Marieke van Rosmalen

Magnetic resonance imaging to characterize chronic inflammatory neuropathies Marieke H.J. van Rosmalen

Magnetic resonance imaging to characterize chronic inflammatory neuropathies © Marieke H.J. van Rosmalen, 2021 ISBN: 978-94-6416-763-4 All rights reserved. No part of this thesis may be reproduced or transmitted in any form or by any means without written permission of the author. The Prinses Beatrix Spierziekten Foundation funded this work and publication of this thesis. Cover design and layout: © evelienjagtman.com Printing Ridderprint | www.ridderprint.nl

Magnetic resonance imaging to characterize chronic inflammatory neuropathies Het karakteriseren van chronische inflammatoire neuropathieën door middel van magnetic resonance imaging (MRI) (met een samenvatting in het Nederlands) Proefschrift ter verkrijging van de graad van doctor aan de Universiteit Utrecht op gezag van de rector magnificus, prof. dr. H.R.B.M Kummeling, ingevolge het besluit van het college voor promoties in het openbaar te verdedigen op donderdag 4 november 2021 des middags te 14.15 uur door Marieke Helena Johanna van Rosmalen geboren op 27 maart 1990 te Rosmalen

Promotoren prof. dr. W.L. van der Pol prof. dr. J. Hendrikse Copromotoren dr. H.S. Goedee dr. ir. M. Froeling The research described in this thesis was supported by a grant of the Prinses Beatrix Spierfonds (W.OR17-21). Financial support by the Prinses Beatrix Spierfonds for the publication of this thesis is gratefully acknowledged.

Voor papa en mama

CONTENTS Chapter 1 General introduction and thesis outline 9 Chapter 2 Clinical outcomes in multifocal motor neuropathy: a combined crosssectional and follow-up study 25 Chapter 3 Low inter-rater reliability of brachial plexus MRI in chronic inflammatory neuropathies 43 Chapter 4 Quantitative assessment of brachial plexus MRI for the diagnosis of chronic inflammatory neuropathies 55 Chapter 5 MRI of the intraspinal nerve roots in patients with chronic inflammatory neuropathies: abnormalities correlate with clinical phenotypes 75 Chapter 6 Quantitative MRI of the brachial plexus shows specific changes in nerve architecture in CIDP, MMN and motor neuron disease 91 Chapter 7 Quantitative MRI of the brachial plexus in CIDP and MMN: a longitudinal cohort study 111 Chapter 8 General discussion 127 Addendum Nederlandse samenvatting Review comité Dankwoord Publicatielijst Curriculum Vitae 157 165 167 173 175

Chapter 1 General introduction and thesis outline

General introduction and thesis outline 11 1 POLYNEUROPATHIES The peripheral nervous system consists of the alpha motor neurons in the anterior horn of the spinal cord, ventral and dorsal roots, peripheral nerves, the neuromuscular junction and muscle. Peripheral nerves connect the brain and spinal cord with muscle, skin, joints and sensory organs. Dysfunction of the peripheral nerves has many causes, but is generally described as ‘polyneuropathy’. Complaints caused by polyneuropathies include weakness and sensory deficits including numbness, sensory ataxia and changed or increased pain sensations in hands and feet. Common causes for polyneuropathy are summarized in Table 1.1 and include drugs and alcohol, diabetes, liver or renal insufficiency, (vitamin) deficiencies and idiopathic. Rare causes are of a genetic or inflammatory nature. Discriminating between these multiple causes is of great importance, as treatment opportunities and prognosis can vary between neuropathies. In the diagnostic work-up of patients suspected to have a polyneuropathy, laboratory findings and nerve conduction studies are important tools. Rare causes may be more elusive and require special diagnostic techniques, including genetic testing and nerve imaging by means of ultrasound or magnetic resonance imaging (MRI). Imaging techniques may be particularly helpful for the identification of the rare chronic inflammatory neuropathies. Table 1.1. Causes of polyneuropathy Type of origin Causes Carcinoma Lymphoma Hereditary Charcot-Marie-Tooth disease, hereditary neuropathy with liability to pressure palsies, neurofibromatosis Idiopathic Chronic idiopathic axonal polyneuropathy Infectious Leprosy, Lyme’s disease, HIV Inflammatory Chronic inflammatory demyelinating polyneuropathy (typical and variants), GuillainBarré syndrome, multifocal motor neuropathy Metabolic Diabetes mellitus, hypothyroidism, liver insufficiency, porphyria, renal insufficiency, vitamin deficiencies Paraneoplastic Small cell lung cancer Paraproteinemic anti-MAG associated polyneuropathy, IgM-monoclonal gammopathy of unknown significance, polyneuropathy organomegaly endocrinopathy M-protein and skin changes syndrome, Waldenström Systemic diseases Amyloidosis, rheumatoid arthritis, sarcoidosis, Sjögren’s syndrome, systemic lupus erythematosus Toxic Alcohol abuse, drug associated (antimicrobials, amiodarone, chemotherapy, digoxin, immunosuppressants), toxins (botulinum toxin, lead, mercury) Vasculitic Microscopic polyangiitis, non-systemic vasculitic neuropathy, polyarteriitis nodosa An overview of causes of peripheral polyneuropathy. Only some examples are shown per type of origin.

Chapter 1 12 CLINICAL BACKGROUND Chronic inflammatory neuropathies Chronic inflammatory demyelinating polyneuropathy (CIDP) and multifocal motor neuropathy (MMN) are both rare polyneuropathies with an inflammatory cause. CIDP is characterized by slowly progressive (mostly) symmetric pure motor, pure sensory, or mixed deficits that are most pronounced in the legs, while MMN is marked by asymmetric weakness without sensory deficits that dominates in the arms. Both polyneuropathies respond to treatment.1,2 Early treatment can improve muscle strength or sensory symptoms, and prevents progression of symptoms and permanent axonal damage which underlines the importance of a timely diagnosis.1,2 Patients with CIDP and MMN both respond to treatment with immunoglobulins. Patients with CIDP, but not with MMN, also respond to treatment with corticosteroids or plasmapheresis. Another important difference between these disorders is that 26% of patients with CIDP may experience remission that allows discontinuation of treatment, while this is uncommon in MMN.3 Diagnosis of CIDP and MMN is based on diagnostic consensus criteria that use a combination of clinical phenotype, nerve conduction study results and ancillary investigations.4,5 The latter play an important role when nerve conduction studies do not meet the required electrodiagnostic criteria.4–7 They include laboratory findings and imaging abnormalities of the peripheral nerves, in particular MRI of the brachial plexus. TECHNICAL BACKGROUND Principles of MRI physics Magnetic resonance imaging is an imaging technique that is able to visualize pathology of the nervous system. It is widely used in clinical practice for examination of the brain, spinal cord, muscle and more recently also the peripheral nervous system. Physics of MRI is complicated but some knowledge on its principles is essential to correctly assess and interpret the images. In short, all protons in body tissue spin on their own axes (Figure 1.1A). After placing the patient in a static magnetic field, i.e. the MRI scanner, the resulting magnetization of all protons inside the patients’ tissue align parallel to the magnetic field (Figure 1.1B). These protons rotate around the long axis of the primary magnetic field (B0), which is called precession. Precession rate is termed as the Larmor frequency. The average of many protons produces the net magnetization. Then, a radiofrequency pulse is emitted from the scanner which creates a magnetic field perpendicular to B0 (Figure 1.1C). When the radiofrequency pulse is at resonance, it creates a phase coherence in the precession of all the proton spins. The net magnetization of all protons rotating in Larmor frequency generates an electric current in the receiving coil, i.e. an electrical conductor, that is placed in the vicinity of the tissue of interest. This current is the nuclear magnetic resonance (NMR) signal. The NMR

General introduction and thesis outline 13 1 signal weakens due to two simultaneous relaxation processes that cause a loss of coherence of the spin system. The NMR signal decreases (loss of transverse magnetization or dephasing) with a time constant called the transverse relaxation time (T2, Figure 1.1D). Concurrently, but much slower, the vector relaxes towards its equilibrium position (recovery of longitudinal magnetization) parallel to the magnetic field: this time constant is called the spin-lattice relaxation time (T1, Figure 1.1E). Figure 1.1 Principles of MRI physics All protons in body tissue spin on their own axes (A). After placing the patient in a static magnetic field, i.e. the MRI scanner, the resulting magnetization of all protons align parallel (B) to the magnetic field (B0). The protons rotate around B0 at the Larmor frequency and the average of many protons produces the net magnetization. Then, a radiofrequency pulse (RF pulse) is emitted from the scanner which creates a magnetic field perpendicular to B0 and the net magnetization moves away from the z axis (C). As soon as the RF pulse is switched off, the protons begin to relax back to their equilibrium. The two main features of relaxation are dephasing of the spins or loss of transverse magnetization (T2 relaxation, D) and realignment along the z axis (T1 relaxation) as an umbrella closing up (E). Every type of body tissue has its own T1 and T2 relaxation times which results in different contrasts in the images. Adjustments in the MRI software enables the scanner to generate T1- or T2-weighted images of the tissues of interest. T1- and T2-weighted MRI provides qualitative information on the anatomical structures of interest. The generation of a T1-weighted image or a T2-weighted image depends on the set echo time (TE) and repetition time (TR) of the MRI sequence. T1-weighted images tend to have a short TE and a short TR. In T1-weighted images tissues that have a slow magnetization

Chapter 1 14 realignment appear dark as it does not retain signal (Figure 1.2A). T2-weighted images require a long TE and TR and highlight differences in the T2 relaxation times of tissues. Tissues with a longer T2 relaxation time will retain signal and appear bright (Figure 1.2B). MRI of the peripheral nervous system, e.g. the brachial or lumbosacral plexus, is often based on T2-weighted images. T2-weighted imaging with fat suppression (e.g. spectral presaturation with inversion recovery (or SPIR)) is an excellent technique to visualize pathology of peripheral nerves, and the brachial or lumbosacral plexus (Figure 1.2C).8 Figure 1.2 Basic pulse sequences of MRI Examples of the healthy brachial plexus visualized in a T1 weighted image (A), a T2 weighted image (B) and T2 weighted imaging with fat suppression (C). Quantitative MRI techniques MRI is a versatile technique that can provide qualitative as well as quantitative information on (nervous) tissues. T1- and T2-weighted imaging, as described in the previous paragraph, provides qualitative information on anatomical tissues and generates an image. Advanced quantitative MRI techniques do not only produce an image, but also generate a quantitative parameter. One of these quantitative techniques is diffusion tensor imaging (DTI). DTI gives quantitative information on microstructural integrity that correlates with histological findings.9–11 DTI measures diffusion of water in tissue in a number of different directions. Diffusion rates of biological tissues are not the same in every direction, which means the tissue is not isotropic but rather anisotropic. The direction and magnitude of the diffusion can be expressed by the diffusion tensor. From this diffusion tensor eigenvalues and eigenvectors can be derived. Eigenvectors express the direction of the diffusion, and eigenvalues express the magnitude of the diffusion. In this way, the degree of diffusion of water can be calculated along the main axis (axonal diffusivity, AD) or perpendicular to the nervous tissue (radial diffusivity, RD). AD is determined by the eigenvalue λ1 and RD is determined by the mean of the eigenvalues λ2 and λ3 (Figure 1.3). The mean diffusivity (MD) is calculated by the mean of

General introduction and thesis outline 15 1 the eigenvalues ((λ1 + λ2 + λ3)/3). Anisotropy is expressed as the ‘fractional anisotropy’ (FA) and can be calculated using a mathematical formula that contains all eigenvalues. FA is a scalar value that ranges from 0 to 1 (Figure 1.3). Pure water has isotropic diffusion properties, which means that the water molecules are equally likely to move in any direction, hence FA = 0. For tissues that have very strong anisotropy FA = 1, i.e. diffusion is restricted by the presence of cell membranes and there may be a preferential direction, for example along nerve fibers. As MD and FA are summary measures of the eigenvalues λ1, λ2 and λ3, changes in MD and FA can be driven by changes in either AD or in RD. For example, an increase of AD or a decrease of RD both cause an increase of FA, and a decrease of AD or RD both cause a decrease of MD. Figure 1.3 Principles of diffusion parameters Isotropic diffusion (left): water molecules are equally likely to move in any direction and fractional anisotropy is 0. Anisotropic diffusion (right): water molecules move in a preferential direction and fractional anisotropy is 1. Other quantitative MRI techniques as T2 mapping and fat fraction analysis can provide information on T2 relaxation times and fat fraction percentage of a tissue. T2 mapping relies on the principle that different echo times result in different T2 contrasts in images. By plotting the signal intensity for different echo times an exponential decay curve can be constructed. The T2 relaxation time can be calculated as a constant of the fitted curve. In this way, the T2 relaxation time can be calculated for the tissue of interest. Fat fraction analysis relies on the fact that water and fat contain protons that can be measured using Dixon imaging (chemical shift imaging). Protons in fat rotate at a different Larmor frequency than protons in water. A minimum of two images, i.e. one in phase and one out of

Chapter 1 16 phase, are necessary to calculate the percentage of fat. The first image is acquired when water and fat have the same phase, i.e. they rotate in phase, and the total signal can be calculated by the sum of the signal of water and the signal of fat. Next, a second image is acquired when the protons are in opposed phase, i.e. they are out of phase, and the total signal contains the signal of water minus the signal of fat. In this way, the percentage of water and the percentage of fat of the total signal of a tissue can be calculated. Injured or inflamed tissue may lead to changes in the diffusion parameters, for example by increasing or decreasing AD or RD, and changes in T2 relaxation times and fat fraction. Measuring and comparing quantitative MRI parameters between groups of patients and (healthy) controls, and correlating the found differences to histology or clinical data, may help in diagnosis, pathophysiology or prognosis of disease. Combined, these quantitative techniques inform on structural nerve changes due to pathophysiological processes. CIDP, MMN & MRI Diagnosis In current clinical practice diagnosis of CIDP and MMN is predominantly based on the consensus criteria of the guideline of the European Federation of Neurological Societies/Peripheral Nerve Society and the Utrecht criteria.4,5,7 It is important to differentiate CIDP and MMN from their clinical mimics as an early start of immunomodulatory treatment in CIDP and MMN could prevent irreversible (axonal) damage and worsening of symptoms.1,2 Differential diagnosis of (typical and variants of) CIDP includes Guillain-Barré syndrome, motor neuron diseases, focal compression neuropathies, diabetic neuropathy and Charcot-Marie-Tooth disease.12 MMN is an important clinical mimic of motor neuron diseases, such as amyotrophic lateral sclerosis and progressive muscular atrophy, as they all can present with an asymmetric weakness without sensory deficits. However, treatment and prognoses differ considerably and the use of diagnostic consensus guidelines may help in the differentiation. These guidelines for CIDP and MMN describe that diagnosis is primarily based on a characteristic clinical presentation and specific features on nerve conduction studies, i.e. conduction blocks. These conduction blocks are believed to be caused by demyelination, in particular in CIDP, or axolemmal changes.13 However, diagnosing CIDP or MMN often remains challenging as nerve conduction studies require specific expertise, cost a lot of time and are often burdensome to patients. Conduction blocks could be easily missed, which compromises diagnostic accuracy.6 In more elusive cases, supportive criteria may help in diagnosis. One of the additional diagnostic tools is laboratory examination which may show an increased protein in the cerebrospinal fluid or presence of anti-GM1 antibodies (MMN only).4,5 Second, a good response to immunomodulatory treatment may add to diagnosis although this criterium is seriously hampered by costs, the risk of adverse events in patients and the lack of a clear definition of treatment response. MRI of the brachial

General introduction and thesis outline 17 1 plexus is in the current diagnostic guidelines the last supportive criterium and may show thickening of the cervical nerve roots (Figure 1.4A) or T2 hyperintensity (Figure 1.4B) in patients with CIDP or MMN.14,15 Enhancement of the nerve roots could be seen in patients with CIDP after injection with gadolinium but is less common and the diagnostic value is low.16 Abnormalities on brachial plexus MRI are more frequently present in patients with CIDP (74% of patients) than in patients with MMN (50% of patients).16 Asymmetrical thickening of the nerve roots seems to be more common in patients with MMN compared to patients with CIDP.17 Figure 1.4 Pathology of the brachial plexus in chronic inflammatory neuropathies In the left panel (A) an example of thickening of the cervical nerve roots is shown (for example compared to figure 1.2C) using T2-weighted imaging with fat suppression. In the right panel (B) an example of T2 hyperintensity (yellow arrow) is shown using T2-weighted imaging. A major drawback of current clinical practice is that brachial plexus MRI is qualitatively assessed by (neuro)radiologists. Obviously enlarged cervical nerve roots are easily seen but less evident thickening may result in an uncertain and subjective assessment as clear cut-off values for nerve size are lacking. Furthermore, variability and reliability of these qualitative assessments are unknown which hampers the diagnostic value of brachial plexus MRI even more. A systematic assessment with quantitative cut-off values for cervical nerve root size, and a comparison of interrater reliabilities between qualitative and quantitative assessments, is needed if we want to improve the diagnostic value of brachial plexus MRI for the diagnosis of CIDP and MMN.

Chapter 1 18 More recently, nerve ultrasound has been explored as another imaging technique in diagnosis of chronic inflammatory neuropathies.18,19 These studies showed that a quantitative assessment, i.e. with objective cut-off values for abnormality, results in good test characteristics. Just as MRI, nerve ultrasound may show thickening of the brachial plexus and peripheral nerves. Unfortunately, nerve ultrasound is not yet widely available as its implementation in clinical practice requires experience. However, nerve ultrasound is a promising technique that might be added to the diagnostic criteria in the future. The combined role of nerve ultrasound and MRI in the diagnostic process of chronic inflammatory neuropathies should be evaluated to optimize the diagnostic performance of both imaging modalities. Pathophysiology Autopsy studies, sural nerve biopsy and immunostaining in vitro have tried to provide insight in underlying immunological mechanisms in CIDP and MMN.20–26 The scarce reports on CIDP describe moderate to severe demyelination and remyelination with onion bulbs without loss of axons.20–22,27 Some histological studies on MMN describe axonal loss without demyelination,23–25,28 while others describe de- and remyelination.29–31 There is an obvious need for additional tools to study the condition of peripheral nerves of patients with a chronic inflammatory neuropathy in vivo. Therefore, quantitative MRI techniques that correlate to histological findings, such as DTI and T2 mapping are promising. Previous DTI studies evaluated the peripheral nerves and the brachial and lumbosacral plexus in small cohorts of patients with CIDP or MMN and healthy controls.32–40 These smaller studies already showed differences in diffusion parameters and T2 relaxation times between groups which suggests that quantitative MRI techniques could be helpful to explore pathophysiologies further. However, large and systematic studies are currently lacking. Prognosis and treatment Patients with CIDP and MMN both respond to immunomodulatory treatment. For patients with MMN intravenous or subcutaneous immunoglobulins is the only treatment option; patients with CIDP may also respond to treatment with corticosteroids or plasmapheresis.1,2 Treatment may improve motor and sensory deficits but management of treatment is challenging in current clinical practice. These challenges mainly rely on the fact that treatment response is not easily monitored as it lacks a clear definition. Therefore, it might be difficult to find the right treatment dose in some patients, which could result in over- or undertreatment. In current clinical practice, improvement of muscle strength as measured by MRC scales, myometry or equivalent tests is assumed to be the golden standard of treatment response but strength measurements might be subject to differences between raters. Objective markers that predict course of disease and treatment response are lacking. However, these biomarkers are needed if we want to improve management of CIDP and MMN.

General introduction and thesis outline 19 1 The value of electrophysiology and nerve ultrasound as biomarkers have been studied previously. These efforts did not result in the identification of quantitative measures that correlate to clinical outcomes or prognosis (unpublished data from our center).41–43 Quantitative MRI techniques, such as DTI, are a potentially powerful tool to monitor tissues. DTI has been explored in several studies of the central nervous system. These studies showed differences in diffusion parameters over time and sometimes showed correlations with clinical parameters.44–49 However, there is only a very limited number of studies of the peripheral nervous system. It is therefore unknown if quantitative MRI captures relevant differences in the peripheral nervous system, for example early treatment effects. THESIS OUTLINE The aim of this thesis is to explore the feasibility and value of qualitative and quantitative MRI techniques in diagnosis, pathophysiology, disease course, and treatment response in chronic inflammatory neuropathies. Chapter 2 contains a description of the natural history of MMN and an analysis of the correlates of a progressive disease course. Chapter 3 evaluates the interrater variability of current practice, i.e. a qualitative assessment of nerve thickening on brachial plexus MRI. In Chapter 4 we explore feasibility and diagnostic performance of a quantitative assessment of nerve thickening. In chapter 5 we study involvement of intraspinal roots in CIDP and MMN. In chapter 6 we assess the use of quantitative MRI techniques (i.e. DTI, T2 mapping, and fat fraction analysis) and attempts to study nerve architecture in CIDP and MMN in vivo. In chapter 7 we present data of quantitative MRI after one year of follow-up. Chapter 8 contains a summary and discussion of the main findings of this thesis and provides recommendations for clinical practice.

Chapter 1 20 REFERENCES 1. Oaklander A, Lunn M, Hughes R, et al. Treatments for chronic inflammatory demyelinating polyradiculoneuropathy (CIDP): An overview of systematic reviews. Cochrane Database Syst. Rev. 2017;13:1–33. 2. van Schaik I, van den Berg L, de Haan R, Vermeulen M. Intravenous immunoglobulin for multifocal motor neuropathy. Cochrane Database Syst. Rev. 2005;2:920–921. 3. Kuwabara S, Misawa S, Mori M, et al. Long term prognosis of chronic inflammatory demyelinating polyneuropathy: A five year follow up of 38 cases. J. Neurol. Neurosurg. Psychiatry 2006;77:66–70. 4. van den Bergh PYK, Hadden RDM, Bouche P, et al. European Federation of Neurological Societies/Peripheral Nerve Society Guideline on management of chronic inflammatory demyelinating polyradiculoneuropathy: Report of a joint task force of the European Federation of Neurological Societies and the Peripher. Eur. J. Neurol. 2010;17:356–363. 5. Vlam L, Van Der Pol WL, Cats EA, et al. Multifocal motor neuropathy: Diagnosis, pathogenesis and treatment strategies. Nat. Rev. Neurol. 2012;8:48–58. 6. Allen JA, Lewis RA. CIDP diagnostic pitfalls and perception of treatment benefit. Neurology 2015;85:498–504. 7. Van Schaik IN, Léger JM, Nobile-Orazio E, et al. European Federation of Neurological Societies/Peripheral Nerve Society Guideline on management of multifocal motor neuropathy. Report of a Joint Task Force of the European Federation of Neurological Societies and the Peripheral Nerve Society - First revis. J Peripher Nerv Syst 2010;15:295–301. 8. Chhabra A, Zhao L, Carrino JA, et al. MR Neurography: Advances. Radiol. Res. Pract. 2013;2013:1–14. 9. Song SK, Sun SW, Ramsbottom MJ, et al. Dysmyelination revealed through MRI as increased radial (but unchanged axial) diffusion of water. Neuroimage 2002;17:1429–1436. 10. Song SK, Sun SW, Ju WK, et al. Diffusion tensor imaging detects and differentiates axon and myelin degeneration in mouse optic nerve after retinal ischemia. Neuroimage 2003;20:1714– 1722. 11. Morisaki S, Kawai Y, Umeda M, et al. In vivo assessment of peripheral nerve regeneration by diffusion tensor imaging. J. Magn. Reson. Imaging 2011;33:535–542. 12. Neligan A, Reilly MM, Lunn MP. CIDP: Mimics and chameleons. Pract. Neurol. 2014;14:399– 408. 13. Van Asseldonk JTH, Van Den Berg LH, Kalmijn S, et al. Criteria for demyelination based on the maximum slowing due to axonal degeneration, determined after warming in water at 37°C: Diagnostic yield in chronic inflammatory demyelinating polyneuropathy. Brain 2005;128:880–891. 14. van Es HW, van den Berg LH, Franssen H, et al. Magnetic resonance imaging of the brachial plexus in patients with multifocal motor neuropathy. Neurology 1997;48:1218–1224. 15. Castillo M, Mukherji SK. MRI of enlarged dorsal ganglia, lumbar nerve roots, and cranial nerves in polyradiculoneuropathies. Neuroradiology 1996;38:516–520. 16. Goedee HS, Jongbloed BA, van Asseldonk J-TH, et al. A comparative study of brachial plexus sonography and magnetic resonance imaging in chronic inflammatory demyelinating neuropathy and multifocal motor neuropathy. Eur. J. Neurol. 2017;24:1307–1313. 17. Jongbloed BA, Bos JW, Rutgers D, et al. Brachial plexus magnetic resonance imaging differentiates between inflammatory neuropathies and does not predict disease course. Brain Behav. 2017;7:e00632. 18. Goedee HS, Van Der Pol WL, Van Asseldonk JTH, et al. Diagnostic value of sonography in treatment-naive chronic inflammatory neuropathies. Neurology 2017;88:143–151.

General introduction and thesis outline 21 1 19. Herraets IJT, Goedee HS, Telleman JA, et al. Nerve ultrasound for the diagnosis of chronic inflammatory neuropathy: a multicenter validation study. Neurology 2020;95:e1745–e1753. 20. Matthews WB, Howell DA, Hughes RC. Relapsing corticosteroid-dependent polyneuritis. J. Neurol. Neurosurg. Psychiatry 1970;33:330– 337. 21. Torvik A, Lundar T. A case of chronic demyelinating polyneuropathy resembling the Guillan-Barré Syndrome. J. Neurol. Sci. 1977;32:45–52. 22. Matsuda M, Ikeda SI, Sakurai S, et al. Hypertrophic neuritis due to chronic inflammatory demyelinating polyradiculoneuropathy (CIDP): A postmortem pathological study. Muscle and Nerve 1996;19:163–169. 23. Krarup C, Stewart JD, Sumne AJ, et al. A syndrome of asymmetric limb weakness with motor conduction block. Neurology 1990;40:118–127. 24. Adams D, Kuntzer T, Steck AJ, et al. Motor conduction block and high titres of anti-GM1 ganglioside antibodies: Pathological evidence of a motor neuropathy in a patient with lower motor neuron syndrome. J. Neurol. Neurosurg. Psychiatry 1993;56:982–987. 25. Veugelers B, Theys P, Lammens M, et al. Pathological findings in a patient with amyotrophic lateral sclerosis and multifocal motor neuropathy with conduction block. J. Neurol. Sci. 1996;136:64–70. 26. Harschnitz O, van den Berg LH, Johansen LE, et al. Autoantibody pathogenicity in a multifocal motor neuropathy induced pluripotent stem cell– derived model. Ann. Neurol. 2016;80:71–88. 27. Sasaki M, Ohara S, Oide T, et al. An autopsy case of chronic inflammatory demyelinating polyradiculoneuropathy with respiratory failure. Muscle and Nerve 2004;30:382–387. 28. Taylor B V., Dyck PJB, Engelstad JN, et al. Multifocal Motor Neuropathy: Pathologic Alterations at the Site of Conduction Block. J. Neuropathol. Exp. Neurol. 2004;63:129–137. 29. Auer RN, Bell RB, Lee MA. Neuropathy with Onion Bulb Formations and Pure Motor Manifestations. Can. J. Neurol. Sci. / J. Can. des Sci. Neurol. 1989;16:194–197. 30. Kaji R, Oka N, Tsuji T, et al. Pathological findings at the site of conduction block in multifocal motor neuropathy. Ann. Neurol. 1993;33:152–158. 31. Corbo M, Abouzahr MK, Latov N, et al. Motor nerve biopsy studies in motor neuropathy and motor neuron disease. Muscle and Nerve 1997;20:15–21. 32. Kakuda T, Fukuda H, Tanitame K, et al. Diffusion tensor imaging of peripheral nerve in patients with chronic inflammatory demyelinating polyradiculoneuropathy: A feasibility study. Neuroradiology 2011;53:955–960. 33. Mathys C, Aissa J, Zu Hörste GM, et al. Peripheral Neuropathy: Assessment of Proximal Nerve Integrity By Diffusion Tensor Imaging. Muscle Nerve 2013;48:889–896. 34. Markvardsen LH, Vaeggemose M, Ringgaard S, Andersen H. Diffusion tensor imaging can be used to detect lesions in peripheral nerves in patients with chronic inflammatory demyelinating polyneuropathy treated with subcutaneous immunoglobulin. Neuroradiology 2016;58:745–752. 35. Kronlage M, Pitarokoili K, Schwarz D, et al. Diffusion Tensor Imaging in Chronic Inflammatory Demyelinating Polyneuropathy: Diagnostic Accuracy and Correlation with Electrophysiology. Invest. Radiol. 2017;52:701– 707. 36. Haakma W, Jongbloed BA, Froeling M, et al. MRI shows thickening and altered diffusion in the median and ulnar nerves in multifocal motor neuropathy. Eur. Radiol. 2017;27:2216–2224. 37. Lichtenstein T, Sprenger A, Weiss K, et al. MRI biomarkers of proximal nerve injury in CIDP. Ann. Clin. Transl. Neurol. 2018;5:19–28. 38. Oudeman J, Eftimov F, Strijkers GJ, et al. Diagnostic accuracy of MRI and ultrasound in chronic immune-mediated neuropathies. Neurology 2020;94:e62–e74. 39. Hiwatashi A, Togao O, Yamashita K, et al. Simultaneous MR neurography and apparent T2 mapping in brachial plexus: Evaluation

Chapter 1 22 of patients with chronic inflammatory demyelinating polyradiculoneuropathy. Magn. Reson. Imaging 2018;55:112–117. 40. Hiwatashi A, Togao O, Yamashita K, et al. Lumbar plexus in patients with chronic inflammatory demyelinating polyradiculoneuropathy: evaluation with simultaneous T2 mapping and neurography method with SHINKEI. Br. J. Radiol. 2018;91:20180501. 41. Van Asseldonk JTH, Van den Berg LH, Van den Berg-Vos RM, et al. Demyelination and axonal loss in multifocal motor neuropathy: Distribution and relation to weakness. Brain 2003;126:186–198. 42. Vucic S, Black K, Baldassari LE, et al. Longterm effects of intravenous immunoglobulin in CIDP. Clin. Neurophysiol. 2007;118:1980– 1984. 43. Iijima M, Yamamoto M, Hirayama M, et al. Clinical and electrophysiologic correlates of IVIg responsiveness in CIDP. Neurology 2005;64:1471–1475. 44. Nowranghi M, Lyketsos C, Leoutsakos J-M, et al. Longitudinal, region-specific course of diffusion tensor imaging measures in mild cognitive impairment and Alzheimer’s disease. Bone 2013;9:519–528. 45. Kitamura S, Kiuchi K, Taoka T, et al. Longitudinal white matter changes in Alzheimer’s disease: A tractography-based analysis study. Brain Res. 2013;1515:12–18. 46. Genc S, Steward CE, Malpas CB, et al. Shortterm white matter alterations in Alzheimer’s disease characterized by diffusion tensor imaging. J. Magn. Reson. Imaging 2016;43:627– 634. 47. Mayo CD, Mazerolle EL, Ritchie L, et al. Longitudinal changes in microstructural white matter metrics in Alzheimer’s disease. NeuroImage Clin. 2017;13:330–338. 48. Sampedro F, Martínez-Horta S, Marín-Lahoz J, et al. Longitudinal intracortical diffusivity changes in de-novo Parkinson’s disease: A promising imaging biomarker. Park. Relat. Disord. 2019;68:22–25. 49. Tringale K, Nguyen T, Bahrami N, et al. Identifying early diffusion imaging biomarkers of regional white matter injury as indicators of executive function decline following brain radiotherapy: A prospective clinical trial in primary brain tumor patients. Radiother Oncol 2019;132:27–33.

Clinical outcomes in multifocal motor neuropathy: a combined cross-sectional and follow-up study Chapter 2 Ingrid J.T. Herraets, Marieke H.J. van Rosmalen, Jeroen W. Bos, Ruben P.A. van Eijk, Elies A. Cats, Bas A. Jongbloed, L. Vlam, S. Piepers, J. Thies van Asseldonk, H. Stephan Goedee, Leonard H. van den Berg, W. Ludo van der Pol Neurology, 2020 Oct; 95, e1979 - e1987

Chapter 2 26 ABSTRACT Objective: To assess the clinical course of multifocal motor neuropathy (MMN) in a large cohort of patients and to identify predictive factors of a progressive disease course. Methods: Between May 2015 and February 2016, we collected clinical data from 100 patients with MMN of whom 60 had participated in a nationwide cross-sectional cohort study in 2007. We documented clinical characteristics using standardized questionnaires and performed a standardized neurological examination. We used multiple linear regression analysis to identify factors that correlated with worse outcome. Results: We found that age of diagnosis (45.2 vs. 48.6 years, p < 0.02) significantly increased between 2007 and 2015 – 2016, whereas diagnostic delay decreased with 15 months. Seven out of ten outcome measures deteriorated over time (all p < 0.01). Patients who had a lower Medical Research Council (MRC) sum score and absence of one or more reflexes at the baseline visit showed a greater functional loss at follow up (p = 0.007 and p = 0.016). Conclusion: Our study shows that MMN is a progressive disease. Although 87% of patients received maintenance treatment, muscle strength, reflexes, vibration sense, and the Self-Evaluation Scale significantly deteriorated over time. Lower MRC sum score and absence of reflexes predicted a more progressive disease course.

Clinical outcomes in MMN 27 2 INTRODUCTION Multifocal motor neuropathy (MMN) is a pure motor disorder characterized by slowly progressive asymmetric distal weakness mainly in the hands, the absence of upper motor neuron signs and presence of one or more abnormal ancillary investigations, i.e. abnormal nerve conduction or conduction block (CB), thickening or T2 hyperintensity on magnetic resonance imaging (MRI) of the brachial plexus, sonographic nerve thickening, increased protein content in the cerebrospinal fluid (CSF) or the presence of anti-GM1 IgM antibodies in serum.1–5 Administration of intravenous or subcutaneous immunoglobulins transiently improves muscle strength and maintenance treatment is therefore needed.6–10 Consensus criteria have facilitated diagnosis of MMN and shortened diagnostic delays, but we know less of the disease course and outcome.5,11 Early case reports suggested that its course is not benign in individual patients, but few studies have longitudinally addressed natural history in larger patients cohorts.12–14 Early treatment may improve long-term outcome, but accumulating axonal damage nevertheless results in significant disability in up to one fifth of patients.11,15 More detailed insight in MMN’s clinical course would help to identify correlates of worse outcome and thereby patients at higher risk for developing severe deficits, and eventually to investigate efficacy of other treatment approaches. We have previously reported the characteristics of a relatively large cross sectional cohort of patients with MMN in the Netherlands.11 In order to gain more insight in the clinical course of MMN, we performed a combined cross-sectional and follow-up study in a cohort of 100 patients with the aim to identify factors that predict a progressive disease course of MMN. METHODS Study design and patients This cross-sectional cohort study was performed between May 2015 and February 2016 in the University Medical Center (UMC) Utrecht, a large tertiary referral center for neuromuscular disorders in The Netherlands. We invited all patients listed in the MMN database of the UMC Utrecht who met the following inclusion criteria: 1) a diagnosis of definite, probable or possible MMN according to the EFNS/PNS criteria and 2) age ≥ 18 years.5 A subgroup of our patients previously participated in a similar cross-sectional cohort study in 2007.11 The local medical ethics committee of the UMC Utrecht approved the research protocol (NL50354.041.14). All included patients gave written informed consent.

Chapter 2 28 Neurological examination and questionnaires We documented clinical characteristics of patients with MMN (including but not limited to site of onset and age at symptom onset) using a standardized questionnaire and collected the Overall Disability Sum Score (ODSS), the Self-Evaluation Scale (SES), the Rasch-built Overall Disability Score for MMN (MMN-RODS) and the Fatigue Severity Scale (FSS).16–20 All patients underwent a standardized neurological examination (Supplemental table 2.1). This consisted of bilateral grading of motor function of 18 muscle groups using the Medical Research Council (MRC) scale to calculate the MRC sum score with a maximum of 180 points. Sensory function was tested using a Rydell-Seiffer tuning fork to assess vibration sense in arms and legs bilaterally. Vibration sense was graded from normal (grade 0) to abnormal at the acromioclavicular joint or anterior superior iliac spine (grade 4).21 Tendon reflexes of biceps, triceps, knee and ankle were performed on both sides and scored as normal, brisk or absent. We used data obtained during a previous study in 2007 as baseline data.11 To minimize inter-observer variability, one of the authors (EAC) who collected clinical data during the 2007 study trained the author (BAJ) who performed the clinical examination in 2015 – 2016, with special emphasis on the interpretation of MRC and Rydell-Seiffer scales. Nerve conduction studies and other ancillary investigations One of the authors (HSG) evaluated available nerve conduction study results using the EFNS/PNS criteria for CB and other abnormalities.5 All patients underwent nerve conduction studies (NCS) using a standardized protocol and stimulation was up to Erb’s point.22 CB was defined as definite CB (compound muscle action potential (CMAP) area reduction of at least 50%) or probable CB (CMAP area reduction of 30-50%), and axonal loss as a decreased distal CMAP (distal CMAP amplitude below the lower limit of normal) in ≥ 1 nerves, including the median, ulnar, radial, musculocutaneous, peroneal, and tibial nerves.5,23 We also collected all available results of laboratory studies (in particular the presence of anti-GM1 IgM antibodies in serum and analysis of cerebrospinal fluid) and of MRI of the brachial plexus. Statistical analyses MMN cohort data We stratified the patients with MMN into two groups: (1) patients diagnosed before our previous study in 2007, and (2) patients diagnosed after 2007) to explore differences in clinical characteristics. Depending on the distribution of the variable, we compared groups using the Mann-Whitney U test (for continuous data) and the χ2 test (for categorical data). To account for right skew in timerelated covariates, we log-transformed (natural) duration of treatment, months untreated and time to diagnosis. Univariate linear regression analyses were performed to identify changes in clinical characteristics over calendar time. Dependent variables were age at diagnosis, time to diagnosis (log-transformed) and age at onset of symptoms. The independent variable was the year of diagnosis. Subsequently, we calculated the mean MRC score per muscle group for patients with longer and shorter disease duration (defined as equal to or larger than the median disease duration). We corrected

Clinical outcomes in MMN 29 2 the obtained p values for multiple testing using the Benjamini-Hochberg method. Multiple linear regression analysis was used with backward elimination based on p value selection to predict the MRC sum score 2015 – 2016 based on sex, symptom onset in a leg, presence of anti-GM1 IgM antibodies, FSS (0 – 63), duration of treatment in months (log-transformed), months untreated (logtransformed) and age at onset of symptoms in years). Longitudinal follow-up data The mean yearly rate of decline of each outcome measure was estimated between visit 1 (2007) and visit 2 (2015 – 2016) and tested using a one sample t test (i.e. assessing whether the yearly rate of decline is other than zero). Multiple linear regression analysis was performed with backward elimination based on p value selection to predict the yearly rate of decline in MRC sum score based on sex, presence of anti-GM1 IgM antibodies, symptom onset in leg, months untreated (logtransformed), age at onset of symptoms in years, ODSS (0 – 8), MRC sum score (0 – 180) and sum score of reflexes (0 – 8). The last three variables were analysed with data of the first visit (2007). Patients were asked to describe their disease course as stable, gradually but slowly progressive, gradually progressive, stepwise progressive or gradually improving. RESULTS Patients We identified a total of 142 patients with MMN. Hundred patients (70.4%) agreed to participate of whom 60 patients previously participated in a nationwide cross sectional cohort study in 2007.11 Reasons for not participating are shown in Figure 2.1. Clinical characteristics Patient characteristics (sex, age at onset of symptoms, MMN diagnosis according to EFNS/PNS criteria and additional investigations i.e. NCS, MRI brachial plexus, CSF protein and presence of anti-GM1 IgM antibodies) between participants (n = 100) and non-participants (n = 42), were not significantly different, except for the onset of muscle weakness (p = 0.04). Median age at onset of symptoms and age of diagnosis were significantly higher in patients diagnosed after 2007 (p < 0.01 and p = 0.02; Table 2.1). We performed univariate linear regression analysis with year of diagnosis as independent variable. Both median age at onset of symptoms and median age of diagnosis significantly increased over time (both p < 0.01) (Figure 2.2). Median time from symptom onset to diagnosis (i.e. diagnostic delay) decreased over time (6.4 years (range 1 – 27) in period 1996 to 2000; 1.8 years (range 1 – 29) in period 2011 – 2015) but was significantly longer for patients with onset of symptoms in a leg and for patients with higher age at diagnosis (p = 0.01 and p < 0.01). We use a starting dose of 0.4 g/kg immunoglobulins per 3-4 weeks and then tailor the dose (if needed up to 1 g/kg) until patients remain stable during the treatment interval.2 The starting dose

Chapter 2 30 was significantly higher for patients diagnosed before 2007 (p < 0.01), probably due to a different treatment regime with repeated loading doses of immunoglobulins in the period before 1995 rather than lower-dosed weekly to monthly maintenance therapy. We found no significant differences in clinical characteristics between males and females. Figure 2.1 Flowchart of study MMN cohort in 2015 n = 152 Not reached n = 9 Deceased n = 8 No time/interest n = 11 Nationwide cohort study 2007 n = 88 Not reached n = 3 Missing data n = 1 No time/interest n = 10 "new" MMN patients n = 54 Follow-up study n = 60 Follow-up study n = 40 Cross-sectional cohort study n = 100 Abbreviations: MMN = multifocal motor neuropathy. Weakness, sensory function and tendon reflexes The distribution of muscle weakness was distal more than proximal and more pronounced in hand than in foot or lower leg muscles (Supplemental table 2.2). Finger flexion and plantar foot flexion were relatively spared compared to hand and finger extension and dorsal foot flexion. Patients with longer disease duration had significantly more weakness in hand and lower leg or foot muscles

Clinical outcomes in MMN 31 2 compared to patients with shorter disease duration (all p < 0.05) (Figure 2.3 and Supplemental table 2.2). We found abnormal vibration sense on the toes in 57 patients (57.6%). Median disease duration was longer in these patients compared to those without sensory findings (median 16.1 years, range 1.3 – 46.5 vs. median 11.5 years, range 1.9 – 30.5; p = 0.03). We found at least one absent reflex in 79 patients (79%). Sixteen of these patients (20%) had generalized areflexia (Supplemental table 2.3). We did not find a relation between the presence of conduction block (definite and/or probable) and the absence of reflexes (p > 0.10). Figure 2.2 Clinical characteristics over time Median age at onset of symptoms and median age of diagnosis over time. Error bars are 95% confidence intervals. Nerve conduction studies and laboratory investigations One or more definite CBs were present in 74 patients (74.0%), only probable CB in 19 patients (19.0%) and no CB in 7 patients (7.0%). The diagnosis of MMN in these 7 patients without CB was based on the presence of anti-GM1 IgM antibody titers (4 patients; 57.1%), abnormal CSF protein concentrations (protein level > 0.4 gram/litre (g/L); 2 patients; 28.6%), an abnormal MRI of the brachial plexus (3 patients; 42.9%), and response to immunoglobulin therapy in all patients. We found evidence of axonal damage during NCS in 71 patients (71.0%), the presence of anti-GM1 IgM antibodies in 55/90 patients (61.1%) and abnormal CSF protein concentrations (>0.4 g/L) in 20/26 (76.9%) patients.

Chapter 2 32 Table 2.1 Clinical characteristics Parameter Diagnosis before 2007 (n = 64) Diagnosis in or after 2007 (n = 36) p Male 46 (72%) 29 (81%) 0.34 Age at onset of symptoms 40.3 (21.4 – 53.8) 45.2 (30.1 – 67.2) < 0.001 Age at diagnosis 45.2 (25.2 – 71.1) 48.6 (30.9 – 73.5) 0.02 Time to diagnosis (months)* 42.0 (3.0 – 433.0) 27.0 (6.0 – 345.0) 0.10 Time from disease onset until treatment (months) 42.0 (3.0 – 435.9) 27.5 (3.9 – 346.0) 0.09 Maintenance treatment with immunoglobulins 55 (86%) 32 (89%) 0.67 Starting dose ivIg maintenance therapy per week (gram) 10.0 (5.0 – 33.0) 8.0 (4.0 – 12.0) < 0.001 Onset of muscle weakness Distal arm 41 (64%) 25 (70%) 0.08 Proximal arm 3 (4%) 3 (8%) Distal leg 18 (28%) 4 (11%) Proximal leg 1 (2%) 0 (0 %) Distal symmetrical 1 (2%) 4 (11%) Number of affected limbs at inclusion 0 2 (3%) 1 (3%) 0.15 1 7 (11%) 8 (22%) 2 12 (19%) 12 (33%) 3 18 (28%) 7 (20%) 4 25 (39%) 8 (22%) Electrophysiological criteria** Definite 45 (70%) 29 (81%) 0.32 Probable 15 (23%) 4 (11%) Negative 4 (6%) 3 (8%) NCS with axonal degeneration 31 (48%) 13 (36%) 0.23 Abnormalities brachial plexus MRI 22/43 (51%) 8/17 (47%) 0.77 Laboratory findings Increased CSF protein 12/16 (75%) 8/10 (80%) 0.77 Presence of anti-GM1 IgM antibodies 38/61 (62%) 17/29 (59%) 0.74 Data are shown in median (range) or number of patients (%), unless stated otherwise. * log transformed variable ** according to the EFNS/PNS criteria5 Abbreviations: ivIg = intravenous immunoglobulins; NCS = nerve conduction studies; MRI = magnetic resonance imaging; CSF = cerebrospinal fluid.

Clinical outcomes in MMN 33 2 Figure 2.3 Correlation of MRC grade and disease duration per muscle group The boxplots provide the variability in disease duration per MRC grade (0 – 5). Disease duration is defined as years from onset of symptoms until first study visit. Abbreviations: MRC = Medical Research Council. Disability questionnaires Results of the disability questionnaires are shown in Supplemental table 2.3. Median ODSS of the arms was 2 (range 0 – 4), of the legs 1 (range 0 – 5) and of arms and legs combined 3 (range 0 – 8). Twelve patients (12%) reported no disability of the arms and 34 patients (34%) did not experience disability of the legs. Correlates of outcome Multiple linear regression analysis showed that a lower MRC sum score correlated with longer disease duration without treatment, presence of anti-GM1 IgM antibodies and lower age at onset of symptoms (p = 0.024, p = 0.046 and p = 0.006 respectively). Outcome measures over time Mean differences between visit 1 (2007) and visit 2 (2015 – 2016) of different outcome measures are shown in Table 2.2. Except for ODSS, FSS and vigorimetry of the left hand, all outcome measures deteriorated over time (all p < 0.01). The difference in MRC sum score between 2015 and 2007 was significantly larger in patients with axonal damage compared to patients without axonal damage (5.2 points vs. 13.8 points; p = 0.014). Most patients indicated that their disease course was stable (25.0%) or mildly progressive (61.7%). The dose of immunoglobulin treatment significantly increased over time (p < 0.001).

Chapter 2 34 Table 2.2 Outcome measures over time Parameter MD per year 95% CI p ODSS (0 – 12 points) -0.004 0.03 – -0.04 0.81 MRC sum score (0 – 180 points) -1.361 -0.97 – -1.75 < 0.001 SES (0 – 25 points) 0.352 0.54 – 0.16 < 0.001 FSS (0 – 63 points) -0.940 -0.25 – -1.63 < 0.001 Vibration sense (abnormal in 0 – 4 limbs) 0.121 0.15 – 0.09 < 0.001 Reflexes arm (absence in 0 – 4 reflexes) 0.055 -0.02 – -0.09 < 0.001 Reflexes leg (absence in 0 – 4 reflexes) 0.072 -0.03 – -0.11 < 0.001 Reflexes sum score (absence in 0 – 8 reflexes) 0.121 --0.06 – -0.18 < 0.001 Grip strength right (kPa) -1.127 -0.39 – -1.87 < 0.001 Grip strength left (kPa) -0.770 0.04 – -1.58 0.06 Number of affected muscle groups 0.465 0.36 – 0.58 < 0.001 Mean difference per year was calculated as the difference between visit 1 (2007) and visit 2 (2015-2016) divided by the follow-up duration. Absence of reflexes arm: biceps and triceps reflexes (0 – 4). Absence of reflexes leg: knee and ankle reflexes (0 – 4). Abbreviations: MD = mean difference; CI = confidence interval; ODSS = Overall Disability Sum Score; MRC = Medical Research Council; SES = Self-evaluation Scale; FSS = Fatigue Severity Scale. Predictors of progression Multiple linear regression showed that faster progression, i.e. a larger difference of the MRC sum score of visit 1 (2007) and visit 2 (2015 – 2016) per year correlated with the reflexes sum score (i.e. absent reflexes) and a lower MRC sum score in 2007 (p = 0.016 and p = 0.007 respectively). DISCUSSION This study aimed to document clinical outcomes of patients with MMN and identify predictors of disease progression. We combined cross-sectional data with longitudinal data with a mean duration between visits of eight years. Our clinical observations confirmed that MMN is a progressive disorder in the large majority of patients even when they receive immunoglobulin maintenance treatment. Virtually all selected outcome measures significantly deteriorated over time. Factors with prognostic value of a progressive disease course were absence of reflexes and a lower MRC sum score at baseline. A previous study described the natural history of 38 treatment-naive patients with MMN retrospectively. Patients with longer disease duration (n = 10) had significantly lower MRC sum scores and a higher number of affected regions. None of the patients experienced spontaneous improvement or a relapsing remitting course.13 Taylor et al. longitudinally assessed 18 patients

www.ridderprint.nl

Made with FlippingBook

RkJQdWJsaXNoZXIy MTk4NDMw