parameters. 27 An overview of the pipeline is shown in figure 1.

[2] 151w Before processing, we resampled the 3D TSE SPIR to a 2x2x2 mm 3 isotropic resolution. Subsequently, we manually drew a rough mask of the brachial plexus area using ITK SNAP (10 minutes per data set). 28 These masks were drawn in the resampled 3D TSE SPIR and in the diffusion-weighted image to guide the registration and fiber tract selection (figure 1B). The first automated part of the processing pipeline comprised data denoising, affine registration to correct for subject motion and eddy currents, b-spline registration to correct for echo-planar imaging distortions, tensor estimation using an iterative weighted linear least squares algorithm and whole volume fiber tractography (seed point resolution 1x1x1 mm 3 , step size 1 mm, seed FA threshold 0.15, FA track range 0.1-0.8, fiber length range 20-200 mm, angle threshold 15 o per step; figure 1C). This first automated processing step required approximately 35 minutes per data set to complete.

[3] 86w Next, we manually defined slices with starting and ending points of tracts. Starting points were located next to the ganglion of nerve root C5, C6 and C7, ending points were located 5 slices further in the distal direction (5 minutes per data set). This aided a second algorithm to find all tract locations of the nerves using a tract density map (figure 1D) and specifies the appropriate region of interests (ROI's) for nerve segmentation (figure 1E). To pair ROI's in the proximal starting and distal ending

[4] 112w We enrolled 137 participants based on the in-and exclusion criteria. We had to exclude 11 more patients: 2 patients (1.5%) because of claustrophobia during scanning, 2 patients (1.5%) because of a changed diagnosis after inclusion, 1 patient (0.7%) due to movement artifacts that led to low data quality and an additional 6 patients (4.4%; CIDP=3, MMN=1, controls=2) after processing and tract segmentation due to insufficient data quality. We used data from 126 study participants for further analysis (CIDP=47, MMN=29, MND=40 (ALS=19, PMA=21), healthy controls=10). The patient characteristics are summarized in table 2. Patients with MMN were younger than patients with CIDP and PMA (p<0.001). Other characteristics did not differ significantly between groups.

[5] 41w After automated processing, we identified 92.9% of all nerve roots (93.3%, 98.4% and 86.9% for C5, C6 and C7 respectively), which increased to 96.0% of C5 nerve roots, 99.6% of C6 nerve roots and 95.6% of C7 after additional manual adjustments.

[6] 50w Diffusion parameters, T2 relaxation times and fat fraction did not differ between right and left side of the nerve roots, except for FA in nerve root C7 (p<0.001). This only significant finding did not influence our data and we therefore decided to combine right and left sides in further analysis.

[7] 41w The same applies to patients with ALS and PMA. There were no significant differences in characteristics between patients with ALS and PMA (p values ranged from 0.075-0.999). We therefore present these data as one group of patients (MND) in further analysis.

[8] 9w This article is protected by copyright. All rights reserved

[9] 71w The means of all quantitative parameters are summarized in table 3 and visualized in figure 2. We found a lower mean FA in patients with CIDP (0.27 (standard deviation (SD) 0.05)) compared to healthy controls (0.30 (SD 0.05); p<0.001), patients with MND (0.28 (SD 0.04); p=0.010) and MMN (0.30 (SD 0.06); p<0.001). FA in patients with MMN and healthy controls was higher compared to patients with MND (p=0.002 and p=0.038, respectively).

[10] 40w We found a higher mean MD in patients with CIDP (1.40x10 -3 mm 2 /s (SD 0.20)) compared to patients with MND (1.35x10 -3 mm 2 /s (SD 0.20); p=0.008) and MMN (1.35x10 -3 mm 2 /s (SD 0.23); p=0.027).

[11] 46w Also, the mean RD was higher in patients with CIDP (1.20x10 -3 mm 2 /s (SD 0.19)) compared to healthy controls (1.12 (SD 0.17); p=0.015), patients with MND (1.14x10 -3 mm 2 /s (SD 0.16); p=0.001) and MMN (1.13x10 -3 mm 2 /s (SD 0.20); p<0.001).

[12] 11w We did not find any significant differences in AD between groups.

[13] 43w For all four study groups (CIDP, MMN, MND, and healthy controls) we only found weak correlations between the MRI metrics, i.e. diffusion parameters, T2 relaxation time and fat fraction, and the clinical covariates, i.e. age, MRC sum score and disease duration (figure 3).

[14] 73w For the correlation with age, r ranged from -0.19-0.14 for patients with CIDP, -0.35-0.31 for patients with MMN, -0.20-0.34 for patients with MND, and -0.18-0.24 for healthy controls. For MRC sum score, r ranged from -0.23-0.10 for patients with CIDP, -0.08-0.24 for patients with MMN, and -0.13-0.20 for patients with MND. For disease duration, r ranged from -0.24-0.25 for patients with CIDP; -0.21-0.23 for patients with MMN, and 0.02-0.23 for patients with MND.

[15] 9w This article is protected by copyright. All rights reserved

[1] 436w With this study, we show that quantitative MRI techniques reveal differences in the brachial plexus between patients with CIDP, MMN, MND and healthy controls. CIDP is characterized by lower FA and higher RD than MMN, MND and healthy controls, whilst MMN is characterized by higher FA values than CIDP and MND. Patients with MMN and healthy controls did not differ. These differences between CIDP and MMN are the most remarkable and important finding as they emphasize important differences in the underlying pathophysiologies. This is the first comparative quantitative MRI study in a relatively large cohort of patients with CIDP and MMN. Diffusion parameters obtained from the sciatic, tibial, median, ulnar and radial nerves were previously reported in smaller cohorts. [14][15][16][17]19 The absolute differences of the measured parameters are around 2% between study groups, which indicates that differences are probably only found in larger groups. However, the finding of a decreased FA and an increased RD in our study patients with CIDP is in agreement with previous findings, indicating that this DTI profile is characteristic for CIDP and can be found throughout the peripheral nervous system. [14][15][16][17]19 Experimental animal studies showed that increased RD may correspond with loss of myelin integrity. [10][11][12] The combination of decreased FA and increased RD has also been reported in patients with Guillain-Barré syndrome (GBS) and demyelinating types of Charcot-Marie-Tooth (CMT), which corroborates that this reflects the disturbance of myelin integrity in peripheral nerves caused by inflammation. [32][33][34][35] The longer T2 relaxation times and lower fat fraction in the CIDP cohort compared to the MND cohort indicate the presence of free water, which may also be due to inflammation, and have also been reported at the lumbosacral plexus. 21,22,36 Although decreased FA in combination with increased RD is a robust finding in patients with CIDP, absolute diffusion values differ between proximal and distal nerve sites. [14][15][16][17]19 This is probably explained by the proximal to distal decrease in the diameter of fascicles with a corresponding increase in the density of the perifascicular connective tissue. 37 In the well-organized tissues of the distal peripheral nerves, water molecule movement is more restricted in specific directions, which results in larger isotropic diffusion and a higher FA. The FA values of the brachial plexus in our study were lower than in previous studies of distal peripheral nerves in arms and legs, which is in line with this hypothesis. [14][15][16][17]19 FA and MD are summary measures from eigenvalues. Changes in FA and MD are therefore driven by changes in AD or RD. In CIDP, the increase in RD seems to drive the changes in FA and MD. RD

[2] 127w This article is protected by copyright. All rights reserved indicates less hindrance of diffusion for water perpendicular to the nervous tissue. This can be the result of various cellular mechanisms, such as demyelination or a disturbance of the cytoskeleton caused by a loss of neurofilaments and microtubules. We think that our findings may reflect demyelination rather than a disturbance of the cytoskeleton as histological studies reported myelin detachment and myelin loss without damage to axons induced by macrophages around the (inter)nodal regions in patients with CIDP. [3][4][5][38][39][40][41] The mechanism of paranodal myelin detachment is present in some patients with CIDP, as described in earlier electrophysiological studies. [42][43][44] Taken together, the changes in FA, MD and RD in our CIDP group most likely reflect the loss of myelin.

[3] 144w The absence of increased RD values in patients with MMN indicates that the underlying pathophysiological mechanism is different from that in CIDP and that demyelination is probably not the dominant pathophysiological process. Patients with MMN seem to have comparable quantitative MRI parameters as healthy controls. Scarce histological reports describe normal myelin sheets. [6][7][8] Electrophysiological studies may support the idea of changed axon structure with largely intact myelin sheets. 44 However, it is rather remarkable that such different DTI profiles are found between patients with MMN and CIDP while diagnostic tools used in clinical practice, such as nerve conduction studies, nerve ultrasound and qualitative MRI of the brachial plexus may show similar abnormalities, e.g. conduction blocks and thickening of the nerves. The differences in DTI profiles indicate that these abnormalities are more likely to present common endpoints of different pathophysiological mechanisms rather than comparable etiologies.

[4] 107w In a previous study we found lower AD in the median and ulnar nerves in the forearm of patients with MMN compared to healthy controls and patients with ALS. 18 It is assumed that AD correlates with axonal loss, e.g. due to axonal swelling due to the breakdown or change in the permeability of the axolemma, which is an important feature of MMN. 11,45,46 We did not detect differences in AD between groups at the brachial plexus in this study, which can be explained by the fact that longer axons and distal parts of axons are more susceptible to injury than short and proximal parts of axons.

[5] 82w Consequently, AD may remain relatively unchanged in the proximal spinal nerve roots of the brachial plexus. In the previous study we did not find a significant difference in FA between patients with MMN and ALS, although absolute values of FA were higher in patients with MMN. 18 We found a significantly higher FA in patients with MMN compared to patients with MND in this study, which can be explained by the larger sample size and higher statical power in the current study.

[6] 13w Correlations between clinical data and quantitative MRI parameters were weak. We refrained from

[7] 325w This article is protected by copyright. All rights reserved studying correlations of nerve conduction studies and imaging results since the measurement sites did not match. More in general, imaging and electrophysiological studies may reveal different pathophysiological dimensions. Previous studies found that imaging results did not correlate with nerve conduction study results in cohorts of patients with inflammatory neuropathies. [47][48][49][50][51] A limitation of our study is the effect of partial volume, which may lead to an underestimation of diffusion parameters and fat fraction, and varying SNR which may lead to higher FA and lower RD in case of lower SNRs. 52 However, the influence of partial volume effects and different SNR values were probably small as our results in DTI analysis, T2 mapping and fat fraction analysis are consistent with each other, and scans were performed in random order with the same software versions. Another limitation might be the registration step in the processing pipeline. Due to an imperfect registration some tracts were not or incompletely found, particularly in nerve root C7 due to strong susceptibility artifacts caused by the lungs. Our healthy control group is small but we think the number of healthy controls is sufficient as standard deviations of the means of the quantitative MRI parameters were small and comparable to the other three study groups, indicating low levels of variation between individuals. Moreover, the diffusion parameters that we observed were similar to those previously reported in literature. 53 We analyzed relatively short segments of the brachial plexus, since analysis of longer tracts resulted in a significant dropout of data due to poor data quality. We therefore decided to only analyze the first centimeter next to the ganglion in order to maximize the number of datasets. Although we could not include the more distal parts of the brachial plexus, the advantage of this approach is a well-powered study that provides information on a large patient population derived with an automated pipeline without subjective bias.

[8] 244w In conclusion, our study gives insight into the nerve architecture of the brachial plexus in a relatively large cohort of patients with CIDP, MMN, MND and healthy controls. Our study shows that diffusion parameters differ between CIDP and MMN, which may reflect differences in the underlying pathophysiological mechanisms. Future studies should combine assessments of the brachial plexus and distal nerves and assess correlations between quantitative MRI parameters in roots, fascicles and peripheral nerves and specific clinical deficits. They should also address whether changes occur in disease course or after treatment. A diffusion-weighted image and a resampled 3D TSE SPIR are obtained (A, upper and lower image respectively). After manually drawn masks of the brachial plexus area (B) the automatic processing pipeline results in whole volume fiber tractography (C). Nerve locations are found in a tract density map (D) which specifies region of interests (E). A connectivity analysis results in reconstruction of nerve roots (F) and subsequently in nerve root segments from which diffusion parameters are derived (G). Boxplots of diffusion parameters, T2 relaxation times and fat fraction with grand mean are shown. MD, AD and RD values are x10 -3 mm 2 /s, T2 relaxation time is in milliseconds and fat fraction is a percentage. Abbreviations: FA = fractional anisotropy; MD = mean diffusivity; AD = axial diffusivity; RD = radial diffusivity; mm = millimeters; s = seconds; CIDP = chronic inflammatory demyelinating polyneuropathy; MMN = multifocal motor neuropathy; MND = motor neuron disease.

[9] 9w This article is protected by copyright. All rights reserved

[10] 70w Table legends Table 1. MRI parameters Table 2. Patient characteristics *Age differs significantly between patients with MMN and patients with CIDP, and between patients with MMN and patients with PMA. Age, disease duration and MRC sum score are mean. Abbreviations: CIDP = chronic inflammatory demyelinating polyneuropathy; MMN = multifocal motor neuropathy; ALS = amyotrophic lateral sclerosis; PMA = progressive muscular atrophy; SD = standard deviation; MRC = Medical Research Council.

[11] 12w Table 3. Quantitative MRI parameters per study group and per nerve root.

[12] 19w This article is protected by copyright. All rights reserved Article FA AD RD MD Fat fraction T2 relaxation time

[1] 45w We performed a cross-sectional study in patients with CIDP, MMN, MND and healthy controls. We performed quantitative MRI in all patients and used an automated processing pipeline to obtain parameters on microstructural integrity. We compared these parameters between groups and explored correlations with clinical data.

[2] 9w This article is protected by copyright. All rights reserved

[3] 122w We performed DTI in a transversal slice orientation to obtain diffusion parameters, T2 mapping in a coronal slice orientation to obtain T2 relaxation times and T1 Dixon in a transverse slice orientation to obtain fat fraction. As an anatomical reference we used a 3D turbo spin-echo (TSE) spectral presaturation with inversion recovery (SPIR) sequence in a coronal slice orientation. The acquisition parameters are shown in Table 1. We performed a data quality check after enrollment of 43 participants that showed a higher-than-expected frequency (> 5%) of insufficient data due to low signal to noise ratios (SNR). Therefore, we performed DTI twice in all the following participants to improve data quality. These two acquisitions were combined in a later stage during data processing.

[4] 19w We exclude scans with low quality, for example due to movement or the presence of artifacts, from further processing.

[5] 106w This article is protected by copyright. All rights reserved slices, the algorithm performs a connectivity analysis for all defined ROI's. Every pair of ROIs with high connectivity is then defined as tract bundles which results in a reconstruction of the nerve roots (figure 1F). Subsequently, the nerve root segments were constructed, using the predefined starting and ending slice (figure 1G). These nerve root segments were used to standardize the site of extraction of diffusion parameters (FA, MD, AD, RD), i.e. next to the ganglion over a distance of 1 cm. This second automated part of the pipeline required approximately 5 minutes per data set to complete.

[6] 76w Finally, we visually identified and labeled the selected tracts as the left and right nerve roots of C5, C6, or C7 (5 minutes per data set). If necessary, manual ROI's were placed to optimize the result of the automated data processing (5 minutes per data set). When no tracts were found, nor with the algorithm, nor manually, the data set was excluded from further analysis. Finally, diffusion parameters per fiber tract were calculated using tract-based analysis.

[7] 15w For statistical analysis we used IBM SPSS Statistics (Version 25, Armonk, New York, United States).

[8] 106w To compare patient characteristics, we used one-way analysis of variance (ANOVA) for numerical data and a Chi-squared test for categorical data. We compared diffusion parameters, T2 relaxation times, and fat fraction per side (i.e. right/left) using a paired sample t test and corrected for multiple testing using the Bonferroni method. To analyze diffusion parameters, T2 relaxation times, and fat fraction between groups we used an univariate general linear model with the MRI parameters as the dependent variable and the study group as a fixed factor. Tukey HSD was used to correct for multiple testing. A p value <0.05 was considered significant. We analyzed diffusion parameters, T2

[9] 65w This article is protected by copyright. All rights reserved relaxation times, and fat fraction of all nerve roots together and per nerve root (i.e. C5, C6, C7) separately. Correlations between the quantitative parameters and clinical data were analyzed using the Pearson correlation coefficient r. We considered r≤0.35 as a weak correlation, 0.36-0.70 as moderate, 0.70-0.89 as high and ≥0.90 as a very high correlation. 31

[10] 9w This article is protected by copyright. All rights reserved

[1] 145w Consecutive patients with CIDP, MMN and MND were included at the outpatient clinic of the University Medical Center Utrecht (UMCU). Alle prevalent and incident patients with an established diagnosis of CIDP or MMN (definite, probable, possible), according to the predefined consensus criteria of the European Federation of Neurological Societies/Peripheral Nerve Society, were eligible for inclusion. 2,24,25 Patients with MND (i.e. amyotrophic lateral sclerosis (ALS) or progressive muscular atrophy (PMA)), according to the Brooks criteria, were enrolled as disease controls. 26 Healthy controls were included if they had no history of neuromuscular disorders, neuropathy, nerve root injuries or other cervical spine disorders. We excluded patients aged <18 years, patients with atypical forms of CIDP (e.g. Lewis Sumner Syndrome) and patients with MND that had a bulbar onset of symptoms to minimalize heterogeneity in these groups, and participants who met one of the routine contraindications for MRI.

[2] 94w We documented demographic and clinical data from all patients, including muscle strength expressed as a Medical Research Council (MRC) sum score. We tested the following 12 muscle groups on both sides: finger flexion, finger extension, finger abduction, wrist flexion, wrist extension, elbow flexion, elbow extension, shoulder abduction, hip flexion, knee flexion, knee extension and foot dorsiflexion. We calculated MRC sum scores of these 24 measurements, ranging from 0 to 120 (normal). The medical ethical committee of the UMCU approved this study (18-349/NL 62866.041.17). This study conforms with the World Medical Association Declaration of Helsinki.

[3] 9w Written informed consent was obtained from all study participants.

[4] 30w All participants underwent an MRI scan of the brachial plexus bilaterally in supine position on a 3.0 Tesla MRI scanner (Philips Healthcare, Best, the Netherlands) using a 24-channel head-neck coil.

[5] 117w Dixon fat fraction maps were calculated using the water and fat image reconstructions of the vendor software. The data obtained with T2 mapping was processed using an extended phase graph fitting approach considering inhomogeneous B 1 +. 29,30 This method accounts for different T2 relaxation times for the water and fat component with the T2 of the fat component fixed to a value calibrated on the subcutaneous fat. Quantitative values of the T2 mapping and T1 DIXON were obtained using the same tract-based analysis used for DTI data. Data underwent registration to the same anatomical space (3D TSE SPIR image) as the DTI data. We obtained T2 relaxation time in milliseconds (ms) and fat fraction in percentages.

[6] 53w We found a longer mean T2 relaxation time in patients with CIDP (42.37 ms (SD 5.36)) compared to patients with MND (41.02 ms (SD 4.81); p=0.023). The fat fraction was lower in patients with CIDP (40.09% (SD 9.61); p<0.001) and MMN (39.44% (SD 9.07); p< 0.001) compared to patients with MND (43.62% (9.74)).