Spinal Cord in Motor Neuron Diseases Using a Multi-Parametric MRI Approach

Université Pierre et Marie Curie Cerveau Cognition Comportement Laboratoire d’Imagerie Biomédicale / Systèmes Dynamiques Anatomo-Fonctionnels chez l’Homme

Thèse de Doctorat en Neurosciences

Présentée par Mohamed-Mounir EL MENDILI

Analysis of the structural integrity of the spinal cord in motor neuron diseases using a multi-parametric MRI approach

Dirigée par Véronique MARCHAND-PAUVERT et Pierre-François PRADAT

Présentée et soutenue le 13 Décembre 2016 devant le jury composé de Devant un jury composé de :

M. Peter BEDE (PU-PH) Trinity College Dublin, Irland Rapporteur

Mme. Virginie CALLOT (DR) Aix-Marseille Université Rapporteur

M. Philippe CORCIA (PU-PH) Université François-Rabelais, Tours Examinateur

M. Stéphane LEHERICY (PU-PH) Université Paris VI Examinateur

Mme. Véronique MARCHAND-PAUVERT (DR) Université Paris VI Directeur de thèse

M. Pierre-Francois PRADAT (PH) Université Paris VI Co-directeur de thèse

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

À tous les patients atteints d’une des maladies du motoneurone

Contents

Contents ...... i

Remerciements ...... iii

List of Tables ...... iv

List of Figures ...... v

Glossary ...... vi

Abstract ...... vii

Publications ...... ix Articles arising from this work ...... ix Additional contributions ...... x

1. Introduction ...... 1

2. Background ...... 2 Motor neuron diseases ...... 2 Historical setting ...... 2 Amyotrophic lateral sclerosis ...... 3 Spinal muscular atrophy ...... 14 Spinal cord MRI ...... 19 Principals of MRI ...... 19 Challenges ...... 20 Clinical setting ...... 24 Research setting ...... 25 Advanced MRI ...... 26

3. Objectives ...... 38

4. Results ...... 39 Methodological studies ...... 39 Study 1 - Validation of a semi-automated spinal cord segmentation method ...... 39 Study 2 - Fast and accurate semi-automated segmentation method of spinal cord MR images at 3T applied to the construction of a cervical spinal cord template ...... 48 Multi-parametric MRI studies in MND ...... 71

- i -

Study 3 - Electrophysiological and spinal imaging evidence for sensory dysfunction in amyotrophic lateral sclerosis ...... 71 Study 4 - Multi-parametric spinal cord MRI as potential progression marker in amyotrophic lateral sclerosis ...... 83 Study 5 – Cervical spinal cord atrophy profile in adult SMN1-linked SMA ...... 92

5. Discussion ...... 106 Methodological developments ...... 106 Multi-parametric MRI studies in MND ...... 108 Perspectives ...... 111

Annex A ...... 113

Annex B ...... 124

References ...... 142

- ii -

Remerciements

Le Papillon et la Tulipe

Sur la tulipe, un papillon se pose. « D’où vient, dit-il, la douce odeur Qui s’éxhale aujourd’hui de ta charmante fleur ? » La tulipe répond : « C’est que je suis éclose À coté d’une rose. »

Pour acquérir politesse et bonté, Fréquentons les meilleurs de la société.

Bataille (Frédéric). Poète, né à Mandeure (Doubs) en 1850.

Je souhaire remercier à travers cette fable tous les meilleurs de la société que j’ai pu fréquenter durant toutes ces 7 dernières années passées dans la recherche scientifique aux : LIF, LIB, CENIR, ICM, Service de Neuroradiologie, Département des Maladie du Système Nerveux et l’IRME. Je suouhaite aussi remercier les meilleurs de la société que j’ai pu fréquenter en dehors de l’envirronement de la recherche scientifique tout au long de ces 16 dernières années d’immigration à Paris, Hambourg, Madrid, Quito, Montréal, Grenoble, Strasbourg, Tours et Nantes. Je remercie particulierement mes anciens et actuels maîtres de la recherche scientifique : Olivier David, Habib Benali, Julien Cohen-Adad, Serge Rossignole et Véronique Marchand-Pauvert. Je remerci infiniment mon maître et ami Pierre-François Pradat pour son soutien scientifique et moral, pour ses encouragements et particulièrement pour sa confiance en mes idées même celles des plus folles, tu es l’un des meilleurs de la société que j’ai fréquenté et que je compte bien continuer à fréquenter. Je remercie mes étudiants et d’une certaine manière mes maîtres aussi : Noémie, Brice et Raphaël. Je remercie mes amis et/ou amours pour leur soutien sans faille tout au long de cette thèse et au- delà : Tania, Lala Salma et Sidi Waldo, Sarita, Nadjia et Raphaël, Clara, Mathieu, Emma et Eric. Enfin, je remerci ma famille, ma mère, mes deux frères et mon plus grand maître, mon père.

- iii -

List of Tables

Table 1. Principal characteristics of ALS phenotypes ...... 10 Table 2. The SMN2 gene copy number and the related motor performance in SMA ...... 15 Table 3. The spin density and the relaxation time in some tissues ...... 19 Table 4. Detected spinal cord MRI abnormalities in “mimic” MND ...... 25 Table 5. Correlations between DTI indexes and white matter microstructures...... 35

Tableau 1 Tableau 2 Tableau 3 Tableau 4 Tableau 5

- iv -

List of Figures

Figure 1. The human motor system ...... 4 Figure 2. The worldwide crude incidence distribution of ALS ...... 5 Figure 3. Patterns of motor involvement in different ALS phenotypes ...... 8 Figure 4. Cervical spine MRI abnormalities in a case of cervical spondylotic myelopathy ...... 13 Figure 5. Overview of non-neuromuscular systemic pathology in SMA ...... 15 Figure 6. A diagnostic algorithm for SMA ...... 18 Figure 7. Spinal cord content, and surrounding tissues and organs ...... 21 Figure 8. Comparison of 3T and 7T C-spine images ...... 22 Figure 9. MRI ghosting artifacts caused by patient motion ...... 23 Figure 10. Image distortion in DTI sequence ...... 24 Figure 11. Slices, saturation bands and shim prescriptions for DTI acquisition ...... 26 Figure 12. Derived indexes from cross-sectional approach ...... 28 Figure 13. Angular correlations in SCI patients ...... 29 Figure 14. VBM analysis - Regional distribution of cord atrophy in MS patients ...... 31 Figure 15. TBM analysis - Regional distribution of cord atrophy related to aging ...... 32 Figure 16. Model of the dynamic of axonal and grey matter degeneration in ALS ...... 110

Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 Figure 12 Figure 13 Figure 14 Figure 15 Figure 16

- v -

Glossary

ALS Amyotrophic lateral sclerosis ALSFRS-R Revised Amyotrophic Lateral Sclerosis Functional Rating scale ASM active surface method CNR Contrast-to-noise ratio CNS Central nervous system CSF Cerebrospinal fluid CST Corticospinal tract DTBM Double threshold-based method DTI Diffusion tensor imaging FA Fractional anisotropy FOV Field-of-view HARDI High-angular resolution imaging LMN Lower motor neuron MD Mean diffusivity MEP Motor evoked potetial MMT Manual muscle testing MND Motor neuron diseases MRI Magnetic resonance imaging MT Magnetization transfer MTR Magnetization transfer ratio PLS Primary lateral sclerosis PMA Primary muscular atrophy RD Radial diffusivity SBM Surface-based morphometry SCI Spinal cord injury SEP Sensory evoked potentials SNR Signal-to-noise ratio SMA Spinal muscular atrophy SMN1 Survival motor neuron-1 gene TMS Transcranial magnetic stimulation UMN Upper motor neuron

- vi -

Abstract

Degenerative motor neuron diseases (MND) are characterized by a progressive dysfunction and loss of ventral horn motor neurons of the spinal grey matter. Beyond this common anatomical susceptibility, which is responsible for a progressive and diffuse weakness, other neurological systems are also impaired. The corticospinal tract (CST) degeneration is a classical feature of amyotrophic lateral sclerosis (ALS), which is the most common adult onset motor neuron disease, but a more widespread multisystem involvement is now well recognized. In particular, early sensory system involvement has been demonstrated in animal models of ALS and also of survival motor neuron 1 gene linked spinal muscular atrophy (SMN1-linked SMA). In human patients, magnetic resonance imaging (MRI) has emerged as the most powerful approach at the brain level to extract quantitative data on neuronal loss, axonal degeneration and demyelination in degenerative conditions. Studies at the spinal cord levels are scarce mainly because of technical and methodological difficulties.

The objective of the present project was to use a multi-parametric MRI approach at the spinal cord level to analyze grey and white matter structures that are impaired in MND, their temporal alterations during the disease course and the functional correlates, as assessed by clinical and electrophysiological examinations. In ALS patients, we showed that spinal cord degeneration extends beyond the motor neurons and the CST and encompasses the ascending sensory axonal tract both structurally, as shown by diffusion tensor imaging (DTI) MRI, and functionally, as shown by somatosensory evoked potentials. Using a longitudinal approach, we showed a significant worsening of the spinal cord atrophy, a metric that mainly reflects grey matter loss, but no significant CST changes as assessed by DTI. These findings support the concept that white matter lesions may be relatively constant during the symptomatic period and that clinical progression is mainly dependent on the extension of grey matter degeneration. Finally, in adult SMN1-linked SMA we demonstrated that MRI atrophy predominates in spinal segments that innervate proximal muscles, which fits with the classical proximal predominance of muscle weakness in SMN1-linked SMA. However, the presence of a clear distal weakness together with the missing correlations between spinal atrophy and muscle deficit suggest that loss of motor neuron cell bodies may not recapitulate all mechanisms responsible for weakness. A longitudinal study, with a multimodal MRI with new MRI developments and electrophysiological evaluation, is ongoing to better analyze the structures impaired in SMA and their temporal alterations in relation with the clinical changes.

Keywords: Human, motor neuron diseases, amyotrophic lateral sclerosis, spinal muscular atrophy, spinal cord, motoneurons, sensory afferents, sensory evoked potentials, magnetic resonance

- vii -

imaging, diffusion tensor imaging, magnetization transfer imaging, segmentation, atrophy measurements, surface-based morphometry.

- viii -

Publications

Articles arising from this work

1. Grolez G, Moreau C, Danel-Brunaud V, Delmaire C, Lopes R, Pradat PF, El Mendili MM, Defebvre L, Devos D. The value of magnetic resonance imaging as a biomarker for amyotrophic lateral sclerosis: a systematic review. BMC Neurol. 2016 Aug 27;16(1):155.

2. El Mendili MM, Lenglet T, Stojkovic T, Behin A, Guimarães-Costa R, Salachas F, Meininger V, Bruneteau N, Le Forestier N, Laforêt P, Lehéricy S, Benali H, Pierre-François Pradat PF. Cervical Spinal Cord Atrophy Profile in Adult SMN1-Linked SMA. PLoS One. 2016 Apr 18;11(4):e0152439.

3. El Mendili MM, Chen R, Tiret B, Villard N, Trunet S, Pélégrini-Issac M, Lehéricy S, Pradat PF, Habib Benali H. Fast and Accurate Semi-Automated Segmentation Method of Spinal Cord MR Images at 3T Applied to the Construction of a Cervical Spinal Cord Template. PLoS ONE 2015; 10(3):e0122224.

4. El Mendili MM, Chen R, Tiret B, Pélégrini-Issac M, Cohen-Adad J, Lehéricy S, Pradat PF, Benali H. Validation of a semiautomated spinal cord segmentation method. J Magn Reson Imaging 2015; 41(2):454-9.

5. Iglesias C, Sangari S, El Mendili MM*, Benali H, Marchand-Pauvert V, Pradat PF. Electrophysiological and spinal imaging evidences for sensory dysfunction in amyotrophic lateral sclerosis. BMJ Open 2015; 5:e007659. *2nd author

6. El Mendili MM, Cohen-Adad J, Pélégrini-Issac M, Rossignol S, Morizot-Koutlidis R, Marchand-Pauvert V, Iglesias C, Sangari S, Katz R, Lehéricy S, Benali H, Pradat PF. Multi- parametric spinal cord MRI as potential progression marker in amyotrophic lateral sclerosis. PLoS ONE 2014; 9(4):e95516.

7. Pradat PF, El Mendili MM. Neuroimaging to investigate multisystem involvement and provide biomarkers in amyotrophic lateral sclerosis. Biomed Res Int. 2014;2014:467560.

8. Pradat PF, El Mendili MM, Cohen-Adad J, Morizot-Koutlidis R, Lehericy S, Benali H. IRM multi-paramétrique de la moelle épinière médullaire dans les pathologies du motoneurone. Revue Neurologique 2014; 170, A222-A223. (French article) - ix -

Additional contributions

9. Sangari S, Iglesias C, El Mendili MM, Benali H, Pradat PF, Marchand-Pauvert V. Impairment of sensory-motor integration at spinal level in amyotrophic lateral sclerosis. Clin Neurophysiol. 2016 Apr;127(4):1968-77.

10. Hanna-Boutros B, Sangari S, Louis-Solal G, El Mendili MM, Lackmy-Valee A, Marchand- Pauvert V, Knikou M. Corticospinal and Reciprocal Inhibition Actions on Human Soleus Motoneuron Activity During Standing and Walking. Physiol Rep 2015; 3(1):e12276.

- x -

1. Introduction

The spinal cord is the entry for sensory information toward the Central Nervous System (CNS) and the exit for motor commands from the CNS to the limbs and the trunk. Lesions of the spinal cord due to neurodegenerative diseases thus lead to major sensory and/or motor consequences whose severity depends on the extent of the lesions. Degenerative motor neuron diseases (MND) form a subgroup of neurodegenerative diseases, and are characterized by a progressive dysfunction and loss of ventral horn motoneurons of the spinal grey matter. Beyond this common anatomical susceptibility, which is responsible for a progressive and diffuse weakness, other neurological systems are also impaired. The corticospinal tract (CST) degeneration is a classical feature of amyotrophic lateral sclerosis (ALS), which is the most common adult onset MND, but a more widespread multisystem involvement is now well recognized. In particular, sensory system involvement has been demonstrated in animal models of ALS. Proximal spinal muscular atrophy (SMA) is the second most common MND and shares with ALS a common feature of ventral horn motoneurons loss. The reason of the classical proximal predominance of the deficit in SMA remains unknown.

In human patients, magnetic resonance imaging (MRI) has emerged as the most powerful approach at the brain level to extract quantitative data on neuronal loss, axonal degeneration and demyelination in degenerative conditions. Studies at the spinal cord levels are scarce mainly because of technical and methodological difficulties. Despite, clinical conventional MRI remains extremely useful in describing a large range of spinal cord abnormalities. However, conventional MRI lacks sensitivity and specificity in providing quantitative data on the integrity of white matter tracts and grey matter layers in patients with sensory and/or motor signs detected during clinical and electrophysiological evaluation, which is particularly the case in MND. More advanced and quantitative MRI techniques are used in the research setting, enabling to capture at the spinal level sensory-motor system involvement in MND.

The objective of the present project was to optimize and use advanced multi-parametric MRI approach at the spinal cord level to analyze grey and white matter structures that are impaired in MND, their temporal alterations during the disease course and the functional correlates, as assessed by clinical and electrophysiological examinations.

The manuscript is organized as follows: In Chapter 2, a literature review will present: i) motor neuron diseases with an especial focus on ALS and SMA; ii) spinal cord MRI with an especial focus on advanced MRI techniques that are commonly used in the pathological context. The objectives of this thesis will be presented in Chapter 3. Five published articles, which constitute the core of the present thesis, will be presented in Chapter 4. Finally, Chapter 5 will discuss our findings and future directions.

- 1 -

2. Background

Motor neuron diseases

Historical setting Amyotrophic lateral sclerosis Spinal muscular Atrophy

Spinal cord MRI

Principals of MRI Challenges Clinical setting Research setting Advanced MRI

Motor neuron diseases

Historical setting

In the second half of the 19th century, with the contribution of other clinicians, Jean-Martin Charcot, a prominent neurologist from Pitié-Salpêtrière Hospital, recognized and described the clinical, anatomical and physiological features of a large spectrum of MND, including ALS, progressive muscular atrophy (PMA), and primary lateral sclerosis (PLS) (Goetz, 2000). Theses characterizations were based on a rigorous description of 20 cases and five autopsies. J.M. Charcot highlighted the major contribution of lower motor neuron (LMN) loss in muscles weakness and amyotrophy as well as the association with pyramidal tract involvement (“lateral sclerosis”) causing spasticity and contractures (Goetz, 2000; Katz et al., 2015). Dispite the historical association of ALS recognition with Jean-Martin Charcot, Jacob Augustus Lockhart Clarke, a prominent British physician, described the clinical and the pathophysiological features of an ALS case in a paper published with Charles Bland Radcliffe (Radcliffe and Lockhart Clarke, 1862), 3 years ahead of Charcot’s description of this disease (Turner et al., 2010). In parallel, several familial forms of MND were also described. For instance, by the end of the 19th century, Johan Hoffman and Guido Werdnig described the clinical and the anatomopathological features - 2 -

of new forms of hereditary infantile paralysis, known as infantile SMA (Dubwitz, 2009). Their spinal cord autopsies showed notably ventral roots atrophy and loss of ventral horn motoneurons. Later in the 20’s century, Victor Dubwitz, Erik Klas Hendrik Kugelberg, Lisa Welander and Klauss Zerre described the other forms of familial SMA, i.e. intermediate, juvenile and adult SMA (Dubwitz, 1964, 1967; Kugelberg and Welander, 1986; Zerre et al., 1995, 1997). DNA sequencing in the 90’s allowed associating some familial forms of MND with specific genes mutations, notably in hereditary SMA with survival motor neuron 1 gene deletion (Lefebvre et al., 1995), and in ALS with superoxyde dismutase 1 (SOD1), followed by the identification of an increasing number of causative genes (TAR DNA-binding protein (TDP-43), FUsed in Sarcoma (FUS), chromosome 9 open reading frame 72 (C9ORF72)…) (Renton et al., 2014).

In more than 150 years, rigorous and intensive effort was made by several generations of clinicians and researchers to: i) describe the demographical, clinical, pathophysiological and recently the genetic features of the MND forms; ii) figure out if some MND forms are part of a single entity, i.e. progressive bulbar palsy and ALS (Joseph Dejerine in 1883), or could be considered as a continuum spectrum of MND, i.e. PLS, ALS and PMA (Van Den Berg 2003; Swinnen and Robberecht, 2014); iii) develop tools enabling accurate diagnosis of MND as well as predict and monitor disease progression. However, the pathophysiological mechanisms of neurodegeneration are not fully understood in MND, notably in ALS and SMA, which are the two most common forms of human MND observed in adults and in children/young adults, respectively (D'Amico et al., 2011;Wijesekera and Leigh, 2009).

The advent of neuroimaging and particularly magnetic resonance imaging (MRI) by the end of 80’s provided new insights into the pathophysiology of MND (Goodin et al., 1988). MRI allows a noninvasive and in-vivo investigation of the structural (structural imaging) and functional (functional imaging) integrity of the CNS during the course of MND (Unrath et al., 2010; Shen et al., 2015; Turner and Verstraete, 2015). Notably, MRI enables investigation of the motor system structures that are classically impaired in MND (Rocha et al., 2012). It also allows assessing the anatomo-functional degenerative process beyond the motor system (van der Graaff et al., 2008; Chiò et al., 2012). Particularly, many studies used MRI to investigate the anatomic substrate of cognitive symptoms in ALS (Rajagopalan and Pioro, 2014; Schuster et al., 2014; Ambikairajah et al., 2014).

Amyotrophic lateral sclerosis

Anatomopathology

- 3 -

Amyotrophic lateral sclerosis is the most common MND. ALS is characterized by a progressive loss of upper motor neurons (UMN), located at primary motor cortex, i.e. pyramidal cells, and LMN, which are the brainstem motor nuclei and alpha motoneurons located in the anterior horn of the spinal cord (Figure 1).

Figure 1. Schematic diagram of the cortical motor system, the CST generated by corticospinal motor neuron (CSMN), descending from the motor cortex, through the internal capsule. Most CST fibers (more than 80% in humans) decussate in the caudal medulla, before descending further through the spinal cord and connecting at the respective height to the alpha motor neuron (AMN). The AMN then projects toward the muscles to form neuromuscular junctions. Modified from Winner et al. (2014).

At the spinal cord level, in addition to the lateral CST degeneration, the anterior/ventral CST is also involved (Swash et al., 1988). At the brain level, animal and human studies show evidence of involvement of other structures beyond primary motor cortex, such as the prefrontal cortex, the supplemental motor area, the primary and secondary somatosensory cortex, the basal ganglia and the cerebellum. All these findings support the concept of ALS as a multi-system neurodegenerative disease - 4 -

(Bede et al., 2016).

Incidence and prevalence

The worldwide annual incidence of ALS is 1.75 (range = 1.55-1.96)/100,000 individuals, 2.03 (1.79-2.37) in men and 1.45 (1.25-1.64) in women (Marin et al., 2016) (Figure 2). In Europe, crude incidence ranges from 1.92 (1.49-2.34) in northern Europe to 2.35 (1.79-2.92) in western Europe per 100,000 individuals (Marin et al., 2016). ALS incidence increases with age, with a peak between 60 and 75 years of age (Marin et al., 2016; Chiò et al., 2013).

The peak of incidence for men is at 70-74 years and at 65-69 years for women (Logroscino et al., 2015). The gender ratio of incidence between men and women is 1.3:1 (Logroscino et al., 2015).

Figure 2. Distribution of ALS worldwide: crude incidence. PYFU: person-years of follow-up. From Marin et al. (2016).

The prevalence of ALS is subject to large variations among studies, ranging from 1.1/100,000 population in Yugoslavia (Alcaz et al., 1996) to 8.2/100,000 in Faroe Islands (Joensen, 2012). Prevalence doubles by a range of 5 years: it reaches 25/100,000 individuals between 60 and 70 years. Prevalence in men is higher than in women but the gender ratio decreases with age (McCombe et al., 2010; Manjaly et al., 2010).

- 5 -

Genetic features

The role of genetic factors in ALS has been demonstrated by the identification of the SOD1 gene mutation in 1993 (Rosen et al., 1993). Since then, more than twenty gene mutations have been found in familial ALS (Yamashita and Ando, 2015), the most important being C9ORF72 which was found in nearly 40% of familial cases and about 6% of sporadic forms. Familial ALS represents about 10% of ALS cases (Bayrn et al., 2011). The inheritance is mostly autosomal dominant with rare autosomal recessive and recessive X-linked modes (Andersen et al., 2011).

Familial ALS forms are characterized by an earlier age at onset compared to sporadic forms (49.7 ± 12.3 versus 61.8 ± 3.8 years), with a median survival of 2.7 years, a gender ratio of 1 to 1.3:1, and a predominance of lower limbs onset (Bali et al., 2016; Camu et al., 1999). Familial ALS disease duration has a bimodal profile: a small number of patients have a very poor prognosis (< 2 years), and most have better prognosis than the sporadic form (> 5 years) (Camu et al., 1999).

Pathophysiology

Multiple pathogenic mechanisms implicated in neurodegeneration in ALS were identified. The most likely neurodegeneration mechanisms are: protein aggregation, neuronal inflammation, oxidative stress, and motoneurons hyper-excitability induced by glutamate excess. In parallel, several mechanisms were proposed to explain the spreading of ALS symptoms.

Protein aggregate

ALS shares the presence of protein aggregates that form inclusions in motoneurons with many neurodegenerative diseases (Ross and Poirier, 2004; Kumar et al., 2016). The TDP-43 is the main component of these protein aggregates, which is also mutated in some familial forms of ALS (Sreedharan et al., 2008). The TDP-43 accumulates in neurons and non-neuronal cells in sporadic forms and the familial forms, except for familial ALS with SOD1 mutation (Mackenzie et al., 2007). Other proteins form these aggregates: cystatin C and transferrin in Bunina bodies, SOD1 or protein FUS/Translated in liposarcoma (FUS/TLS). These aggregates are secondary to impaired protein degradation components (proteasome or autophagy mechanisms).

Altered ARN metabolism

Altered ARN mechanism was suggested in ALS because TDP-43 and FUS, which are mutated in ALS, are involved in ARN splicing, transport and translation (Verma, 2011). Focal accumulations of RNA in nuclei exist in the C9ORF72 gene mutation, which block transcription factors and disrupt cell viability (Ishiura and Tsuji, 2015).

- 6 -

Neurotransmitters and motoneurons afferent abnormalities

Neuronal excitotoxicity was the topic of significant research, despite the fact this mechanism is considered as a delayed mechanism in neurodegeneration (King et al., 2016). The main arguments supporting neuronal excitotoxicity are: increased glutamate level in the cerebrospinal fluid (CSF) (Spreux-Varoquaux et al., 2002), and the effect of riluzole (Lacomblez et al., 1996). Neuronal death was linked to calcium accumulation in neuronal cells due to a relative defect of calcium degradation systems.

Recently a great attention has been given to motoneuron afferent and neurotransmitter disturbances in ALS (Pradat and Dupuis, 2014). Several afferents project directly or through interneurons into LMN and UMN, which determines their excitability condition, function and survival. Dependent GABA inhibitory cortical pathway abnormalities have been demonstrated in Positron Emission Tomography (PET) (Lloyd et al., 2000) and NMR spectroscopy (Foerster et al., 2012). Abnormal spinal glycinergic inhibitory systems could lead to an increase in motoneurons excitability (Chang and Martin, 2011). Human and animal model studies suggested that impaired afferents from neuromuscular spindles contribute to excitability aggravation (Malaspina et al., 2010; Iglesis et al., 2015).

Implication of non-neuronal cells and neural inflammation

Animal models showed that neurodegeneration in ALS is not a cell-autonomous process (Clement et al., 2003) but depends on astrocytes, microglia, oligodendrocytes and Schwann cells. It is important to mention that inflammatory reaction related to microglia can be toxic but also neuroprotective (Beers et al., 2006; Xiao et al., 2007). Moreover, several studies highlighted the possible implication of the neuromuscular junction and muscle in the pathogenesis of motoneuron degeneration (Dupuis et al., 2002; Fischer et al., 2004; Bruneteau et al., 2013; Pradat et al., 2007, 2011).

Spread mechanisms

It has been recently suggested that the regional spread of neurodegeneration during the disease progression is associated with prion-like (proteinaceous infectious particles) propagation mechanisms (Pradat et al., 2015, Juker et al., Nature 2013). In ALS, prion-like mechanisms could explain the intercellular spread by neuroanatomical networks, the trans-synaptic passing of mutated proteins or by the spreading of toxic factors throughout the cerebrospinal fluid or blood (Pradat et al., 2015).

Additional mechanisms

Additional mechanisms are involved in motoneurons loss: oxidative stress, mitochondrial

- 7 -

dysfunction, disruption of axonal transport and endoplasmic reticulum stress (Robberecht and Philips, 2013).

Phenotypes

Amyotrophic lateral sclerosis is a heterogeneous condition with eight recognized phenotypes (Ludolph et al., 2015; Chiò et al., 2011), according to the site of degeneration (bulbar, cervical or lumbar) and the balance between UMN and LMN degeneration (Figure 3): 1) classical phenotype (as described by Charcot); 2) bulbar phenotype; 3) flail arm phenotype; 4) flail leg phenotype; 5) respiratory phenotype; 6) predominant UMN phenotype; 7) pure LNM phenotype; 8) pure UMN phenotype. Principal characteristics of the eight phenotypes are summarized in Table 1.

Figure 3. Patterns of motor involvement in different ALS phenotypes. Red indicates LMN involvement, blue indicates UMN involvement. Darker shading indicates more-severe involvement. From Swinnen and Robberecht (2014).

Classical (Charcot’s) phenotype

Charcot initially described the classical ALS form. The mean age at onset is 63 years, with a gender ratio of incidence between men and women of 1.65:1. This form is characterized by progressive amyotrophic motor weakness in the upper limbs, often asymmetrical, associated with a pyramidal syndrome without important lower limbs atrophy. Bulbar involvement occurs later in the disease - 8 -

course, and is associated with an emotional lability and sometimes with a pseudo-bulbar syndrome. Median of survival from onset is 2.6 years with 10% survival rate after 10 years from the symptoms onset (Chiò et al., 2011).

Flail arm

Flail arm phenotype is characterized by upper limbs deficit due to motoneurons degeneration. This phenotype is associated with delayed moderate signs of UMN involvement. The mean age at onset is 63 years, with a gender ratio of incidence between men and women of 4:1. Disease progression is slow, with a median of survival from the onset of 4.0 years and a survival rate of 17.4% at 10 years (Chiò et al., 2011).

Flail leg phenotype or Marie-Patrikios pseudo-polyneuritic form

Flail leg phenotype is characterized by unilateral and distal weakness and wasting in the lower limbs. Involvement starts at the distal part of the leg and progresses to the proximal one. The contralateral leg is involved after few weeks or months from the onset of symptoms. Disease progression is slow, with a median survival from the onset of 3.0 years and a survival rate of 12.8% at 10 years, which is similar to the Charcot’s form (Chiò et al., 2011).

Bulbar phenotype

Disease onset is characterized by a dysphagia or dysarthria, rapidly associated with pseudo-bulbar signs with yawning, spasmodic laughter and cries. The mean age at onset is 69 years, higher than in the classical syndrome, without gender incidence predominance in men. Median of survival from the onset is 2.0 years, shorter than the classical syndrome, with only 3.4% survival rate at 10 years from the symptoms onset (Chiò et al., 2011).

Respiratory phenotype

Respiratory impairment is rarely the predominant symptom at onset in ALS (Mitsumoto et al., 1998). However, in some cases respiratory impairment can be the revealing symptom. The mean age at onset in these respiratory forms is 62 years, with predominance in men (gender ratio of incidence 6:1). Respiratory insufficiency is revealed by signs of chronic alveolar hyperventilation (Howard et al., 1989) or by a severe acute respiratory failure that requires rapid initiation of artificial ventilation (de Carvalho et al., 1996). Neurological examination frequently reveals other signs of motoneurons involvement (Howard et al., 1989), although they can be absent in some cases. This phenotype is highly progressive, with a median of survival of 1.4 years and no patients survive after 10 years from the onset (Chiò et al., 2011). - 9 -

Predominant UMN phenotype or pyramidal phenotype

Pyramidal signs are predominant and are associated with a moderate motor deficit and a marked hypertonia in the lower limbs. Patients show slow progression compared to those with the classical phenotype, with a median of survival of 6.3 years and 31.9% survival rate at 10 years from the onset (Chiò et al., 2011).

Pure LMN phenotype or PMA

The phenotype is clinically characterized by diffuse and exclusive LMN degeneration. However, in a study on 91 PMA patients (Kim et al., 2009), 20 patients developed UMN signs after a median follow-up of 23.4 months from diagnosis. Post-mortem studies show additional evidence of the CST and UMN degeneration in PMA (Leung et al., 1999; Ince et al., 2003), which lead to the inclusion of PMA in the ALS spectrum. Progression in PMA is slow, with a median survival of 7.3 years (Chiò et al., 2011). Gender ratio of incidence is higher in men (2.6:1) (Kim et al., 2009).

Pure UMN phenotype or PLS

Upper motor neurons degeneration clinical signs are exclusive in this phenotype: pyramidal syndrome predominating in lower limbs and pseudo-bulbar syndrome, without significant muscular atrophy. Patients show slow disease progression, and at least 4 years delay is needed to observe all previous clinical signs, and generally LMN degenerating occurs before this delay (Gordon et al., 2006). Post-mortem examination showed evidence of LMN in most cases, which suggest that PLS is part of the ALS spectrum (D’Amico et al., 2013). Among all ALS phenotypes, PLS has the longest disease evolution, with median survival of 13 years and 71% survival rate at 10 years from symptom onset (Chiò et al., 2011).

Table 1. Principal characteristics of ALS phenotypes from an Italian register of 1332 ALS patients. Modified from Chiò et al. (2011).

Number of cases Age at onset Median survival Phenotype (%) Mean ± SD in years Classic 404 (30.3) 62.8 ± 11.3 2.6 Flail arm 74 (5.5) 62.6 ± 11.8 > 4 Flail leg 173 (13.0) 65.0 ± 9.6 3 Bulbar 456 (43.2) 68.8 ± 9.7 2 Respiratory 14 (1.1) 62.2 ± 8.6 1.5 Pyramidal 120 (9.1) 58.3 ± 13.5 6.3 PMA 38 (2.9) 56.2 ± 11.3 7.3 PLS 53 (4.0) 58.9 ± 10.9 13 Total 1332 64.3 ± 11.3 −

- 10 -

Epidemiology

Amyotrophic lateral sclerosis is classically described as late-onset disease, with a mean ± SD age at onset of 61.8 ± 3.8 years (range 54-67) (Chiò et al., 2013). Mean age at onset for men is 60.8 years and 61.9 years for women (McCombe and Henderson, 2010). Median survival from onset ranges from 20 to 48 months (Marin et al., 2016; Chiò et al., 2002, 2013; Gordon 2013). However, in 10 to 26.6% cases survival rate exceeds 5 years and, in 5 to 13.3%, 10 years from symptom onset (Chiò et al., 2009; Pupillo et al., 2014).

Age is a strong prognostic factor in ALS, with a survival time inversely correlated with age at onset (Chiò et al., 2009). For instance, patients with onset before the age of 40 years have longer median survival, often more than 10 years (Testa et al., 2004; del Aguila 2003), and after 80 years, median survival is substantially reduced to less than 2 years (Forbes et al., 2004).

Clinical evolution

Progressive disability worsening is an important criterion for ALS diagnosis. The median survival from symptom onset ranges from 20 to 48 months (Marin et al., 2016; Chiò et al., 2002, 2013; Gordon 2013). Disease duration in ALS is heterogeneous and mainly depends on the phenotype and age at onset; some ALS patients will have a very fast disease progression and others a very long and slow one, with 10 to 26.6% of cases surviving more than 5 years from onset (Chiò et al., 2009; Pupillo et al., 2014). Consequently, ALS diagnosis should systematically be considered even with the absence of worsening after a 12 months from the onset.

The most common causes of death in ALS patients are a respiratory failure (77 %), which is usually linked to a terminal respiratory insufficiency (58%), and pneumonia favored by deglutition disorders (14 %) (Gil et al., 2008). However, causes of death are not known in 13% of cases (Gil et al., 2008).

Rulizole is the only treatment that has proved effective to slow ALS disease progression (Lacomblez et al., 1996). Treatment action mechanisms are poorly understood and are probably numerous, involving at least altered glutamate metabolism and sodium channel function modulation (Cheah et al., 2010; Vucic et al., 2013). Taken orally twice a day, a 100 mg/day dose (50 mg morning and evening), riluzole decreases the relative risk of death or tracheostomy by 35% over a period of 18 months after adjustment for risk factors (Lacomblez et al., 1996).

- 11 -

Para-clinical examinations

Electromyography and sensory evoked potentials

Electrophysiology examination is the key tool for the positive and differential diagnosis of ALS. Electromyography produces quantitative measures of LMN degeneration, confirming detected signs at the clinical examination at an early stage even in patients with predominant or pure UMN phenotype (de Carvalho et al., 2008). Needle examination includes the detection of abnormal activity at rest reflecting denervation such as fibrillation and fasciculation potentials. According to the new diagnosis criteria of Awaji-Shima, involvement signs in different anatomical regions detected by electromyography (bulbar, cervical, thoracic and lumbar) have the same diagnosis value as those from clinical examination (de Carvalho et al., 2008). Conduction velocity is normal and conduction blocks are not detected, excluding a classic differential diagnostic of multifocal motor neuropathy with conduction blocks. The decrease of sensory evoked potentials (SEP), especially in the lower limbs, lead to suspect a spinal and bulbar muscular atrophy and to carry out a genetic testing (Pradat, 2014). Sensory evoked potentials are useful in the differential diagnosis measure but the detection of abnormalities does not exclude the diagnosis of an ALS (Iglesias et al., 2015)

Motor evoked potential

Transcranial magnetic stimulation (TMS) is a noninvasive brain stimulation technique. This technique generates a changing magnetic field pulses that penetrates the scalp and the skull to reach the small cortex area below the coil center. These pulses induce a secondary ionic current flow that activates neurons at the stimulated site, e.g. primary motor cortex in ALS, and evoking indirect waves throughout their axons up to the muscle (Hallett et al., 2007). TMS is used to investigate the excitability of brain, brainstem and spinal cord regions, i.e. UMN and LMN. Resulting measures from TMS are motor evoked potential (MEP) conduction time, amplitude and threshold, recruitment curve, cortical silent period and intracortical inhibition and facilitation (Hallett et al., 2007).

Transcranial magnetic stimulation is used to reveal UMN degeneration at an early stage of ALS even in predominant LMN form, which is clinically difficult to detect. However, the sensitivity of TMS, especially at an early stage of the disease, is debated and this technique is highly dependent on the experience of the examiner, limiting its generalization as a diagnostic tool in clinical practice (de Carvalho et al., 2008). Triple stimulation technique is a variant of TMS, and was developed to reduce the variability of MEP amplitude and to increase the sensitivity of MEP measures compared to conventional methods (de Carvalho et al., 2008).

- 12 -

MRI

Conventional brain and spinal cord MRI is mainly used to exclude “ALS-mimic” syndromes such as cervical spondylotic myelopathy, which is a very common syndrome in seniors (narrow spinal canal associated with a myelopathy intramedullary T2-hypersignal) or, more rarely, a Hirayama disease (“snake eyes” T2*-hypersignal appearance on the cervical anterior horns) (Ellingson et al., 2015; Sasaki, 2015; Lebouteux et al, 2014; Filippi et al., 2010) (Figure 4). Several studies, using various modalities (T2*, fluid-attenuated inversion recovery (FLAIR) or fast-spin echo proton density-weighted imaging), have shown hyperintensity in the white matter along the CST, from the centrum semiovale to the brainstem (Graham et al., 2004; Hecht et al., 2001; Comi et al., 1999; Goodin et al., 1988; Hofmann et al., 1998; Kato et al., 1997; Tanabe et al., 1998; Mirowitz et al., 1989; Waragai et al., 1997; Cheung et al., 1996; Thorpe et al., 1996). However, such abnormalities are rare, non-specific, not readily quantifiable and do not correlate with disease severity or rate of progression (Winhammar et al., 2005). A cortical atrophy, which is predominant in the frontal region, and a characteristic T2* or FLAIR hypointensity at the level of the primary motor cortex was described in the literature but is exceptionally detectable in clinical practice (Hecht et al., 2001; Oba et al., 1993). Conventional MRI lacks sensitivity and specificity to detect abnormalities in ALS, which limits its usefulness in clinical practice. More advanced and quantitative MRI techniques are used in the research setting, enabling investigation of new aspects of the CNS structure and function involvement in ALS (Agosta et al., 2010; Turner et al., 2015).

Figure 4. Cervical spine MRI of a 45-year-old woman with cervical spondylotic myelopathy demonstrating extensive spinal cord atrophy and T2 weighted signal change at the C5 and C6 levels despite the absence of frank spinal cord compression. Modified from Ellingson et al. (2015).

- 13 -

In two recent papers, we performed a systematic review of MRI findings in ALS from conventional and advanced imaging, and their clinical correlations, and sought to establish whether particular techniques may have value as surrogate biomarkers in this disease as well as their potential use in clinical routine:

Annex A: Neuroimaging to investigate multisystem involvement and provide biomarkers in amyotrophic lateral sclerosis.

Annex B: The value of the magnetic resonance imaging as a biomarker in ALS: a systematic review.

Spinal muscular atrophy

Spinal muscular atrophy is a group of autosomal-recessive hereditary MND characterized by LMN loss, resulting in predominantly proximal muscle atrophy and weakness (Van der Berg et al., 2003). Muscle weakness is symmetrical with lower limbs more affected than upper limbs (Van der Berg et al., 2003). The disorder is caused in 95% of cases by deletions or mutations in the telomeric copy of the survival motor neuron gene-1 (SMN1) (Lefebvre et al., 1995).

Several studies, especially those from animal models and in-vitro cellular cultures, show evidence of the involvement of several non-motor cells as well as organs inside and outside the CNS in SMA pathology (Simone et al., 2015; Hamilton and Gillingwater, 2013):

- Inside the CNS: interneurons, sensory neurons, astrocytes and microglia.

- Outside the CNS: Schwann cells, neuromuscular junction and muscle cells, muscle specific macrophage cells, heart, liver, vascular system, bones, pancreas, lung and intestine.

Previous findings support the concept of SMA, as for ALS (see previous section), as a multisystem disorder (Figure 5).

- 14 -

Figure 5. Overview of non-neuromuscular systemic pathology in SMA. A summary of tissues and organs currently known to be affected in SMA. From Hamilton and Gillingwater (2013).

Genetics of SMA

SMA is caused in 95% of cases by homozygous deletions or mutations of exon 7/8 of the SMN1 gene on chromosome 5q13, producing a non-functional SMN protein (Lefebvre et al., 1995; Rodrigues et al., 1995). The SMN gene exists in two forms: a telomeric form (SMN1) and a centromeric form (SMN2). The SMN2 gene differs from the SMN1 gene by the exception of a C to T substitution in exon 7 that modifies splicing modulator and results in the exclusion of exon 7 in 90% of SMN2 mRNA transcription. The SMN2 gene mostly produces a truncated SMN protein, lacking exon 7, and is quickly degraded. However, the SMN2 gene produces a 10% full-length SMN protein, and thus partially compensates the lack of functional SMN produced by SMN1 gene. Several copies of the SMN2 gene exist in chromosome 5. Hence, the SMN2 gene copy number mainly defines the clinical severity of SMA (Lefebvre et al., 1997; McAndrews et al., 1997; Mailman et al., 2002; Zarkov et al., 2015) (Table 2). Indeed, the SMN2 gene copy number is inversely correlated with disease severity (Swoboda et al., 2005). However, the molecular and cellular mechanisms linking SMN1 gene mutation with the “preferential” LMN loss in SMA remains unclear (d’Errico et al., 2013).

Tableau 2. Association between the SMN2 gene copy number and current motor performance in the patients with SMA (11 SMA type II, 17 SMA type III and 8 SMA type IV). Modified from Zarkov et al. (2015).

SMN2 gene copy number Motor performance Mean ± SD Not able to sit 2.8 ± 0.5 Sitting with support 3.0 ± 0.0 Sitting independently 3.5 ± 0.8 Walking with support 3.6 ± 0.9 Walking independently 3.8 ± 1.0

- 15 -

SMN gene discovery led to the development of several animal models that mimic pathophysiological changes in SMA patients, ranging from severe to complete rescue SMA-like phenotypes, i.e. mouse, pig, zebrafish and drosophilia (Schmid and DiDonato, 2007; Monami and De Vivo, 2014). These animal models give insight into the genetic and molecular mechanisms involved in SMA pathology and enable the consideration of therapies mainly based on the compensation of SMN protein deficiency (Kolb and Kissel, 2015; Monami and De Vivo, 2014).

Clinical features and evolution

SMA is classified into four types based on the age at onset and the maximum achieved gross motor function (Zerres et al., 1995, 1997). The spectrum of SMA ranges from the most severe (type I), in which patients never manage the ability to sit unassisted and the majority died before two years, to late- onset (types IIIb and IV), in which patients have slow disease progression and normal life expectancy (Zerres et al., 1995; Piepers et al., 2008). In all SMA types, cognition remains intact during the course of the disease.

SMA types

The International SMA Consortium meeting (26-28 June 1992, Bonn, Germany), defined three homogeneous groups of SMA disease (types), based on age at onset of symptoms, disease severity and progression criteria (Munsat and Davies, 1992). This classification was further enriched by two population-based studies on SMA (Zerres et al., 1995, 1997). Four SMA types are actually recognized:

SMA type I

SMA type I, infantile-onset Werdnig–Hoffman disease, is the most common genetic cause of infantile death. SMA type I is the most severe and common among SMA types (60% of total SMA). The annual incidence for SMA type I ranges from 3.5 to 7.1/100,000 live births with a prevalence ranging from 0.1 to 0.15/100,000 total population (Jones et al., 2015). SMA type I starts before the achievement of walking, with an onset before 6 months. Type I is clinically characterized by profound hypotonia and a severe motor deficit extended to the four limbs and trunk. Patients are not able to sit unassisted, develop tongue-swallowing, weakness, and usually die from respiratory failure within the first 2 years (32% and 8% of survival probability at 2 and 10 years, respectively).

SMA type II

SMA type II, intermediate-onset Dubowitz disease, starts between 6 and 18 months (27% of total SMA). Patients are clinically characterized by a progressive weakness in legs more than in arms, associated with hypotonia and areflexia. Patients are able to sit unassisted but are not able to walk - 16 -

independently. Survival probability at 5 and 25 years are 98.5% and 68.5%, respectively.

SMA type III

SMA type III, mild-onset Kugelberg–Welander disease, starts after 18 months (12% of total SMA patients). Patients are clinically characterized by a progressive proximal leg weakness more than arms. Patients learn to walk without support but may need a wheelchair at some point of the disease course. Life expectancy is almost normal. The SMA type III group is heterogeneous in terms of maintaining the ability to walk during the course of the disease, thus Zerres et al., (1995) proposed a subdivision of this type into two subgroups: IIIa with an onset from 18 months to 3 years, and IIIb with an onset after 3 years. The probability of being ambulatory after 10, 20 and 40 years from the onset is 70.3%, 33.5% and 22%, respectively, in SMA type IIIa; and 96.3%, 84% and 58.7% in SMA type IIIb.

SMA type IV

SMA type IV refers to adult-onset (>18 years) and represents less than 5% of total SMA patients. Patients are clinically characterized by mild proximal muscle weakness and wasting. SMA type IV symptoms are similar to those of type IIIb but with less motor weakness severity. Patients with SMA type IV have normal life expectancy.

Incidence and prevalence

The worldwide annual incidence for all types of SMA ranges from 5.1 to 16.6 cases per 100,000 live births. Annual incidence for SMA types I, II and III, ranges from 3.5 to 7.1, 1.0 to 5.3 and 1.5 to 4.6 cases per 100,000 live births, respectively (Jones et al., 2015).

Prevalence per 100,000 total population for SMA types I, II and III, ranges from 0.1 to 0.15, 0.57 to 0.67, and 0.64 to 1.05, respectively (Jones et al., 2015).

Para-clinical examinations

As stated by the Consensus Statement for Standard of Care in SMA, the first diagnostic test for a patient suspected to have SMA should be the SMN gene deletion test (Wang et al., 2007). This test is routinely performed by diagnostic laboratories, and the result can be obtained within 2 to 4 weeks. The test has up to 95% sensitivity and nearly 100% specificity (Rodrigues et al., 1995; Lefebvre et al., 1998). Other diagnostic tests are recommended if the molecular genetic testing is negative and a diagnostic algorithm for SMA has been proposed by Consensus Statement for Standard of Care in SMA (Wang et al., 2007) (Figure 6).

- 17 -

Figure 6. Diagnostic evaluation for SMA. A diagnostic algorithm for SMA and other neuromuscular disorders. EMG, electromyography; NCS, nerve conduction study; RNS, repetitive nerve stimulation; CK, creatine kinase; NMJ, neuromuscular junction; MRI, magnetic resonance imaging; SMARD, SMA with respiratory distress; X-SMA, X-link SMA. Modified from Wang et al. (2007).

- 18 -

Spinal cord MRI

Principales of MRI

Uniform magnetic field within the scanner is continuously applied, aligning the net magnetic moment M of the tissues protons, mainly hydrogen atoms, which are most abundant protons in the body. Brief radiofrequency pulses of a variable forms tilts M, leading to their free induction decay, and generates current within the received coil around tissues of interest. The decay between the excitation and the return to equilibrium of M is exploited to generate contrast difference between tissues. MR signal is detected using dedicated coils. The protons’ density, T1 (spin-lattice), T2 (spin-spin) and T2* (local magnetic field inhomogeneities) relaxations vary from one tissue to another, giving rise to various contrasts and sequences, i.e. T1-, T2-, T2*-weighted sequences) (Table 3).

Table 3. The relaxation time constants and relative fully relaxed signals, i.e. their spin densities of some tissues. Modified from Atlas SW (2009) and Smith et al. (2008).

Tissue Relative spin 1.5T 3T densitiesa T1 (ms) T2 (ms) T1 (ms) T2 (ms) Brain Grey matter 83 1000 100 (T2), 65 (T2*) 1331 85 (T2), 42 (T2*) White matter 71 710 80 (T2), 78 (T2*) 832 70 (T2), 49(T2*) CSF 100 4000 2200 4000 2200 Spinal cord at C3 Grey matter − − − 972b, 994c 76 Lateral column − − − 863b, 830c 73 Dorsal column − − − 900b, 853c 72 Skeletal muscle 80 1060 35 1420 32 Venous blood 82 1435 150 1584 80 Arterial blood 82 1435 235 1664 165 Fat 90 300 165 380 133 a 100 denote the 100% spin density; b inversion recovery sequence; c gradient echo sequence.

By using different MR parameters, e.g. echo time (TE) and repetition time (TR), different weightings can be obtained, hence revealing contrast deference between tissues. T1- and T2-weighted sequences designate different measurements of protons’ relaxations after the application of radiofrequency pulses. The T1-weighted sequence gives arise to a high contrast difference between white and grey matter. T2-weighted sequence is more sensitive to tissues’ water content, thus it generates an excellent contrast difference between CSF and white matter. T2*-weighted sequences, derived from T2 decay, considered as “observed” or, “effective” T2 decay, and resulting principally from magnetic field inhomogeneities. T2*-weighted imaging generates high contrast difference between CSF,

- 19 -

white and grey matter.

Despite its expensive price, MRI has several advantages over other imaging modalities such as X- ray, CT, SPECT and PET: (i) noninvasive nature; (ii) multiplanar imaging possibilities; (iii) high soft tissues conspicuity; (iv) ability to picture at macroscopic scale neuronal content integrity. Over the years, MRI has established itself as the modality of choice to evaluate the CNS integrity in a large range of pathologies, particularly at the brain level. However, imaging the spinal cord is more challenging than that of the brain because of (i) the small cross sectional area of the cord (1 cm diameter) requiring higher spatial resolution and thus decreasing the signal-to-noise ratio (SNR) (see foot-note on next page), (ii) motion from the cerebrospinal fluid flow with each cardiac and respiratory cycle, (iii) partial volume effects at the interface between white/grey matter and white matter/CSF, (iv) susceptibility to artifacts from surrounding tissues (Stroman et al., 2014).

Challenges

Adult human spinal cord has the geometry of a bent elliptic cylinder with 0.8-1.4 cm transverse diameter, 0.5-0.9 cm anterior-posterior diameter, and 43-45 cm length (Fradet et al., 2013; Goto and Otsuka, 1997). The spinal cord is surrounded by the CSF, vertebrae and cartilaginous disks, and is close to visceral organs. The major imaging difficulties come from the fact that the spinal cord has tiny cross- sectional dimensions in the axial plane, is long in the sagittal/coronal planes, is surrounded by tissues that have different magnetic susceptibilities, and has nearby organs that are prone to motion (Figure 7). In this regard, the spinal cord is considered as one of the most challenging environments for using MR in the human body.

- 20 -

Figure 7. T2-weighted mid-sagittal (a) and T2*-weighted axial sections at C2 (b) and T12 (c) vertebral levels showing the spinal cord content, and surrounding tissues and organs. A, Anterior; L, Left; P, Posterior; R, Right; S, Superior; I, Inferior.

Several studies (Cohen-Adad et al., 2007, 2011; Stroman et al., 2014) defined the major challenges for spinal cord imaging as: i) the partial volume effect; ii) the inhomogeneous magnetic field environment; iii) the physiological and patient motion.

Partial volume effect

The partial volume appears when several tissues are contributing to the same voxel. For the spinal cord imaging it is the case when a voxel is at the interface CSF/white matter, white/grey matter, and eventually CSF/white matter/blood vessels. The MRI signal is then the sum of all signals originated by tissues with different amounts of spins, which leads to a blurry image at boundaries. However, partial volume can be limited by reducing volume sampling, i.e. increasing spatial resolution1, but this can result in lower SNR2 and contrast-to-noise3 ratios (CNR). An increase of MRI magnet strength (1.5T, 3T or 7T) and/or the number of phased-array coil if parallel imaging is used to decrease the TE and to

1Spatial resolution: The smallest volume unit detectable by an MRI (voxel). Voxel size is defined by the acquisition field-of- view size, divided by the acquisition matrix size. 2Signal-to-noise ratio: The ratio between the amplitude of the received signal, divided by the background noise signal that contaminates the image. 3Contrast-to noise ratio: The signal intensity difference between different tissues in the image, scaled to image noise. - 21 -

increase the number of signal accumulation (4, 16, 19, 32 or 64 channels), or limiting the influence of physiological motion during the acquisition, are beneficial for spatial resolution, SNR and CNR (Zhao et al., 2013; Taso et al., 2014) (Figure 8).

Figure 8. Comparison of 3T and 7T C-spine images (C5/C6). a: 3T image using the 19-channel commercial coil with 0.5 " 0.5 " 3 mm resolution (acquisition time = 1 min 44 sec). b: 7T image using the 19-channel receive array and four-channel transmit array with 0.5 " 0.5 " 3 mm resolution (acquisition time = 1 min 44 sec). c: 7T image using the 19-channel receive array and 4-channel transmit array with 0.3 mm " 0.3 mm " 3 mm resolution (no interpolation) (acquisition time = 3 min 43 sec). d: 3T image at the mid-vertebra C4 using 64-channel head and neck commercial coil (Siemens 3T Magnetom Prisma), with 0.4 " 0.4 " 6 mm resolution (no interpolation) (repetitions = 3; acquisition time = 5 min 46 sec). Modified from Zhao et al. (2013).

Physiological and patient motion

Due to its proximity to visceral organs, the lungs and the heart, almost the entire spine can undergo periodical movements provoked by the respiration and cardiac pulses (Kharbanda et al., 2006; Clark et al., 2000; Verma and Cohen-Adad, 2014). The spinal cord also moves within the spinal canal as the surrounding CSF moves. The CSF motion increases with distance to the head (A-P displacement up to 0.5 mm at the mid cervical level and 1.5 at the upper thoracic level) (Stroman et al., 2006, Figley and Stroman, 2007; Verma and Cohen-Adad, 2014). Image deterioration can also be caused by swallowing or by patient motion due to an uncomfortable position within the scanner during the long MR acquisition. These movements can create ghosting artifacts (Verma and Cohen-Adad, 2014) (Figure 9). By gating the acquisition, i.e. synchronizing with the respiratory and cardiac cycles, the consequences of

- 22 -

the periodical movement can be avoided but this technique implies a longer acquisition time (Taso et al., 2014, Cohen-Adad et al., 2011). Motion artifacts can also be reduced using “saturation bands” that cover the chest and abdomen, by attenuating signals from moving structures so that it does not corrupt the signal from the spinal cord (Figure 9). Velocity compensating gradients sequences and signal averaging across multiple phases of motion can also be applied to minimize motion artifacts. Reducing acquisition time using fast sequences, i.e. fast-spin-echo, paralleled imaging that increases acquisition speed by factors of 1.5 to 3, i.e. SENSE/GRAPPA-type reconstructions, and partial Fourier imaging enable reducing physiological as well as subject motion (Jaermann et al., 2004; Glockner et al., 2005; Noebauer-Huhmann et al., 2007; Fruehwald-Pallamar et al., 2012). A MRI compatible cervical collar could be used to limit involuntary subject motion. The cervical collar enables subjects to lie comfortably in the scanner for a long acquisition time while minimizing artifacts (Yiannakas et al., 2012). Coregistration of all data when dealing with multiple series acquisition, e.g. diffusion tensor imaging and functional MRI, could also be performed to limit inconsistency in derived maps (Mohammadi et al., 2013; Vahdat et al., 2015).

Figure 9. MRI ghosting artifacts caused by patient motion. Signal from shoulders and the chest was attenuated using saturation bands.

Inhomogeneous magnetic field environment

The spinal cord is surrounded by bones and cartilaginous disks. It is also close to the lungs, each containing air. These three elements each have a different magnetic susceptibility. The magnetic field in

- 23 -

the spinal cord region is therefore inhomogeneous, leading to geometric distortions and signal intensity loss. These artifacts can be counteracted with shimming. Shimming consists of compensating for the main field inhomogeneities by creating an auxiliary magnetic field via shim coils (Anderson, 1961; Romeo and Hoult, 1984). However, while volume shimming improves the field homogeneity, it is limited to smooth variations across space and cannot fully compensate for small and localized field variations, such as at cartilaginous discs between vertebral bodies. Echo planar imaging sequence, i.e. diffusion tensor imaging (DTI) (Figure 10), is particularly sensitive to geometric distortions in regions around vertebral disks. In addition to shimming, parallel imaging as well as slices positioning should be considered to limit magnetic field inhomogeneity, i.e. slices centered in the middle of each vertebral body and perpendicular to the spinal cord (Cohen-Adad et al., 2011). The geometry of the magnetic field inhomogeneity should be considered to correct its effect (Zeng et al., 2002; Samson et al., 2013).

Figure 10. Image distortion in DTI sequence at the interface between the spinal cord, vertebrae and intervertebral disks.

For an exhaustive description of spinal cord MR imaging challenges, the reader is referred to the book by J. Cohen-Adad et al. (Cohen-Adad and Wheeler-Kingshott, 2014) and the review from Stroman et al., Neuroimage 2014.

Clinical setting

Clinical neuroradiology is based on the visual inspection of the CNS MR images, to search for abnormalities that might explain signs detected on clinical and electrophysiological evaluation (Table 4). Conventional clinical MRI spinal cord protocols are usually composed by fast MRI sequences; two 2D sagittal T1- and T2-weighted sequences with large FOV and low in-plan resolution (typically 1-2 mm2 and 4-5 mm slice thickness) to visually identify and localize involved spinal cord regions, associated with a 2D T2*-weighted sequence with limited FOV and high in-plan resolution (typically 0.5"0.5 mm and 4-5 mm slice thickness) to visually identify the involved spinal cord compartments (white matter, grey matter or nerve roots). Sequences’ parameters adjustments are usually limited to FOV positioning without a focal shimming and no saturation bands or gating are used, which limit spinal cord - 24 -

abnormalities identification. Despite this, conventional MRI remains extremely useful in describing a large range of spinal cord abnormalities such as traumatic and demyelination lesions, narrow spinal canal, nerve roots degeneration and cord atrophy. However, conventional MRI cannot provide data on the integrity of the remaining white matter tracts and grey matter layers around and beyond involved spinal cord regions nor in the normal appearing spinal cord in patients with sensory and/or motor signs detected during clinical and electrophysiological evaluation. In the clinical diagnosis of MND, conventional MRI has only limited value in the absence of visually identified spinal cord abnormalities and is mainly used to detect pathologies that may “mimic” MND (Rocha and Maia Júnior, 2012; Lebouteux et al., 2014; Pradat, 2016) (Table 4).

Table 4. Detected abnormalities from spinal cord MRI in “mimic” MND.

Involvement site “Mimic” MND Spinal cord MRI abnormalities Peripheral nerve Neuropathies No abnormalities Spinal cord Kennedy Syndrome No abnormalities Hirayama disease “snake eyes” T2*-hypersignal appearance on the cervical anterior horns Cervical spondylotic Narrow spinal canal associated with an myelopathy intramedullary T2-hypersignal Hereditary spastic No abnormalities paraplegia Syringomyelia Thin light T2-Hypersignal inside the spinal cord corresponding to CSF obstruction and collection within the spinal cord. Transverse myelitis T1-hypointensity/T2-hyperintensity corresponding to demyelination lesions, that commonly extended for 3-4 spinal segments Poliomyelitis Bilateral T2-hypersignal in the anterior horn of the spinal cord. HTLV-I associated Cord swelling and high intensity in the central myelopathy portion along the spinal cord in T2-sagittal images

Neuromyelitis optica T1-hypointensity/T2-hyperintensity extended for more than 3 vertebral bodies associated with a central grey matter involvement and cord atrophy.

Dorsal root ganglionopathy T2-hyperintensity at the dorsal column and cord atrophy.

Spinocerebellar ataxia Spinal cord atrophy.

Research setting

In the research setting, more advanced MRI techniques are used to image the spinal cord. In

- 25 -

contrast to clinical context, advanced MRI techniques’ acquisition take into account spinal cord imaging constraints, i.e. limiting the influence of motion, partial volume and magnetic field inhomogeneity (Figure 11) (Cohen-Adad et al., 2011), which results in higher resolution, SNR and CNR compared to MR images produced in the clinical setting. Furthermore, advanced MRI techniques are based on mathematical MR signal modeling and processing of individual MR image voxels, and provide, in addition to visual abnormalities identification, quantitative indexes of white matter motor and sensory pathways as well as grey matter layers integrity in the pathological context. Its typical applications are those sensitive to: i) tissue surface, volume or density variations, giving an indirect access to axonal and neuronal loss or atrophy localization, whether global (Cross-sectional area and volume), or local, at the voxel level or its direction (tensor-based morphometry (TBM), voxel-based morphometry (VBM), and surface-based morphometry (SBM)); ii) water movement distribution along and within white matter fibers, giving an indirect access at the voxel level to fiber bundles degeneration, loss and/or their demyelination (DTI, and its extension, high angular resolution diffusion imaging (HARDI), and tractography); iii) white matter macromolecules content, giving an indirect access at the voxel level to fiber bundles myelin loss (magnetization transfer (MT) imaging) (Stroman et al., 2014).

Figure 11. Localizer image demonstrating typical patient positioning with minimum lordosis. A: Slice prescription for DTI acquisition, covering C2 to T2 vertebral levels. Saturation bands were set ventrally and dorsally to the spinal cord to limit aliasing and ghosting artifacts. B: Box prescription for manual shimming, ensuring good B0 homogeneity within the imaged portion of the spinal cord. The acquisition was cardiac-gated with slices acquired during the quiescent phase of cardiac-related motion of the spinal cord. From Cohen-Adad et al. (2011).

Advanced MRI

The spinal cord is a common site for development of several MND that are mainly represented by - 26 -

ALS and SMA. Furthermore, the spinal cord can be affected by inflammatory diseases such as multiple sclerosis (MS) and by spinal cord injuries (SCI). In all of these disorders, the involvement of the spinal cord reflects a combination of several microscopic factors including axonal degeneration, demyelination or motoneurons loss. All these diseases represent challenges for diagnosis, prognosis, follow-up and monitoring of the effect of investigational drugs.

The present section focused on advanced MRI techniques that are currently used to quantify spinal cord white and grey matter integrity in the pathological spinal cord and particularly in MND: i) high- resolution anatomic data to visually identify the extent of the spinal cord lesions and more importantly to infer on white and grey matter atrophy; ii) DTI and MT imaging to infer on axonal degeneration, loss and demyelination.

Spinal cord atrophy quantification from MR images

Common atrophy indexes

In spinal cord disorders, axons and/or motoneurons loss can be assessed in vivo indirectly from MR images by quantifying spinal cord atrophy at a given cord level. The estimation of spinal cord cross- sectional area at given level was, for more than a decade, the most used approach to investigate atrophy in the pathological context, mainly in MS, that represents the main body of the scientific research on spinal cord pathologies. The cross-sectional approach consists of estimating a mean cord surface over a representative number of slices at a given vertebral level, typically five continuous thick slices perpendicular to the spinal cord at C2 vertebral level (Losseff et al., 1996). There were three main reasons that explain such a choice of vertebral level: i) the high contrast between the CSF and the spinal cord, as well as the large CSF space, low cross-sectional area variability, low cerebrospinal fluid flow speeds and less cord motion, which increases atrophy quantification methods’ reproducibility and accuracy at this level using a variety of magnetic strength, coils and MRI sequences; ii) The brain and the upper cervical spinal cord could be imaged in same FOV with relatively high resolution, SNR and CNR; iii) Pathological studies using cross-sectional area approach have shown pronounced cord atrophy from MR images at C2 vertebral level correlating with clinical disability mainly in MS (Kearney et al., 2015) and, to a lesser extent, in ALS (Valsasina et al., 2007; Branco et al., 2014), SCI (Cohen-Adad et al., 2011; Freund et al., 2011, 2013; Lundell et al., 2011), Huntington's disease (Mühlau et al., 2014), Friedreich’s Ataxia (Chevis et al., 2013), and spinocerebellar ataxia (Lukas et al., 2008). For instance, apart from our MND studies, the two existing cross-sectional studies in sporadic ALS which investigated cord atrophy at C2 vertebral level (Valsasina et al., 2007; Branco et al., 2014), showed a high significant difference in cross-sectional area between ALS patients and age and sex matched controls (28/25 ALS

- 27 -

patients versus 20/43 age- and gender-matched healthy controls; p < 0.001), with moderate correlation with the ALS Functional Rating Scale (ALSFRS) (Valsasina et al., 2007), revised ALSFRS (ALSFRS-R) and ALS severity scale (p # 0.05) and high correlation with disease duration (p < 0.001) (Branco et al., 2014). The only existing longitudinal study in ALS (Agosta et al., 2009), showed a significant cord area change between the baseline and a mean follow-up of 9 months (17 ALS patients; p = 0.003) with no correlations with the ALSFRS changes. Alternatively, some studies proposed a variety of indexes derived from cross-sectional approach (Figure 12):

- Cord anterior-posterior width (APW) and left-right width (LRW) (Lundell et al., 2011). These indexes provide, in the pathological context, information on cord flattening.

- Cord eccentricity (CE), which is defined as the square root of 1 – (APW/LRW)2. CE index provides, in the pathological context, information on cord flattening (Fahl et al., 2015; Branco et al., 2014).

- Cord radial distance, which is defined as the angular variation of the spinal cord radius over the full circle (Lundell et al., 2011).

Figure 12. Derived indexes from cross-sectional approach. APW, anterior-posterior width; CE, cord eccentricity; LRW, left-right width.

The eccentricity index was used in the context of sporadic ALS and spinocerebellar ataxia type 3. No cord flattening was found in ALS patients (43 patients versus 43 age and gender-matched healthy controls) (Branco et al., 2014). In spinocerebellar ataxia type 3 patients (48 patients versus 48 age and gender-matched healthy controls), there was a significant antero-posterior cord flattening (p = 0.005), with no correlation with the clinical scale for assessment and rating of ataxia (Fahl et al., 2015).

Cord widths and radial distance indexes were used in the context of chronic SCI (19 patients versus

- 28 -

16 age and gender-matched healthy controls) (Lundell et al., 2011). The authors showed a preferential anterior-posterior cord atrophy at C2 vertebral level or cord flattening as consequence of the cord trauma. In SCI patients, the APW and the LRW were exclusively correlated with sensory scores and motor score, i.e. light touch and pinprick for sensory scores, and muscle strength for motor score) (p # 0.01), consecutively. Radii in the anterior and posterior parts of the spinal cord were correlated to sensory scores (p < 0.01) (Figure 13). Radii in the lateral and ventro-lateral parts were correlated to motor score (p < 0.01) (Figure 13). The anterior and the posterior parts of the spinal cord represent the main ascending sensory pathways, and the lateral and ventro-lateral parts of the spinal cord represent the main descending motor pathways and motoneurons. This study was the first to propose a focal atrophy investigation of spinal cord sensory-motor systems at C2 vertebral level. This study represents one of the most innovative approaches to quantify cord atrophy because of the intuitiveness and simplicity of its method as well as its potential generalization to explore specifically white matter sensory and motor pathways atrophy as well as motoneurons loss in large regions of the spinal cord in MND.

Figure 13. The directional correlations between the spinal cord radius and the motor score (solid), pinprick sensory score (PPSS) (dark grey) and light touch sensory score (LTSS) (light grey). Correlations are plotted for directions with high significance (p < 0.01). The radial axis shows the correlation coefficient and the spinal cord drawing illustrates the main directions in the plot. A, anterior; L, left; P, posterior; R, right. From Lundell et al. (2011).

Thanks to the recent improvement in MRI technology, scanners with higher magnet strength and the large number of phased-array coils, as well as the gained technical/methodological knowledge in imaging the spinal cord, lead to feasible picturing of large spinal cord regions, i.e. cervical and upper thoracic spinal or the whole thoracic and lumbar spinal cord, with high resolution, SNR and CNR

- 29 -

(typically 3D sequences with voxel size ≤ 1 mm3) or white and grey matter tissues at specific spinal cord levels with high resolution (typically 2D sequences with thick slices 4-5 mm, and in plane resolution ≤ 0.5×0.5 mm2) (Figure 7). The production of such high quality spinal cord MR images was accompanied with adapted and innovative methodological developments to process such images. New image processing algorithms using more advanced approaches emerged, and permit exploration of the topography of spinal cord atrophy in the pathological context. Such explorations give a new insight into the pathophysiological mechanisms involved in spinal cord tissue degeneration. For instance, in our previous cross-sectional study in ALS patients (Cohen-Adad et al., 2013) (25 ALS patients versus 20/43 age-matched controls), we quantified cross-sectional area from high-resolution MR images at the lower cervical spinal cord. Our study suggested a specific atrophy profile along the lower cervical spinal cord in ALS patients. In addition, we showed that cord atrophy at C7 vertebral level (equivalent to C8 spinal level) was correlated with the MEP amplitude obtained after stimulation at the level of C8 nerve root, which reflects impairment of lower motor neurons, not CST. Our findings suggested that regional spinal cord atrophy as measured using MRI reflects anterior horn cells degeneration in ALS. In MS context, one major impact spinal cord MRI study (Klein et al., 2011), explored the susceptibility of the spinal cord levels to atrophy in four MS subtypes using volume measurement (two patients with clinically isolated syndrome (CIS), 27 with relapsing-remitting MS (RRMS), two with secondary-progressive MS (SPMS), and four with primary-progressive MS (PPMS) phenotypes versus age and gender-matched healthy controls; the spinal cord was imaged from C1 to T12). Patients with PPMS/SPMS had a different atrophy profile compared with RRMS/CIS patients. Patient with PPMS/SPMS showed upper cervical cord atrophy. However, RRMS/CIS patients showed an increased volume at the cervical and thoracic levels, which most likely mirror inflammation or edema-related cord expansion.

Advanced atrophy indexes

Cross-sectional and volume measurements produce global atrophy indexes. Alternatively, some studies proposed more advanced and complex methods for a focal cord atrophy quantification, derived from volumetric approach. Voxel-based morphometry method was widely used to localize atrophied brain regions in the pathological context (O'Callaghan et al., 2016; Tavazzi et al., 2015; Byun et al., 2015; Ille et al., 2011). Voxel-based morphometry method is based on statistical analysis of the difference in white/grey matter brain tissues concentration between groups. The latter method was introduced for the first time to analyze cord atrophy related to aging by Valsasina et al. (2012). Their methodological study was a proof of concept of the feasibility of VBM to produce reliable results on cervical spinal cord atrophy due to aging (90 healthy subjects). Their study showed that most voxels significantly associated with a lower probability of containing cord tissue were located in the ventral

- 30 -

Figure 14. Regional distribution of cervical cord atrophy in different study groups. Sagittal and axial MR images at various cervical cord levels show clusters of significant cord atrophy (color-coded for t values; P < 0.001, uncorrected only for display purposes). (A) Patients with MS versus healthy control subjects (HC). (B) Patients with RR MS versus HC. (C) Patients with SP MS versus those with RR MS. A, anterior, L, left; P, posterior; R, right. Modified from Valsasina et al. (2013). portion of the cord. The latter result is consistent with post-mortem studies, which found that aging involves grey matter ventral horns neurons loss (Morrison, 1959; Cruz-Sánchez et al., 1998). Thereafter, the same research group used VBM to map the regional distribution of atrophy at the cervical spinal cord level in MS patients with different clinical phenotypes (Valsasina et al., 2013; Rocca et al., 2013) (Figure 14). The authors showed differences in the atrophy profile between MS phenotypes, i.e. CIS, PPMS, benign MS, SPMS and RRMS. In these two studies, VBM enabled a better characterization of the clinical heterogeneity in MS patients. However, the latter studies have a major limitation, which is their inability to study regional white and grey matter atrophy, which would require the use of higher MR imaging resolution sequences with an optimal contrast between white and grey matter. Despite this, VBM could be of interest in characterizing regional atrophy in MND, i.e. loss of spinal motoneurons, as well as the clinical heterogeneity in MND and particularly in ALS.

As previously mentioned, the new generation of MR scanners and coils enables the production of high resolution MR images that allow picturing white and grey matter tissues at specific spinal cord

- 31 -

levels, thus a specific tissue atrophy quantified in spinal cord diseases. There are mainly two ways to process these MR images: i) by estimating a mean white and grey matter areas within a representative number of slices at a given cord level, similar to the cross-sectional area measurement; ii) delineating white and grey matter tissues then using them as inputs in an advanced morphological analysis method based on structures shape, i.e. TBM. The first approach was used in two major impact MS studies (Schlaeger et al., 2014, 2015). In the first study, white and grey matter areas were measured at the disc levels C2/C3, C3/C4, T8/9, and T9/10 in 142 MS patients (99 relapsing and 43 progressive MS patients) and 20 healthy controls. The authors showed thoracic grey matter atrophy with absence of white matter atrophy in relapsing MS patients. Grey matter atrophy was more pronounced in progressive MS than in relapsing MS and correlated with clinical disability and lower limb function. Cervical grey matter areas had the strongest correlation with clinical disability followed by thoracic cord grey matter area and brain grey matter volume. More interestingly, their study showed the presence of grey matter atrophy at the cervical and thoracic levels, which may reflect a diffuse grey matter degeneration along the whole spinal cord. The second study was in 113 MS patients (25 progressive and 88 remitting MS patients) and 20 healthy controls. The authors showed for the first time that C2/C3 disk level spinal cord grey matter area was the strongest correlations of disability in MS using multivariate models that included brain grey and white mater volumes, fluid-attenuated inversion recovery lesion load, T1 lesion load, C2/C3 disk level white matter area, number of T2 lesions, age, sex, and disease duration. Moreover, brain and spinal grey matter independently contributed to disability. However, the latter MRI studies did not investigate which grey matter region was involved in MS. Lower motor neurons loss could likely contribute to grey matter atrophy as showed in electrophysiological and histopathological studies in MS patients (Vogt et al., 2009; Davison et al., 1934).

Tensor-based morphometry method is based on statistical analysis of structure shape variations between groups. This method was widely used to map brain structure atrophy in a large range of pathological conditions (Hua et al., 2016; Tessa et al., 2014; Agosta et al., 2009; Kipps et al., 2005). This method was proposed, for the first time, to map spinal cord white and grey matter alterations occurring with age (Taso et al., 2015). The authors showed that regional spinal cord atrophy was mostly localized in the anterior horns in elderly healthy subjects (n =25) compared to young ones (n = 40) (the spinal cord was imaged at the middle of C1 to C6 vertebral bodies) (Figure 15). This result is consistent with in-vivo imaging and post-mortem studies described above (Valsasina et al., 2012; Cruz-Sánchez et al., 1998; Morrison, 1959). This method holds great promise for investigating motoneurons loss occurring in MND.

- 32 -

Figure 15. Results of the TBM analysis between young and elderly volunteers (2 sample t-test, p < 0.05, FDR corrected). The red/yellow clusters indicate regions of atrophy in the elderly subjects. From Taso et al. (2015).

Axonal degeneration, loss and demyelination quantification from diffusion-weighted and magnetization transfer imaging.

Diffusion-weighted imaging

Diffusion results from the random movement of molecules in-vivo. Diffusion of water in structures such as the CSF or grey matter is isotropic, i.e. identical in all directions, whereas in white matter it occurs preferentially along the axis of orientation of the fiber bundles. Diffusion tensor imaging allows the measurement of water diffusion anisotropy along and within white matter axons (Yang et al., 2011; Mori and Zhang, 2006; Beaulieu, 2002; Le Bihan et al., 2001), and gives an indirect localization of white matter fiber bundles as well as their macroscopic geometrical arrangement. The application of diffusion gradients in several directions, with a minimum of six noncollinear diffusion-encoded directions, results in DWI. The most widely used mathematical model describing water diffusion is the tensor model. Derived imaging application from the diffusion tensor model is known as DTI (Pierpaoli and Basser, 1996). Water diffusion in the tensor model is presumed to be Gaussian and unidirectional. Diffusion tensor (D) is 3"3 matrix that describes tissue anisotropy at each image voxels:

!!! !!" !!" !" !! !" ! ! ! ! ! Diffusion!!" !!" tensor! !!decomposition to principal diffusion magnitudes and directions results in ordered three eigenvalues ("1, "2, "3) and eigenvectors (V1, V2, V3). The eigenvalue "1 represents the diffusion

- 33 -

magnitude along V1, i.e. along fiber bundles, and λ1 and λ2 the diffusion magnitude along V2 and V3, respectively, i.e. on the plane perpendicular to fiber bundles. Therefore, diffusion anisotropy indexes that characterize different aspects of water diffusion distribution at each image voxel, i.e. ellipsoid, can be computed from those eigenvalues, regardless of directionality. Principal diffusion indexes are mean diffusivity (MD), fractional anisotropy (FA), axial and radial diffusivities (AD and RD):

!! + !! + !! !" = 3

! ! ! 3 (!! − !") + (!! − !") + (!! − !") !" = ! ! ! 2 !! + !! + !!

!" = !!

!! + !! !" = 2

The model behind DTI, which represents water diffusion mainly along and perpendicular to fiber bundles planes, is simple and shows its limit in representing the complex diffusion processes occurring, for example, in non-orthogonal crossed fibers or multiple fibers orientations (Campbell et al., 2005; Tuch et al., 2002). New methods have been developed to correctly represent complex diffusion in biological tissues. These methods are called high angular diffusion resolution imaging (HARDI). Q-ball imaging (QBI) is one of the popular HARDI reconstructions methods (Zhan and Yang, 2006; Campbell et al., 2005; Tuch, 2004). The QBI imaging reconstructs the diffusion orientation distribution function (ODF) from HARDI data on a sphere. The HARDI data can also be used to fit the diffusion tensor model. The HARDI technique has proved their usefulness in imaging white matter micro-diffusion properties in the human brain and in injured spinal cats (Mukherjee et al, 2008; Cohen-Adad et al., 2008; Tuch et al., 2005). The HARDI technique was used, for the first time by our research group, to characterize lesion extend in the normal appearing white matter sensory and motor pathways in SCI patients (Cohen-Adad et al., 2011).

Several pathological imaging studies in humans and animal models, as well as those combining diffusion-tensor imaging and histology, have shown that diffusion indexes reflect a combination of white matter microstructures, i.e. axonal density, axonal diameter, myelin thickness and fibers orientation and coherence, etc. (Takahashi et al., 2002; Song et al., 2003; Concha et al., 2006; Budde et al., 2007; Sun et al., 2008; Zhang et al., 2009; Concha et al., 2010; Choe et al., 2012; Qin et al., 2012; Liu et al., 2013) (Table 5). For instance, AD reflects axonal damage, particularly fragmentation (Song et al., 2003; Concha et al., 2006; Budde et al., 2007; Sun et al., 2008; Zhang et al., 2009; Liu et al., 2013), while

- 34 -

RD diffusivity provides information regarding axonal density, myelin integrity, axonal diameter, and fiber coherence (Song et al., 2003; Sun et al., 2006; Concha et al., 2010). Fractional anisotropy and MD could be seen as a combination of RD and AD (see the previous equations), and thus reflects, at different degrees, a combination of all the micro-architectural components explained by AD and RD. Furthermore, diffusion indexes could be influenced by other components that are not related to axons and myelin, i.e. glial cells, ectopic neurons and tumor cells (Larriva-Sahd et al., 2002; Budde et al., 2010; DeFelipe et al., 2010; Suarez-Sola et al., 2009; Ronen et al., 2013). The link between diffusion indexes and white matter microstructures should carefully be interpreted. Therefore, we can only conclude that diffusion indexes, whether estimated using tensor or q-ball models, allow a general evaluation of white-matter pathways integrity. For an exhaustive description of the correlations between diffusion indexes and white matter tissue microstructures, the reader is referred to the review from Concha (2014).

Table 5. Pearson’s correlations corrected p-values for multiple comparisons (p-values without correction) between DTI indexes and white matter microstructures derived from electron microscopy of a small specimen of the fimbria-fornix. Modified from Concha et al. (2010).

White matter microstructure FA MD AD RD Axonal density 0.13 (0.0495*) 0.83 (0.3) 0.66 (0.2) 0.47 (0.12) Inner axon diameter 0.64 (0.3) 0.68 (0.21) 0.66 (0.2) 0.74 (0.24) Outer axon diameter 0.76 (0.4) 0.83 (0.3) 0.74 (0.24) 0.94 (0.44) Myelin thickness 0.11 (0.04) 0.64 (0.19) 0.92 (0.4) 0.29 (0.064) Cumulative axon membrane 0.02 (0.007)* 0.52 (0.14) 0.83 (0.3) 0.11 (0.023*) Circumference Myelin fraction 0.47 (0.2) 0.83 (0.3) 0.96 (0.47) 0.68 (0.21) * P-values > 0.05

Diffusion tensor imaging was widely used at brain level in MND and almost in ALS, and reported change in DTI indexes in parts of the CST as well as outside the motor regions (Turner and Verstraete, 2015; Agosta et al., 2010). However, at the spine cord level, only a small number of studies used DTI to investigate white matter integrity in MND, and they were exclusively in ALS (Cohen-Adad et al., 2013; Wang et al., 2014; Nair et al., 2010; Agosta et al., 2009; Valsassina et al., 2007). This was mainly related to spinal cord imaging difficulties (see spinal cord imaging challenges), as well as the rarity of MND diseases. In our cross-sectional study in ALS patients (Cohen-Adad et al., 2013) (29 ALS patients versus 21 age-matched controls), we showed changes in FA and RD in the CST and dorsal column (sensory afferents). Fractional anisotropy correlated with ALSFRS-R (p = 0.04) and TMS motor threshold (p = 0.02). The three other cross-sectional ALS studies, showed a significant change in FA, - 35 -

RD and MD at the cervical spinal cord (p < 0.05) (28 ALS patients versus 20 age-matched controls in Valsassina et al., 2007; 14 ALS patients versus 15 age-matched controls in Nair et al., 2010), and more specifically FA and MD in the CST in ALS patients compared to healthy controls (24 ALS patients versus 16 age-matched controls in Wang et al., 2014). In ALS patients, FA was positively correlated with lower ALSFRS (p < 0.001) (Valsassina et al., 2007), finger and foot tapping speed (Nair et al., 2010). RD was negatively correlated with the ALSFRS-R, finger and foot tapping speed (p < 0.05) as well as the respiratory function (p < 0.01), as measured by the force vitality function scale (Nair et al., 2010). In the longitudinal study from Agosta et al. (2009) (17 ALS patients versus 20 age-matched controls), ALS patients showed a significant longitudinal decrease in FA, and an increase in MD at the cervical spinal cord level (p = 0.01). Longitudinal change of diffusivity indexes did not correlated with the ALSFRS changes. For an exhaustive description of DTI applications in spinal cord diseases, the reader is referred to the review from Martin et al. (2013).

Magnetization transfer imaging

Hydrogen nuclei linked to macromolecules such as proteins and lipids constituted of axons myelin sheet (70-80% of lipids and 20-30% of proteins), have an extremely short T2 signal (~10-5 s). These macromolecules are not directly detectable with standard MRI sequences. First, macromolecular spins are saturated using an off-resonance RF pulse, then transfer of magnetization between bound and free pools occurs by means of cross relaxation processes (dipole-dipole interactions, chemical exchange, etc.). This transfer to the « free » pool induces a signal decrease that can be indirectly measured, known as MT imaging (Filippi and Rocca, 2007; Smith et al., 2006; Wolf and Balaban, 1989). The latter interaction and/or exchange can be quantified using a magnetization transfer ratio (MTR), which is calculated as the percentage difference of MT images with macromolecules signal saturation (MS) and one without (M0):

!! − !! !"# = 100× !!

The MTR enables inferring on myelin content, axonal count and density as shown by three MS histological studies at the brain and spinal cord levels, and thus to assess demyelination/remyelination and degeneration (Schmierer et al., 2007, 2004; Mottershead et al., 2003).

Magnetization transfer imaging was widely used at brain and spinal cord levels in demyelinating diseases. By contrast, only few studies used MT imaging at the brain level in the context of ALS, and reported change in MTR index of the CST (Carrara et al., 2012; Tanabe et al., 1998; Kato et al., 1997). At the spinal cord level, to our knowledge, there is only one study in ALS patients that used MT imaging - 36 -

and is from our research group (Cohen-Adad et al., 2013). In our study, we detected a significant difference between patients (n = 29) and age-matched controls (n = 21) for MTR in the CST and dorsal column of the cervical spinal cord. However, there were no correlations between MTR measured in the CST and ALSFRS-R, neither between MTR and TMS measurements. For an exhaustive description of MT imaging applications in spinal cord diseases, the reader is referred to the review from Martin et al. (2015).

- 37 -

3. Objectives

The spinal cord represents the entry for sensory information and the exit for motor commands for the limbs and the trunk. Lesions of the spinal cord due to trauma or neurodegenerative diseases thus lead to major sensory and motor consequences whose severity depends on the extent of the lesions. It is therefore extremely important to establish the extent of the spinal lesions as precisely as possible by using several complementary approaches to clinical and electrophysiological examinations such as imaging of the spinal cord, which was the main topic of the SPINE 1 Imaging Project.

Our research group significantly contributed to the development of a multi-parameter MRI approach to studying spinal cord diseases, particularly MND, which was achieved in the context of the SPINE 1 Imaging Project. The primary objective of the SPINE 1 project was to transfer the acquired knowledge on multi-parametric MRI from previous animal and human studies at the brain and upper cervical spinal cord levels to the entire human cervical and thoracic spinal cord (Cohen-Adad et al., 2008; Filippi and Agosta, 2007; Rodegerdts et al., 2006). The second objective was to capture sensory- motor system involvement in two MND: sporadic ALS and SMN1-linked SMA as well as to quantify the extent of spinal lesions and of the remaining connectivity in traumatic injuries of the spinal cord. The SPINE 1 project included cutting-edge advanced MRI techniques: i) high-resolution anatomic data to visually identify the extent of the spinal lesions and more importantly to infer on white matter and/or grey matter atrophy; ii) high-resolution HARDI and MT imaging techniques to infer on axonal degeneration, loss and demyelination.

The objective of my thesis was to optimize and use the multi-parametric MRI approach at the spinal cord level to analyze grey and white matter structures that are impaired in MND, their temporal alterations during the disease course and the functional correlates, as assessed by clinical and electrophysiological examinations. To achieve such an ambitious objective, imaging, clinical and electrophysiological expertise were combined. In this respect, several processing tools were designed, especially those enabling accurate spinal cord atrophy quantification. The developed tools are detailed in Study 1 and 2, and their applications in MND are presented in Study 4 and 5. Supplementary tools, such as DTI motion correction, are directly integrated in our MND studies (Study 3 and 4).

- 38 -

4. Results

Methodological studies

Study 1 - Validation of a semi-automated spinal cord segmentation method Study 2 - Fast and accurate semi-automated segmentation method of spinal cord MR images at 3T applied to the construction of a cervical spinal cord template

Multi-parametric MRI studies in MND

Study 3 - Electrophysiological and spinal imaging evidence for sensory dysfunction in amyotrophic lateral sclerosis Study 4 - Multi-parametric spinal cord MRI as potential progression marker in amyotrophic lateral sclerosis Study 5 - Cervical spinal cord atrophy profile in adult SMN1-linked SMA

Methodological studies

Study 1 - Validation of a semi-automated spinal cord segmentation method

Background and objectives

Atrophy quantification from MR images requires as a first step the separation of the spinal cord from surrounding tissues, i.e. CSF, and to delineate white and grey matter if they were visible on the images. Regions-of-interest delineation can be achieved manually or by using semi-automated or automated segmentation methods. Several segmentation methods with varying degrees of manual intervention, reproducibility and accuracy have been proposed to assess spinal cord atrophy from MR images (Horsfield et al., 2010; Zivadinov et al., 2008; McIntosh and Hamarneh, 2006; Tench et al., 2005; Hickman et al. 2004; Coulon et al. 2002; Losseff et al., 1996). Most published studies have used low-resolution T1-weighted 1.5T images and mainly investigated the upper cervical spinal cord in MS patients. However, validation of cord atrophy as a biomarker in spinal cord diseases requires evaluating these segmentation methods in large segments of the spinal cord using high-resolution anatomical MR images generated from actual 3T MRI systems as well as in diseases other than MS. In this respect, MND are characterized by selective neuronal damage that provides relevant models. Indeed, ALS and SMA are

- 39 -

characterized by diffuse degeneration of LMN localized in the anterior horn of all spinal cord levels (Katsuno et al., 2010; Wijesekera and Leigh, 2009). The entire CST also degenerates in ALS (Wijesekera and Leigh, 2009). On the other hand, SCI at the chronic phase offers a model of secondary white matter tract lesions due to Wallerian degeneration (Wyndaele and Wyndaele, 2006).

The semi-automated threshold-based segmentation method proposed by Losseff et al. (1996) is one of the most used approaches to quantify atrophy related to its high reproducibility and accuracy. This method has to be initialized manually by drawing two contours, one to delineate the spinal cord and the other surrounding the CSF. However, this method has only been validated at C2 vertebral level using 1.5T images in MS patients (Horsfield et al., 2010; Losseff et al., 1996).

To validate cord atrophy measurement at 3T using Losseff’s segmentation method, we evaluated its reproducibility, accuracy and robustness to manual initialization at both the cervical and thoracic levels and in neurological disorders other than MS.

Material and methods

A large dataset of 111 subjects including 49 healthy subjects, 29 ALS, 19 adult SMN1 gene-linked SMA and 14 SCI patients were enrolled in the study. Losseff’s segmentation method was used to measure the spinal cord area from high-resolution T2-weighted 3D turbo spin echo images (52 sagittal slices, FOV=280 mm, voxel size=0.9×0.9×0.9 mm3, acquisition time ~6 min), in the cervical and thoracic spinal cord (from C2 to T9). Method reproducibility and accuracy were evaluated. The robustness of the method to initialization was assessed by simulating modifications to the contours drawn manually prior to segmentation.

Results synthesis

This study showed the high reproducibility and accuracy of the semi-automated method from Losseff et al. (1996) for measuring a cross-sectional area of the spinal cord in both cervical and thoracic regions at 3T in a large number of normal subjects and a variety of patients with spinal cord disorders (Table 2). Improved image spatial resolution obtained at 3T allowed higher segmentation performances than reported previously (Losseff et al., 1996; Horsfield et al., 2010). By conducting numerical simulations, we have shown that method was highly robust to its initial contours. More specifically, simulations suggested optimal segmentation results would be obtained with initial contours representing as little as 10% to 30% of manually drawn areas for both the spinal cord and the CSF (Figure 2). Therefore, full automation of the method could be envisaged by automatically extracting CSF and cord regions, while allowing large range of error rate in contours extraction. Finally, as the method was efficient at different levels and in various diseases of the spinal cord, it is expected that it may be applied - 40 -

in a wide range of neurological diseases for a reproducible and accurate quantification of spinal cord atrophy. The high performances of the Losseff’s method make us confident about the accuracy of spinal cord atrophy quantification in the context of MND (Study 4).

Erratum: In Table 1, the French abbreviation SLA was used instead of ALS.

- 41 -

Study 2 - Fast and accurate semi-automated segmentation method of spinal cord MR images at 3T applied to the construction of a cervical spinal cord template

Background and objectives

In Study 1, we showed that the semi-automated method by Losseff et al. (1996) demonstrated high reproducibility and accuracy for measuring spinal cord cross-sectional area in both cervical and thoracic regions at 3T in normal subjects and in patients with MND and SCI. Improved image spatial resolution obtained at 3T allowed higher segmentation performances than reported previously. More interestingly, full automation of the method seems possible given its low sensitivity to the initialization steps. Losseff’s segmentation method is a semi-automated method, and thus needs a large manual intervention from an operator, which restrains its use to quantify atrophy in few representative slices of the cord region of interest.

In clinical practice, it is often difficult to visually determine spinal cord atrophy. The growing interest in assessing spinal cord atrophy using MRI therefore underlines the need for a fast and accurate segmentation method of the spinal cord with limited manual intervention. This method would also enable segmentation of large data sets in a limited time, thus facilitating investigating in large cord regions using either volume or cross-sectional area measurements. Particularly, this method would enable investigation of the topography of spinal cord atrophy in spinal cord diseases using morphometry methods (Morra et al., 2009; Wang et al., 2010, 2011). Furthermore, to be used, advanced processing methods require an additional step of data normalization. The idea behind data normalization is to adjust for intersubject variability that is not related to the studied pathophysiological process (Mazziotta et al., 2001), as well as to realign all data in the same reference, i.e. template, for statistical comparisons between groups (Vasasina et al., 2012). Several methods have been proposed to construct a template of the cervical spinal cord (Fonov et al., 2014; Valsasina et al., 2012; Stroman et al., 2008). However, the proposed spinal cord templates lacked resolution and accuracy in the construction procedure or used a limited sample of healthy volunteers.

This second study was designed to address two main objectives: i) to construct and validate a new fast and accurate segmentation method with minimal manual intervention benefitting from the high robustness of Losseff’s method to its initialization procedure; ii) as an application, we extended the existing work at 1.5 and 3T (Fonov et al., 2014; Valsasina et al., 2012) to data normalization and the construction of an accurate MRI template of the cervical spinal cord using a large sample of 3T images.

- 48 -

Material and methods

A semi-automated double threshold-based method (DTbM) was proposed enabling both cross- sectional and volumetric measures from 3D T2-weighted turbo spin echo MRI scans of the spinal cord at 3T (52 sagittal slices, FOV=280 mm, voxel size=0.9×0.9×0.9 mm3, acquisition time ~6 min). Eighty- two healthy subjects, 10 patients with ALS, 10 with adult SMN1 gene-linked SMA and 10 with SCI were studied. The DTbM was compared with active surface method (ASM) (Horsfield et al., 2010) and Losseff’s segmentation method (Losseff et al., 1996), two of the most used segmentation methods to quantify atrophy (De leener et al., 2016a), taking manual outlining from an experienced operator (with four years’ experience) as the ground truth. An independent expert neuroradiologist (with seven years’ experience) visually scored DTbM and ASM segmentations in cervical and thoracic cord regions. The operator was blinded to the type of data and methods. Accuracy was also quantified at the cervical and thoracic levels as well as at the C2 vertebral level. Fifty-nine MR images from healthy subjects were used to construct a cervical spinal cord template and a standardization pipeline was designed.

Results synthesis

In this second methodological study we designed a segmentation method with minimal manual intervention as well as a data normalization pipeline. Visual scoring showed better performance for DTbM than for ASM in both healthy and pathological groups (Table 3). More importantly, the neuroradiologist did not find any visual abnormalities in most cases of ALS and SMA data, justifying the need for quantitative methods to analyze MND data, which represented one of my main scientific focus during my thesis. Quantitative accuracy analysis showed that DTbM outperformed ASM at both cervical and thoracic spinal cord levels as well as at C2 vertebral level in both healthy and pathological groups, confirming the visual evaluation results (Tables 4 and 5). Our method showed similar accuracy compared with Losseff’s method, but with the advantage of limited manual interaction. An accurate data normalization pipeline was designed that included DTbM, enabling the construction of an accurate straight cervical spinal cord template and a tissue probability map (Figure 3). Our segmentation method may also accurately measure cross-sectional area and spinal cord volume in a large range of pathologies. More interestingly, the proposed tools could be used to investigate the topography of spinal cord atrophy in spinal cord diseases using morphometry methods (Morra et al., 2009; Valsasina et al., 2012, 2013; Rocca et al., 2013). Among these methods, surface-based morphometry was used to map cord atrophy in the context of adult SMN1-linked SMA (Study 5).

- 49 - RESEARCH ARTICLE Fast and Accurate Semi-Automated Segmentation Method of Spinal Cord MR Images at 3T Applied to the Construction of a Cervical Spinal Cord Template

Mohamed-Mounir El Mendili1*, Raphaël Chen1☯, Brice Tiret1☯, Noémie Villard1☯, Stéphanie Trunet2, Mélanie Pélégrini-Issac1, Stéphane Lehéricy2,3, Pierre-François Pradat1,4, Habib Benali1 a11111 1 Sorbonne Universités, UPMC Univ Paris 06, CNRS, INSERM, Laboratoire d’Imagerie Biomédicale, F- 75013, Paris, Île de France, France, 2 APHP, Groupe Hospitalier Pitié-Salpêtrière, Service de Neuroradiologie, F-75013, Paris, Île de France, France, 3 Sorbonne Universités, UPMC Univ Paris 06, UMR-S975, Inserm U975, CNRS UMR7225, Centre de recherche de l’Institut du Cerveau et de la Moelle épinière—CRICM, Centre de Neuroimagerie de Recherche—CENIR, F-75013, Paris, Île de France, France, 4 APHP, Groupe Hospitalier Pitié-Salpêtrière, Département des Maladies du Système Nerveux, F-75013, Paris, Île de France, France

OPEN ACCESS ☯ These authors contributed equally to this work. Citation: El Mendili M-M, Chen R, Tiret B, Villard N, * [email protected] Trunet S, Pélégrini-Issac M, et al. (2015) Fast and Accurate Semi-Automated Segmentation Method of Spinal Cord MR Images at 3T Applied to the Construction of a Cervical Spinal Cord Template. Abstract PLoS ONE 10(3): e0122224. doi:10.1371/journal. pone.0122224 Objective

Academic Editor: Bogdan Draganski, Centre To design a fast and accurate semi-automated segmentation method for spinal cord 3T MR Hospitalier Universitaire Vaudois Lausanne - CHUV, images and to construct a template of the cervical spinal cord. UNIL, SWITZERLAND

Received: May 2, 2014 Materials and Methods

Accepted: February 19, 2015 A semi-automated double threshold-based method (DTbM) was proposed enabling both

Published: March 27, 2015 cross-sectional and volumetric measures from 3D T2-weighted turbo spin echo MR scans of the spinal cord at 3T. Eighty-two healthy subjects, 10 patients with amyotrophic lateral scle- Copyright: © 2015 El Mendili et al. This is an open access article distributed under the terms of the rosis, 10 with spinal muscular atrophy and 10 with spinal cord injuries were studied. DTbM Creative Commons Attribution License, which permits was compared with active surface method (ASM), threshold-based method (TbM) and man- unrestricted use, distribution, and reproduction in any ual outlining (ground truth). Accuracy of segmentations was scored visually by a radiologist medium, provided the original author and source are in cervical and thoracic cord regions. Accuracy was also quantified at the cervical and tho- credited. racic levels as well as at C2 vertebral level. To construct a cervical template from healthy Data Availability Statement: Anonymized data or subjects’ images (n=59), a standardization pipeline was designed leading to well-centered representative images from the present study, developed methods and the findings are fully straight spinal cord images and accurate probability tissue map. available without restriction upon request to the authors. Results Funding: This study was supported by the Visual scoring showed better performance for DTbM than for ASM. Mean Dice similarity Association Française contre les Myopathies (AFM) coefficient (DSC) was 95.71% for DTbM and 90.78% for ASM at the cervical level and and the Institut pour la Recherche sur la Moelle épinière et l'Encéphale (IRME). The research leading 94.27% for DTbM and 89.93% for ASM at the thoracic level. Finally, at C2 vertebral level, to these results has also received funding from the mean DSC was 97.98% for DTbM compared with 98.02% for TbM and 96.76% for ASM.

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 1 / 21 Spinal Cord Segmentation Method

program “Investissements d’avenir” ANR-10-IAIHU- DTbM showed similar accuracy compared with TbM, but with the advantage of limited 06. manual interaction. Competing Interests: The authors have declared that no competing interests exist. Conclusion A semi-automated segmentation method with limited manual intervention was introduced and validated on 3T images, enabling the construction of a cervical spinal cord template.

Introduction Spinal cord atrophy is a neuroimaging feature that is associated with motor neuron diseases such as amyotrophic lateral sclerosis (ALS) and spinal muscular atrophy (SMA) [1–5], inflammatory diseases such as multiple sclerosis (MS) [6–8], and spinal cord injuries (SCI) [9–11]. In the past two decades, several segmentation methods with varying degree of manual intervention have been proposed to assess cord atrophy from MR images [12–19]. These methods have been validated using mostly 1.5T images and mainly investigated cord atrophy in MS patients, at the upper cervical level where the spinal cord is most atrophied [20]. In the context of spinal cord diseases, there is clear interest in investigating larger portions of the spinal cord. In- deed, degenerative diseases of the spinal cord, such as ALS and SMA, are characterized by diffuse degeneration of lower motor neurons localized in the anterior horn of all spinal cord levels [21, 22]. The entire corticospinal tract also degenerates in ALS [22]. Spinal cord injuries can also affect all spinal cord levels [23]. In clinical practice, it is often difficult to visually determine spinal cord atrophy. The growing interest in assessing spinal cord atrophy using MRI therefore underlines the need for a fast and accurate method of spinal cord segmentation with limited manual intervention. Such method would indeed facilitate the investigation of large regions of the spinal cord using either volume or cross-sectional area (CSA) measurements. This method would also enable segmentation of large data sets in limited time. Furthermore, this would open the way to the construction of a template that could be used for cord normalization in group analyses. Several methods have been proposed to construct a template of the cervical spinal cord [24–26]. In [24, 26], less than 20 subjects were used to construct a T2-weighted 3T MRI template and such small sample was not representative enough of the anatomical variability of the spinal cord. In [24], a large amount of manual intervention was needed to straighten the spinal cord (one had to draw manually a line along the anterior edge of the spinal cord for each subject), while we in- troduce in this paper a segmentation method with minimal manual intervention. The cord straightening approach proposed in [24] leads to a better alignment of the anterior cord regions than the posterior cord regions, which is not the case in the present study as we use the cord centerline to straighten the spinal cord. Recently, a T1-weighted 1.5T MRI template has been proposed for the cervical spinal cord to quantify age-related atrophy as well as disease-related atrophy in MS patients [25, 27, 28]. However the authors used low-resolution 1.5T images and a semi-automated method, namely the active surface method (ASM) [14], which requires important manual intervention and lacks accuracy (CSA was reported to be overestimated by approximately 14% at C2 vertebral level compared with manual outlining). In previous studies, the semi-automated threshold-based method (TbM) proposed in [18] proved highly accurate and reproducible at 1.5T (i.e., inter-, intra-operator and scan-rescan reproducibility) [14, 18, 29]. Recently, TbM has been validated at 3T in a large dataset of 111 subjects including healthy volunteers, ALS, SMA and SCI patients [30]. TbM underestimated CSA

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 2 / 21 Spinal Cord Segmentation Method

by 3.29% at the cervical level and 1.70% at the thoracic level. Interestingly, TbM proved highly robust to initialization suggesting potential automation. The present study therefore has two main objectives: 1) To construct and validate a new fast and accurate segmentation method with minimal manual intervention benefitting from the high robustness of TbM to its initialization procedure; 2) As an application, to extend existing work at 1.5 and 3T [25, 26] to the construction of an accurate MRI template of the cervical spinal cord using a large sample of 3T images.

Materials and Methods MRI acquisition Subjects were positioned head-first supine, with a 2-centimeter-thick pillow to lift the head and no pillow below the neck. This strategy was used to limit the natural cervical cord lordosis at around C3–C4, i.e., excessive cord curvature in the antero-posterior direction. Subjects were positioned as comfortably as possible, centered in the scanner’s reference, and were systematically asked not to move during the acquisition in order to minimize motion artifacts. Scans were performed at CENIR, Pitié-Salpétrière Hospital, Paris, France, using a 3T MRI system (TIM Trio 32-channel, Siemens Healthcare, Erlangen, Germany). The spinal cord was imaged using a three-dimensional T2-weighted turbo spin echo (TSE) sequence with a slab- selective excitation pulse (SPACE sequence: Sampling Perfection with Application optimized Contrasts using different flip angle Evolution). Imaging parameters were: voxel size = 0.9×0.9×0.9 mm3; Field of View (FOV) = 280×280mm²; 52 sagittal slices; Repetition time (TR)/ Echo time (TE) = 1500/120ms; flip angle = 140°; generalized autocalibrating partially parallel acquisition (GRAPPA) with acceleration factor R = 3; turbo factor = 69; acquisition time ~6 min. This sequence provides high signal-to-noise ratio due to 3D acquisition, high resolution due to isotropic acquisition, short acquisition times by combining parallel acquisition with high turbo factors. No specific absorption rate limitation was observed in any of our acquisitions. In 1.5T spinal cord MRI studies, SPACE sequence showed an absence of signal loss due to pulsatile CSF motion and no artifacts due to swallowing or vessel pulsation [31], improving clinical diagnosis potential of MRI [32] and an excellent assessment of anatomy and resistance to artifacts in imaging difficult anatomy such as in scoliosis and spondylolysis [33], compared with conventional MRI sequences. At 3T, SPACE sequence showed an increased signal to noise ratio compared with 1.5T [34, 35] and allowed better inter-observer agreement for detecting foraminal stenosis, central spinal stenosis, and nerve compression compared with 2D T2-weighted TSE sequence [36].

Segmentation method

Preprocessing. Data were corrected for B1 non-uniformity intensity using Minc-Toolkit N3 [37]. Preprocessing of MR images using N3 has been shown to substantially improve the accuracy of anatomical analysis techniques such as tissue classification and segmentation as well as cortical surface extraction [38, 39]. The operator had to choose a region of interest inter- actively using an in-house Matlab script (The Mathworks Inc., Natick, MA, USA), as follows: The operator just had to display the sagittal slices where the spinal cord was the most median in the FOV, and to select with two mouse clicks the upper and lower limits of the spinal cord region to be segmented; the images were then automatically cropped to focus on this region of interest. This was the only manual intervention required in the whole segmentation procedure. The produced images were then resampled to a voxel size of 0.3×0.3×0.3 mm3 using 3D cubic interpolation in order to maximize segmentation accuracy [11, 30].

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 3 / 21 Spinal Cord Segmentation Method

Segmentation: Volumetric approach. Conventional TbM [18] requires manual initialization since the operator has to delineate in each axial slice to be segmented the contours of the CSF and spinal cord regions. TbM then uses the mid intensity of the mean intensities in these regions as a threshold to classify voxels into two classes, namely spinal cord and CSF. By conducting numerical simulations, we have shown previously that TbM was highly robust to its initial contours [30]. More specifically, simulations suggested optimal segmentation results would be obtained with initial contours representing as little as 10% to 30% of manually drawn areas for both spinal cord and CSF. Full automation of the TbM method could therefore be envisaged by automatically extracting CSF and cord regions, while allowing 10% to 30% error rate as compared with manually drawn areas. This led us to design a double threshold-based method with minimal manual intervention (DTbM) implemented using Matlab, which involved the following steps in each slice of the region of interest to be segmented (see Fig. 1 for a flowchart of the method):

Fig 1. Overview of DTbM steps (S). Slice 1 represents an example of a resampled axial image. Slices 2 and 3 represent the progressive contrast enhancement of slice 1. doi:10.1371/journal.pone.0122224.g001

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 4 / 21 Spinal Cord Segmentation Method

S1: The slice was segmented using Otsu’s global thresholding method (first threshold of DTbM) [40]. This method seeks for the optimal grey level threshold that maximizes the inter- class variance between the foreground and the background of the MRI images. The optimal threshold enables to select pixels of highest intensity, corresponding mainly to CSF.

S2: A binary mask MCSF of the CSF was obtained by selecting from these pixels the cluster with the highest intensity and a minimum size of 50 connected pixels.

S3: A flood-fill algorithm [41] was applied in order to fill the inner part of MCSF, yielding to a mask MF. S4a: If MF-MCSF was zero, this meant that the border of the CSF mask was not completely closed and there was no inner region to fill. The global contrast of the initial MRI slice was then enhanced by truncating the image histogram by 0.1% of its upper end and then steps S1 to

S3 were repeated until MF-MCSF was not zero, which meant that the border of the CSF mask was closed and the inner part had been completely filled.

S4b: If MF-MCSF was not zero, this difference between MF and MCSF corresponded to the inner part of the CSF mask, i.e. a binary mask of the spinal cord (MSC). S5: MSC and MCSF were eroded (disk radius of 3 pixels) in order to lie between 10% to 30% error rate as compared with the expected manually drawn areas [30]. S6: CSF and spinal cord regions obtained by masking the initial image by the binary masks from S5 were taken as initial inputs to the threshold-based method (TbM) proposed by Losseff et al. [18], as demonstrated in [30]. TbM uses the mid intensity of the mean intensities in the two regions as a threshold (second threshold of DTbM) to further classify region pixels into two classes, namely the spinal cord and the CSF [18]. CSA was defined as the mean area in mm2 of the resulted masks. S7: The segmented spinal cord regions obtained from S6 in all slices were corrected in order to adjust for possible missing or inaccurate segmentations. First, the centerline of the spinal cord was extracted by locating the center of mass of the spinal cord region in each slice. The centerline curve was smoothed using robust locally weighted regression [42]. The derivative of the centerline and polynomial curve fitting were used to detect aberrant discontinuities in the centerline then to remove them (a threshold of 15 voxels was set). Missing points in the centerline were then estimated using cubic spline interpolation. The radius from the center of mass of each axial mask to its borders was measured with an angular resolution of 5° by mapping Cartesian coordinates to polar coordinates (radius every 5°) [11]. To limit nerve roots contribution to the final segmentation, the length of the radiuses was smoothed for each angle using robust locally weighted regression across slices. Then, radiuses of the missing masks were estimated for each angle using cubic spline interpolation across slices. Finally, the contours of the spinal cord were calculated by converting polar coordinates (center of mass and radius every 5°) back to Cartesian coordinates. Finally, contours were converted to binary masks leading to a 3D mask covering the whole spinal cord region. Segmentation: cross-sectional area approach. When only a few slices needed to be segmented, the volumetric approach was replaced by a cross-sectional area approach, which merely consisted of segmenting preprocessed images using steps S1 to S6 as described above.

Comparison with existing methods First, the volumetric measures obtained from DTbM in all spinal cord vertebral levels were compared with those obtained from ASM [14] using visual scoring. Second, the CSA measures obtained from DTbM at C2 vertebral level were compared with those obtained from ASM and TbM [18] using quantitative criteria, taking manual outlining as the ground truth. All segmentations were achieved by an experienced operator (i.e. four years’ experience) using in-house

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 5 / 21 Spinal Cord Segmentation Method

Matlab scripts for manual outlining, DTbM and TbM and a trial version of Jim 6.0 (Xinapse Systems, UK; http://www.xinapse.com) for ASM. ASM is based on image smoothing followed by an active surface model constrained by intensity gradient image information. The operator placed landmarks on the center of the spinal cord in the axial plane at the extremes of the cord region to be segmented and (i) every 10 axial slices for volumetric approach and (ii) in all axial slices for CSA measures. The parameters of the method were: nominal cord diameter: range = 7–10 mm; number of shape coefficients = 24; order of the longitudinal variation = 12. For parameter adjustment, the experienced operator was guided by the user’s manual of the method (http://www.xinapse.com/Manual/), data characteristics as well as visual assessment of segmentation results.

Subjects One hundred and twelve subjects were randomly selected from our database of spinal cord 3T MRI scans. There were 82 healthy volunteers, 10 patients with ALS, 10 patients with SMA and 10 patients with SCI. Four healthy subjects were excluded from our study due to disk protru- sion compressing the spinal cord (n = 2, at C5/C6 level) or inducing loss of CSF space and con- tacting the spinal cord (n = 2, at C3/C4, C4/C5 and C5/C6 levels). The local Ethics Committee of our institution approved all experimental procedures (Paris-Ile de France Ethical Committee under the 2009-A00291-56 registration number), and written informed consent was obtained from each participant. MRI experiments were conducted according to the principles expressed in the latest revision of the Declaration of Helsinki.

Evaluation Groups. MR images from the healthy subjects comprised two fields of view, one covering the whole cervical and the upper thoracic spinal cord regions (n = 60) and the other including partly the cervical and the thoracic spinal cord regions (n = 18). MR images from the patients covered the whole cervical and upper thoracic spinal cord regions. Therefore, subjects were split into two groups: • Group 1 included 7 subjects randomly chosen from the 60 healthy volunteers with images covering the whole cervical and the upper thoracic spinal cord, 3 subjects randomly chosen from the 18 healthy volunteers with images including partly the cervical and the thoracic spinal cord (mean age of selected healthy subjects ± SD: 41.9 ± 16.5 years, 4 females), 10 patients with ALS (mean age ± SD: 54.5 ± 10.9 years, 3 females), 10 patients with SMA (mean age ± SD: 36.1± 10.9 years, 8 females), and 10 patients with SCI (mean age ± SD: 46.4±18.2 years, 2 females). Table 1 shows the segmented spinal cord region for each subject. Their images were used for visual evaluation of DTbM and ASM segmentations using the volumetric approach at the cervical and thoracic spinal cord regions. They were also used for quantitative evaluation of CSA accuracy measured at the middle of each available spinal cord vertebral level (Table 1). • Group 2 included 10 healthy volunteers randomly chosen from the 60 healthy subjects with images covering the whole cervical spinal cord (mean age ± SD: 45.7 ± 15.8 years, 6 females), and the same ALS, SMA and SCI patients as in Group 1. Their images were used for quantitative comparison of DTbM, TbM and ASM segmentations at C2 vertebral level. This region had high contrast between CSF and spinal cord as well as large CSF spaces and low CSA variability, and therefore segmentation was expected to be more reproducible and accurate at this level. Furthermore, recent studies measuring CSA have shown pronounced cord atrophy at C2 vertebral level in ALS and SCI patients that correlates with clinical disability [1, 10, 11].

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 6 / 21 Spinal Cord Segmentation Method

Table 1. Segmented spinal cord regions for each group of subjects (Upper–Lower vertebral level limits).

Controls ALS SMA SCI C2–T5 C2–T2 C2–T5 C2–T4 C7–T9 C2–T3 C2–T4 C2–T3 C2–T4 C2–T5 C2–T6 C2–C7 C3–T8 C2–T4 C2–T5 C2–T3 C2–T1 C2–T1 C2–T6 C2–T1 C3–T8 C2–T5 C2–T6 C2–T3 C2–T4 C2–T5 C2–T5 C2–T4 C2–T3 C2–T5 C2–T5 C2–T2 C2–T4 C2–T6 C2–T4 C2–T4 C2–T6 C2–T4 C2–T5 C2–T4 doi:10.1371/journal.pone.0122224.t001

All data were preprocessed using the method described in the “Preprocessing” section above. Visual evaluation (volumetric approach). Images from Group 1 were segmented using the volumetric version of DTbM (steps S1 to S7) and ASM [14]. Accuracy of DTbM and

Table 2. Criteria and scores for the visual evaluation.

Criteria Score 1- Global overlap agreement Insufficient to moderate 0 Substantial 1 Perfect 2 2- Overlap agreement by region Normal appearing spinal cord Insufficient1 0 Moderate2 1 Substantial3 2 Perfect4 3 Nerve roots Included 0 Not included 1 T2-hyperintensity Insufficient to moderate 0 Substantial to perfect 1 Atrophy Insufficient to moderate 0 Substantial to perfect 1 Narrow spinal canal Insufficient to moderate 0 Substantial to perfect 1

1Insufﬁcient to moderate overlap in at least 3 vertebral levels. 2Insufﬁcient to moderate overlap in less than 3 vertebral levels. 3Global substantial overlap agreement. 4Global perfect overlap agreement.

doi:10.1371/journal.pone.0122224.t002

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 7 / 21 Spinal Cord Segmentation Method

ASM segmentations was evaluated visually by an independent expert neuroradiologist (with seven years’ experience) using the OsiriX Imaging Software (http://www.osirix-viewer.com/). The operator was blinded to the type of data and methods. The evaluation used a combined score based on neuroradiological diagnosis and overlap criteria (Table 2): The operator evaluated visually the overlap between the spinal cord region and the mask resulting from each segmentation method in the axial plane, in the whole spinal cord in the FOV (Table 1, global

overlap) and in specific regions of the spinal cord that were subject to T2-hyperintensity, atrophy, or narrow spinal canal, which are usually present in clinical routine MR images of the pathological spinal cord. The operator also evaluated segmentation accuracy in visually normal appearing spinal cord regions and nerve roots regions. The overlap criteria were: • Insufficient: The method failed to segment accurately the spinal cord. The resulted axial masks underestimated/overestimated by a large number of voxels the visually expected masks. • Moderate: The resulted axial masks were close to but partially larger or smaller than the expected spinal cord region. • Substantial: The resulted axial masks underestimated/overestimated the visually expected spinal cord region by only few voxels. • Perfect: Resulted axial masks fitted perfectly the spinal cord region. • Included: Resulted axial masks included voxels from nerve roots. • Not included: Resulted axial masks did not include voxels from nerve roots. Quantitative evaluation. Quantitative evaluation of CSA measures obtained on the whole spinal cord using the volumetric approach would have required manual segmentation of a large number of slices to get a ground truth, which was untractable. Nevertheless, to try and evaluate the performance of the volumetric approach at all cord levels, a ground truth was obtained by manually segmenting the middle slice of each available spinal cord vertebral level (Table 1) for all subjects of Group 1. In the SCI group, only vertebral levels rostral and caudal to the spinal cord lesions were considered [10, 30]. CSA values obtained in this middle slice using DTbM (see section “Segmentation: volumetric approach” above) and ASM were compared with manual outlining. On the other hand, we carried out a specific evaluation of the cross-sectional approach at C2 vertebral level using images from Group 2, in order to conform to segmentation evaluations largely conducted in the literature. This required straightening the spinal cord before segmentation, as follows: the cord centerline was extracted (i.e. the line linking the centers of mass of all axial masks) and further smoothed using robust locally weighted regression. Spinal cord images were resampled in planes perpendicular to the centerline using 3D cubic interpolation leading to straight spinal cord images. Five three-millimeter-thick slices were selected with the most inferior slice passing centrally through the C2/C3 disk, as largely made in MS and SCI studies at C2 vertebral level [7–10, 14, 18, 43–46]. Unlike what has been proposed in [16, 43], we chose to resample data to account for partial volume between the spinal cord and the CSF as proposed in [11, 30]. In addition, cord straightening enabled to reduce partial volume due to the cord orientation between the CSF and the spinal cord as well as blurring at the interface between the two tissues, and thus increase segmentation accuracy [16, 43]. Images were segmented using manual segmentation, TbM, DTbM (as described in “Cross-sectional area approach” section above), and ASM.

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 8 / 21 Spinal Cord Segmentation Method

Accuracy was evaluated in two ways: (1) by using the Dice similarity coefficient (DSC), which evaluates the performance of segmentation methods by measuring their spatial overlaps [47]; the DSC is defined as follows: 2×|GT S| ⁄ (|GT| + |S|), where |GT| and |S| are the CSA in \ mm2 obtained by the manual segmentation (ground truth) and the segmentation method to be evaluated (DTbM, TbM or ASM), respectively. |GT S| is the area in mm2 common to both \ segmentation results; (2) by computing the relative CSA estimation error made by DTbM, TbM and ASM, considering the CSA from manual segmentation made by the experienced operator as the ground truth.

Template construction The 60 healthy subjects with images covering the whole cervical spinal cord (mean age ± SD: 44.3 ± 16.2 years, range: 21–72 years, 31 females) were used to construct the cervical spinal cord template. Preprocessing. The Minc-Toolkit N3 tool [37] was used to correct MR images for intensity non-uniformity then to apply intensity normalization. The experienced operator defined manually two landmarks in the sagittal slice where the spinal cord was the most median in the FOV to delimit the cervical spinal cord (i.e. from the upper limit of C2 vertebral level to the middle vertebral body C7/T1). After cropping and resampling the images as described in the “Volumetric approach: preprocessing” section above, a second non-uniformity intensity correction was applied using N3 in order to remove residual artifacts. Segmentation. The cervical spinal cord was segmented using the volumetric version of DTbM (steps S1 to S7) and the cord centerline was extracted (i.e. the line linking the centers of mass of all axial masks) and further smoothed using robust locally weighted regression. Standardization. Spinal cord images and cord masks were resampled in planes perpendicular to the centerline using 3D cubic interpolation for cord images and 3D nearest neighbor interpolation for masks, the centerline being centered in the FOV, which avoided registration between individuals as done in [25]. Thereby, the spinal cord images were always centered in the antero-posterior and left-right directions. Resulting images were standardized in length by rescaling them to the median cord length using 3D cubic interpolation for cord images and 3D nearest neighbor interpolation for masks (median length across subjects = 402 slices) as proposed in [25]. Template and probability tissue map. Straightened cord images were averaged across subjects and smoothed using 3D Gaussian filter (FWHM = 0.3×0.3×0.6 mm) to insure the va- lidity of the Gaussian random theory, as well as to correct for possible imperfections in the standardization process, as showed in [25], which yielded a template of the cervical spinal cord. Straightened cord masks were averaged, which yielded a tissue probability map for the spinal cord, where voxel value from 0 to 1 is the probability for the voxel to belong to the spinal cord.

Results Statistical analyses were conducted with Matlab software. Wilcoxon signed-rank test was used to compare accuracy measurement of DTbM, TbM and ASM at C2 vertebral level, and DTbM and ASM at the cervical and thoracic levels; p values less than 0.05 were considered significant.

Computational time The mean number (± SD) of segmented slices was 701±96 slices/subject. Computation time for the whole procedure was 3.9±1.2 min/subject (mean ± SD), which included preprocessing time: 38±4 s/subject, segmentation time: 1.9±1.1 min/subject, correction (step S7) time: 1.5

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 9 / 21 Spinal Cord Segmentation Method

±0.3 min/subject. This was achieved using a 64-bit Quad-core (Intel Xeon, processor speed: 2.67 GHz) workstation. Computation time was not optimized as the whole procedure was coded in Matlab language. Fig. 2 shows an example of segmentation result from DTbM in an ALS patient.

Fig 2. Segmentation and cord straightening. (a) T2-weighted mid-sagittal section in an ALS patient. The yellow plus signs delimit the spinal cord region to be preprocessed and segmented. (b) Mid-sagittal section of the preprocessed data and (c) resulted segmentation mask (orange). (d) Mid-coronal and (e) sagittal sections of the straightened spinal cord. (f) Mid-sagittal and (g) coronal sections of the straightened mask (orange). A, anterior; I, inferior; L, left, P, posterior, R, right, S, superior. doi:10.1371/journal.pone.0122224.g002

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 10 / 21 Spinal Cord Segmentation Method

Table 3. Results of the visual evaluation.

Criteria Number of scores for DTbM/Number of scores for ASM

Controls ALS SMA SCI 1- Global overlap agreement Insufficient to moderate 0/2 0/2 0/3 0/1 Substantial 2/8 1/8 1/7 3/8 Perfect 8/0 9/0 9/0 7/1 2- Overlap agreement by region Normal appearing spinal cord Insufficient 0/5 0/5 0/7 0/4 Moderate 0/4 0/3 0/2 1/2 Substantial 3/1 2/2 3/1 3/4 Perfect 7/0 8/0 7/0 6/0 Nerve roots Included 0/1 1/1 0/2 0/0 Not included 10/9 9/9 10/8 10/10 T2-hyperintensity Insufficient to moderate 0/1* 0/1 1/0 4/5 Substantial to perfect 1/0* 1/0 0/1 4/3 Atrophy Insufficient to moderate - 0/2 0/0 4/4 Substantial to perfect - 2/0 1/1 2/2 Narrow spinal canal Insufficient to moderate - 0/1 1/1 2/1 Substantial to perfect - 1/0 1/1 1/2

*Asymptomatic syrinx revealed by the visual evaluation. doi:10.1371/journal.pone.0122224.t003

Visual evaluation (volumetric approach) Accuracy results are shown in Table 3 for each group of subjects. DTbM outperformed ASM in both healthy and pathological groups for global overlap (number of scores for DTbM/ASM: 33/1 perfect, 7/31 substantial and 0/ 8 insufficient to moderate), overlap in normal appearing spinal cord (DTbM/ASM: 28/0 perfect, 11/8 substantial, 1/11 moderate and 0/21 insufficient) and nerve roots regions (DTbM/ASM: 39/36 not included and 1/4 included in the final seg-

mentations). Both DTbM and ASM showed similar overlap performance in regions with T2- hyperintensity (DTbM/ASM: 6/4 substantial to perfect and 5/7 insufficient to moderate), atrophy (DTbM/ASM: 5/3 substantial to perfect and 4/6 insufficient to moderate) and narrow spinal canal (DTbM/ASM: 3/3 substantial to perfect and 3/3 insufficient to moderate). The individual visual scoring is shown in S1 Table.

Quantitative evaluation Volumetric approach. Table 4 shows accuracy measurements at the cervical and thoracic levels. At the cervical level, the mean DSC value was 95.71% for DTbM and 90.78 for ASM (manual outlining was considered as the ground truth). At the thoracic level, the mean DSC value was 94.27% for DTbM and 89.93% for ASM. When considering separately each group of subjects, the mean DSC value was 96.07% for DTbM and 90.89% for ASM in the control group. In the ALS group, the mean DSC value was 95.71% for DTbM and 90.06% for ASM. In

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 11 / 21 Spinal Cord Segmentation Method

Table 4. Accuracy of mean cross-sectional area measured at the cervical and thoracic levels.

Cervical level Thoracic level

Method Method

Accuracy measurements DTbM ASM DTbM ASM Mean (SD) dice similarity coefﬁcient (%) Whole population 95.71 (1.00) 90.78 (2.85) 94.27 (4.31) 89.93 (4.21) Controls 96.07 (0.81) 90.89 (3.62) 95.33 (0.8) 89.57 (3.55) ALS 95.71 (1.19) 90.06 (2.75) 94.91 (1.04) 90.74 (2.57) SMA 95.30 (1.01) 91.07 (2.51) 95.12 (0.84) 91.03 (1.60) SCI 95.75 (0.93) 91.10 (2.73) 91.07 (8.95) 88.00 (7.65) Mean (SD) error estimation (%) Whole population -4.86 (3.68) 13.53 (6.73) -6.49 (3.42) 9.47 (7.87) Controls -5.05 (2.41) 14.34 (7.32) -4.81 (3.00) 16.08 (9.49) ALS -4.38 (4.20) 15.44 (6.48) -6.87 (2.37) 14.99 (6.92) SMA -4.39 (5.20) 13.88 (9.05) -6.72 (2.62) 15.26 (4.28) SCI -5.61 (2.63) 11.08 (4.28) -7.54 (4.96) 12.08 (8.47)

Manual segmentation by the experienced operator was the ground truth for accuracy measurements. doi:10.1371/journal.pone.0122224.t004

the SMA group, the mean DSC value was 95.30% for DTbM, 91.07% for ASM. Finally, in the SCI group, the mean DSC value was 95.75% for DTbM and 91.10% for ASM. DTbM underestimated the mean CSA measured by manual segmentation (considered as the ground truth) by 4.86% and 6.49% and ASM overestimated the mean CSA measured by manual segmentation by 13.53% and 9.47% at the cervical and thoracic levels, respectively. When considering separately each group of subjects, DTbM underestimated the mean CSA measured by manual segmentation by 5.05% and 4.81% and ASM overestimated the mean CSA by 14.34% and 16.08% at the cervical and thoracic levels, respectively. In the ALS group, DTbM underestimated the mean CSA measured by manual segmentation by 4.38% and 6.87% and ASM overestimated the mean CSA by 15.44% and 14.99% at the cervical and thoracic levels, respectively. In the SMA group, DTbM underestimated the mean CSA measured by manual segmentation by 4.39% and 6.72% and ASM overestimated the mean CSA by 13.88% and 15.26% at the cervical and thoracic levels, respectively. Finally, in the SCI group, DTbM underestimated the mean CSA measured by manual segmentation by 5.61% and 7.54% and ASM overestimated the mean CSA by 11.08% and 12.08% at the cervical and thoracic levels, respectively. The behaviour of DTbM method is illustrated in S1 Fig. Based on DSC values, DTbM showed a significantly higher accuracy for the whole population than ASM at both the cervical and thoracic levels (Wilcoxon signed-rank test, p<0.001). Cross-sectional area approach at C2 vertebral level. Table 5 shows mean CSA measurements made by the experienced operator using manual outlining, as well as DTbM, TbM and ASM results. Results of accuracy are also given. The mean DSC value was high for DTbM (97.98%), TbM (98.02%) and ASM (96.76%). When considering separately each group of subjects, the mean DSC value was 98.18% for DTbM, 98.2% for TbM and 96.63% for ASM in the control group. In the ALS group the mean DSC value was 97.86% for DTbM, 97.92% for TbM and 96.83% for ASM. In the SMA group the mean DSC value was 98.01% for DTbM, 98.11% for TbM and 96.83% for ASM. Finally, in the SCI group, the mean DSC value was 97.86% for DTbM, 97.87% for TbM and 96.77% for ASM.

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 12 / 21 Spinal Cord Segmentation Method

Table 5. Accuracy of mean cross-sectional area measured at C2 vertebral level.

Method Method

Accuracy measurements Manual DTbM TbM ASM Mean (SD) CSA (mm2) 81.07 (13.3) 80.97 (14.08) 80.86 (13.96) 85.19 (14.52) Mean (SD) Dice similarity coefﬁcient (%) Whole population – 97.98 (0.43) 98.02 (0.38) 96.76 (0.85) Controls – 98.18 (0.33) 98.2 (0.31) 96.63 (1.14) ALS – 97.86 (0.53) 97.92 (0.44) 96.82 (0.98) SMA – 98.01 (0.45) 98.11 (0.39) 96.83 (0.64) SCI – 97.86 (0.34) 97.87 (0.34) 96.77 (0.65) Mean (SD) error estimation (%) Whole population – -0.25 (2.13) -0.36 (1.90) 5.02 (3.62) Controls – 0.34 (1.87) 0.10 (1.78) 6.21 (2.76) ALS – 0.25 (2.64) 0.11 (2.37) 5.07 (3.69) SMA – -0.63 (2.08) -0.74 (1.64) 4.66 (3.75) SCI – -0.96 (1.88) -0.91 (1.79) 4.13 (4.35)

Manual segmentation by the experienced operator was the ground truth for accuracy measurements. doi:10.1371/journal.pone.0122224.t005

DTbM and TbM underestimated the mean CSA measured by manual segmentation (considered as the ground truth) by 0.25% and 0.36%, respectively. ASM overestimated the mean CSA measured by manual segmentation by 5.02%. When considering separately each group of subjects, DTbM, TbM and ASM overestimated the mean CSA measured by manual segmentation by 0.34%, 0.10% and 6.21%, respectively, in the control group. In the ALS group, DTbM, TbM and ASM overestimated the mean CSA measured by manual segmentation by 0.34% for DTbM, 0.10% for TbM and 6.21% for ASM. In the ALS group, DTbM, TbM overestimated the mean CSA measured by manual segmentation by 0.25% for DTbM and 0.11% for TbM and 5.07%. In the SMA group, DTbM, TbM underestimated the mean CSA measured by manual segmentation by 0.63% for DTbM and 0.74% for TbM. ASM overestimated the mean CSA measured by manual segmentation by 4.66%. Finally, in the SCI group, DTbM, TbM underestimated the mean CSA measured by manual segmentation by 0.96% for DTbM and 0.91% for TbM. ASM overestimated the mean CSA measured by manual segmentation by 4.13%. Based on DSC values, DTbM showed a significantly higher accuracy for the whole population than ASM (Wilcoxon signed-rank test, p<0.001) and similar accuracy to TbM (Wilcoxon signed-rank test, not significant).

Template Fig. 2 shows an example of cord straightening for images from an ALS patient. Visual evaluation revealed an asymptomatic syrinx in the MR images of one healthy subject, who was subse- quently excluded from template construction, leading to a final group of 59 healthy volunteers (mean age ± SD: 44.6 ± 16.2 years, range: 21–72 years, 31 females). During the template construction step, the experienced operator had to adjust visually poor segmentations in rare occa- sions (error rate = 2.57%, defined as: 100 × (the number of mis-segmented axial slices)/(the total number of segmented axial slices)) using TbM [18]. Finally, images from all subjects were centered and straightened. Template and probability tissue map of the cervical spinal cord for the 59 healthy subjects are shown in Fig. 3.

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 13 / 21 Spinal Cord Segmentation Method

Fig 3. Cervical spinal cord template and probability tissue map. (a) Coronal (left), sagittal (middle) and axial views (right) of the cervical spinal cord template through mid-vertebral level from C2 to C7. (b) Probability tissue map. Views are arranged in the same order and orientation as in (a). A, anterior; I, inferior, L, left, P, posterior, R, right; S, superior. doi:10.1371/journal.pone.0122224.g003

Discussion DTbM accuracy was evaluated visually by a neuroradiologist at the cervical and thoracic spinal cord levels and compared with ASM. Besides, DTbM accuracy was visually and quantitatively evaluated at the cervical and thoracic levels and compared with ASM. In addition, DTbM accuracy was quantitatively evaluated at the C2 vertebral level and compared with ASM and the

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 14 / 21 Spinal Cord Segmentation Method

well-established method for CSA measurements, namely TbM. A template and a tissue probability map of the cervical spinal cord were constructed.

Comparison of segmentation methods DTbM showed higher accuracy for segmenting both the normal and the pathological spinal cord than ASM for the cervical and thoracic spinal cord levels. The lower accuracy of ASM can be explained by the fact that this method relies on hypotheses that are not fulfilled in the pathological spinal cord. These hypotheses are that (a) the cross-sectional shape of the cord varies only slowly in the superior-inferior direction and (b) in cross-sections, the radius from the estimated center to the edges varies smoothly around the edge [14]. Such hypotheses are valid when considering short regions of the spinal cord, for instance C2/C5 vertebral levels where the method has been validated and used to quantify atrophy in MS patients [14]. However, when considering large regions of the spinal cord (Table 1), the first hypothesis is no more valid even in healthy subjects as shown in post-mortem studies [48, 49] and recently in vivo [50]. In neurodegenerative diseases, ALS is characterized by cell loss in the anterior horn and corticospinal tract degeneration [22]. Corticospinal tract degeneration induces atrophy in the lateral direction and loss of alpha motor neurons results in atrophy in the ventral direction. SMA is characterized by degeneration of cell bodies in the anterior horn [21], which potentially induces a pronounced ventral atrophy. In SCI patients, more pronounced atrophy was reported in the antero-posterior direction than in the left/right direction at C2 vertebral level [11]. Furthermore, trauma induced large spinal cord shape deformation at lesion level and around it. Therefore the second hypothesis may not be valid in ALS, SMA and SCI patients and this could explain the lack of ASM accuracy in segmenting pathological populations. As DTbM depends only on image contrast, it has the advantage of being less sensitive to variations in spinal cord shape as compared with ASM. Quantitative evaluation of DTbM and ASM compared with manual segmentation was not possible for volumetric approach in this study due to the large number of slices to be segmented. For this reason, we resorted to a visual evaluation strategy. In one control subject, visual

evaluation revealed a central and focal T2-hypersignal at C5 vertebral level corresponding to an asymptomatic syrinx. This subject was therefore excluded from template construction. Fur- thermore, accuracy results of the visual evaluation were confirmed by the quantitative evaluation of CSA measurement by using DTbM and ASM at the middle of each available spinal cord vertebral level (Table 1, Table 4). The lower accuracy at the thoracic level than at the cervical level could be explained by the lower signal-to-noise ratio inherent to the thoracic coil compared with the cervical coil as well as the choice made for shimming box positioning to focus on cervical and upper thoracic levels [2, 3, 5, 10]. A quantitative evaluation at C2 vertebral level in healthy and pathological subjects was also achieved (Table 5). Even at C2 vertebral level, our analysis showed a higher accuracy of DTbM compared with ASM and similar accuracy for DTbM and TbM. ASM showed improved accuracy than previously reported in [14]. After data preprocessing, DTbM was fully automatic and hence operator independent, which is not the case for TbM and ASM [14, 18, 30, 51]. In regions where CSF spaces were reduced (i.e. regions of narrow spinal canal), Otsu’s global thresholding method failed to extract the CSF correctly. Thus, global contrast enhancement was needed. However, this strategy did not enable initializing contours optimally with TbM leading to misestimating the spinal cord region [30]. In case this strategy failed, correction step

S7 and reduction in nerve roots contribution and in the small T2-hyperintense region in the final segmentation result were useful (1 control, 1 ALS, 4 SCI patients). However, segmentation

and correction step S7 failed in regions of large areas of T2-hyperintensity in the spinal cord (1

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 15 / 21 Spinal Cord Segmentation Method

SMA patient), large regions of spinal canal narrowing with limited CSF space (1 SMA patient)

as well as at the lesion level in 4 SCI patients with T2-hypersignal and atrophy of the spinal cord and narrow spinal canal.

Template construction DTbM was used to construct a spinal cord template. Despite the high spatial resolution of imaging data, we were not able to determine the precise position of the nerve roots in all subjects, which would ideally enable to standardize all subjects spinal level by spinal level. Alternatively, two ways of standardization can be considered: 1) standardization to the whole median cord length [25]; 2) standardization to the median vertebral level for each vertebral level. As demonstrated previously for the cervical spinal cord [52], nerve root position showed less variation than vertebral body position. Therefore, the first strategy was expected to better preserve nerve roots position than the second. Furthermore, this strategy enabled to remove the effect of cord length in the standardized images [25]. Compared with previous studies that used ASM to construct low-resolution 1.5T MRI cervical spinal cord templates [25, 27, 28], the segmentation method proposed here using high-resolution 3T MR images showed improved accuracy. In addition, the new method resulted in a centered and straight template and probability tissue map. In a recent study, a spinal cord template from 3T T2-weighted MR images and a small sample of 16 subjects was proposed with minimal manual intervention [26]. The template included the cervical and mid-thoracic regions (i. e., from C2 to T6 vertebral levels). MR images were segmented using a graph-cut segmentation method that is known to lack validation [53]. The standardization procedure consisted of a nonlinear registration of each subject’s spinal canal into an average space as it is usually done when spatially normalizing brain MR images [54– 56]. Given the small size of the spinal cord compared with CSF as well as the high inter-subject CSA variability for CSF in comparison to the spinal cord [57], nonlinear registration of the whole spinal canal is expected to increase bias when registering subjects into an average space, while in the present study we focussed on standardizing the spinal cord only. Unlike what we proposed, the standardization procedure proposed in [26] did not take inter-individual nerve roots position variability into account [58]. Besides, the error in CSA area estimation using nonlinear registration (mean error = 16%) would make the method too inaccurate for atrophy quantification in the pathological context [2, 3, 5, 10].

Atrophy quantification in neurodegenerative diseases and trauma Grey and white matter could not be differentiated in the acquired images due to insufficient contrast inside the spinal cord. However, as mentioned before, atrophy may predominate in a particular direction in neurodegenerative diseases and trauma. This preferential direction of cord atrophy may be detected in group analysis using surface-based morphometry techniques on spinal cord masks normalized to the template space instead of using voxel-based morphometry. For instance, morphological changes have been successfully detected in the hippocampus and ventricles in Alzheimer’s disease using radial distance approach and tensor-based morphometry [59, 60] and in the lateral ventricles in HIV/AIDS patients using multivariate tensor- based morphometry [61], as well as in the basal ganglia in ALS patients using surface-based vertex approach [62]. Our segmentation method may also measure accurately CSA and spinal cord volumes in a large range of pathologies.

Limitations and future directions Manual outlining was taken as the ground truth for methods evaluation, however the accuracy of this procedure strongly depends on the operator’s experience degree [63]. Nevertheless, we

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 16 / 21 Spinal Cord Segmentation Method

believe that segmentations performed by the trained operator (four years’ experience) are close to the ground truth and therefore bias should be minimal. Future studies could use majority voting or simultaneous truth and performance level estimation methods to produce ground truth segmentation for method evaluation [64, 65]. We did not have the opportunity to conduct a scan-rescan evaluation of DTbM. However, DTbM is based on the well-established method TbM [18], which showed low scan-rescan variability [29]. Thus, DTbM is expected to have the same order of variability as TbM, but this is worth further investigating. Future studies could also integrate two anatomical imaging acquisitions for accurate nerve

root localization [66] and for white and grey matter delineation using high-resolution T2Ã- weighted 2D gradient recalled echo sequence, a multi-echo gradient-echo sequence (MGE) or a gradient echo with a multi-shot echo-planar imaging sequence (GE-MS EPI) [67, 57]. For instance, MGE sequence has already enabled construction of a probabilistic cervical and thoracic spinal cord atlas of 15 subjects [57]. However, because of acquisition time constraint, only one slice per vertebral level was acquired (0.5×0.5 mm2 in-plane resolution and 5 mm slice thickness). Furthermore, manual segmentation was used for grey and white matter delineation, a procedure that is time consuming and the accuracy of which strongly depends on the operator’s experience degree [30, 63]. GE-MS EPI sequence also enabled the construction of an atlas of the cervical spinal cord. However, multiple manual outlining was needed to construct a “gold standard” segmentation, which is extremely time consuming [68], despite the benefit in

accuracy. As T2Ã-weighted images present a spinal cord/CSF contrast similar to that of T2- weighted images, DTbM would likely be able to segment the spinal cord accurately using T2Ã- weighted data. Grey matter could be extracted from such images using an atlas-based segmentation approach via the template from Taso et al. (2014). Resulted masks from T2-weighted and T2Ã-weighted images could be fused to construct a probabilistic spinal cord atlas of white and grey matter as recently proposed in [26]. Besides, as DTbM segments each axial slice independently, parallelization of segmentations could be envisaged [69], which would drastically reduce time computation (few milliseconds per slice).

Conclusion We developed a fast and accurate semi-automated segmentation method of the spinal cord at both cervical and thoracic levels from 3T T2-weighted MR images, applied to healthy subjects and patients with various spinal cord diseases. DTbM showed higher accuracy than ASM and similar accuracy to TbM, however it clearly facilitated the whole processing by limiting the manual intervention. This enabled the construction of a cervical spinal cord template and tissue probability map.

Supporting Information S1 Fig. Illustration of the behaviour of DTbM method. Good agreement of DTbM method: (a,b) in a spinal cord region with T2-hypersignal in a SCI patient; (c,d) in an atrophied spinal cord region in an ALS patient; (e,f) in a narrow spinal canal region in an SMA patient. Failure of DTbM method: (g,h) in a region with T2-hypersignal in a SCI patient; (i,j) in an atrophied spinal cord region in an ALS patient; (k,l) in a narrow spinal canal region in an ALS patient. A, anterior; I, inferior, L, left, P, posterior, R, right; S, superior. (TIFF)

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 17 / 21 Spinal Cord Segmentation Method

S1 Table. Results of the individual visual scoring. (XLSX)

Acknowledgments We are grateful to all subjects and their relatives. We thank Kevin Nigaud, Romain Valabrègue, Alexandre Vignaud, Frédéric Humbert, Christelle Macia, Mélanie Didier and Eric Bardinet from CENIR for helping with the acquisitions. We thank Drs Mark A. Horsfield for providing us the trial version of Jim 6.0. We thank the clinicians of the Paris ALS Center, Gaëlle Brune- teau, Lucette Lacomblez, Marie-Violaine Lebouteux, Timothée Lenglet, François Salachas and Vincent Meininger. We thank the clinicians of Pitié-Salpêtrière Hospital, Drs Anthony Behin, Tanya Stojkovic, Pascal Laforêt and Bruno Eymard for their contribution. We thank also Thierry Albert from Coubert Hospital, Caroline Hugeron from Raymond-Poincaré Hospital, Robert Lafaye de Micheaux and Isabelle Petit from Invalides National Institute for their contribution.

Author Contributions Conceived and designed the experiments: MMEM PFP HB. Performed the experiments: MMEM. Analyzed the data: MMEM PFP RC BT NV ST. Contributed reagents/materials/analysis tools: MMEM PFP RC BT NV MPI. Wrote the paper: MMEM RC BT NV ST MPI SL PFP HB. MRI platform availability: SL.

References 1. Branco LM, De Albuquerque M, De Andrade HM, Bergo FP, Nucci A, França MC Jr (2014) Spinal cord atrophy correlates with disease duration and severity in amyotrophic lateral sclerosis. Amyotroph Later- al Scler Frontotemporal Degener 15(1–2): 93–7. 2. El Mendili MM, Cohen-Adad J, Pelegrini-Issac M, Rossignol S, Morizot-Koutlidis R, Marchand-Pauvert V, et al. (2014) Multi-parametric spinal cord MRI as potential progression marker in amyotrophic lateral sclerosis. PLoS ONE 9(4):e95516. doi: 10.1371/journal.pone.0095516 PMID: 24755826 3. Cohen-Adad J, El Mendili MM, Morizot-Koutlidis R, Lehéricy S, Meininger V, Blancho S, et al. (2013) In- volvement of spinal sensory pathway in ALS and specificity of cord atrophy to lower motor neuron degeneration. Amyotroph Lateral Scler Frontotemporal Degener 14(1): 30–38. doi: 10.3109/17482968. 2012.701308 PMID: 22881412 4. Agosta F, Rocca MA, Valsasina P, Sala S, Caputo D, Perini M, et al. (2009) A longitudinal diffusion tensor MRI study of the cervical cord and brain in amyotrophic lateral sclerosis patients. J Neurol Neuro- surg Psychiatry 80(1): 53–55. doi: 10.1136/jnnp.2008.154252 PMID: 18931009 5. El Mendili MM, Cohen-Adad J, Blancho S, Morizot-Koutlidis R, Lenglet T, Lehéricy S, et al. (2012) Spi- nal MRI atrophy study in amyotrophic lateral sclerosis and spinal muscular atrophy. In: Proceeding of the 10th annual meeting of The European Network for the Cure of ALS (ENCALS), Dublin, Ireland, Ab- stract 01. 6. Lukas C, Sombekke MH, Bellenberg B, Hahn HK, Popescu V, Bendfeldt K, et al. (2013) Relevance of spinal cord abnormalities to clinical disability in multiple sclerosis: MR imaging findings in a large cohort of patients. Radiology 269(2): 542–552. doi: 10.1148/radiol.13122566 PMID: 23737540 7. Cohen AB, Neema M, Arora A, Dell'oglio E, Benedict RH, Tauhid S, et al. (2012) The relationships among MRI-defined spinal cord involvement, brain involvement, and disability in multiple sclerosis. J Neuroimaging 22(2): 122–128. doi: 10.1111/j.1552-6569.2011.00589.x PMID: 21447024 8. Lin X, Tench CR, Evangelou N, Jaspan T, Constantinescu CS (2004) Measurement of spinal cord atrophy in multiple sclerosis. J Neuroimaging 14(3 Suppl): 20S–26S. PMID: 15228756 9. Freund P, Weiskopf N, Ashburner J, Wolf K, Sutter R, Altmann DR, et al. (2013) MRI investigation of the sensorimotor cortex and the corticospinal tract after acute spinal cord injury: a prospective longitudinal study. Lancet Neurol 12(9): 873–881. doi: 10.1016/S1474-4422(13)70146-7 PMID: 23827394 10. Cohen-Adad J, El Mendili MM, Lehéricy S, Pradat PF, Blancho S, Rossignol S, et al. (2011) Demyelin- ation and degeneration in the injured human spinal cord detected with diffusion and magnetization

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 18 / 21 Spinal Cord Segmentation Method

transfer MRI. Neuroimage 55(3): 1024–1033. doi: 10.1016/j.neuroimage.2010.11.089 PMID: 21232610 11. Lundell H, Barthelemy D, Skimminge A, Dyrby TB, Biering-Sørensen F, Nielsen JB (2011) Independent spinal cord atrophy measures correlate to motor and sensory deficits in individuals with spinal cord injury. Spinal Cord 49(1): 70–75. doi: 10.1038/sc.2010.87 PMID: 20697420 12. Bergo FPG, Franca MC, Chevis CF, Cendes F (2012) SpineSeg: A segmentation and measurement tool for evaluation of spinal cord atrophy. In: Proceeding of the 7th Iberian Conference on Information Systems and Technologies (CISTI), Madrid, Spain, 2: 400–403. 13. McIntosh C, Hamarneh G, Toom M, Tam RC (2011) Spinal cord segmentation for volume estimation in healthy and multiple sclerosis subjects using crawlers and minimal paths. Proceedings of the First IEEE International Conference on Healthcare Informatics, Imaging and Systems Biology (HISB) 25–31. 14. Horsfield MA, Sala S, Neema M, Absinta M, Bakshi A, Sormani MP, et al. (2011) Rapid semi-automatic segmentation of the spinal cord from magnetic resonance images: application in multiple sclerosis. Neuroimage 50(2): 446–455. 15. Zivadinov R, Banas AC, Yella V, Abdelrahman N, Weinstock-Guttman B, Dwyer MG (2008) Compari- son of three different methods for measurement of cervical cord atrophy in multiple sclerosis. AJNR Am J Neuroradiol 29(2): 319–325. PMID: 17974604 16. Tench CR, Morgan PS, Constantinescu CS (2005) Measurement of cervical spinal cord cross-sectional area by MRI using edge detection and partial volume correction. J Magn Reson Imaging 21(3): 197–203. PMID: 15723367 17. Hickman SJ, Hadjiprocopis A, Coulon O, Miller DH, Barker GJ (2004) Cervical spinal cord MTR histogram analysis in multiple sclerosis using a 3D acquisition and a B-spline active surface segmentation technique. Magn Reson Imaging 22(6): 891–895. PMID: 15234459 18. Losseff NA, Webb SL, O'Riordan JI, Page R, Wang L, Barker GJ, et al. (1996) Spinal cord atrophy and disability in multiple sclerosis. A new reproducible and sensitive MRI method with potential to monitor disease progression. Brain 119(Pt 3): 701–708. PMID: 8673483 19. Chen M, Carass A, Oh J, Nair G, Pham DL, Reich DS, et al. (2013) Automatic magnetic resonance spinal cord segmentation with topology constraints for variable fields of view. Neuroimage 83: 1051–1062. doi: 10.1016/j.neuroimage.2013.07.060 PMID: 23927903 20. Klein JP, Arora A, Neema M, Healy BC, Tauhid S, Goldberg-Zimring D, et al. (2011) A 3T MR imaging investigation of the topography of whole spinal cord atrophy in multiple sclerosis. AJNR Am J Neurora- diol 32(6): 1138–1142. doi: 10.3174/ajnr.A2459 PMID: 21527570 21. Katsuno M, Banno H, Suzuki K, Adachi H, Tanaka F, Sobue G (2010) Clinical features and molecular mechanisms of spinal and bulbar muscular atrophy (SBMA). Adv Exp Med Biol 685: 64–74. PMID: 20687495 22. Wijesekera LC, Leigh PN (2009) Amyotrophic lateral sclerosis. Orphanet J Rare Dis 4: 3. doi: 10.1186/ 1750-1172-4-3 PMID: 19192301 23. Wyndaele M, Wyndaele JJ (2006) Incidence, prevalence and epidemiology of spinal cord injury: what learns a worldwide literature survey? Spinal Cord 44(9): 523–529. PMID: 16389270 24. Stroman PW, Figley CR, Cahill CM (2008) Spatial normalization, bulk motion correction and coregistration for functional magnetic resonance imaging of the human cervical spinal cord and brainstem. Magn Reson Imaging 26(6): 809–814. doi: 10.1016/j.mri.2008.01.038 PMID: 18499380 25. Valsasina P, Horsfield MA, Rocca MA, Absinta M, Comi G, Filippi M (2012) Spatial normalization and regional assessment of cord atrophy: voxel-based analysis of cervical cord 3D T1-weighted images. AJNR Am J Neuroradiol 33(11): 2195–2200. doi: 10.3174/ajnr.A3139 PMID: 22678848 26. Fonov VS, Le Troter A, Taso M, De Leener B, Lévêque G, Benhamou M et al. (2014) Framework for integrated MRI average of the spinal cord white and gray matter: the MNI-Poly-AMU template. Neuro- image 102 Pt 2:817–27. doi: 10.1016/j.neuroimage.2014.08.057 PMID: 25204864 27. Rocca MA, Valsasina P, Damjanovic D, Horsfield MA, Mesaros S, Stosic-Opincal T, et al. (2013) Voxel-wise mapping of cervical cord damage in multiple sclerosis patients with different clinical phenotypes. J Neurol Neurosurg Psychiatry 84(1): 35–41. doi: 10.1136/jnnp-2012-303821 PMID: 23064100 28. Valsasina P, Rocca MA, Horsfield MA, Absinta M, Messina R, Caputo D, et al. (2013) Regional cervical cord atrophy and disability in multiple sclerosis: a voxel-based analysis. Radiology 266(3): 853–861. doi: 10.1148/radiol.12120813 PMID: 23192778 29. Leary SM, Parker GJ, Stevenson VL, Barker GJ, Miller DH, Thompson AJ (1999) Reproducibility of magnetic resonance imaging measurements of spinal cord atrophy: the role of quality assurance. Magn Reson Imaging 17(5):773–6. PMID: 10372531 30. El Mendili MM, Chen R, Tiret B, Pélégrini-Issac M, Cohen-Adad J, Lehéricy S, et al. (2014) Validation of a semi-automated spinal cord segmentation method. J Magn Reson Imaging. doi: 10.1002/jmri.24571

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 19 / 21 Spinal Cord Segmentation Method

31. Meindl T, Wirth S, Weckbach S, Dietrich O, Reiser M, Schoenberg SO, et al. (2009) Magnetic resonance imaging of the cervical spine: comparison of 2D T2-weighted turbo spin echo, 2D T2*weighted gradient-recalled echo and 3D T2-weighted variable flip-angle turbo spin echo sequences. Eur Radiol 19(3): 713–721. doi: 10.1007/s00330-008-1175-7 PMID: 18813933 32. Lichy MP, Wietek BM, Mugler JP, Horger W, Menzel MI, Anastasiadis A, et al. (2005) Magnetic resonance imaging of the body trunk using a single-slab, 3-dimensional, T2-weighted turbo-spin-echo sequence with high sampling efficiency (SPACE) for high spatial resolution imaging: initial clinical experiences. Invest Radiol 40(12): 754–760. PMID: 16304477 33. Tins B, Cassar-Pullicino V, Haddaway M, Nachtrab U (2012) Three-dimensional sampling perfection with application-optimised contrasts using a different flip angle evolutions sequence for routine imaging of the spine: preliminary experience. Br J Radiol 85(1016): e480–489. doi: 10.1259/bjr/25760339 PMID: 22374284 34. Del Grande F, Chhabra A, Carrino JA (2012) Getting the most out of 3 tesla MRI of the spine. Rheuma- tology Network. 35. Rodegerdts EA, Boss A, Riemarzik K, Lichy M, Schick F, Claussen CD, et al. (2006) 3D imaging of the whole spine at 3T compared to 1.5T: initial experiences. Acta Radiol 47(5): 488–493. PMID: 16796312 36. Lee S, Jee WH, Jung JY, Lee SY, Ryu KS, Ha KY (2014) MRI of the lumbar spine: comparison of 3D isotropic turbo spin-echo SPACE sequence versus conventional 2D sequences at 3.0 T. Acta Radiol “In press”. 37. Sled JG, Zijdenbos AP, Evans AC (1998) A nonparametric method for automatic correction of intensity nonuniformity in MRI data. IEEE Trans Med Imaging 17(1): 87–97. PMID: 9617910 38. Boyes RG, Gunter JL, Frost C, Janke AL, Yeatman T, Hill DL, et al. (2008) Intensity non-uniformity correction using N3 on 3-T scanners with multichannel phased array coils. Neuroimage 39(4): 1752–1762. PMID: 18063391 39. Zheng W, Chee MW, Zagorodnov V (2009) Improvement of brain segmentation accuracy by optimizing non-uniformity correction using N3. Neuroimage 48(1): 73–83. doi: 10.1016/j.neuroimage.2009.06.039 PMID: 19559796 40. Otsu N (1979) A threshold selection method from gray-level histograms. Systems, Man and Cybernet- ics, IEEE Transactions on 1(9): 62–66. 41. Soille P (2003) Morphological image analysis: principles and applications. Springer-Verlag New York, Inc. Secaucus, NJ, USA. 128 p. 42. Cleveland WS (1981) LOWESS: A program for smoothing scatterplots by robust locally weighted regression. The American Statistician 35(1): 54. 43. Carbonell-Caballero J, Manjón JV, Martí-Bonmatí L, Olalla JR, Casanova B, de la Iglesia-Vayá M, et al. (2006) Accurate quantification methods to evaluate cervical cord atrophy in multiple sclerosis patients. MAGMA 19(5): 237–246. PMID: 17115124 44. Freund PA, Dalton C, Wheeler-Kingshott CA, Glensman J, Bradbury D, Thompson AJ, et al. (2010) Method for simultaneous voxel-based morphometry of the brain and cervical spinal cord area measurements using 3D-MDEFT. J Magn Reson Imaging 32(5): 1242–1247. doi: 10.1002/jmri.22340 PMID: 21031531 45. Freund PA, Weiskopf N, Ward NS, Hutton C, Gall A, Ciccarelli O, et al. (2011) Disability, atrophy and cortical reorganization following spinal cord injury. Brain 134(Pt 6): 1610–1622. doi: 10.1093/brain/ awr093 PMID: 21586596 46. Freund P, Wheeler-Kingshott CA, Nagy Z, Gorgoraptis N, Weiskopf N, Friston K, et al. (2012) Axonal integrity predicts cortical reorganisation following cervical injury. J Neurol Neurosurg Psychiatry 83(6): 629–637. doi: 10.1136/jnnp-2011-301875 PMID: 22492214 47. Dice LR (1945) Measures of the amount of ecologic association between species. Ecology 26: 297–302. 48. Ko HY, Park JH, Shin YB, Baek SY (2004) Gross quantitative measurements of spinal cord segments in human. Spinal Cord 42(1): 35–40. PMID: 14713942 49. Kameyama T, Hashizume Y, Sobue G (1996) Morphologic features of the normal human cadaveric spinal cord. Spine (Phila Pa 1976) 21(11): 1285–1290. PMID: 8725917 50. Fradet L, Arnoux PJ, Ranjeva JP, Petit Y, Callot V (2014) Morphometrics of the Entire Human Spinal Cord and Spinal Canal Measured From In Vivo High-Resolution Anatomical Magnetic Resonance Im- aging. Spine 39(4): E262–269. doi: 10.1097/BRS.0000000000000125 PMID: 24253776 51. Kearney H, Yiannakas MC, Abdel-Aziz K, Wheeler-Kingshott CA, Altmann DR, Ciccarelli O, et al. (2013) Improved MRI quantification of spinal cord atrophy in multiple sclerosis. J Magn Reson Imaging 39(3): 617–623. doi: 10.1002/jmri.24194 PMID: 23633384

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 20 / 21 Spinal Cord Segmentation Method

52. Cadotte D, Fehlings M, Mikulis D, Nugaeva N, Wilson J (2012) Creating an MRI anatomic atlas of the human cervical spine: Implication for spinal fMRI. In: Proceeding of the 18th annual meeting of The Or- ganization of the Human Brain Mapping (OHBM), Beijing, China, Abstract 794. 53. Boykov Y and Funka-Lea G (2006) Graph cuts and efficient ND image segmentation. International journal of computer vision 70(2), 109–131. 54. Collins DL, Holmes CJ, Peters TM, Evans AC (1995) Automatic 3-D model-based neuroanatomical segmentation. Human brain mapping 3(3), 190–208. 55. Collins DL, Neelin P, Peters TM, Evans AC (1994) Automatic 3D intersubject registration of MR volumetric data in standardized Talairach space. J Comput Assist Tomogr 18(2):192–205. PMID: 8126267 56. Fonov VS, Evans AC, Botteron K, Almli CR, McKinstry RC, Collins DL et al. (2011) Unbiased average age-appropriate atlases for pediatric studies. Neuroimage 54(1):313–27. doi: 10.1016/j.neuroimage. 2010.07.033 PMID: 20656036 57. Taso M, Le Troter A, Sdika M, Ranjeva JP, Guye M, Bernard M, et al. (2014) Construction of an in vivo human spinal cord atlas based on high-resolution MR images at cervical and thoracic levels: preliminary results. MAGMA 27(3): 257–267. doi: 10.1007/s10334-013-0403-6 PMID: 24052240 58. Cadotte DW, Cadotte A, Cohen-Adad J, Fleet D, Livne M, Wilson JR et al. (2014) Characterizing the Lo- cation of Spinal and Vertebral Levels in the Human Cervical Spinal Cord. AJNR Am J Neuroradiol “In press”. 59. Morra JH, Tu Z, Apostolova LG, Green AE, Green AE, Avedissian C, Madsen SK, et al. (2009) Auto- mated mapping of hippocampal atrophy in 1-year repeat MRI data from 490 subjects with Alzheimer's disease, mild cognitive impairment, and elderly controls. Neuroimage 45(1 Suppl): S3–S15. doi: 10. 1016/j.neuroimage.2008.10.043 PMID: 19041724 60. Wang Y, Song Y, Rajagopalan P, An T, Liu K, Chou YY, et al. (2011) Surface-based TBM boosts power to detect disease effects on the brain: an N = 804 ADNI study. Neuroimage 56(4): 1993–2010. doi: 10. 1016/j.neuroimage.2011.03.040 PMID: 21440071 61. Wang Y, Zhang J, Gutman B, Chan TF, Becker JT, Aizenstein HJ, et al. (2010) Multivariate tensor- based morphometry on surfaces: application to mapping ventricular abnormalities in HIV/AIDS. Neuro- image 49(3): 2141–2157. doi: 10.1016/j.neuroimage.2009.10.086 PMID: 19900560 62. Bede P., Elamin M., Byrne S., McLaughlin RL, Kenna K, Vajda A, et al. (2013) Basal ganglia involvement in amyotrophic lateral sclerosis. Neurology 81(24): 2107–2115. doi: 10.1212/01.wnl. 0000437313.80913.2c PMID: 24212388 63. El Mendili MM, Tiret B, Chen R, M Pélégrini-Issac M, S Lehéricy S, Pradat P-F, et al. (2014) Spinal cord atrophy quantification: comparison of segmentation methods for 3T MRI T2-weighted images. In: Pro- ceeding of the 20th annual meeting of The Organization of the Human Brain Mapping (OHBM), Ger- many, Hamburg, Abstract 2944. 64. Heckemann RA, Hajnal JV, Aljabar P, Rueckert D, Hammers A (2006) Automatic anatomical brain MRI segmentation combining label propagation and decision fusion. Neuroimage 33(1):115–26. PMID: 16860573 65. Warfield SK, Zou KH, Wells WM (2004) Simultaneous truth and performance level estimation (STA- PLE): an algorithm for the validation of image segmentation. IEEE Trans Med Imaging 23(7):903–21. PMID: 15250643 66. Kwon JW, Yoon YC, Choi SH (2012) Three-dimensional isotropic T2-weighted cervical MRI at 3T: comparison with two-dimensional T2-weighted sequences. Clin Radiol 67(2): 106–113. doi: 10.1016/j.crad. 2011.06.011 PMID: 22142499 67. Korzan JR, Gorassini M, Emery D, Taher ZA, Beaulieu C (2002) In vivo magnetic resonance imaging of the human cervical spinal cord at 3 Tesla. J Magn Reson Imaging 16(1): 21–27. PMID: 12112499 68. Asman AJ, Bryan FW, Smith SA, Reich DS, Landman BA (2014) Groupwise multi-atlas segmentation of the spinal cord's internal structure. Med Image Anal 18(3): 460–471. doi: 10.1016/j.media.2014.01. 003 PMID: 24556080 69. Bellec P, Lavoie-Courchesne S, Dickinson P, Lerch JP, Zijdenbos AP, Evans AC (2012) The pipeline system for Octave and Matlab software (PSOM): a lightweight scripting framework and execution en- gine for scientific workflows. Front Neuroinform 6: 7. doi: 10.3389/fninf.2012.00007 PMID: 22493575

PLOS ONE | DOI:10.1371/journal.pone.0122224 March 27, 2015 21 / 21

Multi-parametric MRI studies in MND

Study 3 - Electrophysiological and spinal imaging evidence for sensory dysfunction in amyotrophic lateral sclerosis

Background and objectives

Animal and human studies in ALS showed evidence of involvement of spinal and cortical interneurones, and structures such as corpus callosum, basal ganglia and the cerebellum (Prell and Grosskreutz, 2013; Thivart et al. 2010; Pradat et al., 2009; van der Graaff et al. 2011; Carr et al. 2010; Özdinler et al., 2011; Shefner and Logigian, 1998). In addition, post-mortem examinations of spinal cords in ALS patients have shown axonal loss within the dorsal roots suggesting an alteration of the sensory afferent pathways at late stages of the disease (Kawamura et al., 1981; Averback and Crocker, 1983; Bradley et al., 1983). Peripheral and central degeneration of sensory pathways have also been reported in presymptomatic ALS mice, suggesting that the sensory defect occurs at a very early stage of the degenerative process (Sábado et al., 2014; King et al., 2012; Guo et al., 2009). Electrophysiological studies showed evidence of sensory abnormalities that increase with disease severity in ALS patients (Constantinovici, 1993; Gregory et al., 1993; Cosi et al., 1984). Therefore, the question of whether these alterations are consecutive to motoneurons loss, or could act as a primary debilitating mechanism of the disease remains an open question.

In a previous study of ALS patients (Cohen-Adad et al., 2013), by using DTI and MT imaging, we showed that the dorsal column, i.e. pathways conveying peripheral sensory information to supraspinal structures, were altered. Diffusion tensor imaging also revealed dorsal column abnormalities in a subset of 12 patients with early onset (median/maximum disease duration 9/12.6 months). This latter, and unexpected, result suggests that the sensory afferent pathways to spinal and cortical structures might already be affected at an early stage in ALS, and thus that the sensory alteration may indeed participate in ALS pathogenesis. However, MRI does not allow a functional exploration of the pathways, spinal and/or cortical, fed by sensory afferents. Therefore, it is now necessary to verify if the anatomical changes observed with MRI in ALS patients are correlated to a dysfunction of sensory feedback integration.

The objective of the present study was to investigate the anatomo-functional properties of sensory pathways in ALS patients with distal motor dysfunctions in upper limbs and no sensory symptoms, combining spinal DTI and SEP.

- 71 -

Material and Methods

Twenty-one ALS patients with no clinical sensory signs and 21 age-matched controls with no history of neurological disease were recruited. The disease duration from the onset of weakness in ALS patients was 26.6±3.6 months. The spinal cord was imaged using 3T cardiac-gated high-resolution diffusion tensor imaging (DTI) (C5-T1 vertebral level, voxel size = 1×1×5 mm3, acquisition time ~10 min). Manual ROIs were drawn in the spinal cord to isolate the dorsal column, containing sensory afferents. Sensory evoked potentials were recorded after median and ulnar nerve stimulations.

Results synthesis

We coupled spinal DTI and electrophysiological recordings for the first time to evaluate the anatomo-functional properties of sensory pathways in humans. This combined approach revealed subclinical sensory defects in 85% of ALS patients, while separated methods revealed abnormal values in 60% (Figure 5). The anatomical damage of dorsal column revealed by spinal DTI was correlated to the depression of the peripheral afferent volley revealed by the lower amplitude of cervicobrachial SEP. All together these results strongly suggest that sensory defects have been underestimated in previous studies. Therefore, subclinical sensory impairments can manifest in ALS and should no longer cast doubt on the diagnosis. The defect was correlated to the disease duration but a follow-up study would confirm the progressive sensory affection. Together with our previous MRI study (Cohen-Adad et al., 2013), these results demonstrated for the first time that advanced spinal cord MRI techniques are powerful tools to investigate extra-motor system involvement in MND.

- 72 - Downloaded from http://bmjopen.bmj.com/ on February 26, 2015 - Published by group.bmj.com

Open Access Research Electrophysiological and spinal imaging evidences for sensory dysfunction in amyotrophic lateral sclerosis

Caroline Iglesias,1 Sina Sangari,1 Mohamed-Mounir El Mendili,1 Habib Benali,1 Véronique Marchand-Pauvert,1 Pierre-François Pradat1,2

To cite: Iglesias C, Sangari S, ABSTRACT Strengths and limitations of this study El Mendili M-M, et al. Objectives: The prevalence of sensory impairment at Electrophysiological and an early stage of amyotrophic lateral sclerosis (ALS) is spinal imaging evidences for ▪ For the first time, spinal diffusion tensor imaging still debated. The study aim was to investigate the sensory dysfunction (DTI) and electrophysiological recordings were in amyotrophic lateral anatomofunctional properties of sensory pathways in coupled to evaluate the anatomofunctional prop- sclerosis. BMJ Open 2015;5: patients with ALS, combining spinal diffusion tensor erties of sensory pathways in humans. This com- e007659. doi:10.1136/ imaging (DTI) and somatosensory evoked potentials bined approach and the normalisation of bmjopen-2015-007659 (SEPs). somatosensory evoked potentials revealed sub- Design: Case–control study. clinical sensory defects in 85% patients with ▸ Prepublication history for Settings: ALS referral centre and laboratory of amyotrophic lateral sclerosis (ALS). this paper is available online. biomedical imaging (Paris, France). ▪ The anatomical damages of dorsal columns To view these files please Participants: Well-characterised group of 21 patients revealed by spinal DTI were correlated to the visit the journal online with ALS with moderate disability (mean amyotrophic depression of the peripheral afferent volley (http://dx.doi.org/10.1136/ lateral sclerosis Functional Rating Scale (ALSFRS) score revealed by the lower amplitude of N9 which bmjopen-2015-007659). 39.3±1.0) and no clinical sensory signs and control further supports the fact that sensory defects group of 21 gender and age-matched healthy subjects. can occur at an early stage of ALS. Outcome measures: Fractional anisotropy and ▪ Subclinical sensory impairments can manifest in CI and SS are co-first diffusivity of the dorsal columns at C5-T1 levels (DTI ALS and should no longer cause doubt on the authors; VM-P and P-FP are metrics) and SEPs after median and ulnar nerve diagnosis. The defect was correlated to the co-last authors. stimulations (latency and amplitude of N9 and N20 disease duration but a follow-up study would components). confirm the progressive sensory affection. Received 13 January 2015 Results: Abnormal DTI metrics indicated anatomical Revised 29 January 2015 damages of ascending sensory fibres in 60% of Accepted 2 February 2015 ∼ patients (p<0.05). Raw SEPs (μV) were smaller in INTRODUCTION ∼40% of patients but the difference with healthy Amyotrophic lateral sclerosis (ALS) is the subjects was not significant (p>0.16). Their most common motor neuron disease charac- normalisation to prestimulus activity strengthened the terised by degeneration of upper as well as difference between groups (p<0.05) and allowed lower motor neurons. Postmortem anatomo- identification of ∼60% of patients with abnormal values. pathology has also revealed degeneration of According to N9 latency, the peripheral conduction time sensory ﬁbres but to a lesser extent than was normal in patients (p>0.32) but based on N20 motor axons, and it has been suggested that latency, the central conduction time (between spinal cord and parietal cortex) was found to be slower the sensory system might be similarly affected (p<0.05). Significant correlation was found between DTI as the motor system, but the degeneration is 1–3 metrics and N9 amplitude (p<0.05). Altered SEPs were less advanced at the time of death. also correlated with the disease duration (p<0.05). Peripheral and central degeneration of Taken together, spinal imaging and electrophysiology sensory pathways have also been reported in helped to identify ∼85% of patients with subclinical presymptomatic ALS mice, suggesting that sensory defect while separated methods revealed the sensory defect occurs at a very early stage For numbered affiliations see abnormal values in ∼60%. of the degenerative process.4–6 Accordingly, end of article. Conclusions: Sensory impairments have been sensory nerve biopsy7 and multiparametric underestimated at early stages of ALS. These results spinal imaging of the dorsal column8 have Correspondence to show for the first time the interest to combine revealed subclinical anatomical damage in electrophysiology and imaging to assess non-motor Dr Véronique Marchand- patients, including those at an early disease Pauvert; system involvement in ALS. stage after diagnosis. From a pathophysio- veronique.marchand@upmc. Trial registration number: IDRCB2012-A00016-37. fr logical perspective, the role of sensory

Iglesias C, et al. BMJ Open 2015;5:e007659. doi:10.1136/bmjopen-2015-007659 1 Downloaded from http://bmjopen.bmj.com/ on February 26, 2015 - Published by group.bmj.com

Open Access involvement in the degenerative process in ALS is still proprioception and nociception), (3) motor weakness in unknown. From a clinical perspective, while discreet hand muscles (median and ulnar nerve territories) on sensory clinical impairment does not exclude ALS diag- the electrophysiologically explored side and (4) absence nosis, it is declared atypical or even excluded when of medical conditions associated with peripheral neur- sensory abnormalities are major.9 10 Because of these opathy (eg, diabetes, alcoholism, neurotoxic drugs, uncertainties, better characterisation of the sensory nerve entrapment, neuropathy). Twenty one patients involvement in well-characterised populations of ALS is with ALS were enrolled in the study (all but one spor- needed. adic; 5 females, 56.3±2 years old, range 37–76) and Sensory nerve action potentials (SNAPs) and somato- 21 aged- and gender-matched neurologically intact par- sensory evoked potentials (SEPs), routinely used to ticipants (controls; 5 females, 56.6±2.1 years old, range assess sensory pathways, have been investigated exten- 33–73). The disease duration from the onset of weak- sively in patients with ALS. Except in three studies,11–13 ness was 26.6±3.6 months. All patients but one were treated abnormal SEPs were reported, on average, in about 1/3 with riluzole (100 mg/day) and α-tocopherol (1000 mg/ patients without clinical sensory deficits.7 14–22 The day). Clinical measures included manual muscle testing, abnormality increases with disease severity,20 23–28 and is total Amyotrophic Lateral Sclerosis Functional Rating Scale more common when stimulating lower limb peripheral (ALSFRS)-R and assessment of hand motor performance nerves than those of upper limbs.16 17 25 29 However, the by the ALSFRS-R subscore for handwriting, hand muscles anatomical substrate of altered SEPs demonstrated in testing and the nine hole peg test (table 1). previous studies remains disputed. Indeed, altered cortical SEPs were not always associated with abnormal Spinal diffusion tensor imaging SNAPs and/or cervicobrachial SEPs (N9). Moreover, Image acquisitions were performed using a 3 T MRI altered SNAPs were most often observed in patients system (TIM Trio, Siemens Healthcare, Erlangen, exhibiting sensory signs or symptoms, or were some- Germany) and neck/spine coil. Cardiac-gated diffusion times related to associated medical conditions (poly- tensor imaging (DTI) was performed using a single shot neuropathy, median or ulnar nerve entrapment echo-planar imaging (EPI) sequence with monopolar neuropathies).12 20 25 27–30 In line with this, the various scheme. The axial slice was positioned perpendicular to components of cortical SEPs are not altered unequally, the spinal cord, in the middle of each vertebral level whether in amplitude and/or latency, and given the between C2-T2 vertebral levels (8 slices) but the acquisi- multiple origins of the SEP components,31 some tion was optimised for C5-T1 (figure 1A). Imaging para- authors have suggested that abnormal SEPs in ALS are meters were: field of view =128 mm, TR/TE=700/96 ms, not due to sensory defects but to abnormal cortico- voxel size =1×1×5 mm3, parallel acquisition: R=2, b-value cortical interactions.24 32 1000 s/mm2, 64 directions, 2 repetitions, acquisition Until now, it has therefore been difficult to establish a time 10 min. For a comprehensive description of the link between anatomical and functional alterations of MRI acquisition parameters, the readers are invited to sensory pathways in ALS, given that the latter are based refer to a previous paper.34 on SEPs, that depends on peripheral afferent inputs and cortico-subcortical excitability.31 Moreover, the origin of Electrophysiology the neural sources, discrimination of which is based on Electrophysiological investigations were performed the the latency of SEP components, becomes uncertain when same day as the MRI in each participant. Median and SEPs are delayed. Thus, the aim of the present work was ulnar nerves were stimulated (1 ms rectangular electrical to test, in typical patients with ALS with distal motor dys- stimulation; DS7A, Digitimer Ltd, Hertfordshire, UK) at functions in upper limbs and no sensory symptoms, the wrist level using bipolar surface electrodes (0.5 cm2, whether abnormal metrics of dorsal column imaging are 1 cm apart). All controls were stimulated on the domin- associated with altered SEPs, with a focus on N9 and N20 ant side, according to The Edinburgh handedness inven- components, which depend mainly on the peripheral tory;35 right side in 20 participants and left side in afferent volley.31 In addition, to control for experiment- 1. Patients with ALS were stimulated on the most affected dependent variability of electrophysiological measure- side (distal hand weakness) or on the dominant side ments, a limitation for the interpretation of previous when both sides were equally affected: 17 patients were studies, we propose to normalise SEP amplitude to stimulated on the dominant right side, 1 patient on the improve evaluation of interindividual differences, as per- dominant left side and 3 patients on the left non- formed when studying muscular recordings.33 dominant but more affected side. Perceptual and motor thresholds (PT and MT, respectively; figure 2) were estimated in both nerves. Stimulus intensity was then METHODS adjusted at 9×PT in each subject. A recording surface Participants electrode was stuck in the supraclavicular fossa (Erb’s Inclusion criteria were: (1) probable or definite ALS point) ipsilateral to the stimulations with a reference elec- according to the El Escorial criteria, (2) absence of trode on the contralateral supraclavicular fossa. A needle sensory symptoms or signs (normal touch, electrode was inserted in the scalp, 4 cm lateral and 2 cm

2 Iglesias C, et al. BMJ Open 2015;5:e007659. doi:10.1136/bmjopen-2015-007659 Downloaded from http://bmjopen.bmj.com/ on February 26, 2015 - Published by group.bmj.com

Open Access

Table 1 Clinical data ALSFR-S Muscular testing Duration El Escorial Total Hand Total UL Hand Patient Onset (months) criteria Atrophy (/48) (/4) (/150) (/45) (/10) 1.M.68 UL 74 Probable + 42 1 148 43 8 2.M.50 UL 28 Probable + 42 1 145 40 5 3.M.54 LL 28 Probable + 41 4 120 41 8 4.M.59 LL 24 Probable + 34 3 102 39 7 5.M.62 B 12 Probable + 43 3 147 44 9 6.M.48 UL 23 Probable + 39 2 145 42 8 7.M.61 LL 33 Probable + 39 4 105 43 8 8.M.44 LL 21 Probable − 40 4 131 44 9 9.F.76 LL 26 Lab-supported − 34 4 105 41 8 10.M.66 UL 34 Probable + 37 2 144 41 6 11.M.45 B 15 Definite + 31 1 137 41 8 12.M.55 LL 58 Probable + 30 2 79 38 6 13.F.37 UL 39 Probable + 41 3 140 39 6 14.M.51 UL 10 Probable + 45 4 134 43 8 15.M.52 UL 15 Lab-supported + 46 4 145 40 8 16.F.50 UL 14 Probable + 44 4 145 44 9 17.F.56 UL 15 Probable + 40 3 144 42 7 18.M.58 LL 43 Probable + 33 3 114 42 8 19.F.62 LL 8 Probable − 40 3 145 42 8 20.M.65 UL 17 Probable + 45 3 145 40 8 21.M.63 LL 19 Lab-supported + 40 3 129 43 8 Patient: rank.gender (F, female; M, male).age (years old); Onset: first clinical signs manifested in upper limb (UL) or lower limb (LL) or bulbar (B); Duration of the disease in months; El Escorial criteria for amyotrophic lateral sclerosis (ALS): definite ALS (Definite), clinically probable ALS (Probable), Clinically probable—laboratory supported ALS (Lab-supported); Atrophy observed (+) or not (−) on the side explored during the electrophysiological investigations; ALSFR-S: total score of the revised ALS functional resting scale (Total/maximal score 48) and for handwriting (Hand/4); Muscular testing: in upper and lower limbs and in the neck (Total/150), on the side explored during the electrophysiological investigations (UL/45) and in intrinsic hand muscles (Hand/10). posterior from Cz, in front of the primary somatosensory classification, the operator opted for a conservative cortex (S1) contralateral to the stimulations with a approach by excluding pixels that were at the cord/corti- surface electrode for a reference on the ipsilateral ear cospinal fluid or white matter/grey matter interfaces. lobe. Electrophysiological signals were filtered (band- width 30–3000 Hz) and amplified (×10 000; D360 fi 8-Channel Patient Ampli er, Digitimer Ltd, Electrophysiology Hertfordshire, UK) before being digitally stored (2 kHz Peak latencies of N9 (Erb’s point) and N20 (parietal) fl sampling rate) on a personal computer for of ine ana- were evaluated in each subject. The peripheral conduc- lysis (Power 1401 and Signal Software, CED, Cambridge, tion velocity (CVp) was calculated according to the UK). Each recording session consisted of 200 ulnar nerve distance between stimulating electrodes and C7 vertebra and 200 median nerve stimuli alternated randomly and N9 latency. The central conduction velocity (2 Hz). Three recording sessions were performed for (CVc) was calculated according to the distance between each participant. C7 and parietal cortex and the difference between N9 and N20 latencies. Peak-to-peak amplitude of N9 Data analysis (figure 3A, B) and amplitude of the first positivity of DTI N20 (figure 3C, D) were measured and normalised to Motion correction was applied slice-by-slice using FSL the prestimulus activity (calculated over a 100 ms period, FLIRT with three degrees of freedom (Tx, Ty, Rz). excluding the stimulus artefact). Given the variability of Diffusion tensor and its related metrics, fractional anisot- prestimulus activity, the SD of prestimulus activity was fi ropy (FA), axial (λ⁄⁄) and radial (λ⊥) and mean diffusivity calculated and multiplied by 2 (2×SD) to de ne the (MD), were estimated on a voxel-wise basis. The region of positive and negative limits of background activity interest (ROI), limited to the hemidorsal region ipsilat- (dotted lines in figure 3A, D). This procedure is com- eral to the electrical stimulations, was manually defined monly used to determine when the activity significantly on each slice using geometry-based information. To avoid changes from background activity after stimuli. Because any user bias, the ROI was defined on the mean diffusion these limits give a good estimation of prestimulus activity weighted images by an experienced segmentation oper- amplitude, we used the procedure to normalise the ator (MMEM; figure 1A).8 34 Furthermore, during ROIs amplitude of SEPs.

Iglesias C, et al. BMJ Open 2015;5:e007659. doi:10.1136/bmjopen-2015-007659 3 Downloaded from http://bmjopen.bmj.com/ on February 26, 2015 - Published by group.bmj.com

Open Access

Figure 1 Spinal diffusion tensor imaging metrics. (A) T2-weighted MRI sagittal slice (rostro (R)—caudal (C) axis) and mean diffusion-weighted images (C6 vertebra level) in one patient with amyotrophic lateral sclerosis (ALS). For each level between C5 and T1 (dashed white lines), the posterior (P, in blue) and left (L) and right (R, in red) region of interests (ROIs) were drawn manually on the mean diffusion-weighted images to include most of the posterior column-medial lemniscus pathway (sensory pathways) and lateral corticospinal tracts (motor pathways), respectively. The posterior ROI was split in two parts for the right and left side of the spinal cord, according to the antero (A)—posterior axis and the slice centre of mass. Box plot charts showing distribution of data for the radial (B) and mean diffusivity (C) in controls (black rectangles) and in patients with ALS (red rectangles): the top and bottom line of the box correspond to the 75th centile (top quartile) and 25th centile (bottom quartile), respectively, and the open circle in the box, to the 50th centile (median). The two bars extend from the 10th centile (bottom centile) to the 90th centile (top centile). The bar caps delimit outliers. The top and bottom filled circles indicate the maximum and minimum values, respectively. The cross within the box indicates the arithmetical mean. *p<0.05. Statistics medial lemniscus pathway ipsilateral to electrical stimuli First, each DTI and electrophysiological metric was inde- delivered when investigating SEPs (see below), in 20 pendently compared between groups using unpaired controls and 19 patients (due to artefacts, 1 control and t tests. Then, the metrics found to be significantly modi- 2 patients were excluded). On average, both metrics fied in patients were retained for further analyses to test were significantly higher in patients than in controls the relationships between these metrics and clinical fea- (unpaired t test, p<0.05). Abnormal values of λ⊥ and tures for the group of participants. Thus, multiple regres- MD were observed in 11/19 patients (57.9%). The dif- sion analyses were performed to test the link between λ⊥, ferences in λ⊥ and MD were not accompanied by a sig- MD, N9, N20 and CVc according to the subject groups. nificant change in FA or λ⁄⁄ (p=0.29 and 0.06, In addition, Pearson correlation analyses were performed respectively). to investigate the relationships between clinical features Similar analyses were also performed on the lateral and λ⊥, MD, N9 and N20. The 95% CI of control data region corresponding to the lateral corticospinal tract. was calculated for DTI and electrophysiological metrics. On average, λ⊥ and MD were also higher in patients These intervals were used to fix the limits of normal than in controls (p<0.01 and 0.05, respectively) and FA values in order to identify the patients exhibiting abnor- was smaller (p<0.05); λ⁄⁄ was not significantly different mal data with a risk of error less than 5%. Statistical ana- between groups (p=0.42). lysis was conducted using StatEL software (http://www. adscience.eu) and the significance level was set at p value Electrophysiological measurements <0.05. Arithmetical means are indicated ±1SEM. Figure 2 shows the mean PT and MT for median and ulnar nerves in controls and patients. While PT for both RESULTS nerves was similar between controls and patients Spinal DTI metrics (p=0.93 and 0.73 for median and ulnar nerves, respect- The box plot charts in figure 1B, C show the distribution ively; figure 2A), MT was higher in patients, reaching the fi of λ⊥ and MD, respectively, in the hemidorsal region of statistically signi cant level for the ulnar nerve (p<0.05 the spinal cord, corresponding to the posterior column- and 0.12 for the median nerve; figure 2B). The stimulus

4 Iglesias C, et al. BMJ Open 2015;5:e007659. doi:10.1136/bmjopen-2015-007659 Downloaded from http://bmjopen.bmj.com/ on February 26, 2015 - Published by group.bmj.com

Open Access

median and ulnar nerve, respectively, and p<0.05 for N20; figure 3E, F), which confirmed that the peripheral volley (N9) and the first component of early cortical SEP (N20) were smaller in the patient group. No significant correlation was observed between the amplitudes of N9 and N20 (Pearson, p=0.67). Abnormal N9 was observed in 17/20 patients (85%) for the ulnar nerve and in 9/20 patients (45%) for the median nerve. Abnormal N20 was observed in 11/21 patients (52.4%) for the ulnar nerve and in 10/21 patients (47.6%) for the median nerve. On average, the SEP normalisation helped to identify 57.5% of patients with abnormal values while 41.3% exhibited abnormal raw data (in μV).

Relationships between DTI and electrophysiological metrics The scatter plots in figure 4 show the dispersion of N9 and N20 amplitudes according to MD (dorsal column ipsilateral to stimulation site) for controls and patients, for median as well as ulnar nerves (Dixon test did not detect any aberrant values). Multiple regression analyses further confirmed the significant difference between groups, in N9 and N20 amplitudes. In addition, significant relationships were found between DTI metrics and N9 (except for λ⊥ and N9 from ulnar) but not with N20 (see statistics in table 3). Similarly, significant relation- Figure 2 Perceptual and motor threshold for median and ships were observed between the CVc and λ⊥ (p<0.05) ulnar nerve stimuli. Mean perceptual (A) and motor (B) and MD (p<0.01) for the median nerve. thresholds for median nerve (left columns) and for ulnar nerve Figure 5 shows that a larger proportion of patients (right columns) estimated in the group of controls (open exhibited abnormal DTI metrics and/or evoked poten- columns) and of patients with amyotrophic lateral sclerosis tials (84.7±3.8% on average; red portion) compared to (red columns). Vertical bars are ±1 SEM. *p<0.05. patients having normal metrics (white portion). Among the patients exhibiting at least one abnormal metric, intensity used for evoking SEPs (9×PT) was above MT 44.4±7.1% have both abnormal DTI and evoked poten- for both nerves in all controls and in 16/21 patients for tials (purple subportions). Similar proportions were the median nerve and in 18/21 patients for the ulnar found when comparing the DTI metrics and the CVc. nerve. fi Signi cant difference between groups was not Correlation with clinical features observed in peak latency for N9 (for which one patient Multiple regression analyses were performed to investi- was excluded because of artefacts) nor for N20 (table 2; gate the relationships between DTI and electrophysio- 0.250.3). were smaller in the patient compared to the control. On average, the raw amplitude (in μV) of both N9 and N20 was smaller in patients than in controls, but the differ- DISCUSSION ence did not reach statistical significance (0.16

Iglesias C, et al. BMJ Open 2015;5:e007659. doi:10.1136/bmjopen-2015-007659 5 Downloaded from http://bmjopen.bmj.com/ on February 26, 2015 - Published by group.bmj.com

Open Access

Figure 3 N9 and N20 amplitude. Raw signal (in μV) from recordings at the Erb point level (A and B) and at cortical level (in front of S1, C and D) in one healthy subject (control, A and C; black lines) and in one patient with amyotrophic lateral sclerosis (ALS, B and D; red lines). Dashed lines indicate the 0 level. Upper and lower dotted lines indicate +2×SD and −2×SD levels, respectively. Group data showing the mean amplitude of N9 (left) and of N20 (right) normalised to prestimulus activity in controls (open columns) and in patients with ALS (red columns) after median (E) and ulnar nerve stimulation (F). Vertical bars are±1 SEM. *p<0.05, **p<0.01 and ***p<0.001. limb could be identified. Considered separately, each exhibited less motor weakness especially in proximal technique revealed abnormal values in less than 60% upper limb muscles. Accordingly, λ⊥ for the lateral corti- patients. The second novelty of this study is that cospinal tracts was lower in the present study compared normalisation helped to identify more patients with to the previous one, which supports a less severe impair- abnormal SEP from a 40% detection rate when ment of the motor pathways in the patients investigated expressed in μV, to 60% when taking into account the here. Moreover, while the relationship between motor level of prestimulus activity. This study has also shown a and sensory deficits has not been characterised so far, significant relationship between DTI metrics and N9, some studies have reported increased SEP alteration and a significant depression of evoked potentials with with the disease severity.20 23–28 Given that the degener- disease duration. ation is likely to be slower in sensory pathways than in 2 motor pathways, one would expect a smaller λ⊥ for the Reliability of spinal DTI dorsal columns in the present study, compared to the Spinal DTI metrics were found to be significantly altered previous one,8 which was the case. Thus, the comparison in patients, which gives further support for early anatom- between the two studies indicates that spinal DTI is a ical damage of the dorsal columns in ALS. Compared to reliable and very sensitive method for detecting anatom- our previous study,8 the patients in the present study ical tract defects at spinal level.

6 Iglesias C, et al. BMJ Open 2015;5:e007659. doi:10.1136/bmjopen-2015-007659 Downloaded from http://bmjopen.bmj.com/ on February 26, 2015 - Published by group.bmj.com

Open Access

signals, which accounts for the huge interindividual and Table 2 N9 and N20 characteristics intraindividual variability of EEG recordings.36 Some Median Ulnar studies on visual evoked potentials in humans have Controls ALS Controls ALS revealed the influence of the level of prestimulus EEG N9 activity on the size of the evoked potentials and reported Latency 10.7±0.2 10.8±0.2 11.4±0.2 11.3±0.2 better reliability with normalised data.37 38 Here, SEPs CVp 65.9±0.9 66.6±0.7 62.1±0.9 63.5±1.1 were normalised to 2×SD of prestimulus activity, which is Amplitude 6.1±0.5 5.1±0.5 5.3±0.8 4.2±0.4 commonly used to evaluate the level of activity before N20 conditioning, and to determine when the poststimulus Latency 19.0±0.4 19.5±0.3 19.2±0.3 19.6±0.3 signal deviates significantly from background activity. CVc 31.5±1.2 28.3±1.0* 33.4±1.4 30.3±1.2 Normalisation of evoked potentials to this value allowed Amplitude 7.4±0.2 7.2±0.2 7.4±0.2 7.0±0.2 the basal activity to be taken into account and reduced Latency: mean peak latency (ms±1 SEM) of N9 recorded at the Erb point, and of N20 recorded at the level of primary the variability due to noise. Comparison of normalised somatosensory cortex (S1) for median and ulnar nerves, in data strengthened the differences between groups, controls and patients with amyotrophic lateral sclerosis. CVp: helped to identify more patients with abnormal values peripheral conduction velocity (m/s±1 SEM) according to N9 peak latency and the distance between stimulation site and C7 vertebra. and indicated a greater mean suppression from 20% (in CVc: central conduction velocity (m/s±1 SEM) according to the raw data) to 50% (when normalised) for N9 and <5% to difference between N20 and N9 peak latencies and the distance 20% for N20 (according to the mean values in controls). between C7 vertebra and recording site, in front of S1. Amplitude: peak-to-peak amplitude of N9 and peak amplitude of N20 (μV). The comparison of raw and normalised data suggests *p<0.05. that abnormal SNAPs and SEPs in ALS might have been previously underestimated. Normalisation of electrophysiological measurements The amplitude of SNAPs and SEPs produced by mixed Alteration of peripheral volleys nerve stimulation are usually compared using the raw Histological evidence for sensory nerve degeneration signals expressed in μV. However, it is well known in elec- has only been found in lower limbs with sural nerve trophysiology that intrinsic (background activity, individ- biopsies.3 In upper limbs, abnormal SNAPs were found ual bioelectrical characteristics) and extrinsic factors mainly in patients with clinical sensory (single-use electrodes, whose properties and position signs.7 12 20 25 27 29 30 To some extent, we found analo- can slightly differ across experiments) can influence the gous results because the characteristics of N9 (Erb’s

Figure 4 Relationships between diffusion tensor imaging and electrophysiological metrics. N9 (A and C) and N20 (B and D) amplitude (% 2×SD of the prestimulus activity) plotted against the mean diffusivity (×10−3 mm2/s) for median nerve (A and B) and for ulnar nerve (C and D). Each open circle represents one control and each red circle, one patient with amyotrophic lateral sclerosis. Iglesias C, et al. BMJ Open 2015;5:e007659. doi:10.1136/bmjopen-2015-007659 7 Downloaded from http://bmjopen.bmj.com/ on February 26, 2015 - Published by group.bmj.com

Open Access

Table 3 Results of multiple regression analyses

Regression Group λ⊥ MD Residuals Median N9 0.27 212.3 2.04×10+6 −1.96×10+6 673.4 0.013* 0.025* 0.05* 0.024* 0.14 N20 0.13 290.8 1.87×10+6 −1.21×10+6 491.4 <0.05* <0.014* 0.14 0.24 0.38 Ulnar N9 0.39 136.4 7.51×10+5 −8.40×10+5 392.0 <0.001*** <0.01** 0.1 0.028* 0.06 N20 0.21 241.8 1.96×10+6 −1.34×10+6 629.9 <0.034* <0.014* 0.06 0.12 0.18 Regression:R2 and p value of multiple regression for N9 and N20 produced by median or ulnar nerve stimuli. Group: effect of the group of participants (amyotrophic lateral sclerosis vs controls), coefficient and p value. λ⊥: Effect of the radial diffusivity, coefficient and p value. MD: effect of the mean diffusivity, coefficient and p value. Residuals: coefficient and p value of the constant. *p<0.05, **p<0.01 and ***p<0.001. point) were as in controls: similar peak latency and hand motor weakness, we found a higher MT in patients similar CVp with no signiﬁcant change in the mean raw and, in some, the conditioning stimuli used to evoke amplitude (in μV). However, abnormal raw values were SEPs were below MT. However, these patients did not observed in 20–50% of patients and the normalisation exhibit the smallest N9 and there was no link between strengthened the difference between groups, which abnormal N9 values and MT. Lastly, the relationship further supports subclinical peripheral abnormalities.27 between spinal DTI and N9 further supports the fact Our group of patients was very homogenous in regard that the latter depends mainly on the peripheral sensory to hand muscle weakness in the median and ulnar nerve afferent volley31 and its alteration in ALS. territories, which might account for the higher proportion of patients exhibiting abnormal N9 compared to Origin of the impairments along the sensory pathways previous studies.21–23 Alternatively, a reduced antidromic The present changes in spinal DTI metrics give further volley in motor axons may account for N9 depression. evidence for anatomical impairments of sensory path- However, a study performed in patients with pro- ways in ALS. The alteration of N9 amplitude, without nounced or complete loss of motor axons in the median changes in peripheral conduction velocity, suggests loss nerve failed to reveal any change in N9.39 According to of peripheral sensory axons.3–6

Figure 5 Patient data distribution. Proportion of patients exhibiting normal (white portion) and abnormal diffusion tensor imaging (DTI) metrics and/or abnormal amplitude of N9 (A and B) or N20 (C and D; red portion) for the median (A and B) and ulnar nerve (C and D). Among the patients with abnormal metrics (red portion), the proportion of patients with abnormal DTI metrics only (and thus normal evoked potentials) is illustrated by the blue subportion, the proportion of patients with abnormal evoked potential only (and thus normal DTI) is illustrated by the green subportion, and the proportion of patients with abnormal DTI metrics and abnormal evoked potentials is illustrated by the purple subportion.

8 Iglesias C, et al. BMJ Open 2015;5:e007659. doi:10.1136/bmjopen-2015-007659 Downloaded from http://bmjopen.bmj.com/ on February 26, 2015 - Published by group.bmj.com

Open Access

The first cortical component N20 was delayed and 2Département des Maladies du Système Nerveux, AP-HP, Hôpital Pitié- depressed to a lesser extent than N9. The absence of cor- Salpêtrière, Paris, France relation between the alterations of N9 and N20 can be Acknowledgements The authors thank Dr N Leforestier, Dr T Lenglet, explained by the fact that at a stimulus intensity producing Dr F Salachas, Dr G Bruneteau and Dr V Meininger for their help in patient maximal N20, N9 reached half its maximal size, and this recruitment, and Dr R Morizot-Koutlidis for her help in the experimental intensity can be at or slightly above MT.40 Therefore, at design for SEP recordings. They also thank C Macia, K Nigaud, F Humbert the stimulus intensity we used (9×PT, which was above MT and R Valabrègue for their valuable technical assistance during MRI in most participants), N20 was likely to be maximal and, acquisitions, which were performed at CENIR-ICM Hôpital Pitié-Salpêtrière. Their gratitude also goes to S Blancho for her help to obtain the approval of because of saturation, the measurements might have been the local ethics committee and for monitoring the project. At last, they deeply less sensitive at a cortical level than at a peripheral level. thank Dr Anna Hudson (Neuroscience Research Australia) for scrutinising the However, N20 not only depends on the sensory afferent English. 40 inputs, but also to central processing. Accordingly, a Contributors VM-P and P-FP were involved in conception and design of recent cerebral MRI study has shown atrophy in the thal- research and P-FP in patient selection. CI, SS and VM-P were responsible for amus,41 suggesting that the thalamocortical relay for acquisition and analysis of electrophysiological data, CI and M-MEM for MRI sensory information may be altered in ALS. In line with acquisitions, and M-MEM for DTI analysis. CI, SS, M-MEM and VM-P was responsible for statistical analysis; HB for supervision of MRI and statistical this, we found a slower CVc in patients than in controls, analysis; CI, SS, M-MEM, VM-P and P-FP for interpretation of the data; and without any change in CVp. Therefore, abnormal N9 and CI, M-MEM and VM-P for drafting of the manuscript and figures. All authors N20, together with impaired spinal DTI reported here, have approved the final version of the manuscript. give strong evidence for subclinical sensory defects in ALS Funding This work was supported by IRME (ID RCB 2012-A00016-37) and at peripheral as well as central levels. ANR (ANR-12-JSV4-0007-01). CI was supported by a grant from FRM, Comparing median nerve and skin stimuli suggests that M-MElM by IHU-A-ICM Pitié-Salpêtrière (ANR-10-IAIHU-06), and SS by a N20 mainly depends on cutaneous inputs to the 3b grant from UPMC. 31 42–44 area. Thus, altered N20 from median nerve stimu- Competing interests None. fi lation could suggest impaired cutaneous bres. However, Patient consent Obtained. if low-threshold cutaneous afferents were affected, this fl Ethics approval The study conformed to the standards set by the latest would have in uenced PT, which was found unchanged. revision of the Declaration of Helsinki and has been approved by the ethics Alternatively, changes could be due to proprioceptive committee of Pitie-Salpetriere Hospital (CPP-Ile-de-France VI). afferents. Cutaneous fields innervated by median nerve Provenance and peer review Not commissioned; externally peer reviewed. are larger than those innervated by ulnar nerve but 75% intrinsic hand muscles are innervated by ulnar nerve. Data sharing statement No additional data are available. Additionally, SEPs from mixed nerves in lower limbs, Open Access This is an Open Access article distributed in accordance with probably more motor than the median nerve (such as the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, ulnar nerve), are mainly produced by proprioceptive which permits others to distribute, remix, adapt, build upon this work non- 45 commercially, and license their derivative works on different terms, provided afferents. Therefore, impaired proprioceptive inputs the original work is properly cited and the use is non-commercial. See: http:// would affect SEPs produced by ulnar nerve more than creativecommons.org/licenses/by-nc/4.0/ those produced by median nerve, which was the case. The absence of clinical sensory signs, of correlation between sensory defect and motor symptoms and the fact that DTI metrics for the lateral corticospinal tract REFERENCES 1. Kawamura Y, Dyck PJ, Shimono M, et al. Morphometric comparison were more affected than those for the dorsal columns of the vulnerability of peripheral motor and sensory neurons in suggest that the impairment of sensory pathways was less amyotrophic lateral sclerosis. J Neuropathol Exp Neurol than that of motor systems. However, the correlation 1981;40:667–75. 2. Averback P, Crocker P. Regular involvement of Clarke’s nucleus in with disease duration suggests that the degeneration of sporadic amyotrophic lateral sclerosis. Arch Neurol 1982;39:155–6. sensory pathways is progressive. 3. Bradley WG, Good P, Rasool CG, et al. Morphometric and biochemical studies of peripheral nerves in amyotrophic lateral sclerosis. Ann Neurol 1983;14:267–77. 4. Guo YS, Wu DX, Wu HR, et al. Sensory involvement in the CONCLUSIONS SOD1-G93A mouse model of amyotrophic lateral sclerosis. Exp Mol The combination of normalised electrophysiological Med 2009;41:140–50. measurements and spinal DTI metrics suggest that 5. King AE, Blizzard CA, Southam KA, et al. Degeneration of axons in spinal white matter in G93A mSOD1 mouse characterized by NFL sensory deficits in ALS have been previously underesti- and alpha-internexin immunoreactivity. Brain Res mated and gives further evidence for early degeneration 2012;1465:90–100. 6. Sabado J, Bado J, Casanovas A, et al. Accumulation of misfolded in sensory pathways at peripheral as well as central SOD1 in dorsal root ganglion degenerating proprioceptive sensory levels. These results show, for the first time, the benefit neurons of transgenic mice with amyotrophic lateral sclerosis. of combining electrophysiology and biomedical imaging BioMed Res Int 2014;2014:e852163. 7. Hammad M, Silva A, Glass J, et al. Clinical, electrophysiologic, and to assess non-motor system involvement in ALS. pathologic evidence for sensory abnormalities in ALS. Neurology 2007;69:2236–42. 8. Cohen-Adad J, Mendili MME, Morizot-Koutlidis R, et al. Involvement Author affiliations of spinal sensory pathway in ALS and specificity of cord atrophy to 1 Sorbonne Universités, UPMC Univ Paris 06, CNRS, INSERM, Laboratoire lower motor neuron degeneration. Amyotroph Lateral Scler Front d’Imagerie Biomédicale, Paris, France Degener 2013;14:30–8.

Iglesias C, et al. BMJ Open 2015;5:e007659. doi:10.1136/bmjopen-2015-007659 9 Downloaded from http://bmjopen.bmj.com/ on February 26, 2015 - Published by group.bmj.com

Open Access

9. De Carvalho M, Dengler R, Eisen A, et al. Electrodiagnostic criteria 29. Georgesco M, Salerno A, Camu W. Somatosensory evoked for diagnosis of ALS. Clin Neurophysiol 2008;119:497–503. potentials elicited by stimulation of lower-limb nerves in amyotrophic 10. Pouget J. Critères électro-neuro-myographiques de la sclérose lateral sclerosis. Electroencephalogr Clin Neurophysiol latérale amyotrophique. Rev Neurol Paris 2008;162:4S34–42. 1997;104:333–42. 11. Wirguin I, Brenner T, Argov Z, et al. Multifocal motor nerve 30. Schulte-Mattler WJ, Jakob M, Zierz S. Focal sensory nerve conduction abnormalities in amyotrophic lateral sclerosis. J Neurol abnormalities in patients with amyotrophic lateral sclerosis. J Neurol Sci 1992;112:199–203. Sci 1999;162:189–93. 12. De Carvalho M, Swash M. Nerve conduction studies in amyotrophic 31. Mauguiere F. Somatosensory evoked potentials: normal responses, lateral sclerosis. Muscle Nerve 2000;23:344–52. abnormal waveforms, and clinical applications in neurological 13. Machii K, Ugawa Y, Kokubo Y, et al. Somatosensory evoked diseases. In: Niedermeyer E, Lopez da Silva F. eds. potential recovery in Kii amyotrophic lateral sclerosis/parkinsonism– Electroencephalography: basic principles, clinical applications, and dementia complex (Kii ALS/PDC). Clin Neurophysiol related fields. Philadelphia, PA, USA: Lippincott Williams & Wilkins, 2003;114:564–8. 2005:54.1–54.53. 14. Anziska BJ, Cracco RQ. Short-latency somatosensory evoked 32. Palma V, Guadagnino M, Brescia Morra V, et al. Multimodality potentials to median nerve stimulation in patients with diffuse evoked potentials in sporadic amyotrophic lateral sclerosis: neurologic disease. Neurology 1983;33:989–93. a statistical approach. Electromyogsr Clin Neurophysiol 15. Bosch EP, Yamada T, Kimura J. Somatosensory evoked potentials 1993;33:167–71. in motor neuron disease. Muscle Nerve 1985;8:556–62. 33. De Luca CJ. A practicum on the use of sEMG signals in movement 16. Matheson JK, Harrington HJ, Hallett M. Abnormalities of sciences. 2008. http://www.delsys.com/KnowledgeCenter/Practicum. multimodality evoked potentials in amyotrophic lateral sclerosis. html. Arch Neurol 1986;43:338–40. 34. Cohen-Adad J, El Mendili MM, Lehéricy S, et al. Demyelination and 17. Radtke RA, Erwin A, Erwin CW. Abnormal sensory evoked degeneration in the injured human spinal cord detected with potentials in amyotrophic lateral sclerosis. Neurology diffusion and magnetization transfer MRI. NeuroImage 1986;36:796–801. 2011;55:1024–33. 18. Zanette G, Polo A, Gasperini M, et al. Far-field and cortical 35. Oldfield RC. The assessment and analysis of handedness: the somatosensory evoked potentials in motor neuron disease. Muscle Edinburgh inventory. Neuropsychologia 1971;9:97–113. Nerve 1990;13:47–55. 36. Arieli A, Sterkin A, Grinvald A, et al. Dynamics of ongoing activity: 19. Shefner JM, Tyler HR, Krarup C. Abnormalities in the sensory action explanation of the large variability in evoked cortical responses. potential in patients with amyotrophic lateral sclerosis. Muscle Nerve Science 1996;273:1868–71. 1991;14:1242–6. 37. Klistorner AI, Graham SL. Electroencephalogram-based scaling of 20. Constantinovici A. Abnormal somatosensory evoked potentials in multifocal visual evoked potentials: effect on intersubject amplitude amyotrophic lateral sclerosis. Romanian J Neurol Psychiatry Rev variability. Invest Ophthalmol Vis Sci 2001;42:2145–52. Roum Neurol Psychiatr 1993;31:273–8. 38. You Y, Thie J, Klistorner A, et al. Normalization of visual evoked 21. Georgesco M, Salerno A, Carlander B, et al. [Somatosensory potentials using underlying electroencephalogram levels improves evoked potentials in amyotrophic lateral sclerosis and primary lateral amplitude reproducibility in rats. Invest Ophthalmol Vis Sci sclerosis]. Rev Neurol (Paris) 1994;150:292–8. 2012;53:1473–8. 22. Ogata K, Tobimatsu S, Furuya H, et al. Sporadic amyotrophic lateral 39. Sonoo M, Hatanaka Y, Tsukamoto H, et al. N10 component in sclerosis showing abnormal somatosensory evoked potentials: median nerve somatosensory evoked potentials (SEPs) is not an a report of three cases. Fukuoka Igaku Zasshi 2001;92:242–50. antidromic motor potential. Clin Neurophysiol 2004;115: 23. Lawyer T Jr, Netsky MG. Amyotrophic lateral sclerosis. AMA Arch 2645–9. Neurol Psychiatry 1953;69:171–92. 40. Gandevia SC, Burke D. Saturation in human somatosensory 24. Dustman RE, Snyder EW, Callner DA, et al. The evoked response as pathways. Exp Brain Res 1984;54:582–5. a mesure of cerebral dysfunction. In: Begleiter H, ed. Evoked brain 41. Bede P, Elamin M, Byrne S, et al. Basal ganglia involvement in potentials and behavior. New York: Plenum Press, 1979:321–63. amyotrophic lateral sclerosis. Neurology 2013;81:2107–15. 25. Cosi V, Poloni M, Mazzini L, et al. Somatosensory evoked potentials 42. Gandevia SC, Burke D. Projection of thenar muscle afferents to in amyotrophic lateral sclerosis. J Neurol Neurosurg Psychiatry frontal and parietal cortex of human subjects. Electroencephalogr 1984;47:857–61. Clin Neurophysiol 1990;77:353–61. 26. Gregory R, Mills K, Donaghy M. Progressive sensory nerve 43. Allison T, McCarthy G, Wood CC, et al. Potentials evoked in human dysfunction in amyotrophic lateral sclerosis: a prospective clinical and monkey cerebral cortex by stimulation of the median nerve a and neurophysiological study. J Neurol 1993;240:309–14. review of scalp and intracranial recordings. Brain 27. Mondelli M, Rossi A, Passero S, et al. Involvement of peripheral 1991;114:2465–503. sensory fibers in amyotrophic lateral sclerosis: electrophysiological 44. Kunesch E, Knecht S, Schnitzler A, et al. Somatosensory evoked study of 64 cases. Muscle Nerve 1993;16:166–72. potentials elicited by intraneural microstimulation of afferent nerve 28. Hamada M, Hanajima R, Terao Y, et al. Median nerve fibers. J Clin Neurophysiol 1995;12:476–87. somatosensory evoked potentials and their high-frequency 45. Gandevia S, Burke D, McKeon B. The relationship between the size oscillations in amyotrophic lateral sclerosis. Clin Neurophysiol of a muscle afferent volley and the cerebral potential it produces. 2007;118:877–86. J Neurol Neurosurg Psychiatry 1982;45:705–10.

10 Iglesias C, et al. BMJ Open 2015;5:e007659. doi:10.1136/bmjopen-2015-007659

Study 4 - Multi-parametric spinal cord MRI as potential progression marker in amyotrophic lateral sclerosis

Background and objectives

Recent works by our team and others showed that multi-parametric MRI approaches (DTI, MT and atrophy measurement) allow quantification of lesions in ALS (Cohen-Adad et al., 2013; Agosta et al., 2009; Valsasina et al., 2007) and detection of infraclinical lesions (Study 3). In addition, we showed that some metrics correlated with LMN involvement, which is reflected by cord atrophy, as assessed by transcranial magnetic stimulation (TMS) results (Cohen Adad et al., 2013). Importantly, we showed a correlation between cord atrophy in ALS patients and clinical disability (Cohen-Adad et al., 2013), that was later confirmed in an independent cohort (Branco et al., 2014). In the context of ALS, which among the neurodegenerative diseases is characterized by its fast progression, investigating the progression of lesions is a major challenge. From a pathophysiological perspective, there is a growing interest in the mechanisms of propagation of ALS with the emerging hypothesis of prion-like propagation (Pradat et al., 2015). Neuroimaging methods appear as necessary tools to study in-vivo the temporal and spatial pattern of propagation (Pradat et al., 2015). Longitudinal studies are also mandatory to test and validate biomarkers of progression that are an unmet need for monitoring disease progression at the individual level and to provide surrogate markers of efficacy in clinical trials (Annex A). The development of high motor disability and respiratory insufficiency with orthopnea are classical limitations for performing longitudinal studies in ALS patients.

The main objective of the present study was to test the hypothesis that spinal cord MRI was able to detect in-vivo the progression of the neurodegenerative lesions in ALS. The second objective was to test whether the changes of MRI metrics were clinically meaningful, that is, a major criteria for biomarker development, by investing correlations with functional decline.

Material and Methods

After a first MRI (Cohen-Adad et al., 2013), 29 ALS patients were clinically followed during 12 months; 14/29 patients underwent a second MRI at 11±3 months. Cross-sectional area that has been shown to be a marker of LMN degeneration was measured in cervical and upper thoracic spinal cord from 3D T2-weighted images (Cohen-Adad et al., 2013). Diffusion tensor imaging and MT imaging metrics were measured within the lateral CST in the cervical region (from C2 to C7). Patients were clinically assessed and scored on the ALSFRS-R and manual muscle testing (MMT) score.

- 83 -

Results synthesis

Our longitudinal study showed that it was possible to detect a change of MRI metrics over time. We were able to detect a significant decrease in the lateral segment of the spinal of MTR by 0.33 units/month and by 0.14 mm²/month for cross-sectional area. In the context of ALS, it outlines the accuracy of our multi-parametric MRI approach in detecting the progression of neurodegeneration in ALS. Interestingly, we were not able to detect a significant change over time in DTI metrics that reflected UMN degeneration.

Our results suggest that the cross-sectional area is the most reliable marker of longitudinal neurodegeneration in the spinal cord compared to MMT or DTI metrics such as fractional anisotropy or radial diffusivity. Our study also demonstrated a strong relationship between the spinal cord atrophy rate of change and clinical progression (ALFRS-R and MMT). These results are promising for development of biomarkers in ALS. Here again, further longitudinal studies in a larger population are needed to validate spinal cord atrophy as a surrogate marker of disease progression. An important aspect is to evaluate the sensitivity of multi-parametric MRI metrics to change over time, i.e. whether MRI is able to detect significant changes earlier than classical clinical scores, e. g. ALSFRS-R and MMT.

- 84 - Multi-Parametric Spinal Cord MRI as Potential Progression Marker in Amyotrophic Lateral Sclerosis

Mohamed-Mounir El Mendili1,2,3, Julien Cohen-Adad4, Me´lanie Pelegrini-Issac1,2,3, Serge Rossignol5, Re´gine Morizot-Koutlidis6, Ve´ronique Marchand-Pauvert1,2,3, Caroline Iglesias1,2,3, Sina Sangari1,2,3, Rose Katz1,2,3, Ste´phane Lehericy7,8, Habib Benali1,2,3, Pierre-François Pradat1,2,3,9* 1 Sorbonne Universite´s, UPMC Univ Paris 06, UM CR 2, Laboratoire d’Imagerie Biome´dicale, Paris,ˆ Ile-de-France, France, 2 CNRS, UMR 7371, Laboratoire d’Imagerie Biome´dicale, Paris,ˆ Ile-de-France, France, 3 INSERM, U 1146, Laboratoire d’Imagerie Biome´dicale, Paris,ˆ Ile-de-France, France, 4 Ecole Polytechnique de Montreál, De´partement de Geńie E´lectrique, Montreál, Que´bec, Canada, 5 Universite´ de Montreál, GRSNC, Faculty de Me´decine, Montreál, Que´bec, Canada, 6 AP-HP, Groupe Hospitalier Pitie´-Salpeˆtrie`re, De´partement d’Explorations Fonctionnelles Neurologiques, Paris,ˆ Ile-de-France, France, 7 Inserm U975, UPMC Univ Paris 6, UMR-S975, CNRS UMR7225, Centre de recherche de l’Institut du Cerveau et de la Moelle e´pinie`re – CRICM, Centre de Neuroimagerie de Recherche – CENIR, Paris,ˆ Ile-de-France, France, 8 APHP, Groupe Hospitalier Pitie´-Salpeˆtrie`re, Service de Neuroradiologie, Paris,ˆ Ile-de-France, France, 9 APHP, Groupe Hospitalier Pitie´-Salpeˆtrie`re, De´partement des Maladies du syste`me Nerveux, Paris,ˆ Ile-de-France, France

Abstract

Objective: To evaluate multimodal MRI of the spinal cord in predicting disease progression and one-year clinical status in amyotrophic lateral sclerosis (ALS) patients.

Materials and Methods: After a first MRI (MRI1), 29 ALS patients were clinically followed during 12 months; 14/29 patients underwent a second MRI (MRI2) at 1163 months. Cross-sectional area (CSA) that has been shown to be a marker of lower motor neuron degeneration was measured in cervical and upper thoracic spinal cord from T2-weighted images. Fractional

anisotropy (FA), axial/radial/mean diffusivities (lH, l//, MD) and magnetization transfer ratio (MTR) were measured within the lateral corticospinal tract in the cervical region. Imaging metrics were compared with clinical scales: Revised ALS Functional Rating Scale (ALSFRS-R) and manual muscle testing (MMT) score.

Results: At MRI1, CSA correlated significantly (P,0.05) with MMT and arm ALSFRS-R scores. FA correlated significantly with leg ALFSRS-R scores. One year after MRI1, CSA predicted (P,0.01) arm ALSFSR-R subscore and FA predicted (P,0.01) leg ALSFRS-R subscore. From MRI1 to MRI2, significant changes (P,0.01) were detected for CSA and MTR. CSA rate of change (i.e. atrophy) highly correlated (P,0.01) with arm ALSFRS-R and arm MMT subscores rate of change.

Conclusion: Atrophy and DTI metrics predicted ALS disease progression. Cord atrophy was a better biomarker of disease progression than diffusion and MTR. Our study suggests that multimodal MRI could provide surrogate markers of ALS that may help monitoring the effect of disease-modifying drugs.

Citation: El Mendili M-M, Cohen-Adad J, Pelegrini-Issac M, Rossignol S, Morizot-Koutlidis R, et al. (2014) Multi-Parametric Spinal Cord MRI as Potential Progression Marker in Amyotrophic Lateral Sclerosis. PLoS ONE 9(4): e95516. doi:10.1371/journal.pone.0095516 Editor: Cedric Raoul, Inserm, France Received February 7, 2014; Accepted March 27, 2014; Published April 22, 2014 Copyright: ß 2014 El Mendili et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This study was supported by the Association Franc¸aise contre les Myopathies (AFM) and the Institut pour la Recherche sur la Moelle e´pinie`re et l’Ence´phale (IRME). The research leading to these results has also received funding from the program ‘‘Investissements d’avenir’’ ANR-10-IAIHU-06. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * E-mail: [email protected]

Introduction new neuroimaging approaches, including diffusion tensor imaging (DTI) and magnetization transfer imaging, may provide promising Amyotrophic Lateral Sclerosis (ALS) is the most frequent motor biomarkers in ALS [7–10]. neuron disease characterized by degeneration of both upper and MRI studies in ALS mostly investigated neurodegenerative lower motor neurons. Disease progression is characterized by changes in the brain and thus only investigated upper motor worsening of weakness and physical disability with a median neuron involvement [11–15]. Spinal cord MRI has the advantage survival ranging from 2.5 to 3 years [1,2]. There is no curative of investigating the two motor system components that are treatment and the only available neuroprotective drug is riluzole involved in ALS, i.e the lower motor neuron (via gray matter that showed only a modest effect on survival but no evidence of atrophy) and upper motor neuron (via degeneration of corticospi- effect on the progression of disability [3]. There is an unmet need nal tract). However, spinal cord imaging has technical limitations for biomarkers in amyotrophic lateral sclerosis, not only for better due to the small diameter of the spinal cord, physiological motions diagnosis but also to assess disease progression and the effect of (respiratory and cardiac movements) and susceptibility artifacts disease-modifying drugs in clinical trials (surrogate markers) [16,17,18]. Recently, progress in neuroimaging protocols using [4,5,6]. There is an increasing body of knowledge suggesting that

PLOS ONE | www.plosone.org 1 April 2014 | Volume 9 | Issue 4 | e95516 Spinal Cord MRI Progression Marker in ALS

Table 1. Patients demographics and clinical features in the whole population and in the subgroup of patients who had a second MRI (mean 6 SD).

At baseline At follow up

Characteristics Whole population Sub-group Sub-group Delta Rate of change

Gender 9 Females/20 Males 4 Females/10 Males – – – Age 53.169.8 years 52.6611 years 53.5611 years – – Recruitment period Feb 2010– Feb 2011 Feb 2010– Feb 2011 Jun 2011– Jul 2012 – – Disease duration 26.8626.9 months 28.7627.6 months 39.8628.8 months 11.162.7 months – Site of onset 16 Upper; 11 Lower; 6 Upper; 8 Lower – – – 2 Upper+Lower ALSFRS Arm (/8) 5.0062.25 5.7162.30 2.9362.89 248.75% 20.25 units/month Leg (/8) 4.7262.47 4.9362.23 3.5062.71 228.99% 20.12 units/month Total (/48) 37.1466.35 39.9366.18 30.5769.40 221.47% 20.76 units/month MMT Arm (/70) 52.90611.90 56.79611.11 45616.44 222.14% 21.06 units/month Leg (/70) 59.72617.85 61.14616.94 53.79617.13 212.03% 20.58 units/month Total (140) 112.62623.80 117.93618.29 98.00624.95 216.90% 21.64 units/month TMS Threshold (ms) 68.01620.82 63.65618.46 – – – Amplitude (mV) 1.9261.38 1.6360.97 – – –

doi:10.1371/journal.pone.0095516.t001 multi-parametric MRI approaches (DTI, MT and atrophy and handling utensils, and a leg ALSFRS-R subscore (walking, measurement) allowed to detect abnormalities in the cervical cord climbing stairs). The patients were also scored on manual muscle of ALS patients [10,19–22], some metrics being correlated with testing (MMT) using the Medical Research Council score [24]. upper or lower motor neuron involvement as assessed by MMT was performed at baseline in all ALS patients and at one transcranial magnetic stimulation (TMS) results [10]. Two year in the subgroup of patients who had a second MRI. Seven transversal MRI studies showed a correlation between CSA and muscles in each limb were assessed. Information about patients’ clinical disability [10,22]. For MRI metrics to be used as surrogate demographics and clinical disability are given in Table 1. It marker of neurodegeneration, longitudinal studies are mandatory includes disease duration, which was defined as the delay between to determine whether these metrics are sensitive to changes over the onset of weakness and MRI examination. Patients were scored time and correlate with clinical progression. To address this issue on the day of each MRI. we performed a spinal cord MRI longitudinal study in a series of ALS patients. MRI Acquisition Acquisitions were performed using a 3T MRI system (TIM Materials and Methods Trio, Siemens Healthcare, Erlangen, Germany), with a body coil for signal excitation and a neck/spine coil for signal reception. Subjects Anatomical imaging was performed at the cervical and upper Twenty-nine patients with probable or definite ALS were thoracic levels. DTI and magnetization transfer imaging were enrolled in the study and underwent a baseline spinal MRI performed from vertebral C2 to T2 using the following protocol: examination (MRI1). A follow-up MRI (MRI2) was performed in Anatomical data. A sagittal T2-weighted three-dimensional 14 of these patients after a mean delay of 11.1 months (62.7 (3D) turbo spin echo (TSE) image with slab selective excitation was months). Follow-up MRI was not feasible in other patients due to acquired. Imaging parameters were: isotropic voxel size 0.9 mm3; orthopnea (n = 9), worsening of the general condition rendering FOV = 2806280 mm2; 52 sagittal slices; TR = 1500 ms; transportation and MRI examination impossible (n = 1), loss of TE = 120 ms; acceleration factor = 3; acquisition time , 6 min. follow-up (n = 2), death (n = 2) and refusal (n = 1). The local Ethics Diffusion weighted-imaging. DTI data were acquired Committee of our institution approved all experimental proce- using a single shot EPI sequence with monopolar diffusion- dures (Paris-Ile de France Ethical Committee under the 2009- weighting scheme. The acquisition was cardiac-gated. Eight axial A00291–56 registration number), and written informed consent slices covering C2 to T2 vertebral levels were acquired. Imaging was obtained from each participant. All clinical investigations have parameters were: voxel size = 16165 mm3; FOV = 1286128 been conducted according to the principles expressed in the mm2; TR = 700 ms; TE = 96 ms; acceleration factor = 2; b- Declaration of Helsinki. value = 1000 s/mm2; 64 diffusion encoding directions; 4 averages; Revised ALS Functional Rating Scale [23] data were obtained acquisition time , 15 min. at MRI1, one year after MRI1 in all 29 ALS patients and at MRI2 Magnetization transfer. 3D gradient echo images with slab- for the 14 ALS patients. We calculated an arm ALSFRS-R selective excitation were acquired with and without magnetization subscore pertaining to functions such as handwriting, cutting food transfer (MT) saturation pulse (Gaussian envelope, duration

PLOS ONE | www.plosone.org 2 April 2014 | Volume 9 | Issue 4 | e95516 Spinal Cord MRI Progression Marker in ALS

Figure 1. Regions of interest definition. (A) T2-weighted MRI mid-sagittal slice for an ALS patient. (B,C) T1-weighted data for MTR (B) and mean diffusion-weighted data for DTI (C). (D,E) Defined anatomical ROIs. ROIs were selected in the dorso-lateral (red) aspect of the spinal cord to include most of the CST. (F,G) T2-wheighted data before (F) and after (G) cord outlining. A, anterior, I, inferior; L, left; R, right; P, posterior; S, Superior. doi:10.1371/journal.pone.0095516.g001

9984 ms, frequency offset 1200 Hz). Imaging parameters were: DATA Processing 3 2 voxel size = 0.960.962 mm ; FOV = 2306230 mm ; axial orien- Cord atrophy. Cord cross-sectional area (CSA) was mea- tation with 52 slices (covering the same C2-T2 region as the DTI sured by an experienced operator on the segmentation technic (i.e. scans), TR = 28 ms; TE = 3.2 ms; acquisition time ,5 min per four years experience) on the T2-TSE images in the middle of volume. vertebral levels from C2 to T6 using the semi-automatic method For an exhaustive description of the MRI acquisition param- that has been shown to be accurate and with a low inter-observer eters, the reader is referred to [25]. variability [28,29,30]. The plane perpendicular to the spinal cord was resampled to minimize partial volume between spinal cord TMS Examination and cerebrospinal fluid [31]. To increase reproducibility and TMS was performed within two weeks of the MRI1 using a accuracy of the used segmentation method, images were MAGSTIM 200 device, delivering monophasic stimulation segmented by an experienced operator on the technic. through a round coil (9 cm diameter). Responses of hand muscle DTI. Data were corrected for motion slice-by-slice using FSL (adductor digiti minimi) were recorded with surface electrodes FLIRT [32] with three degrees of freedom (Tx, Ty, Rz). Diffusion using a KPnet system (Natus/Dantec, Denmark). The stimulation metrics were estimated voxel-wise using FSL DTIFIT: fractional was first applied at the cervical level (C7-D1) to excite the root anisotropy (FA), radial diffusivity (lH), axial diffusivity (l//) and near its exit from the spinal cord [26]. The motor evoked potential mean diffusivity (MD). amplitude was measured to assess lower motor neurons impair- Magnetization transfer. Gradient echo volumes with and ment. The motor evoked potential threshold was measured to without magnetization transfer pulse were coregistered using FSL assess CST degeneration [27] (Table 1). FNIRT non-linear algorithm. Magnetization transfer ratio (MTR) was computed voxel-wise following the equation 100 6 ((S0 2 SMT)/S0), where S0 and SMT represent the signal without and with the magnetization transfer pulse, respectively.

Table 2. Results of stepwise linear regressions.

ALSFRS-R MMT

Predictors Arm Leg Total Arm Leg Total

FA 0.7198 0.0003(*) 0.0393(*) 0.7766 0.1280 0.1975

lH 0.6664 0.4428 0.7222 0.5399 0.1439 0.5993 Cord area 0.0493(*) 0.1492 0.4142 0.0193(*) 0.0434(*) 0.8317

Dependent variables were the ALSFRS-R (arm, leg and total scores) and the MMT (arm, leg and total scores) at MRI1. Predictors were gender, age, disease duration, site

of onset, DTI metrics (FA, lH, l//and MD), MTR, CSA, TMS measurements (MEP threshold and amplitude). Significant values are marked with (*). Only significant predictors are represented here. doi:10.1371/journal.pone.0095516.t002

PLOS ONE | www.plosone.org 3 April 2014 | Volume 9 | Issue 4 | e95516 Spinal Cord MRI Progression Marker in ALS

Table 3. Results of stepwise linear regressions.

ALSFSR-R

Predictors Arm Leg Total

FA 0.5342 0.0022(*) 0.0002(*)

lH 0.6667 0.1377 0,0093(*) Cord area 0.0129(*) 0.7775 0.0140(*)

Dependent variables were the ALSFRS-R (arm, leg and total scores) one year after MRI1. Predictors were gender, age, disease duration, site of onset, DTI metrics (FA, lH, l//and MD), MTR, CSA, TMS measurements (MEP threshold and amplitude). Significant values are marked with (*). Only significant predictors are represented here. doi:10.1371/journal.pone.0095516.t003

ROI-based analysis. The lateral portion of the cord was Longitudinal MRI Study delineated manually and using geometry-based information by an Comparisons between baseline and follow-up experienced operator on segmentation. To minimize bias, these MRI. Wilcoxon signed rank test was performed to evaluate regions of interest (ROIs) were defined on the mean diffusion changes in MRI metrics from MRI1 to MRI2. Correlation weighted images (for DTI analysis) and on the 3D gradient-echo between changes in MRI metrics over time and progression of T1-weighted image (for MT analysis) (Figure 1). This method has clinical disability. Spearman’s rank correlation coefficient was used been shown to be sensitive enough to discriminate changes to investigate correlations between MRI metrics and rate of between sensory (posterior) and motor (lateral) tracks in the change of clinical disability scores, defined as the change of a context of ALS [10] and spinal cord injury [25]. For an exhaustive variable between MRI1 and MR2 scans, divided by the time description of ROIs definition, the reader is referred to [25]. interval between the two scans.

Statistics Results Statistical analysis was performed using Matlab (The Math- works Inc, MA, USA). Correlations of MRI Metrics with Clinical Disability at Correlations of MRI metrics with clinical disability at Baseline baseline. Stepwise linear regression was used to find the best Cord CSA was the only significant predictor associated with the predictor of clinical disability at MRI1 in all 29 patients. arm subscore of ALSFRS-R (P = 0.049) and MMT (P = 0.019 for Dependent variables were: ALSFRS-R (total, arm subscore and arm and 0.04 for leg) scales. FA measure in the lateral segment of leg subscore) and MMT. Predictors were: gender, age, site onset, the spinal cord was the only predictor of leg subscore (P = 3.1024) disease duration, FA, lH, l//, MD, MTR, cross-sectional area, and total score (P = 0.039) of the ALSFRS-R. Results of stepwise MEP threshold and amplitude. The probability for a predictor to linear regression analysis are summarized in Table 2 (Bonferroni enter the stepwise model was based on a Fisher’s test, with a P- non corrected, significance level alpha = 0.0056). value set to 0.05. Prediction of clinical disability after one-year follow- Prediction of Clinical Disability after One-year Follow-up up. Stepwise linear regression was used to find the best predictor Results of the stepwise linear regression analysis revealed that of clinical disability in all 29 ALS patients one year after MRI . 1 the mean spinal cord CSA at MRI1 was the only predictor of the Dependent variable was: ALSFRS-R (arm, leg and total scores) arm ALSFRS-R subscore (P=0.013, Table 3). FA measured at after one-year follow-up. Predictors were: gender, age, site onset, MRI1 was the only predictor of the leg ALSFRS-R subscore disease duration, FA, l , l//, MD, MTR, cross-sectional area, H (P=0.002). FA, lH and CSA measurements predicted the total MEP threshold and amplitude measured at baseline. The 24 ALSFSR-R score (PFA = 2.10 , PlH = 0.009, PCSA = 0.014). There probability for a predictor to enter the stepwise model was based was no predictor of the decline of the arm or leg ALSFRS-R on a Fisher’s test, with a P-value set to 0.05. subscores (P.0.05) (Bonferroni non corrected, significance level alpha = 0.0056).

Table 4. MRI metrics at baseline and follow-up in ALS patients (mean 6 SD).

MRI metrics Baseline Follow-up P-value Delta Rate of change

FA 0.52360.054 0.51260.048 0.168 – – 23 2 lH (610 mm /s) 0.74060.128 0.78261.113 0.127 – – 23 2 l// (610 mm /s) 1.74060.117 1.67960.102 0.052 _ – MD (61023 mm2/s) 1.05160.115 1.08760.098 0.068 _ – MTR 33.2262.51 30.0063.15 0.0029(*) 29.68% 20.33 units/month Cord area (mm2) 59.766.7 58.166.9 0.00085(*) 22,73% 20.14 mm2/month

Significant differences are marked with (*). doi:10.1371/journal.pone.0095516.t004

PLOS ONE | www.plosone.org 4 April 2014 | Volume 9 | Issue 4 | e95516 Spinal Cord MRI Progression Marker in ALS

Figure 2. Comparisons between ALS patients at baseline and follow-up for individual cross-sectional area and MTR in the dorso- lateral aspect of the spinal cord that includes mostly the corticospinal tract. Group differences were assessed using Wilcoxon signed rank test. Levels of significance are indicated as: *P,0.01, **P,0.001. doi:10.1371/journal.pone.0095516.g002

Comparisons between Baseline and Follow-up MRI decreased by 0.33 units/month (P=0.003). Figure 2 shows plots of MRI metrics for MRI1 and MRI2 in ALS patients are indicated CSA and MTR at baseline and follow-up. in Table 4. No significant change was detected either for FA (P = 0.168), lH The cord CSA decreased in all patients, except in one case, by a (P = 0.127), l//(P = 0.052) or for MD (P = 0.068) (Bonferroni mean of 0.14 mm2/month (P = 9.1024). Only one patient did not corrected, significance level alpha = 0.008). show a decrease in spinal cord area (minor increase of 0.04 mm2/ month), within the range of sensitivity of the segmentation method [30]. Furthermore, this patient had a very long disease duration (9 years) and was clinically stable between the two MRI. MTR

Figure 3. Correlations between cord area rate of change and the ALSFRS-R arm subscore rate of change and MMT arm subscore rate of change. Correlation coefficients and P-values are derived from Spearman’s correlation. doi:10.1371/journal.pone.0095516.g003

PLOS ONE | www.plosone.org 5 April 2014 | Volume 9 | Issue 4 | e95516 Spinal Cord MRI Progression Marker in ALS

Correlation between Changes in MRI Metrics Over Time TMS index of lower motor neuron involvement, and not with and Progression of Clinical Disability MEP threshold, that is a TMS index of upper motor neuron Results of correlations are shown in Figure 3. The atrophy rate involvement [10]. So as, the observed correlations between of the spinal cord area (i.e., CSA rate of change) was highly atrophy rate of change and arm functional sub-scores decline correlated with worsening in clinical functional status, the suggest that lower motor neuron degeneration was the correlations being higher when considering the arm subscore of determinant factor of the arm disability progression. However, the ALSFRS-R (R = 0.702, P = 0.005) and the arm subscore of the future study using higher resolution enabling separation between MMT (R = 0.717, P=0.004). No correlations were detected grey and white matter may provide more arguments to support between the atrophy rate of change and rates of changes in the these assumptions. leg subscore of the ALSFRS-R (R = 0.376, P = 0.185), the total It is possible that cord atrophy partly contributed to the ALSFSR-R (R = 0.481, P = 0.084), and the leg subscores of the longitudinal decrease in MTR due to partial volume effects with MMT (R = 0.153, P = 0.601). adjacent CSF and gray matter, which may have lowered the No correlations were detected between the rate of change in apparent MTR calculation from averaging effects. However, MTR and the rates of changes in the arm ALSFRS-R several arguments support that partial volume effect minimally (R = 0.037, P = 0.914), leg ALSFRS-R (R = 20.432, P = 0.184), contributed to the detected effect. Firstly, although cord area was arm MMT (R = 20.059, P = 0.863) and leg MMT (R = 0.009, significantly smaller at follow-up, the variation was about 1.6 mm2, P = 0.979) subscores (Bonferroni corrected, significance level which corresponds to a change in radius of less than five microns alpha = 0.025). (using circular approximation of the cord). Given that pixel size is 161 mm2, the partial volume effect resulting from the coarse grid- Discussion based definition of the ROIs–which only encompass 10–15 voxels over the corticospinal region–dominates in comparison to the Results show that spinal cord atrophy and MTR are sensitive change in cord diameter. In other words, the variability the progression of the tissue neurodegenerative process in the introduced by the coarse sampling of MRI markers is orders of spinal cord in ALS patients. The cord area at baseline was magnitude higher than the systematic bias introduced by a small predictive of the functional ability of the arms after a one-year change in the cord size. Secondly, during delineation of the cord follow-up duration. Between MRI1 and MRI2, MTR decreased we opted for a conservative approach, by leaving out pixels that significantly in the lateral segments of the spinal cord by 0.33 were at the cord/CSF or white matter/gray matter interfaces. units/month and the cross sectional area decreased significantly by 2 Although this approach has flaws, this is the currently adopted 0.14 mm /month. Furthermore, the atrophy rate of change was approach for quantifying MRI metrics in the spinal cord. Thirdly, strongly correlated with the rates of change in the arm ALSFRS-R if it were a pure partial volume effect, differences would also be and MMT subscores. observed in the diffusion metrics submitted to the same processing Significant decrease in cord area in ALS patients followed (spatial resolution was almost the same for DTI and MTR). longitudinally was previously reported [20], although no correla- However, neither significant difference nor trends were observed tion with the ALSFRS score was found. Several factors may for diffusion metrics between baseline and follow-up. Although explain this lack of correlation in the study of Agosta et al. First, previous studies introduced atrophy as a confound for isolating the the lower field strength (1.5T versus 3T) might have been effect of a dependent variable [41], it is often a delicate process due associated with lower resolution and/or lower signal-to-noise ratio, to the high correlation between atrophy and other MRI metrics yielding less precision in delineating the spinal cord. Second, the (not due to partial volume effects, but due to degenerative present study includes a larger rostro-caudal portion of the spinal processes). cord, which may have increased the sensitivity to detect motor neuron degeneration affecting the ALSFRS score. Third, the follow-up period was shorter (9 versus 11 months in the present Conclusion study) and patients showed a slower ALSFRS decline (0.5 versus In conclusion, we demonstrated a significant relationship 0.8 in the present study). Finally, variability in the population between atrophy rate and disease progression in our cohort of studied, given the small sample sizes, may also account for the ALS patients. In addition, our results suggest that spinal cord differences. Among our cohort of 29 patients, a follow-up MRI cross-sectional area and MTR are more reliable markers of was only feasible in 14 patients. This relatively high level of drop- longitudinal neurodegeneration in the spinal cord compared to out rate (52%) is common in longitudinal studies in ALS (from DTI metrics such as FA or radial diffusivity. However the 27% to 75%, depending on the duration of the follow-up) [20,33– somewhat lower sensitivity of diffusion MRI might be overcome 39]. The development of respiratory insufficiency responsible for thanks to the current improvements in hardware, acquisition and orthopnea and severe disability are classical limitations for processing techniques. Further longitudinal studies in a larger performing a re-scan in ALS patients. population are needed to validate atrophy rate of the spinal cord Our result suggest that changes over time of spinal cord MRI as a valid surrogate marker of disease progression in ALS. metric may reflect to some extent the respective contribution of lower and upper motor neuron degeneration to disability. Acknowledgments Changes of FA measured in the lateral spinal cord containing mostly the cortico-spinal tract were linked to leg disablity. These We are grateful to all subjects and their relatives. We thank Kevin Nigaud, results are reminiscent of clinical observations, since symptoms Romain Valabre`gue, Alexandre Vignaud, Fre´de´ric Humbert, Christelle due to upper motor neuron degeneration, particularly spasticity, Macia and Eric Bardinet for helping with the acquisition and Dr. Henrik mainly affect the lower limb. Stiffness in the lower limb, which Lundell for providing us the code to measure the cord area. We thank the clinicians of the Paris ALS Center, Drs Gae¨lle Bruneteau, Lucette is associated with balance and posture impairment, is a major Lacomblez, Marie-Violaine Lebouteux, Timothee Lenglet and Franc¸ois cause of disability even in ALS patients without noticeable Salachas. motor deficit in the lower limbs [40]. In our previous study, we showed that CSA was correlated with MEP amplitude, that is a

PLOS ONE | www.plosone.org 6 April 2014 | Volume 9 | Issue 4 | e95516 Spinal Cord MRI Progression Marker in ALS

Author Contributions reagents/materials/analysis tools: MMEM HB. Wrote the paper: MMEM JCA MPI SR RMK VMP CI SS RK SL PFP. MRI platform availability: Conceived and designed the experiments: PFP. Performed the experi- SL. ments: MMEM PFP. Analyzed the data: MMEM HB PFP. Contributed

References 1. Chio`A, Mora G, Leone M, Mazzini L, Cocito D, et al. (2002) Early symptom 22. Branco LM, De Albuquerque M, De Andrade HM, Bergo FP, Nucci A, et al. progression rate is related to ALS outcome: a prospective population-based (2013) Spinal cord atrophy correlates with disease duration and severity in study. Neurology 59: 99–103. amyotrophic lateral sclerosis. Amyotroph Lateral Scler Frontotemporal Degener 2. Gordon PH, Salachas F, Bruneteau G, Pradat PF, Lacomblez L, et al. (2012) ‘‘In press’’. Improving survival in a large French ALS center cohort. J Neurol 259: 1788– 23. Cedarbaum JM, Stambler N, Malta E, Fuller C, Hilt D, et al. (1999) The 1792. ALSFRS-R: a revised ALS functional rating scale that incorporates assessments 3. Bensimon G, Lacomblez L, Delumeau JC, Bejuit R, Truffinet P, et al. (2002) A of respiratory function. BDNF ALS Study Group (Phase III). J Neurol Sci 169: study of riluzole in the treatment of advanced stage or elderly patients with 13–21. amyotrophic lateral sclerosis. J Neurol 249: 609–615. 24. Great Lakes ALS Study Group (2003) A comparison of muscle strength testing 4. Bede P, Bokde AL, Byrne S, Elamin M, Fagan AJ, et al. (2012) Spinal cord techniques in amyotrophic lateral sclerosis. Neurology 61: 1503–1507. markers in ALS: diagnostic and biomarker considerations. Amyotroph Lateral 25. Cohen-Adad J, El Mendili MM, Lehe´ricy S, Pradat PF, Blancho S, et al. (2011) Scler 13: 407–415. Demyelination and degeneration in the injured human spinal cord detected with 5. Turner RT, Modo M (2010) Advances in the application of MRI to diffusion and magnetization transfer MRI. Neuroimage 55: 1024–1033. amyotrophic lateral sclerosis. Expert Opin Med Diagn 4: 483–496. 26. Rossini PM, Barker AT, Berardelli A, Caramia MD, Caruso G, et al. (1994) 6. Pradat PF, Attarian S, Camdessanche JP, Carluer L, Cintas P, et al. (2010) Non-invasive electrical and magnetic stimulation of the brain, spinal cord and Research in amyotrophic lateral sclerosis: what is new in 2009? Rev Neurol roots: basic principles and procedures for routine clinical application. Report of (Paris) 166: 683–698. an IFCN committee. Electroencephalogr Clin Neurophysiol 91: 79–92 7. Carrara G, Carapelli C, Venturi F, Ferraris MM, Lequio L, et al. (2012) A 27. Mills KR, Nithi KA (1997) Corticomotor threshold is reduced in early sporadic distinct MR imaging phenotype in amyotrophic lateral sclerosis: correlation between T1 magnetization transfer contrast hyperintensity along the cortico- amyotrophic lateral sclerosis. Muscle Nerve 20: 1137–1141. spinal tract and diffusion tensor imaging analysis. AJNR Am J Neuroradiol 33: 28. Losseff NA, Webb SL, O’riordan JI, Page R, Wang L, et al. (1996) Spinal cord 733–739. atrophy and disability in multiple sclerosis. A new reproducible and sensitive 8. da Rocha, AJ, Oliveira AS, Fonseca RB, Maia AC, Buainain R P, et al. (2004) MRI method with potential to monitor disease progression. Brain 119: 701–708. Detection of corticospinal tract compromise in amyotrophic lateral sclerosis with 29. Leary SM, Parker GJM, Stevenson VL, Barker GJ, Miller DH, et al. (1999) brain MR imaging: relevance of the T1-weighted spin-echo magnetization Reproducibility of magnetic resonance imaging measurements of spinal cord transfer contrast sequence. AJNR Am J Neuroradiol 25: 1509–1515. atrophy: the role of quality assurance. Magn Reson Imaging 17: 773–776. 9. Charil A, Corbo M, Filippi M, Kesavadas C, Agosta F, et al. (2009) Structural 30. El Mendili MM, Chen R, Tiret B, Pe´le´grini-Issac M, Cohen-Adad J, et al. (2014) and metabolic changes in the brain of patients with upper motor neuron Validation of a semiautomated spinal cord segmentation method. J Magn Reson disorders: a multiparametric MRI study. Amyotroph Lateral Scler 10: 269–279. Imaging ‘‘In press’’. 10. Cohen-Adad J, Mendili MM, Morizot-Koutlidis R, Lehe´ricy S, Meininger V, et 31. Lundell H, Barthelemy D, Skimminge A, Dyrby TB, Biering-Sørensen F, et al. al. (2013) Involvement of spinal sensory pathway in ALS and specificity of cord (2011) Independent spinal cord atrophy measures correlate to motor and sensory atrophy to lower motor neuron degeneration. Amyotroph Lateral Scler deficits in individuals with spinal cord injury. Spinal Cord 49: 70–75. Frontotemporal Degener 14: 30–38. 32. Jenkinson M, Bannister P, Brady M, Smith S (2002) Improved optimization for 11. Keil C, Prell T, Peschel T, Hartung V, Dengler R, et al. (2012) Longitudinal the robust and accurate linear registration and motion correction of brain diffusion tensor imaging in amyotrophic lateral sclerosis. BMC Neurosci 13: 141. images. Neuroimage 17: 825–841. 12. Zhang Y, Schuff N, Woolley SC, Chiang GC, Boreta L, et al. (2011) Progression 33. Block W, Karitzky J, Tra¨ber F, Pohl C, Keller E, et al. (1998) Proton magnetic of white matter degeneration in amyotrophic lateral sclerosis: A diffusion tensor resonance spectroscopy of the primary motor cortex in patients with motor imaging study. Amyotroph Lateral Scler 12: 421–429. neuron disease: subgroup analysis and follow-up measurements. Arch Neurol 55: 13. Rajagopalan V, Allexandre D, Yue GH, Pioro EP (2011) Diffusion Tensor 931–936. Imaging Evaluation of Corticospinal Tract Hyperintensity in Upper Motor 34. Suhy J, Miller RG, Rule R, Schuff N, Licht J, et al. (2002) Early detection and Neuron-Predominant ALS Patients. J Aging Res 2011: 481745. longitudinal changes in amyotrophic lateral sclerosis by 1H MRSI. Neurology 14. Yin H, Cheng SH, Zhang J, Ma L, Gao Y, et al. (2008) Corticospinal tract 58: 773–779. degeneration in amyotrophic lateral sclerosis: a diffusion tensor imaging and 35. Rule RR, Suhy J, Schuff N, Gelinas DF, Miller RG, et al. (2004) Reduced NAA fibre tractography study. Ann Acad Med Singapore 37: 411–415. in motor and non-motor brain regions in amyotrophic lateral sclerosis: a cross- 15. Hong YH, Lee KW, Sung JJ, Chang KH, Song IC, et al. (2004) Diffusion tensor sectional and longitudinal study. Amyotroph Lateral Scler Other Motor Neuron MRI as a diagnostic tool of upper motor neuron involvement in amyotrophic Disord 5: 141–149. lateral sclerosis. J Neurol Sci 227: 73–78. 36. Blain CR, Williams VC, Johnston C, Stanton BR, Ganesalingam J, et al. (2007) 16. Xu J, Shimony JS, Klawiter EC, Snyder AZ, Trinkaus K, et al. (2013) Improved A longitudinal study of diffusion tensor MRI in ALS. Amyotroph Lateral Scler 8: in vivo diffusion tensor imaging of human cervical spinal cord. Neuroimage 67: 348–355. 64–76. 17. Andre JB, Bammer R (2010) Advanced diffusion-weighted magnetic resonance 37. Mitsumoto H, Ulug˘ AM, Pullman SL, Gooch CL, Chan S, et al. (2007) imaging techniques of the human spinal cord. Top Magn Reson Imaging 21: Quantitative objective markers for upper and lower motor neuron dysfunction in 367–378. ALS. Neurology 68: 1402–1410. 18. Cohen-Adad J, Benali H, Rossignol S (2007) Methodology for MR diffusion 38. Sage CA, Peeters RR, Go¨rner A, Robberecht W, Sunaert S (2007) Quantitative tensor imaging of the cat spinal cord. Conf Proc IEEE Eng Med Biol Soc 2007: diffusion tensor imaging in amyotrophic lateral sclerosis. Neuroimage 2007; 34: 323–326. 486–499. 19. Nair G, Carew JD, Usher S, Lu D, Hu XP, et al. (2010) Diffusion tensor imaging 39. Unrath A, Ludolph AC, Kassubek J (2007) Brain metabolites in definite reveals regional differences in the cervical spinal cord in amyotrophic lateral amyotrophic lateral sclerosis: A longitudinal proton magnetic resonance sclerosis. Neuroimage 53: 576–583. spectroscopy study. J Neurol 254: 1099–1106. 20. Agosta F, Rocca MA, Valsasina P, Sala S, Caputo D, et al. (2009) A longitudinal 40. Pradat PF, Bruneteau G, Munerati E, Salachas F, Le Forestier N, et al. (2009) diffusion tensor MRI study of the cervical cord and brain in amyotrophic lateral Extrapyramidal stiffness in patients with amyotrophic lateral sclerosis. Mov sclerosis patients. J Neurol Neurosurg Psychiatry 80: 53–55. Disord 24: 2143–2148. 21. Valsasina P, Agosta F, Benedetti B, Caputo D, Perini M, et al. (2007) Diffusion 41. Ciccarelli O, Wheeler-Kingshott CA, McLean MA, Cercignani M, Wimpey K, anisotropy of the cervical cord is strictly associated with disability in amyotrophic et al. (2007) Spinal cord spectroscopy and diffusion-based tractography to assess lateral sclerosis. J Neurol Neurosurg Psychiatry 78: 480–484. acute disability in multiple sclerosis. Brain 130: 2220–2231.

PLOS ONE | www.plosone.org 7 April 2014 | Volume 9 | Issue 4 | e95516

Study 5 – Cervical spinal cord atrophy profile in adult SMN1-linked SMA

Background and objectives

In this part of my thesis, we analyze the extent of neurodegeneration in another MND, i.e. adult SMN1-linked SMA that shares a common feature of LMN loss with ALS. We investigated the topography of LMN degeneration along the spinal cord in SMA. The clinical data with a classical proximal predominance of the deficit suggests that LMN degeneration may predominate in spinal levels innervating proximal muscles. Whether it corresponds to a selective vulnerability of motor neuron pools depending on their location within the spinal cord and/or to peripheral factors, particularly related to different muscle characteristics, remains unknown. Nevertheless, there is no post-mortem study in SMA patients that studies the distribution of motor neuron loss along the spinal cord and establishes correlations with clinical deficits.

New developments in spinal cord MRI provide a unique opportunity to produce an in-vivo index reflecting LMN loss, as it has been recently shown in our studies of ALS patients (Cohen-Adad et al., 2013; Study 4). In our previous work, we developed and validated a segmentation method that accurately quantifies spinal cord cross-sectional area and volume at large cervical and thoracic spinal cord levels (Study 1 and 2). Here we tested the hypothesis that the classical proximal predominance of the motor deficit may be due to the predominant degeneration of motoneurons at specific spinal cord levels. We therefore quantified, using spinal MRI, the degree of atrophy at different cervical levels as well as the motor deficit seen in these patients. For this purpose, a surface-based morphometry method was used for the first time to produce a three-dimensional cord atrophy profile along the cervical spinal cord and to establish comparisons with the distribution of weakness in SMA patients.

Material and Methods

Eighteen SMN1-linked SMA patients (types: 5 IIIa, 10 IIIb and 3 IV) and 18 age/gender-matched healthy volunteers were recruited. Patients were scored on manual muscle testing, motor function measure (Bérard et al., 2005), and the ALSFRS-R. The spinal cord was imaged using high-resolution 3D T2-weighted sequence from 3T MRI system. Surface-based morphometry technique was used to produce a three-dimensional cervical spinal cord atrophy profile (Morra et al., 2008). Radial distance (Lundell et al., 2011; Morra et al., 2008), and cord cross-sectional area measurements in SMA patients were compared to those in controls and correlated with strength and disability scores.

- 92 -

Results synthesis

Manual muscle testing revealed a diffuse weakness in the upper limbs involving the muscles innervated by the C5 to C8 spinal segments. Cross-sectional area measurements revealed a significant cord atrophy gradient mainly located between C3 and C6 vertebral levels with a cord atrophy rate ranging from 5.4% to 23% in SMA patients compared to controls. Radial distance was significantly lower in SMA patients compared to controls in the anterior-posterior direction with a maximum along C4 and C5 vertebral levels. There were no correlations between atrophy measurements, strength and disability scores. No correlation was detected between rostro-caudal cross-sectional area gradient and proximal-distal muscles strength variation. As showed by radial distance and cross-sectional area comparisons between SMA and controls, atrophy predominates in the spinal segments innervating the proximal muscles. However, there was a clear distal weakness and no correlations between MRI atrophy measurements and MMT were detected. Our findings suggest that beyond motor neuron loss, other factors such as neuromuscular junction or intrinsic skeletal muscle defects may play a role in the complex mechanisms underlying weakness in SMA.

Erratum: In the “Clinical examination” sub-section, C7 (wrist flexion/extension) should be replaced by C7 (wrist flexion/extension and elbow extension).

- 93 - RESEARCH ARTICLE Cervical Spinal Cord Atrophy Profile in Adult SMN1-Linked SMA

Mohamed-Mounir El Mendili1, Timothée Lenglet2,3, Tanya Stojkovic4, Anthony Behin4, Raquel Guimarães-Costa4, François Salachas2, Vincent Meininger2, Gaelle Bruneteau2, Nadine Le Forestier2, Pascal Laforêt4, Stéphane Lehéricy5,6, Habib Benali1, Pierre- François Pradat1,2*

1 Sorbonne Universités, UPMC Univ Paris 06, CNRS, INSERM, Laboratoire d’Imagerie Biomédicale, F- 75013, Paris, France, 2 APHP, Hôpital Pitié-Salpêtriere, Département des Maladies du Système Nerveux, Centre référent SLA, Paris, France, 3 APHP, Hôpital Pitié-Salpêtriere, Service d’Explorations Fonctionnelles, Paris, France, 4 APHP, Centre de Référence Maladies Neuromusculaires Paris-Est, Institut de Myologie, Paris, France, 5 APHP, Hôpital Pitié-Salpêtriere, Service de Neuroradiologie, Paris, France, 6 Sorbonne Universités, UPMC Univ Paris 06, UMR-S975, Inserm U975, CNRS UMR7225, Centre de recherche de l’Institut du Cerveau et de la Moelle épinière–CRICM, Centre de Neuroimagerie de Recherche– CENIR, Paris, France a11111 * [email protected]

Abstract

OPEN ACCESS Purpose Citation: El Mendili M-M, Lenglet T, Stojkovic T, The mechanisms underlying the topography of motor deficits in spinal muscular atrophy Behin A, Guimarães-Costa R, Salachas F, et al. (SMA) remain unknown. We investigated the profile of spinal cord atrophy (SCA) in SMN1- (2016) Cervical Spinal Cord Atrophy Profile in Adult SMN1-Linked SMA. PLoS ONE 11(4): e0152439. linked SMA, and its correlation with the topography of muscle weakness. doi:10.1371/journal.pone.0152439

Editor: Cedric Raoul, Inserm, FRANCE Materials and Methods

Received: February 1, 2016 Eighteen SMN1-linked SMA patients type III/V and 18 age/gender-matched healthy volunteers were included. Patients were scored on manual muscle testing and functional scales. Accepted: March 14, 2016 Spinal cord was imaged using 3T MRI system. Radial distance (RD) and cord cross-sec- Published: April 18, 2016 tional area (CSA) measurements in SMA patients were compared to those in controls and Copyright: © 2016 El Mendili et al. This is an open correlated with strength and disability scores. access article distributed under the terms of the Creative Commons Attribution License, which permits Results unrestricted use, distribution, and reproduction in any medium, provided the original author and source are CSA measurements revealed a significant cord atrophy gradient mainly located between credited. C3 and C6 vertebral levels with a SCA rate ranging from 5.4% to 23% in SMA patients Data Availability Statement: A MAT-file containing compared to controls. RD was significantly lower in SMA patients compared to controls individual MRI data, demographics and clinical in the anterior-posterior direction with a maximum along C4 and C5 vertebral levels features, necessary to reproduce all results from our ( < 10−5). There were no correlations between atrophy measurements, strength study, was submitted as a supporting information: p-values DATA_AVAILABILITY.mat. and disability scores.

Funding: The authors have no support or funding to report. Conclusions

Competing Interests: The authors have declared Spinal cord atrophy in adult SMN1-linked SMA predominates in the segments innervating that no competing interests exist. the proximal muscles. Additional factors such as neuromuscular junction or intrinsic skeletal

PLOS ONE | DOI:10.1371/journal.pone.0152439 April 18, 2016 1 / 12 Spinal Cord Atrophy Profile in SMA

muscle defects may play a role in more complex mechanisms underlying weakness in these patients.

Introduction Spinal muscular atrophy (SMA) is a group of degenerative diseases affecting lower motor neurons [1]. SMA diseases are classically separated in proximal and distal forms [1]. The proximal form of SMA is mainly due to deletions or mutations in the telomeric copy of the survival motor neuron gene-1 (SMN1) [2] SMA III is a milder form of SMA starting after the achievement of walking and is subdivided in SMA IIIa with an onset before 3 years and SMA IIIb after 3 years. Eventually, SMA IV refers to an onset in adults (>18 years) [3]. The reason of the classical proximal predominance of the deficit is unknown and its elucidation would provide an important clue to understand the physiopathology of the disease [4]. New developments in spinal cord MRI provide a unique opportunity to investigate in vivo cord atrophy and to produce an index reflecting lower motor neurons loss, which has never been reported in any SMA type. In ALS, another motor neuron disease that combines lower and upper motor neuron degeneration, cord atrophy measurement reflected grey matter loss [5, 6]. We tested here the hypothesis that the classical proximal predominance of the motor deficit in SMA may be due to predominant degeneration of lower motor neurons at specific spinal cord levels. We therefore quantified, using spinal MRI and for the first time a surface based morphometry technic, the degree of atrophy at different cervical levels and establish comparisons with the distribution of weakness in SMA patients.

Material and Methods Subjects SMA patients were recruited from the Myology Institute and the Motor Neuron Disease referral Center (Pitié-Salpêtrière Hospital, Paris, France) between May 2009 and March 2011. Local Ethic Committee approved all experimental procedures (Paris-Ile de France Ethical Committee under the 2009-A00291-56 registration number), and written informed consent was obtained from each participant. All clinical investigations and MRI examinations were conducted according to the principles expressed in the Declaration of Helsinki.

Clinical examination Muscle strength was measured by manual muscle testing (MMT) and performed in all patients by the same experienced neurologist (P.-F. P. among the authors) using the Medical Research Council (MRC) scale (0–5) [7]. Particular care was taken to have the patients relaxed before examinations since it was noted that patients with SMA got fatigue easily, which biased muscle strength measurements. MMT subscores were calculated by grouping muscles according to their motor column locations at cervical spinal cord segments: C5 (shoulder abduction), C6 (elbow flexion), C7 (wrist flexion/extension), C8 (abduction of the 1st finger and distal flexion of the 3rd finger). Motor Function Measure (MFM) scale was performed [8]. MFM provides a measurement of the motor capacity by three functional indexes: standing and transfer (D1), axial and proximal motor function (D2) and distal motor function (D3). In addition, SMA patients were scored on the Revised ALS Functional Rating Scale (ALSFRS-R) [9]. An arm

PLOS ONE | DOI:10.1371/journal.pone.0152439 April 18, 2016 2 / 12 Spinal Cord Atrophy Profile in SMA

ALSFRS-R subscore (handwriting, cutting food and handling utensils), and a leg ALSFRS-R subscore (walking, climbing stairs) were calculated (S1 Dataset).

MRI acquisition Subjects were positioned head-first supine, with a 2-centimeter-thick pillow to lift the head and no pillow below the neck. This strategy was used to limit the natural cervical cord lordosis at around C3–C4, i.e., excessive cord curvature in the antero-posterior (A-P) direction. Subjects were positioned as comfortable as possible, centered in the scanner’s reference and were systematically asked not to move during the acquisition in order to minimize motion artifacts. Scans were performed using a 3T MRI system (TIM Trio 32-channel, Siemens Healthcare, Erlangen, Germany). The spinal cord was imaged using a three-dimensional (3D) T2-weighted turbo spin echo acquisition with a slab-selective excitation pulse (SPACE sequence: sampling perfection with application optimized contrasts using different flip angle evolution). Imaging parameters were: voxel size = 0.9×0.9×0.9 mm3; Field of View (FOV) = 280×280 mm2; 52 sagittal slices; Repetition time (TR)/ Echo time (TE) = 1500/120ms; flip angle = 140°; generalized autocalibrating partially parallel acquisition (GRAPPA) with acceleration factor R = 3; turbo factor = 69; acquisition time was 6 min [5, 6].

Data processing Preprocessing. Data were corrected for non-uniformity intensity using Minc-Toolkit N3 [10]. Two landmarks have been manually defined in the sagittal slice where the spinal cord was the most median in the FOV by an experienced operator on spinal cord MRI segmentation techniques (M.-M. E. M., 5 years of experience, among the authors), to delimit the cervical spinal cord (i.e. from the upper limit of the odontoid process of the C2 vertebrae to the middle vertebral body C7/T1) [11, 12]. Data were cropped and resampled to a voxel size of 0.3×0.3×0.3 mm3 using 3D cubic interpolation in order to maximize segmentation accuracy [12–14]. Fig 1A shows a preprocessed mid-sagittal section in an SMA patient. Segmentation. Data were segmented using a double threshold-based method (DTbM), an improved version of the well-established threshold-based method (TbM) [12, 14, 15]. The same experienced operator as previously, adjusted visually rare poor segmentations using TbM and, when needed, manually [12]. Fig 1B shows an example of segmentation result in an SMA patient. Cord standardization. Spinal cord masks resulting from segmentation were straightened, then standardized in length by rescaling them to the median cord length (median length across subjects = 413 slices) using 3D nearest neighbor interpolation, leading to a straight and well centered spinal cord masks in the A-P and right-left (R-L) directions in the FOV [11, 12]. Fig 1C and 1D show an example of cord and mask straightening results in an SMA patient. Radial distance and cross-sectional area measurements. Radiuses from the center of mass of each axial mask to its borders were measured with an angular resolution of 5° by mapping Cartesian coordinates to polar coordinates (one radius every 5°), defined as radial distance (RD) (Fig 1E)[13]. This distance could be seen as a measurement of spinal cord thickness. RD enabled to detect a preferential A-P atrophy direction in spinal cord injuries (SCI) at C2 vertebral level that correlated with the clinical status of SCI patients, represented by the ASIA motor and sensory [13]. RD has also been used to spatially localize the hippocampus diffuse atrophy in Alzheimer patients, a tissue that has a geometry resembling to a bent elliptic cylinder as for the spinal cord [16]. Cross-sectional area (CSA) was also computed. CSA was defined as the cord area in mm2 covered by each axial mask along the standardized spinal cord images (S1 Dataset).

PLOS ONE | DOI:10.1371/journal.pone.0152439 April 18, 2016 3 / 12 Spinal Cord Atrophy Profile in SMA

Fig 1. Spinal cord segmentation, cord straightening and radial distance illustration. (A) Preprocessed T2-weighted mid-sagittal section in an SMA patient. The yellow dashed lines delimit the cervical spinal cord region. (B) Resulted segmentation mask (orange). (C) Cervical spinal cord straightening. (D) Mask straightening (orange). (E) Radial distance measurements. A, anterior; I, inferior; L, left, P, posterior, R, right, S, superior. doi:10.1371/journal.pone.0152439.g001

Statistical analysis Statistical analysis was performed using Matlab1 (The Mathworks Inc, MA, USA). Comparison between distal and proximal muscles strength. Wilcoxon signed rank test was performed to test the difference in strength between proximal (MMT C5 spinal level; shoulder abduction) and distal muscles (MMT C8 spinal level; abduction of the 1st finger and distal flexion of the 3rd finger) in SMA patients, a significance level alpha = 0.05 was used. Comparisons of MR measurements between patients and controls. RD and CSA along the standardized spinal cord images were compared between SMA patients and controls using

PLOS ONE | DOI:10.1371/journal.pone.0152439 April 18, 2016 4 / 12 Spinal Cord Atrophy Profile in SMA

permutation test (one side, 100.000 permutations) [16]. Permutation tests measure the distribution of atrophy features represented by RD or CSA decrease in SMA patients compared to controls that would occur by chance if patients and controls were randomly assigned to groups [17]. The high number of permutations 100.000 was chosen to control the standard error of the omnibus probability p (SEp) [18]. Given the high number of computed comparisons (72×413 permutation tests for RD comparisons and 413 permutation tests for CSA comparisons), a supra-significance level alpha = 10−3 was used [13, 16]. The previously constructed template [12] was used as a visualization support for the 3D p-values profile resulted from RD comparisons between SMA patients and controls (S1 Dataset). Mean patients CSA was sub- tracted from mean controls CSA then normalized by mean controls CSA to obtain cord atrophy rate along the cervical spinal cord (in %). Correlations between MR measurements, clinical disability and disease duration. Spearman's rank correlation coefficient was used to investigate correlations between RD and CSA along the cervical spinal cord and disease duration, MMT, MFM and ALSFRS-R subscores (non-normally distributed and ordinal data). Due to the high number of comparisons (413 correlation tests), a supra-significance level alpha = 10−3 was used [13, 16]. In addition, CSA gradient from C4 to C7 vertebral levels, corresponding mainly to C5–C8 spinal segments [19] was correlated, using Spearman's rank correlation coefficient, with MMT variation from C5 to C8 spinal segments. Rostro-caudal CSA gradient and proximal-distal muscles strength variation were represented by the slop of the straight line that best fits CSA curve and MMT subscores, resulting from a linear regression model.

Results Demographics and clinical features Eighteen SMN1-linked SMA patients and 18 age and gender matched healthy volunteers have been recruited. Population demographics and clinical futures in SMA patients are summarized in Table 1. MMT revealed diffuse weakness in the upper limbs involving the muscles innervated by C5 to C8 spinal segments. There was no significant difference between proximal (C5) and distal muscle strength (C8) (Wilcoxon signed rank test; p-value = 0.89) in SMA patients.

Study of spinal cord atrophy Comparisons between patients and controls. Fig 2 shows the 3D p-values profile resulting from RD comparisons between SMA patients and controls. RD was significantly lower in SMA patients compared to controls in the anterior-posterior direction and mostly located between C3 and C6 vertebral levels, corresponding to C4–C7 spinal segments, with a maximum along C4 and C5 vertebral levels, corresponding to C5–C6 spinal segments, (p-values < 10−5)[19]. Fig 3A and 3B show CSA profile from SMA patients and controls and cord atrophy rate along the cervical spinal cord, respectively. CSA showed a significant atrophy gradient mainly located between C3 and C6 vertebral levels with a cord atrophy rate ranging from 5.4% to 23%. CSA difference between SMA patients and controls significantly increased from C2 to C3 vertebral level and remained stable along C4 vertebral level with a mean cord atrophy rate equal to 20%. After this plateau, CSA difference decreased slowly and remained significant to the upper part of C7 vertebral level and was no longer significant beyond. Correlations between MR measurements, clinical disability and disease duration. There were no correlations between RD, CSA along the cervical spinal cord and clinical disability subscores MMT, MFM and ALSFRS-R, nor between RD, CSA and disease duration (p-values > 10−3)(S1–S5 Figs). No correlation was detected between rostro-caudal CSA gradient and proximal-distal muscles strength variation (p-value = 0.35).

PLOS ONE | DOI:10.1371/journal.pone.0152439 April 18, 2016 5 / 12 Spinal Cord Atrophy Profile in SMA

Table 1. Population demographics and clinical features in SMA patients.

Characteristics SMA patients Controls Number 18 18 Age 36 ± 11 years 35 ± 11 years Gender M:F 10:8 11:7 Type 5IIIa, 10IIIb, 3IV – Disease duration 26 ± 15 years – MMT Scores Arm (/70) 56.3 ± 10.2 – Leg (/70) 46.7 ± 15.2 – Total (/140) 102.9 ± 24.1 – C5 spinal level (/10) 7.3 ± 1.9 – C6 spinal level (/10) 8.4 ± 1.9 – C7 spinal level (/30) 26.1 ± 4.5 – C8 spinal level (/20) 14.4 ± 2.7 – Percent changes (%) C5 spinal level 73.3 ± 19.4 – C6 spinal level 84.4 ± 18.9 – C7 spinal level 86.9 ± 15.2 – C8 spinal level 72.2 ± 13.5 – MFM D1 (/39) 19.5 ± 11.1 – D2 (/39) 37.4 ± 1.6 – D3 (/39) 38.3 ± 1.5 – ALSFRS-R Arm (/8) 7.1 ± 1.2 – Leg (/8) 3.3 ± 1.4 – Total (/48) 40.2 ± 3.8 – doi:10.1371/journal.pone.0152439.t001

Discussion The main results of the present study are: 1) a three- and two-dimensional cervical spinal cord atrophy profile can be established in SMA types III and IV; 2) the A-P predominance of atrophy suggests that RD, in addition to CSA, could be considered as an index of motor neurons loss in the anterior horns; 3) there were no correlations between RD/CSA and the degree or the topography of clinical deficits. Spinal cord atrophy was quantified using an accurate segmentation method, DTbM [12]. Segmentation accuracy was increased by adjusting rare poor segmentations using TbM and, when needed, manually [12, 14]. Furthermore, a supra-significance level alpha = 10−3 was used to minimize false positive results due to error in RD and CSA estimations [13, 16]. The combination of high segmentation accuracy [12] cord standardization procedure that allowed groups’ comparison for large spinal cord segments [11, 12] enabled producing for the first time an accurate 3D cervical spinal cord atrophy profile in SMA patients using RD [12, 16]. RD analysis showed preferential cord atrophy in the A-P direction of the spinal cord between C3 and C6 vertebral levels, corresponding to C4–C7 spinal segments [19]. The apparent atrophy at dorsal side of the spinal cord was probably an indirect consequence of anterior horns atrophy. In fact, the large atrophy in the ventral region produces a displacement of centers of mass of axial masks towards the dorsal side of the spinal cord. Thus, the measured radiuses from the center

PLOS ONE | DOI:10.1371/journal.pone.0152439 April 18, 2016 6 / 12 Spinal Cord Atrophy Profile in SMA

Fig 2. 3D atrophy profile along the cervical spinal cord in SMA patients. The color-coding indicates regions with significant atrophy (color scale = p- values of the between groups comparison). A, anterior; I, inferior; L, left, P, posterior, R, right, S, superior. doi:10.1371/journal.pone.0152439.g002

Fig 3. 2D atrophy profile along the cervical spinal cord in SMA patients. (A) Cross-sectional area profile in SMA patients and controls. (B) Cord atrophy rate in SMA patients. The color-coding indicates regions with significant atrophy (color scale = p-values of the between groups comparison). doi:10.1371/journal.pone.0152439.g003

PLOS ONE | DOI:10.1371/journal.pone.0152439 April 18, 2016 7 / 12 Spinal Cord Atrophy Profile in SMA

of mass decreased in both dorsal and ventral regions in SMA patients, which may explain the visible atrophy at both sides of the spinal cord. It cannot also been ruled out that the atrophy of the dorsal column may participate to some extent. Besides post-mortem and histological studies that reported lower motor neuron loss in SMA patients at the cervical cord level [20–22], two autopsy cases of SMN1-linked SMA patients type III showed a loss of the fasciculus gracilis myelinated fibers [22, 23]. The fact that atrophy predominates in the C5 and C6 spinal segments fits with the classical proximal predominance of muscle weakness in SMN1-linked SMA. Our results are also in agreement with a study in delta 7 SMN mice that compared the loss of motor neurons at different spinal cord levels [24]. The authors showed that the reduction of the mean somatic area of motor neurons was maximal at the C5–C6 spinal levels innervating proximal forelimb muscles and was less severe, and not statistically significant rostrally at the C3 level (phrenic motor neurons) and caudally at the C8 level (forelimb distal muscles). We showed that RD and CSA were not correlated with clinical deficits. First, we did not find correlations between atrophy at a given spinal level and the deficit in the corresponding innervated muscles. Second, the gradient of spinal cord atrophy did not mirror the distribution of weakness in our series of patients. In addition, MMT at the upper limbs showed the presence of distal weakness, which is in accordance with analysis of the literature in SMN1-linked SMA patients type III/IV [25–29]. Interestingly, a study in a rodent SMA model showed that muscle denervation was located throughout the body involving both proximal and distal regions [30]. In our series, the absence of correlation between SCA and weakness could be related to methodological limits, mainly the small number of patients in this rare disease and a lack of sensitivity of MMT for weakness quantification. Despite of these limitations, our findings are in line with the complexity of SMA pathogenesis, which is related to multiple factors, involving motor neuron cell bodies, motor axons and muscles, converging in affecting the integrity and function of the neuromuscular unit leading to muscle deficits [24, 30, 31]. Particularly, studies in various SMA mouse models pointed out that the disease process may be initiated at the neuromuscular junctions with reduced transmitter release, smaller and immature endplates and neurofilament accumulation at presynaptic terminals [23, 32–34]. A recent study using repetitive nerve stimulation showed a decrement in nearly one half of SMA patients types II and III [35]. Such evidence for dysfunction of neuromuscular junction was detected both in distal and proximal muscle groups of the arm and was predominantly postsynaptic. Interestingly, recent findings in different SMA animal models have highlighted the contribution of intrinsic skeletal muscle defects, with disruption of the myogenic program associated with a decrease in myofi- ber size and an increase in immature myofibers [36]. A study in an SMA feline model provided convincing evidence that motor neuron degeneration begins distal to the cell body and pro- ceeds retrogradely (« dying-back process ») [37]. From a more clinical perspective, the sensitivity of MRI suggests that quantification of spinal cord atrophy could be a potential biomarker of motor neurons loss in SMA, as previously shown in ALS [5, 6, 38]. There is an unmet need for surrogate markers of disease progression in SMA since clinical scales mildly change over time, especially in SMA types III and IV, in which motor disability worsen very slowly [25, 28, 39]. However, longitudinal studies are needed to evaluate if RD and CSA are sensitive to change over time, as shown for CSA in ALS [6] and could be used as a tool to monitor the effect of investigational therapies. These spinal cord MRI metrics are also applicable to other lower motor neuron diseases such as monomelic atrophy (Hirayama disease) to localize local spinal cord flattening and atrophy [40, 41]. Improvement in MRI techniques are promising to increase the sensitivity of the method particularly at the lumbar spinal cord level which remains technically challenging due to increased physiological motions compared to the cervical level [42]. In the future, the

PLOS ONE | DOI:10.1371/journal.pone.0152439 April 18, 2016 8 / 12 Spinal Cord Atrophy Profile in SMA

development of high-resolution MRI sequences offers the opportunity to specifically measure grey matter loss [42, 43]. However, the absence of correlations between MRI measurements and clinical deficits suggests that the combination with other methods, particularly electrophysiology [35, 44], muscle imaging [44] or electrical impedance myography [45], may be more suitable to measure disease progression in respect with the complexity of the disease.

Conclusions This study is a proof of concept that advanced MRI methods can provide indexes of motor neurons loss in the anterior horns in SMN1-linked SMA. Thanks to theses methods, we were able to demonstate that atrophy predominates in the segments innervating the proximal muscles in SMA. In the future, awareness about the potential of advanced MRI methods in this research field should foster collaborative studies that are needed to include larger number of patients and increase the statistical power in transversal and longitudinal studies.

Supporting Information S1 Dataset. Data set underlying the findings in the present study. MAT-file containing individual MRI data, demographics and clinical features. (MAT) S1 Fig. 3D correlation profile between MMT C5 subscore and RD along the cervical spinal cord. The color-coding indicates regions with p-values < 0.05 (color scale = p-values of the correlation between MMT C5 subscore and RD). (TIFF) S2 Fig. 3D correlation profile between MMT C6 subscore and RD along the cervical spinal cord. The color-coding indicates regions with p-values < 0.05 (color scale = p-values of the correlation between MMT C6 subscore and RD). (TIFF) S3 Fig. 3D correlation profile between MMT C7 subscore and RD along the cervical spinal cord. The color-coding indicates regions with p-values < 0.05 (color scale = p-values of the correlation between MMT C7 subscore and RD). (TIFF) S4 Fig. 3D correlation profile between MMT C8 subscore and RD along the cervical spinal cord. The color-coding indicates regions with p-values < 0.05 (color scale = p-values of the correlation between MMT C8 subscore and RD). (TIFF) S5 Fig. 2D correlation profile between MMT C8 subscore and CSA along the cervical spinal cord. The color-coding indicates regions with p-values < 0.05. For correlations between MMT C5–C7 subscores and CSA along the cervical spinal cord, all p-values were higher than 0.05. (TIFF)

Acknowledgments We are grateful to all subjects for their participation to the study. We thank Kevin Nigaud, Romain Valabrègue, Frédéric Humbert, Antoine Burgos, Mélanie Didier and Eric Bardinet from CENIR for helping with the acquisitions.

PLOS ONE | DOI:10.1371/journal.pone.0152439 April 18, 2016 9 / 12 Spinal Cord Atrophy Profile in SMA

Author Contributions Conceived and designed the experiments: PFP MME TL HB. Performed the experiments: PFP MME TL. Analyzed the data: MME PFP. Contributed reagents/materials/analysis tools: MME PFP TL. Wrote the paper: MME TL TS AB RGC FS VM GB NLF PL SL HB PFP. MRI platform availability: SL.

References 1. Van Den Berg-Vos RM, Van Den Berg LH, Visser J, de Visser M, Franssen H, Wokke JH. The spectrum of lower motor neuron syndromes. J Neurol 2003; 250(11):1279–1292. PMID: 14648143 2. Lefebvre S, Bürglen L, Reboullet S, Clermont O, Burlet P, Viollet L, et al. Identification and characterization of a spinal muscular atrophy-determining gene. Cell 1995; 80(1):155–165. PMID: 7813012 3. Zerres K, Rudnik-Schöneborn S. Natural history in proximal spinal muscular atrophy. Clinical analysis of 445 patients and suggestions for a modification of existing classifications. Arch Neurol 1995; 52 (5):518–523. PMID: 7733848 4. Ciciliot S, Rossi AC, Dyar KA, Blaauw B, Schiaffino S. Muscle type and fiber type specificity in muscle wasting. Int J Biochem Cell Biol 2013; 45(10):2191–2199. doi: 10.1016/j.biocel.2013.05.016 PMID: 23702032 5. Cohen-Adad J, El Mendili MM, Morizot-Koutlidis R, Lehéricy S, Meininger V, Blancho S, et al. Involve- ment of spinal sensory pathway in ALS and specificity of cord atrophy to lower motor neuron degeneration. Amyotroph Lateral Scler Frontotemporal Degener 2013; 14(1):30–38. doi: 10.3109/17482968. 2012.701308 PMID: 22881412 6. El Mendili MM, Cohen-Adad J, Pelegrini-Issac M, Rossignol S, Morizot-Koutlidis R, Marchand-Pauvert V, et al. Multi-parametric spinal cord MRI as potential progression marker in amyotrophic lateral sclerosis. PLoS One 2014; 9(4):e95516. doi: 10.1371/journal.pone.0095516 PMID: 24755826 7. Great Lakes ALS Study Group. A comparison of muscle strength testing techniques in amyotrophic lateral sclerosis. Neurology 2003; 61(11):1503–1507. PMID: 14663032 8. Bérard C, Payan C, Hodgkinson I, Fermanian J; MFM Collaborative Study Group. A motor function measure for neuromuscular diseases. Construction and validation study. Neuromuscul Disord 2005; 15(7):463–470. PMID: 16106528 9. Cedarbaum JM, Stambler N, Malta E, Fuller C, Hilt D, Thurmond B, et al. The ALSFRS-R: a revised ALS functional rating scale that incorporates assessments of respiratory function. BDNF ALS Study Group (Phase III). J Neurol Sci 1999; 169(1–2):13–21. PMID: 10540002 10. Sled JG, Zijdenbos AP, Evans AC. A nonparametric method for automatic correction of intensity non- uniformity in MRI data. IEEE Trans Med Imaging 1998; 17(1):87–97. PMID: 9617910 11. Valsasina P, Horsfield MA, Rocca MA, Absinta M, Comi G, Filippi M. Regional cervical cord atrophy and disability in multiple sclerosis: a voxel-based analysis. Radiology 2013; 266(3):853–861. doi: 10. 1148/radiol.12120813 PMID: 23192778 12. El Mendili MM, Chen R, Tiret B, Villard N, Trunet S, Pélégrini-Issac M, et al. Fast and Accurate Semi- Automated Segmentation Method of Spinal Cord MR Images at 3T Applied to the Construction of a Cer- vical Spinal Cord Template. PLoS one 2015b; 10(3):e0122224. 13. Lundell H, Barthelemy D, Skimminge A, Dyrby TB, Biering-Sørensen F, Nielsen JB. Independent spinal cord atrophy measures correlate to motor and sensory deficits in individuals with spinal cord injury. Spi- nal Cord 2011; 49(1):70–75. doi: 10.1038/sc.2010.87 PMID: 20697420 14. El Mendili MM, Chen R, Tiret B, Pélégrini-Issac M, Cohen-Adad J, Lehéricy S, et al. Validation of a semiautomated spinal cord segmentation method. J Magn Reson Imaging 2015a; 41(2):454–459. 15. Losseff NA, Webb SL, O'Riordan JI, Page R, Wang L, Barker GJ, et al. Spinal cord atrophy and disability in multiple sclerosis. A new reproducible and sensitive MRI method with potential to monitor disease progression. Brain 1996; 119(Pt 3):701–708. PMID: 8673483 16. Morra JH, Tu Z, Apostolova LG, Green AE, Avedissian C, Madsen SK, et al. Automated mapping of hippocampal atrophy in 1-year repeat MRI data from 490 subjects with Alzheimer's disease, mild cognitive impairment, and elderly controls. Neuroimage 2009; 45(1 Suppl):S3–15. doi: 10.1016/j.neuroimage. 2008.10.043 PMID: 19041724 17. Thompson PM, Hayashi KM, De Zubicaray GI, Janke AL, Rose SE, Semple J, et al. Mapping hippocampal and ventricular change in Alzheimer disease. Neuroimage 2007; 36(4):1397–1398. 18. Edgington ES, Onghena P. Randomization tests, 4th ed. Chapman & Hall/CRC, Boca Raton, FL, 2007.

PLOS ONE | DOI:10.1371/journal.pone.0152439 April 18, 2016 10 / 12 Spinal Cord Atrophy Profile in SMA

19. Cadotte DW, Cadotte A, Cohen-Adad J, Fleet D, Livne M, Wilson JR, et al. Characterizing the location of spinal and vertebral levels in the human cervical spinal cord. AJNR Am J Neuroradiol 2015; 36 (4):803–810. doi: 10.3174/ajnr.A4192 PMID: 25523587 20. Huang K, Luo Y. Adult spinal muscular atrophy. A report of four cases. J Neurol Sci. 1983 Oct; 61 (2):249–259. PMID: 6644327 21. Araki S, Hayashi M, Tamagawa K, Saito M, Kato S, Komori T, et al. Neuropathological analysis in spinal muscular atrophy type II. Acta Neuropathol. 2003 Nov; 106(5):441–448. PMID: 12898156 22. Winder TR, Auer RN. Sensory neuron degeneration in familial Kugelberg-Welander disease. Can J Neurol Sci. 1989 Feb; 16(1):67–70. PMID: 2924211 23. Kuru S, Sakai M, Konagaya M, Yoshida M, Hashizume Y, Saito K. An autopsy case of spinal muscular atrophy type III (Kugelberg-Welander disease). Neuropathology 2009; 29(1):63–67. doi: 10.1111/j. 1440-1789.2008.00910.x PMID: 18410269 24. d'Errico P, Boido M, Piras A, Valsecchi V, De Amicis E, Locatelli D, et al. Selective vulnerability of spinal and cortical motor neuron subpopulations in delta7 SMA mice. PLoS One 2013; 8(12): e82654. doi: 10. 1371/journal.pone.0082654 PMID: 24324819 25. Carter GT, Abresch RT, Fowler WM Jr, Johnson ER, Kilmer DD, McDonald CM. Profiles of neuromuscular diseases. Spinal muscular atrophy. Am J Phys Med Rehabil 1995; 74(5 Suppl): S150–159. PMID: 7576422 26. Gardner-Medwin D, Hudgson P, Walton JN. Benign spinal muscular atrophy arising in childhood and adolescence. J Neurol Sci 1967; 5(1): 121–158. PMID: 6061756 27. Hausmanowa-Petrusewicz I, Fidzianska-Dolot A. Clinical features of infantile and juvenile spinal muscular atrophy. In: Gamstorp I, Sarnat HB, editors. Progressive spinal muscular atrophies. New York: Raven Press, 1984: 31–42. 28. Piepers S, van den Berg LH, Brugman F, Scheffer H, Ruiterkamp-Versteeg M, van Engelen BG, et al. A natural history study of late onset spinal muscular atrophy types 3b and 4. J Neurol 2008; 255(9): 1400–1404. doi: 10.1007/s00415-008-0929-0 PMID: 18575920 29. Meadows JC, Marsden CD, Harriman DGF. Chronic spinal muscular atrophy in adults. Part 1. The Kugelberg-Welander syndrome. J Neurol Sci 1969; 9: 527–550. PMID: 5367043 30. Ling KK, Lin MY, Zingg B, Feng Z, Ko CP. Synaptic defects in the spinal and neuromuscular circuitry in a mouse model of spinal muscular atrophy. PLoS One 2010; 5(11): e15457. doi: 10.1371/journal.pone. 0015457 PMID: 21085654 31. Hamilton G, Gillingwater TH. Spinal muscular atrophy: going beyond the motor neuron. Trends Mol Med 2013; 19(1): 40–50. doi: 10.1016/j.molmed.2012.11.002 PMID: 23228902 32. Kariya S, Park GH, Maeno-Hikichi Y, Leykekhman O, Lutz C, Arkovitz MS, et al. Reduced SMN protein impairs maturation of the neuromuscular junctions in mouse models of spinal muscular atrophy. Hum Mol Genet 2008; 17(16): 2552–2569. doi: 10.1093/hmg/ddn156 PMID: 18492800 33. Murray LM, Comley LH, Thomson D, Parkinson N, Talbot K, Gillingwater TH. Selective vulnerability of motor neurons and dissociation of pre- and post-synaptic pathology at the neuromuscular junction in mouse models of spinal muscular atrophy. Hum Mol Genet 2008; 17(7): 949–962. PMID: 18065780 34. Kong L, Wang X, Choe DW, Polley M, Burnett BG, Bosch-Marcé M, et al. Impaired synaptic vesicle release and immaturity of neuromuscular junctions in spinal muscular atrophy mice. J Neurosci 2009; 29(3): 842–851. doi: 10.1523/JNEUROSCI.4434-08.2009 PMID: 19158308 35. Wadman RI, Vrancken AF, van den Berg LH, van der Pol WL. Dysfunction of the neuromuscular junction in spinal muscular atrophy types 2 and 3. Neurology 2012; 79(20): 2050–2055. doi: 10.1212/WNL. 0b013e3182749eca PMID: 23115209 36. Boyer JG, Deguise MO, Murray LM, Yazdani A, De Repentigny Y, Boudreau-Larivière C, et al. Myo- genic program dysregulation is contributory to disease pathogenesis in spinal muscular atrophy. Hum Mol Genet 2014; 23(16): 4249–4259. doi: 10.1093/hmg/ddu142 PMID: 24691550 37. Wakeling EN, Joussemet B, Costiou P, Fanuel D, Moullier P, Barkats M, et al. Failure of lower motor neuron radial outgrowth precedes retrograde degeneration in a feline model of spinal muscular atrophy. J Comp Neurol 2012; 520(8): 1737–1750. doi: 10.1002/cne.23010 PMID: 22120001 38. Bede P, Bokde AL, Byrne S, Elamin M, Fagan AJ, Hardiman O. Spinal cord markers in ALS: diagnostic and biomarker considerations. Amyotroph Lateral Scler. 2012 Sep; 13(5):407–15. doi: 10.3109/ 17482968.2011.649760 PMID: 22329869 39. Werlauff U, Vissing J, Steffensen BF. Change in muscle strength over time in spinal muscular atrophy types II and III. A long-term follow-up study. Neuromuscul Disord 2012; 22(12): 1069–1074. doi: 10. 1016/j.nmd.2012.06.352 PMID: 23127961

PLOS ONE | DOI:10.1371/journal.pone.0152439 April 18, 2016 11 / 12 Spinal Cord Atrophy Profile in SMA

40. Hirayama K, Tomonaga M, Kitano K, Yamada T, Kojima S, Arai K. Focal cervical poliopathy causing juvenile muscular atrophy of distal upper extremity: a pathological study. J Neurol Neurosurg Psychia- try. 1987 Mar; 50(3):285–90. PMID: 3559609 41. Bede P, Walsh R, Fagan AJ, Hardiman O. "Sand-watch" spinal cord: a case of inferior cervical spinal cord atrophy. J Neurol. 2014 Jan; 261(1):235–7. doi: 10.1007/s00415-013-7193-7 PMID: 24281771 42. Taso M, Le Troter A, Sdika M, Ranjeva JP, Guye M, Bernard M, et al. Construction of an in vivo human spinal cord atlas based on high-resolution MR images at cervical and thoracic levels: preliminary results. MAGMA 2014; 27(3): 257–267. doi: 10.1007/s10334-013-0403-6 PMID: 24052240 43. Taso M, Le Troter A, Sdika M, Cohen-Adad J, Arnoux PJ, Guye M, et al. A reliable spatially normalized template of the human spinal cord—Applications to automated white matter/gray matter segmentation and Tensor-Based Morphometry (TBM) mapping of gray matter alterations occurring with age. Neuro- image 2015. pii: S1053-8119(15)00414-0. doi: 10.1016/j.neuroimage.2015.05.034 44. Sproule DM, Punyanitya M, Shen W, Dashnaw S, Martens B, Montgomery M, et al. Muscle volume estimation by magnetic resonance imaging in spinal muscular atrophy. J Child Neurol 2011; 26(3): 309– 317.s. doi: 10.1177/0883073810380457 PMID: 20929908 45. Rutkove SB, Gregas MC, Darras BT. Electrical impedance myography in spinal muscular atrophy: a longitudinal study. Muscle Nerve 2012; 45(5): 642–647. doi: 10.1002/mus.23233 PMID: 22499089

PLOS ONE | DOI:10.1371/journal.pone.0152439 April 18, 2016 12 / 12

5. Discussion

Methodological developments

Multi-parametric studies in MND

Perspectives

Methodological developments

In our first methodological study, we validate the semi-automated segmentation method from Losseff’s et al. (1996), which represents one of the most used approaches to quantify atrophy at 1.5T. The method was validated at 3T in the cervical and thoracic spinal cord in a large set of healthy volunteers and three groups of patients with ALS, SMA and SCI. The method showed high reproducibility and accuracy for measuring spinal cord cross-sectional area in both cervical and thoracic regions at 3T. Improved image spatial resolution obtained at 3T and data preprocessing allowed higher segmentation performances than reported previously (Losseff et al., 1996, McIntosh et al., 2011; Horsfield et al., 2010; Hickman et al., 2004). However, a limitation of our study is that the method is based on contrast difference between the CSF and the spinal cord, which potentially limit its use in regions with T2-hypersignal at the CSF/SC boundary, e.g. demyelination lesions, and narrow spinal canal, e.g. cervical spondylotic myelopathy and SCI. In that respect, we proposed additional preprocessing steps to circumvent the method limitations as well as to increase its accuracy, i.e. by correcting for partial volume and segmentation imperfections (Horsfield et al., 2010; Carbonell- Caballero et al., 2006; Tench et al., 2005; Haralick and Shapiro, 1991). Another limitation of our study is that Losseff’s method manual is time-consuming due to its semi-automated characteristic (mean ± SD computational time = 8.0±1.0 min/subject), which restrains its use to investigate atrophy in large portions of the spinal cord in feasible time (Cohen-Adad et al., 2011, 2013), and ultimately its transfer in clinical setting. Hence, the method automation is needed.

Our previous study suggested the possible full automation of Losseff’s method, given its low sensitivity to the initialization steps. This led us to conduct a new study with the aim to design a fast and accurate semi-automated segmentation method for spinal cord 3T MR images, benefitting from the high robustness of Losseff’s method to its initialization procedure. The computation time for DTbM was - 106 -

3.9±1.2 min/subject (mean ± SD), which is two time faster than Losseff’s method. Our method was rigorously evaluated both qualitatively by visual evaluation made by an experienced neuroradiologist, and quantitatively at C2 vertebral level, the cervical and thoracic regions in 3T T2-weighted data from healthy subjects and patients with ALS, SMA and SCI. DTbM outperformed active surface method (ASM) (Horsfield et al., 2010), the second most used segmentation method to quantify atrophy in MS after Losseff’s method (De leener et al., 2016a), for both quantitative and qualitative accuracy evaluation. Our method showed similar accuracy to Losseff’s method and higher accuracy than ASM.

Despite its high accuracy, our method failed to outline the spinal cord in regions of large areas of T2- hyperintensity, large regions of spinal canal narrowing with limited CSF space (one SMA patient), as well as at the lesion level in four SCI patients with T2-hypersignal and atrophy of the spinal cord and narrow spinal canal. Another limitation of our study is that we did not have the opportunity to conduct a scan-rescan evaluation of DTbM. However, DTbM is based on the well-established Losseff’s method, which showed low scan-rescan variability (Leary et al., 1999). Thus, DTbM is expected to have the same order of variability as Losseff’s method. In addition, we designed a data normalization pipeline that includes DTbM, enabling the construction of an accurate straight 3T T2-weighted cervical spinal cord template using MRI data from 59 healthy volunteers. Parallel to our methodological development, De leener et al., (2014) proposed an automated segmentation method that showed lower accuracy and DTbM. The same group proposed a standardization method that leads to an error in cross-sectional area estimation of 16% (4.8% and 6.5% for DTbM at the cervical and thoracic levels, respectively), which would make the method too inaccurate for atrophy quantification in the pathological context Fonov et al. (2015).

Grey and white matter could not be differentiated in the acquired images due to insufficient contrast inside the spinal cord. However, atrophy may predominate in a particular direction in neurodegenerative diseases and trauma. Indeed, ALS is characterized by cell loss in the anterior horn and the CST degeneration (Wijesekera and Leigh, 2009). The CST degeneration and loss of alpha motor neurons may result in atrophy in the lateral and ventral direction, respectively. SMA is characterized by degeneration of cell bodies in the anterior horn (Katsuno et al., 2010), which potentially induces a pronounced ventral atrophy. In SCI patients, more pronounced atrophy was reported in the antero- posterior direction than in the left/right direction at C2 vertebral level (Lundell et al., 2011). This preferential direction of cord atrophy may be detected in group analysis using the surface-based morphometry method (Morra et al., 2009; Lundell et al., 2011). The hypothesis of a preferential ventral direction in SMA patients was confirmed using our data processing tools combined, for the first time at the spinal cord level, with the surface-based morphometry method (Study 5). Parallel to our atrophy investigation, Taso et al. (2015) proposed mapping spinal cord white and grey matter alterations - 107 -

occurring with age using TBM method. The authors showed that regional spinal cord atrophy was mostly localized in anterior horns in elderly healthy subjects. This method holds great promise for investigating motor neuron loss occurring in MND. However, due to technical and time constraints of our imaging protocol, we could not acquire such high resolution MR images enabling to investigate more specifically regional white and grey matter atrophy in MND. The same method could provide supporting evidence of motoneurons loss at the upper cervical spinal cord in progressive MS pathology (Schlaeger et al., 2014, 2015; Vogt et al., 2009; Davison et al., 1934). In SCI and spinocerebellar ataxia type 3 patients, TBM could be used to investigate the regional contribution of white and grey matter atrophy in the preferential anterior-posterior cord flattening at C2 vertebra level (Lundell et al., 2011; Fahl et al., 2015).

Multi-parametric MRI studies in MND

Lesions of the spinal cord due to neurodegenerative diseases lead to major motor and sensory consequences whose severity depends on their extent. In our studies in MND, we attempt to establish the extent of the spinal lesions as precisely as possible by using multi-parametric MRI as a complementary approach to clinical and electrophysiological examinations.

In a previous study in ALS patients (Cohen-Adad et al. 2013), we showed in the spinal cord, by using DTI and MT imaging, that the dorsal column, i.e. those pathways conveying peripheral sensory information to supraspinal structures were altered. Dorsal column was also involved in a selected subgroup of 12 patients with an early onset of disease (median/maximum disease duration 9/12.6 months), which suggested a sensory involvement at an early stage of the disease. In that respect, we coupled spinal DTI and electrophysiological recordings to further investigate the anatomofunctional properties of sensory pathways in ALS patients and to give more evidence of their involvement. The selected patients had no sensory symptoms or signs and medical conditions associated with peripheral neuropathy. This combined approach revealed sub-clinical sensory defects in 85% of ALS patients, while separated methods revealed abnormal values in 60%. Our results strongly suggest peripheral and central sensory defects, which have been underestimated in previous studies. Together with our previous MRI study, these results demonstrated that advanced spinal cord MRI techniques, combined with clinical and electrophysiological measurements, enabled us to investigate motor and extra-motor system involvement in ALS. Our results suggested that sensory afferent pathways to spinal and cortical structures might be already affected at an early stages in ALS, and thus that sensory alterations may indeed participate to ALS pathogenesis. In this respect, a recent electrophysiological study from our group in the same ALS patients suggested that sensory afferents and interneurons involvement leads to

- 108 -

motoneurons hyper-excitability, which may contribute to their degeneration (Sangari et al., 2016). The latter result is in line with several studies in humans and animal models, suggesting that motoneurons loss is probably not due to their selective vulnerability but a consequence of converging factors causing their degeneration (King et al., 2016). These factors did not affect exclusively motoneurons but comprise several neurons and cells that are interconnected with; i.e., sensory neurons, interneurons, glial cells and myocytes (Morrison et al. 1998; Ilieva et al., 2009; Chang and Martin 2011; McGown et al. 2013).

Beyond the topographical extension of the lesions, we investigated their progression over time. Our longitudinal study suggested that LMN degeneration, as reflected by atrophy (Cohen-Adad et al., 2013), was the major determinant of the disease progression. In addition, our study suggests that white matter lesions in the CST and also in the dorsal column (sensory afferents; unpublished data), may be relatively constant during the symptomatic period and, by contrast, clinical progression is mainly dependent on the extent of anterior horn cell degeneration, as assessed by correlations between atrophy changes and functional deterioration. In this respect, a recent study at the brain level suggested that axonal lesions, reflected by DTI changes along the CST, remain relatively constant in the symptomatic period contrasting with the extension of grey matter lesions, as assessed by VBM, and further supported by the lack of change over time in UMN clinical score in nearly all ALS patients (Menke et al., 2014). Altogether, these results suggest that molecular mechanism underlying degeneration of axons and neuronal cell bodies may not progress at the same rate during the course of the disease. We indicate bellow a putative model of the dynamic of axonal and grey matter degeneration (Figure 16).

- 109 -

Figure 16. Putative model of the dynamic of axonal and grey matter degeneration. During the presymptomatic phase, axonal and grey matter lesions progress at slow rate. Close to the onset of clinical signs, grey matter lesions accelerate and play the major role in clinical signs. During the symptomatic period, axonal lesions slowly progress. By contrast, grey matter lesions show rapid progression, possibly in relation to specific propagation mechanisms (i.e. “prion-like”), and functional decline is mainly dependent on the extension of anterior horn cells degeneration. Yellow: CST (motor pathways); blue: grey matter (anterior horn cells); red: dorsal column (sensory pathways).

- 110 -

Using MRI, we analyzed the topographical extension of lesions in adult SMN1-linked SMA that share a common feature of LMN loss with ALS. The key findings of lesions extent analysis in SMA patients may be summarized as follows: first, we showed that cord atrophy predominates in the spinal segments innervating the proximal muscles, in agreement with a study in delta 7 SMN mice that compared the loss of motor neurons at different spinal cord levels (d'Errico et al., 2013). Second, we did not find correlations between atrophy at a given spinal level and the deficit in the corresponding innervated muscles. In addition, the gradient of spinal cord atrophy did not mirror the distribution of weakness in our series of patients. The fact that atrophy predominates the spinal segments innervating the proximal muscles may have fitted with the general assumption that SMA is characterized by proximal muscle weakness. However, muscle strength in the upper limbs showed a more diffuse deficit with the presence of distal weakness, which is in accordance with analysis of the literature in SMN1-linked SMA patients type III/IV (Carter et al., 1995; Gardner-Medwin et al., 1967; Meadows et al., 1969; Hausmanowa-Petrusewicz and Fidzianska-Dolot, 1984; Piepers et al., 2008). Finally, our findings are in line with the concept of SMA as a multi-system disorder, which is related to multiple factors, involving motor neuron cell bodies, motor axons and muscles, converging in affecting the integrity and function of the neuromuscular unit leading to muscle deficits (Hamilton and Gillingwater, 2013; d'Errico et al., 2013; Ling et al., 2010; Kariya et al., 2008; Murray et al., 2008; Kong et al., 2009; Wadman et al., 2012; Boyer et al., 2014).

Perspectives

From a more clinical perspective, the sensitivity of MRI suggests that quantification of spinal cord atrophy could be a potential biomarker of motor neuron loss in both SMA and ALS. There is an unmet need for surrogate markers of disease progression in SMA and ALS. In SMA, clinical scales mildly change over time, especially in SMA types III and IV, in which motor disability worsen very slowly (Carter et al., 1995; Piepers et al., 2008; Werlauff et al., 2012). In both conditions, longitudinal studies are mandatory to evaluate if multi-parametric MRI metrics are sensitive enough to change over time and thus could be used as tools to monitor the effect of investigational therapies. Nevertheless, improvements in MRI techniques are promising to increase the sensitivity of these metrics, particularly at the lumbar spinal cord level, which remains technically challenging due to increased physiological motions compared to the cervical level. In the future, the development of high-resolution MRI sequences will offer the opportunity to specifically measure grey matter loss (Taso et al., 2014, 2015; Schlaeger et al., 2015; Yiannakes et al., 2014). The combination of multi-parametric MRI with other methods, particularly electrophysiology and advanced clinical examinations, may be more suitable to measure disease progression with respect to the complexity of MND. A new project (SPINE 2; Primary - 111 -

investigator: Pierre-François Pradat; Sponsor: IRME (Institut de Recherche sur la Moelle épinière et l’Encéphale)) that includes advanced MRI techniques, benefiting from recent advances in the field of MRI, combined with electrophysiology and tools for quantification of muscle weakness is ongoing. The aim of SPINE 2 project is to pursue our investigation of lesion spread in ALS, SMA and another MND, X-linked spinal bulbar muscular atrophy, at both cervical spinal cord and brain level.

Technological and methodological advances in the field of spinal cord MRI bring new perspectives for the clinical setting. Indeed, more and more clinical centers are equipped by 3T MR systems with adapted coil and advanced MRI techniques (DTI, MT imaging, anatomical imaging) to image the spinal cord with high resolution in feasible duration time for patients. Specific processing tools of spinal cord images (De Leener et al., 2016b; Study 1 and 2) will facilitate the transfer from the research setting to clinical practice, which will improve the diagnosis and monitoring of a large range of spinal cord pathologies.

- 112 -

Annex A

PRADAT PF and El Mendili MM

2015

Neuroimaging to investigate multisystem involvement and provide biomarkers in amyotrophic lateral sclerosis

BioMed Research International

- 113 - Hindawi Publishing Corporation BioMed Research International Volume 2014, Article ID 467560, 10 pages http://dx.doi.org/10.1155/2014/467560

Review Article Neuroimaging to Investigate Multisystem Involvement and Provide Biomarkers in Amyotrophic Lateral Sclerosis

Pierre-François Pradat1,2,3,4 and Mohamed-Mounir El Mendili1,2,3

1 Sorbonne Universites,´ UPMC Universites´ Paris 06,UMR7371,UMR-S1146, LIB, 75005 Paris, France 2 CNRS, UMR 7371, LIB, 75005 Paris, France 3 INSERM, UMR-S 1146, LIB, 75005 Paris, France 4 Departement´ des Maladies du Systeme` Nerveux, Hopitalˆ Pitie-Salp´ etriˆ ere` (AP-HP), 75005 Paris, France

Correspondence should be addressed to Pierre-Franc¸ois Pradat; [email protected]

Received 21 February 2014;Accepted25 March 2014;Published17 April 2014

Academic Editor: Ana Cristina Calvo

Copyright © 2014 P.-F. Pradat and M.-M. El Mendili. Tis is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Neuroimaging allows investigating the extent of neurological systems degeneration in amyotrophic lateral sclerosis (ALS). Advanced MRI methods can detect changes related to the degeneration of upper motor neurons but have also demonstrated the participation of other systems such as the sensory system or basal ganglia, demonstrating in vivo that ALS is a multisystem disorder. Structural and functional imaging also allows studying dysfunction of brain areas associated with cognitive signs. From a biomarker perspective, numerous studies using difusion tensor imaging showed a decrease of fractional anisotropy in the intracranial portion of the corticospinal tract but its diagnostic value at the individual level remains limited. A multiparametric approach will be required to use MRI in the diagnostic workup of ALS. A promising avenue is the new methodological developments of spinal cord imaging that has the advantage to investigate the two motor system components that are involved in ALS, that is, the lower and upper motor neuron. For all neuroimaging modalities, due to the intrinsic heterogeneity of ALS, larger pooled banks of images with standardized image acquisition and analysis procedures are needed. In this paper, we will review the main fndings obtained with MRI, PET, SPECT, and nuclear magnetic resonance spectroscopy in ALS.

1. Introduction delay is increased in patients who present isolated lower motor neuron (LMN) signs [3]. It is needed to rule out Amyotrophic lateral sclerosis (ALS) is a neurodegenerative other LMN pathologies such as spinal muscular atrophy disease characterized by a progressive alteration in the upper, (SMA), bulbospinal muscular atrophy (Kennedy disease), or cortical, motor neurons and lower motor neurons, located or multifocal neuropathies with conduction blocks [1]. Te in the spinal cord and brainstem. Due to its clinical hetero- apparition of central signs during the follow-up may allow geneity and the lack of biological markers to diagnose ALS, confrming the diagnosis of ALS but this is not the case in all the delay between the frst symptoms and the diagnosis is patients, notably because the severity of peripheral signs can evaluated at 9–13 months [1]. Tere is an unmet need to mask central signs. fnd specifcandearlybiomarkerstohelptodiagnoseand Neuroimaging methods allow investigating in vivo the characterize phenotype or progression [2]. extent of neurological systems degeneration. Tere is a Signs of central motor neuron degeneration are ofen growing body of evidence demonstrating that ALS is a difcult to detect in clinical practice, justifying the interest multisystem neurological disorder. Autopsy studies have for objective neuroimaging markers of upper motor neuron shown that degeneration of central nervous system struc- (UMN) involvement. It has been estimated that clinical UMN tures is not restricted to the primary motor cortex and the signs are absent at frst examination in 7 to 10%ofpatients pyramidal tract [5–8]. In addition to motor signs, cognitive who further develop full-blown ALS [3, 4]. Tediagnostic signs are detected by neuropsychological tests in about 50% 2 BioMed Research International of patients with sporadic ALS and typical frontotemporal study showed that FA in motor-related regions of the CC is dementia (FTD) occurs in approximately 10%ofthepatients more afected than in other CC areas in ALS patients [62]. [9, 10]. Association of ALS and FTD occurs in the majority Tractography is a method based on DTI that enables of C9ORF72-linked familial ALS (FALS) [11, 12]. Atypical reconstruction of the three-dimensional geometry of the clinical features can be associated, defning “ALS plus” [1] pyramidal tract [63] and allows establishing an FA profle syndromes with signs and symptoms involving the sensory along the pyramidal tract [64]. A tractography study showed (particularly in SOD1-linked FALS) [13–22], extrapyramidal that the decreases in FA are largely limited to the precentral [14, 23–31], cerebellar [23, 32], ocular [33, 34], and autonomic areas in patients with ALS [52]. Using tractography to [35]systems. segment the corticobulbar tract, lower FA was measured in In this paper, we will review the main fndings obtained patients with bulbar-onset versus limb-onset disease [65]. with diferent modalities of neuroimaging in ALS. We will By using a voxel-by-voxel approach, which consists of especially focus on magnetic resonance imaging (MRI) stud- comparing groups of patients without an aprioriprede- ies and particularly on spinal cord neuroimaging that has fned region of interest, our team and other groups have shown great developments in the last few years. showed that abnormalities on DTI were detectable outside the primary motor regions, thereby confrming that ALS is a multisystem degenerative disorder [53]. Difuse lesions have 2. Magnetic Resonance Imaging also been observed using voxel-based morphometry (VBM), atechnicallowingautomatedsegmentationandquantifca- 2.1.ConventionalMagneticResonanceImaging.Beyond its tion of grey and white matter volumes to study regional role to exclude several “ALS-mimick” syndromes [1], abnor- diferences [66]. Recently authors used high-resolution T1- malities suggestive of upper motor neuron involvement are weighted imaging data for model-based subcortical registra- sometimes detected. Several studies, using various modalities tion and segmentation to explore the involvement of sub- (T , fuid-attenuated inversion recovery (FLAIR) or fast cortical structures [67]. Tey used vertex-wise statistics that spin echo proton density-weighted imaging), have shown ∗ provide quantitative, visual, surface-projected information hyperintensity in the white matter along the corticospinal 2 about the shape of the various subcortical structures. Tis tract, from the centrum semiovale to the brainstem [36– analysis revealed changes afecting the basal ganglia including 46]. However, such abnormalities are rare, nonspecifc, not the superior and inferior aspects of the bilateral thalami, the readily quantifable, and do not correlate with disease severity lateral and inferior portion of the lef hippocampus, and the or rate of progression [47]. A cortical atrophy, which is medial and superior aspect of the lef caudate (Figure 1). Te predominant in the frontal region, and a characteristic T or authors conclude that dysfunction of frontostriatal networks FLAIR hypointensity at the level of the primary motor cortex ∗ is likely to contribute to the unique neuropsychological have been described in the literature but are exceptionally 2 profle of ALS, dominated by executive dysfunction, apathy, detectable in clinical practice [39, 48]. and defcits in social cognition. Te discovery that the presence of a hexanucleotide 2.2.AdvancedMRI. Difusion-based neuroimaging methods expansion in C9orf72 gene [11, 12]wasassociatedwithFTD allow evaluating the degeneration of white matter fber in ALS led to specifcneuroimaginginvestigations.Arecent bundles. Difusion results from the random movement of study used tract-based spatial statistics of multiple white molecules in vivo.Difusion of water in structures such as the matter difusion parameters, cortical thickness measure- cerebrospinal fuid (CSF) or grey matter is isotropic (identical ments, and VBM analyses in C9orf72-negative patients with in all directions), whereas in white matter it occurs prefer- ALS carrying the C9orf72 hexanucleotide repeat expansion entially along the axis of orientation of the fber bundles. [68]. Teresultisthatextensivecorticalandsubcortical Teapplicationofdifusion gradients in several directions frontotemporal involvement was identifed in association results in difusion tensor imaging (DTI). An anisotropy map with the C9orf72 genotype, compared to the relatively limited can thus be obtained, which provides information about the extramotor pathology in patients with C9orf72-negative ALS microstructural organization of the white matter. [67]. In contrast, a DTI study showed that patients with SOD1 Astudypublishedin1999 demonstrated a decrease in gene-linked FALS showed less extensive pathologic white fractional anisotropy (FA) in the intracranial portion of the matter in motor and extramotor pathways compared with corticospinal tract (subcortical white matter, internal capsule, patients with sporadic ALS [69]. and brainstem) [49]. It has been confrmed by numerous In clinical practice, the diagnostic value of FA measure- other studies [50–60]. Abnormalities have been observed ment remains limited mainly because of the overlap between in patients who had no UMN signs at the time of MRI the values measured in ALS patients and those in control investigation but developed pyramidal tract symptoms later subjects. In one study, measurement of FA in the internal in the course of their disease, suggesting that DTI could capsule to detect central motor neuron lesions compared with contribute to earlier diagnosis of ALS in patients with pure healthy subjects had a sensitivity of 95%, but the specifcity LMN involvement [51]. It has been suggested that a decreased was only 71%, with a positive predictive value of 82% fractional anisotropy (FA) in the corpus callosum (CC) [46]. A recent individual patient data (IPD) meta-analysis was a good DTI marker in patients with ALS [61]. Tese using corticospinal tract data suggested that the diagnostic changes may correspond to degeneration of transcallosal accuracy of DTI lacks sufcient discrimination [59]. Of 30 fbers passing between primary motor cortices. Another identifed studies, 11 corresponding authors provided IPD BioMed Research International 3

(a) (b)

Figure 1: Evidence of hippocampal and basal ganglia involvement in amyotrophic lateral sclerosis. Comparative surface-based vertex analyses between healthy controls and C9orf72 hexanucleotide repeat negative ALS patients corrected for age and multiple comparisons reveal signifcant lef hippocampal (a) and thalamic changes (b) (Courtesy of Peter Bede-Trinity College Dublin).

and 221 ALS patients and 187 healthy control subjects were DWI sequence (Syngo RESOLVE) has been proposed [82]. available for the study. Tepooledsensitivitywas0.68 (95% Te RESOLVE sequence allows minimization of suscepti- CI: 0.62–0.75), and the pooled specifcity was 0.73 (95%CI: bility distortions and T blurring. Furthermore, it can be 0.66–0.80). combined with other acquisition∗ strategies such as reduction feld-of-view (rFOV) [832 –85] and parallel imaging [86]to 2.3.SpinalCordImaging.Spinal MRI has the advantage provide fber tractography in large portions of the spinal to investigate the two motor system components that are cord. Such advances open doors to an accurate quantifca- involved in ALS, that is, the lower and upper motor neuron. tion of DTI metrics profle along the corticospinal tract or Although widely applied to the brain, advanced methods sensory tracts. Tis is fundamentally needed to clarify some such as DTI are challenging at the spinal level because of physiopathological aspects of ALS disease, for instance, the (i) the small size of the cord relative to the brain ( 1 cm dying-back versus dying-forward hypotheses and the possi- diameter in the human) requiring higher spatial resolution ble sensory aferents involvement [79]. In parallel, the new generation of 3 TMRIscannersequippedwith300 mT/m and thus decreasing the signal-to-noise ratio, (ii) physiolog-∼ ical motions (respiration, cardiac) that may bias anisotropic gradients [87, 88]providesnewexploratorydimensionsfor difusion coefcient estimation and create ghosting artifacts white matter microstructures in the spinal cord [89, 90]. As [70–72], (iii) partial volume efects that are more problematic showed in vivo as well as ex vivo for the brain, the new in the cord due to the surrounding cerebrospinal fuid [73], MRI scanners improved tissues sensitivity, signal-to-noise (iv) chemical-shif artifacts arising from the epidural fat and ratio (SNR), and spatial and angular difusion resolution in other nearby structures, and (v) geometric distortions arising practical time. from magnetic feld inhomogeneities in nearby intervertebral MRI pathological spinal cord studies were mainly focused disks and lungs. Te latter point is particularly challenging on white matter integrity where grey matter was not for a in difusion MRI since usual sequences based on echo planar great interest until the past few years. Tis was mainly related imaging (EPI) are very sensitive to such artifacts [74, 75]. In to the difculties in imaging the spinal cord due to low SNR past years, with the development of new methods, such as as well as contrastdiference between CSF and white and cardiac and respiratory gating [75–78], it has been shown that grey matter at 1.5 T. Passing to higher magnetic feld strength DTI and magnetization transfer (MT) imaging were feasible (3 Tand7 T) [91–93], the construction of adapted coils for to detect changes in the spinal cord in ALS [79–81]. In a spinal cord imaging [94–96] and the adaptation of existing recent study, we have shown that abnormalities in the spinal sequences [91, 97]bringnewhorizonsforspinalcordanatom- cord using a multiparametric MRI approach combing DTI, ical explorations. Tree recent studies showed the feasibility MT ratio (MTR), and atrophy measurements correlated with of white/grey matter imaging and presented reliable tools for functional impairment [79]. In this study, local spinal cord anatomical structures characterization in controls [98, 99] atrophy was correlated with muscle defcits and with the and multiple sclerosis patients [100]using3 TMRI-systems. motor evoked potential amplitude measured by transcranial In parallel, one study has showed preliminary results at 3 T magnetic stimulation (TMS), an index of LMN dysfunction. of the construction of a probabilistic atlas and anatomical It suggests that regional atrophy is a sensitive biomarker of template of the human cervical and thoracic spinal cord that motor neuron loss in the anterior horns of the spinal cord. included CSF and white and gray matter [97]. Such atlas can Conversely, DTI and MTR changes in the corticospinal be used for atrophy localization in ALS patients using VBM tract correlated with the higher facilitation motor threshold by adapting the methodology proposed in [101]tothecaseof measured by TMS, a parameter that refects the functionality 3 TMRimages(Figure 2). of the pyramidal tract. Interestingly, changes of DTI metrics Such advances bring new perspectives for neurodegen- demonstrated a subclinical involvement of sensory pathways. erative diseases and particularly for ALS. Quantifcations of Recently, new acquisition strategies have been proposed white and grey matter degeneration are currently feasible. to overcome the inherent difculties in difusion-weighted MRI sequences, image processing, and statistical tools have imaging of the spinal cord. A promising high-resolution been already developed [91, 97, 101–103] and are almost ready 4 BioMed Research International

S 3. Nuclear Magnetic Resonance Spectroscopy AP I A Nuclear magnetic resonance spectroscopy (NMRS) allows RL C2 P measuring the neurochemical profle of a particular region of the brain in vivo. Te main peak is N-acetylaspartate (NAA), which is considered as a marker of neuronal integrity. C3 Several studies have demonstrated a decrease in NAA [115, 116] and/or ratios of NAA with choline-containing com- C4 pounds (Cho) and creatine (Cr) [NAA/Cho and NAA/Cr ratios] at the level of the motor cortex of patients with ALS C5 [115, 117]. For diagnosis applications, studies suggested that NMRS C6 spectroscopy may discriminate between patients with classical ALS and patients with progressive muscular atrophy, a C7 disease confned to LMNs [118, 119]. Tediagnosticvalueis limited because of an overlap between those values in ALS patients and healthy controls. It has been suggested that the combination of N-acetylaspartate (NAA) and myo-inositol (a) (b) may improve the specifcity of the test but further studies are Figure 2:(a)T2-weighted turbo spin echo midsagittal section in needed [120]. Using a whole-brain resonance spectroscopic ahealthysubject(male,62 years old), showing the anatomical imaging approach, NAA showed a signifcant relationship landmarks of the cervical spinal cord. (b) T -weighted 2D gradient with disability [121]. A cross-sectional study assessing proton recalled echo axial sections at the vertebral∗ levels C2,C3,C4,and NMRS of the cervical spine showed that NAA/Cr and C5 for the same subject (voxel size 2 mm). Images NAA/Myo ratios were reduced in patients with ALS but have been acquired using a 3 TMRIsystem(TIMTrio32-channel, also in SOD1-positive people at risk for FALS, suggesting Siemens Healthcare, Erlangen, Germany).=0.7 A:× anterior;0.7 × 3 I: inferior; L: that neurometabolic changes occur early in the course of lef; P: posterior; R: right; S: superior. the disease process [122]. Recent advances in high resolution NMRS at 3 Tallowdirectquantifcation of GABA in the cortex [123]. Recently, a study in a small number of subjects showed that decreased levels of GABA were present in the motor cortex of ALS patients compared to healthy controls for use to fgure out some of the mechanisms involved in [124]. It suggests that a loss of inhibition by interneurons spinal cord tissues degeneration in ALS. may play a role in neurodegeneration through excitotoxicity mechanisms.

2.4.FunctionalMRI.Functional MRI (fMRI), by measur- 4. Positron Emission Tomography ing cortical blood oxygen level dependent (BOLD) signal changes, provides a tool to study cortical function and Positron emission tomography (PET) and monophotonic reorganization. Regional modifcations in cerebral blood fow emission tomography (single photon emission comput- have been studied during hand motor tasks [104]. Several erized tomography (SPECT)) are nuclear imaging tech- studies have demonstrated that cerebral activation involved niques which use various tracers to either reveal neu- more extensive cortical regions than in control subjects. ron dysfunction or investigate a pathogenic mechanism However, whether it refects cortical reorganization [104]or it involved in the disease. Using SPECT, several studies have is the result of cortical functional adaptation due to peripheral demonstrated a decrease in cerebral blood fow afer injec- weakness [105]remainsamatterofdebate.Usingasimple tion of hexamethyl-propyleneamine oxime labelled with hand motor task when the motor defcit is still moderate, a technetium-99 m(99 mTc). Tey showed abnormalities in study showed that cerebral activation is correlated with the the primary motor cortex [125–128], which can also extend rate of disease progression suggesting that brain functional in an anterior fashion into the frontal lobes, particularly in rearrangement in ALS may have prognostic implications patients with associated cognitive problems [129]. Studies [106]. using PET with 2-fuoro-2-deoxy-glucose also showed a Unlike the traditional fMRI, the recently developed variable decrease in cerebral glucose metabolism at rest [130– resting-state functional (rfMRI) techniques avoid potential 132]. Interestingly, a PET study found not only hypometabolic performance since rfMRI does not require the subjects to but also hypermetabolic areas in the brain of sporadic ALS perform any task [107, 108]. Studies that have been carried patients, possibly due to increased FDG uptake by astrocytes out in ALS provide abundant evidence for a reorganisation of and/or microglia [133]. In a recent study, the authors report various cerebral networks [108–113]. Interestingly, one study an FDG metabolism study in patients with the C9orf72 showed that rfMRI changes correlated with the rate of disease mutation compared to nonmutated ALS patients (either with progression and duration [114]. or without dementia) [134]. TeconclusionisthatC9orf72 BioMed Research International 5 mutated ALS patients t have a more widespread central ner- are needed. For this purpose, the NeuroImaging Society in vous system involvement than ALS patients without genetic ALS (NISALS) has emerged in 2010 and has an interactive mutations, with or without dementia. web Platform (http://nedigs05.nedig.uni-jena.de/nisals/)to A PET study showed a widespread loss of binding of the provide quality controlled MRI data for the international GABAA ligand [ C]-fumazenil in sporadic ALS patients scientifccommunity. [135]. Because GABAA11 receptors are widely distributed in the cerebral cortex and are located both on pyramidal cells and Conflict of Interests interneurons, [ C]-fumazenil provides a means of detecting motor and extramotor11 dysfunction in ALS. It was shown Teauthorsdeclarethatthereisnoconfict of interests that abnormalities were less extensive in patients with SOD1- regarding the publication of this paper. linked familial ALS patients, suggesting that GABA-ergic neurotransmission may be less severely impaired in these Acknowledgments cases [136]. Nigrostriatal dysfunction has also been shown by both PET [137]andSPECTstudies[138]. TeauthorsthanktheFrenchAssociationforMyopathies PET imaging can also detect infammatory processes (AFM), the Institute for Research on Spinal Cord and Brain that are implicated in the pathogenesis of ALS. Assess- (IRME), and IHU-A-ICM via the program “Investissements ment of microglial activation can be performed through d’avenir” (ANR-10-IAIHU-06)fortheirresearchsupportto neuroimaging of the 18 kDA translocator protein (TSPO) the cited studies performed by the authors of the paper. that is present in activated glial cells, using selective TSPO- selective radioligands such as C-PK11195. Extensive cortical References abnormalities in fxation of this11 tracer, predominantly in the frontal and temporal lobes have been observed in one study [1]P.-F.PradatandG.Bruneteau,“Diferential diagnosis and [139]. TeDPA714 radioligand has a longer half-life, a better atypical subsets of ALS,” Revue Neurologique,vol.162,no.2,pp. bioavailability, and less nonspecifcbindingthanPK11195 S81–S90, 2006. [140]. Tis allows the examination of milder microglial [2] P.-F. Pradat and M. Dib, “Biomarkers in amyotrophic lateral activation by PET. A prospective study showed signifcant sclerosis: facts and future horizons,” Molecular Diagnosis & binding of DPA714 both in motor cortex areas and also in Terapy,vol.13,no.2,pp.115–125, 2009. temporal areas since the earlier stages of the disease [141]. [3] S. Zoccolella, E. Beghi, G. Palagano et al., “Predictors of delay With the continuing development in new radioligands, PET in the diagnosis and clinical trial entry of amyotrophic lateral sclerosis patients: a population-based study,” Journal of the imaging may potentially provide tools to monitor the efect Neurological Sciences,vol.250,no.1-2,pp.45–49, 2006. of drugs targeting infammation. [4] B. J. Traynor, M. B. Codd, B. Corr, C. Forde, E. Frost, and O. M. Hardiman, “Clinical features of amyotrophic lateral sclerosis according to the El Escorial and Airlie House diagnostic criteria: 5. Conclusion a population-based study,” Archives of Neurology,vol.57,no.8, pp. 1171–1176, 2000. In the recent years, thanks to technological and methodolog- [5]K.Iwanaga,S.Hayashi,M.Oyakeetal.,“Neuropathologyof ical developments, neuroimaging was revealed as an indis- sporadic amyotrophic lateral sclerosis of long duration,” Journal pensable research tool to understand the pathophysiology of the Neurological Sciences,vol.146,no.2,pp.139–143, 1997. of ALS. Because ALS is a multisystem disorder and not a [6] P. Castaigne, F. Lhermitte, J. Cambier, R. Escourolle, and P. le pure motor neuron disease, there is a need to investigate in Bigot, “Etude` neuropathologique de 61 observations de sclerose´ vivo the participation of other systems such as the sensory or laterale´ amyotrophique. Discussion nosologique,” Revue Neu- extrapyramidal systems. Neuroimaging studies have already rologique,vol.127,no.4,pp.1–414, 1972. provided insights about the potential role of sensory feed- [7] S. Sasaki, Y. Tsutsumi, K. Yamane, H. Sakuma, and S. back, infammation, and loss of inhibition by interneurons Maruyama, “Sporadic amyotrophic lateral sclerosis with exten- in the pathogenesis of ALS. Studies on presymptomatic sive neurological involvement,” Acta Neuropathologica,vol.84, carriers of mutations responsible for FALS provide the unique no. 2,pp.211–215, 1992. opportunity to study the very early mechanism that triggers [8] Y. Kawamura, P. J. Dyck, and M. Shimono, “Morphomet- the neurodegenerative cascade leading to the loss of motor ric comparison of the vulnerability of peripheral motor and neurons. sensory neurons in amyotrophic lateral sclerosis,” Journal of Among MRI techniques, although DTI initially appeared Neuropathology and Experimental Neurology,vol.40,no.6,pp. 667–675, 1981. as the most promising diagnostic tool, the disappointing [9] C. Lomen-Hoerth, J. Murphy, S. Langmore, J. H. Kramer, R. K. results of a meta-analysis suggest that a multiparametric Olney, and B. Miller, “Are amyotrophic lateral sclerosis patients approach will be required to make neuroimaging a critical cognitively normal?” Neurology,vol.60,no.7,pp.1094–1097, component in the workup of ALS. Development of spinal 2003. cord imaging will be a key element to provide useful biomark- [10] P. H. Gordon, D. Delgadillo, A. Piquard et al., “Terange ers as sensitive and specifc as possible to help diagnosis and clinical impact of cognitive impairment in French patients and to characterize phenotype or progression. Due to the with ALS: a cross-sectional study of neuropsychological test intrinsic heterogeneity of ALS, larger pooled banks of images performance,” Amyotrophic Lateral Sclerosis,vol.12,no.5,pp. with standardized image acquisition and analysis procedures 372–378, 2011. 6 BioMed Research International

[11]M.DeJesus-Hernandez,I.R.Mackenzie,B.F.Boeveetal., [25] A. Verma and W. G. Bradley, “Atypical motor neuron disease “Expanded GGGGCC hexanucleotide repeat in noncoding and related motor syndromes,” Seminars in Neurology,vol.21, region of C9ORF72 causes chromosome 9p-linked FTD and no. 2,pp.177–187, 2001. ALS,” Neuron,vol.72,no.2,pp.245–256, 2011. [26]S.Zoccolella,G.Palagano,A.Fraddosioetal.,“ALS-plus:5 cases [12] A. E. Renton, E. Majounie, A. Waite et al., “A hexanucleotide of concomitant amyotrophic lateral sclerosis and parkinson- repeat expansion in C9ORF72 is the cause of chromosome 9p21- ism,” Neurological Sciences,vol.23,no.2,pp.S123–S124, 2002. linked ALS-FTD,” Neuron,vol.72,no.2,pp.257–268, 2011. [27] J. Desai and M. Swash, “Extrapyramidal involvement in amy- [13] S. S. Gubbay, E. Kahana, N. Zilber, G. Cooper, S. Pintov, and otrophic lateral sclerosis: backward falls and retropulsion,” Y. Leibowitz, “Amyotrophic lateral sclerosis. A study of its Journal of Neurology, Neurosurgery & Psychiatry,vol.67,no.2, presentation and prognosis,” Journal of Neurology,vol.232,no. pp. 214–216, 1999. 5,pp.295–300, 1985. [28]H.Miwa,M.Kajimoto,andT.Kondo,“Choreainmotorneuron [14] H. Mitsumoto, D. A. Chad, and E. P. Pioro, “Clinical features: disease,” Movement Disorders,vol.17,no.6,p.1397, 2002. signs and symptoms,” in Amyotrophic Lateral Sclerosis,H. [29]P.F.Pradat,F.Salachas,L.Lacomblezetal.,“Associationof Mitsumoto, D. A. Chad, and E. P. Pioro, Eds., pp. 47–64,FA chorea and motor neuron disease,” Movement Disorders,vol.17, Davis Company, Philadelphie, Pa, USA, 1998. no. 2,pp.419–420, 2002. [15]T.-M.Li,E.Alberman,andM.Swash,“Comparisonofsporadic [30]U.I.Knirsch,R.Bachus,G.Gosztonyi,R.Zschenderlein,and and familial disease amongst 580 cases of motor neuron A. C. Ludolph, “Clinicopathological study of atypical motor disease,” Journal of Neurology Neurosurgery & Psychiatry,vol. neuron disease with vertical gaze palsy and ballism,” Acta 51,no.6,pp.778–784, 1988. Neuropathologica,vol.100,no.3,pp.342–346, 2000. [16]B.Moulard,W.Camu,A.Malafosse,M.Billiard,andM.Baldy- [31] P.-F. Pradat, G. Bruneteau, E. Munerati et al., “Extrapyramidal Moulinier, “Etude´ clinique des formes familiales de sclerose´ stifness in patients with amyotrophic lateral sclerosis,” Move- laterale´ amyotrophique: revue de la litterature,”´ Revue Neu- ment Disorders,vol.24,no.14,pp.2143–2148, 2009. rologique,vol.153,no.5,pp.314–324, 1997. [32] N. Schimke, K. Krampf, S. Petri, R. Dengler, and J. Bufer, [17] K. Abe, M. Aoki, M. Ikeda, M. Watanabe, S. Hirai, and Y. “Cerebral symptoms with motor neuronal disorders: a special Itoyama, “Clinical characteristics of familial amyotrophic lateral form of ALS-plus syndrome,” Der Nervenarzt,vol.73,no.8,pp. sclerosis with Cu/Zn superoxide dismutase gene mutations,” 751–753, 2002. Journal of the Neurological Sciences,vol.136,no.1-2,pp.108–116, [33]H.Hayashi,S.Kato,T.Kawada,andT.Tsubaki,“Amyotrophic 1996. lateral sclerosis: oculomotor function in patients in respirators,” [18] P. M. Andersen, L. Forsgren, M. Binzer et al., “Autosomal Neurology,vol.37,no.8,pp.1431–1432, 1987. recessive adult-onset amyohophic lateral sclerosis associated [34] R. Sharma, S. Hicks, C. M. Berna, C. Kennard, K. Talbot, and with homozygosity for Asp90A1aCuZn-superoxidedismutase M. R. Turner, “Oculomotor dysfunction in amyotrophic lateral mutationAclinicalandgenealogicalstudyof 36 patients,” Brain, sclerosis: a comprehensive review,” Archives of Neurology,vol. vol. 119,no.4,pp.1153–1172, 1996. 68,no.7,pp.857–861, 2011. [19] W.Camu, J. Khoris, B. Moulard et al., “Genetics of familial ALS [35] T. Shimizu, A. Kawata, S. Kato et al., “Autonomic failure in ALS and consequences for diagnosis,” Journal of the Neurological with a novel SOD1 gene mutation,” Neurology,vol.54,no.7,pp. Sciences,vol.165,no.1,pp.S21–S26, 1999. 1534–1537, 2000. [20] H. F. Jafari-Schluep, J. Khoris, V. Mayeux-Portas, C. Hand, G. [36] G. Cheung, M. J. Gawal, P.W.Cooper, R. I. Farb, and L. C. Ang, Rouleau, and W. Camu, “Les anomalies du gene` superoxyde “Amyotrophic lateral sclerosis: correlation of clinical and MR dismutase 1 dans la sclerose´ laterale´ amyotrophique famil- imaging fndings,” Radiology,vol.194,no.1,pp.263–270, 1995. iale: correlations´ phenotype/g´ enotype´ et implications pratiques. [37] G. Comi, M. Rovaris, and L. Leocani, “Neuroimaging in L’e x p erience´ franc¸aise et revue de la litterature,”´ Revue Neu- amyotrophic lateral sclerosis,” European Journal of Neurology, rologique,vol.160,no.1,pp.44–50, 2004. vol. 6,no.6,pp.629–637, 1999. [21]A.Kawata,S.Kato,H.Hayashi,andS.Hirai,“Prominent [38] D. S. Goodin, H. A. Rowley, and R. K. Olney, “Magnetic sensory and autonomic disturbances in familial amyotrophic resonance imaging in amyotrophic lateral sclerosis,” Annals of lateral sclerosis with a Gly93Ser mutation in the SOD1 gene,” Neurology,vol.23,no.4,pp.418–420, 1988. Journal of the Neurological Sciences,vol.153,no.1,pp.82–85, [39]M.J.Hecht,F.Fellner,C.Fellner,M.J.Hilz,D.Heuss, 1997. and B. Neundorfer,¨ “MRI-FLAIR images of the head show [22] J. Khoris, B. Moulard, V. Briolotti et al., “Coexistence of corticospinal tract alterations in ALS patients more frequently dominant and recessive familial amyotrophic lateral sclerosis than T2-, T1- and proton-density-weighted images,” Journal of with the D90ACu,Znsuperoxidedismutasemutationwithin the Neurological Sciences,vol.186,no.1-2,pp.37–44, 2001. the same country,” European Journal of Neurology,vol.7,no.2, [40] E. Hofmann, G. Ochs, A. Pelzl, and M. Warmuth-Metz, “Te pp. 207–211, 2000. corticospinal tract in amyotrophic lateral sclerosis: an MRI [23] R. J. Davenport, R. J. Swingler, A. M. Chancellor, and C. P.War- study,” Neuroradiology,vol.40,no.2,pp.71–75, 1998. low, “Avoiding false positive diagnoses of motor neuron disease: [41] Y. Kato, K. Matsumura, Y. Kinosada, Y. Narita, S. Kuzuhara, lessons from the Scottish motor neuron disease register,” Journal and T. Nakagawa, “Detection of pyramidal tract lesions in of Neurology Neurosurgery & Psychiatry,vol.60,no.2,pp.147– amyotrophic lateral sclerosis with magnetization-transfer mea- 151, 1996. surements,” American Journal of Neuroradiology,vol.18,no.8, [24]A.EisenandD.B.Calne,“Amyotrophiclateralsclerosis,Parkin- pp. 1541–1547, 1997. son’s disease and Alzheimer’s disease: phylogenetic disorders [42] S. Mirowitz, K. Sartor, M. Gado, and R. Torack, “Focal signal- of the human neocortex sharing many characteristics,” Te intensity variations in the posterior internal capsule: normal Canadian Journal of Neurological Sciences,vol.19,no.1,pp.117– MR fndings and distinction from pathologic fndings,” Radi- 120, 1992. ology,vol.172,no.2,pp.535–539, 1989. BioMed Research International 7

[43] J. L. Tanabe, M. Vermathen, R. Miller, D. Gelinas, M. W.Weiner, [60] M. R. Turner, M. C. Kiernan, P. N. Leigh, and K. Talbot, and W.D. Rooney, “Reduced MTR in the corticospinal tract and “Biomarkers in amyotrophic lateral sclerosis,” Te Lancet Neu- normal T2 in amyotrophic lateral sclerosis,” Magnetic Resonance rology,vol.8,no.1,pp.94–109, 2009. Imaging,vol.16,no.10,pp.1163–1169, 1998. [61] N. Filippini, G. Douaud, C. E. MacKay, S. Knight, K. Talbot, [44]J.W.Torpe, I. F. Moseley, C. H. Hawkes, D. G. MacManus, and M. R. Turner, “Corpus callosum involvement is a consistent W. I. McDonald, and D. H. Miller, “Brain and spinal cord MRI feature of amyotrophic lateral sclerosis,” Neurology,vol.75,no. in motor neuron disease,” Journal of Neurology Neurosurgery & 18,pp.1645–1652, 2010. Psychiatry,vol.61,no.3,pp.314–317, 1996. [62]M.C.Chapman,L.Jelsone-Swain,T.D.Johnson,K.L.Gruis, [45]M.Waragai,“MRIandclinicalfeaturesinamyotrophiclateral and R. C. Welsh, “Difusion tensor MRI of the corpus callosum sclerosis,” Neuroradiology,vol.39,no.12,pp.847–851, 1997. in amyotrophic lateral sclerosis,” Journal of Magnetic Resonance [46] J. M. Graham, N. Papadakis, J. Evans et al., “Difusion tensor Imaging,vol.39,no.3,pp.641–647, 2013. imaging for the assessment of upper motor neuron integrity in [63] O. Abe, H. Yamada, Y.Masutani et al., “Amyotrophic lateral scle- ALS,” Neurology,vol.63,no.11,pp.2111–2119, 2004. rosis: difusion tensor tractography and voxel-based analysis,” [47] J. M. C. Winhammar, D. B. Rowe, R. D. Henderson, and M. C. NMR in Biomedicine,vol.17,no.6,pp.411–416, 2004. Kiernan, “Assessment of disease progression in motor neuron [64] O. Ciccarelli, T. E. Behrens, D. R. Altmann et al., “Probabilistic disease,” TeLancetNeurology,vol.4,no.4,pp.229–238, 2005. difusion tractography: a potential tool to assess the rate of [48]H.Oba,T.Araki,K.Ohtomoetal.,“Amyotrophiclateral disease progression in amyotrophic lateral sclerosis,” Brain,vol. sclerosis: T2 shortening in motor cortex at MR imaging,” 129,no.7,pp.1859–1871, 2006. Radiology,vol.189,no.3,pp.843–846, 1993. [65] S. Aoki, N. K. Iwata, Y. Masutani et al., “Quantitative evaluation [49]C.M.Ellis,A.Simmons,D.K.Jonesetal.,“Difusion tensor of the pyramidal tract segmented by difusion tensor tractog- MRI assesses corticospinal tract damage in ALS,” Neurology, raphy: feasibility study in patients with amyotrophic lateral vol. 53,no.5,p.1051, 1999. sclerosis,” Radiation Medicine,vol.23,no.3,pp.195–199, 2005. [50]P.Valsasina,F.Agosta,B.Benedettietal.,“Difusion anisotropy [66] J. Grosskreutz, J. Kaufmann, J. Fradrich,¨ R. Dengler, H.-J. of the cervical cord is strictly associated with disability in amy- Heinze, and T. Peschel, “Widespread sensorimotor and frontal otrophic lateral sclerosis,” Journal of Neurology, Neurosurgery & cortical atrophy in amyotrophic lateral sclerosis,” BMC Neurol- Psychiatry,vol.78, no. 5,pp.480–484, 2007. ogy,vol.6,no.1, article 17, 2006. [51] M. Sach, G. Winkler, V.Glauche et al., “Difusion tensor MRI of [67]P.Bede,M.Elamin,S.Byrneetal.,“Basalgangliainvolvement early upper motor neuron involvement in amyotrophic lateral in amyotrophic lateral sclerosis,” Neurology,vol.81,no.24,pp. sclerosis,” Brain,vol.127,no.2,pp.340–350, 2004. 2107–2115, 2013. [52]C.A.Sage,R.R.Peeters,A.Gorner,¨ W. Robberecht, and S. [68] P.Bede, A. L. Bokde, S. Byrne et al., “Multiparametric MRI study Sunaert, “Quantitative difusion tensor imaging in amyotrophic of ALS stratifed for the C9orf72 genotype,” Neurology,vol.81, lateral sclerosis,” NeuroImage,vol.34, no. 2,pp.486–499, 2007. no. 4,pp.361–369, 2013. [53] L. Tivard, P.-F. Pradat, S. Lehericy´ et al., “Difusion tensor [69]B.R.Stanton,D.Shinhmar,M.R.Turneretal.,“Difusion imaging and voxel based morphometry study in amyotrophic tensor imaging in sporadic and familial (D90ASOD1)formsof lateral sclerosis: Relationships with motor disability,” Journal of amyotrophic lateral sclerosis,” Archives of Neurology,vol.66,no. Neurology, Neurosurgery & Psychiatry,vol.78,no.8,pp.889– 1,pp.109–115, 2009. 892, 2007. [70] H. S. Kharbanda, D. C. Alsop, A. W. Anderson, G. Filardo, and [54]H.Mitsumoto,A.M.Ulug,ˇ S. L. Pullman et al., “Quantitative D. B. Hackney, “Efects of cord motion on difusion imaging of objective markers for upper and lower motor neuron dysfunc- the spinal cord,” Magnetic Resonance in Medicine,vol.56,no.2, tion in ALS,” Neurology,vol.68,no.17,pp.1402–1410, 2007. pp. 334–339, 2006. [55] J. C. T. Wong, L. Concha, C. Beaulieu, W. Johnston, P. S. Allen, [71] C. A. Clark, G. J. Barker, and P. S. Tofs, “Improved reduction and S. Kalra, “Spatial profling of the corticospinal tract in of motion artifacts in difusion imaging using navigator echoes amyotrophic lateral sclerosis using difusion tensor imaging,” and velocity compensation,” Journal of Magnetic Resonance,vol. Journal of Neuroimaging,vol.17,no.3,pp.234–240, 2007. 142,no.2,pp.358–363, 2000. [56]S.K.Schimrigk,B.Bellenberg,M.Schluter¨ et al., “Difusion [72] T. Verma and J. Cohen-Adad, “Efect of respiration on the B0 tensor imaging-based fractional anisotropy quantifcation in feld in the human spinal cord at 3T,” Magnetic Resonance in the corticospinal tract of patients with amyotrophic lateral Medicine, 2014. sclerosis using a probabilistic mixture model,” American Journal [73] R. G. Nunes, P. Jezzard, and S. Clare, “Investigations on the of Neuroradiology,vol.28, no. 4,pp.724–730, 2007. efciency of cardiac-gated methods for the acquisition of [57] C. R. V.Blain, V.C. Williams, C. Johnston et al., “Alongitudinal difusion-weighted images,” Journal of Magnetic Resonance,vol. study of difusion tensor MRI in ALS,” Amyotrophic Lateral 177,no.1,pp.102–110, 2005. Sclerosis,vol.8,no.6,pp.348–355, 2007. [74]H.U.Voss,R.Watts,A.M.Ulug,ˇ and D. Ballon, “Fiber tracking [58] J. Li, P.Pan, W.Song, R. Huang, K. Chen, and H. Shang, “Ameta- in the cervical spine and inferior brain regions with reversed analysis of difusion tensor imaging studies in amyotrophic gradient difusion tensor imaging,” Magnetic Resonance Imag- lateral sclerosis,” Neurobiology of Aging,vol.33,no.8,pp.1833– ing,vol.24,no.3,pp.231–239, 2006. 1838, 2011. [75] H. Lundell, D. Barthelemy, F. Biering-Sørensen, J. Cohen- [59]B.R.Foerster,B.A.Dwamena,M.Petrouetal.,“Diagnostic Adad, J. B. Nielsen, and T. B. Dyrby, “Fast difusion tensor accuracy of difusion tensor imaging in amyotrophic lateral imaging and tractography of the whole cervical spinal cord sclerosis: a systematic review and individual patient data meta- using point spread function corrected echo planar imaging,” analysis,” Academic Radiology,vol.20,no.9,pp.1099–1106, 2013. Magnetic Resonance in Medicine,vol.69,no.1,pp.144–149, 2012. 8 BioMed Research International

[76] E. Spuentrup, A. Buecker, C. Koelker, R. W. Guenther, and M. [92]J.Cohen-Adad,W.Zhao,B.Keiletal.,“7-T MRI of the Stuber, “Respiratory motion artifact suppression in dufusion- spinal cord can detect lateral corticospinal tract abnormality in weighted MR imaging of the spine,” European Radiology,vol.13, amyotrophic lateral sclerosis,” Muscle & Nerve,vol.47,no.5,pp. no. 2,pp.330–336, 2003. 760–762, 2013. [77] P. Summers, P. Staempfi, T. Jaermann, S. Kwiecinski, and S. S. [93]J.Cohen-Adad,W.Zhao,L.L.Wald,andA.L.Oaklander,“7T Kollias, “A preliminary study of the efects of trigger timing on MRI of spinal cord injury,” Neurology,vol.79,no.22,pp.2217– difusion tensor imaging of the human spinal cord,” American 2217, 2012. Journal of Neuroradiology,vol.27,no.9,pp.1952–1961, 2006. [94] J. Cohen-Adad, A. Mareyam, B. Keil, J. R. Polimeni, and L. L. [78]J.Cohen-Adad,M.-M.ElMendili,S.Lehericy´ et al., “Demyeli- Wald, “32-Channel RF coil optimized for brain and cervical nation and degeneration in the injured human spinal cord spinal cord at 3 T,” Magnetic Resonance in Medicine,vol.66,no. detected with difusion and magnetization transfer MRI,” Neu- 4,pp.1198–1208, 2011. roImage,vol.55,no.3,pp.1024–1033, 2011. [95] J. Cohen-Adad, B. Buchbinder, and A. L. Oaklander, “Cervical [79]J.Cohen-Adad,M.M.ElMendili,R.Morizot-Koutlidisetal., spinal cord injection of epidural corticosteroids: comprehensive “Involvement of spinal sensory pathway in ALS and specifcity longitudinal study including multiparametric magnetic reso- of cord atrophy to lower motor neuron degeneration,” Amy- nance imaging,” Pain,vol.25,no.6,pp.653–676, 2012. otrophic Lateral Sclerosis and Frontotemporal Degeneration,vol. [96] W.Zhao, J. Cohen-Adad, J. R. Polimeni et al., “Nineteen-channel 14,no.1,pp.30–38, 2013. receive array and four-channel transmit array coil for cervical [80]F.Agosta,M.A.Rocca,P.Valsasinaetal.,“Alongitudinal spinal cord imaging at 7T,” Magnetic Resonance in Medicine, difusion tensor MRI study of the cervical cord and brain in 2013. amyotrophic lateral sclerosis patients,” Journal of Neurology, Neurosurgery & Psychiatry,vol.80,no.1,pp.53–55, 2009. [97] M. Taso, A. le Troter, M. Sdika et al., “Construction of an in vivo human spinal cord atlas based on high-resolution MR images [81] G. Nair, J. D. Carew, S. Usher, D. Lu, X. P. Hu, and M. at cervical and thoracic levels: preliminary results,” Magnetic Benatar, “Difusion tensor imaging reveals regional diferences Resonance Materials in Physics, Biology and Medicine,pp.1–11, in the cervical spinal cord in amyotrophic lateral sclerosis,” 2013. NeuroImage,vol.53,no.2,pp.576–583, 2010. [98] R. S. Samson, O. Ciccarelli, C. Kachramanoglou et al., [82] J. Cohen-Adad, “High-resolution DWI in brain and spinal cord “Tissue and column specifcmeasurementsfrommulti with syngo RESOLVE,” MAGNETOM Flash the Magazine of parameter mapping of the human cervical spinal cord at 3 T,” MRI,vol.2,pp.16–23, 2012. NMR in− Biomedicine,vol.− 26,no.12,pp.1823–1830, 2013. − [83]J.Finsterbusch,“Improvingtheperformanceofdifusion- weighted inner feld-of-view echo-planar imaging based on 2D- [99] M. C. Yiannakas, H. Kearney, R. S. Samson et al., “Feasibility of Selective radiofrequency excitations by tilting the excitation grey matter and white matter segmentation of the upper cervical plane,” Journal of Magnetic Resonance Imaging,vol.35,no.4,pp. cord in vivo: a pilot study with application to magnetisation 984–992, 2012. transfer measurements,” NeuroImage,vol.63,no.3,pp.1054– 1059, 2012. [84] J. B. Andre, G. Zaharchuk, E. Saritas et al., “Clinical evaluation of reduced feld-of-view difusion-weighted imaging of the [100] H. Kearney, M. C. Yiannakas, K. Abdel-Aziz et al., “Improved cervical and thoracic spine and spinal cord,” American Journal MRI quantifcation of spinal cord atrophy in multiple sclerosis,” of Neuroradiology,vol.33,no.10,pp.1860–1866, 2012. Journal of Magnetic Resonance Imaging,vol.39,no.3,pp.617– 623, 2013. [85] G. Zaharchuk, E. U. Saritas, J. B. Andre et al., “Reduced feld-of- view difusion imaging of the human spinal cord: comparison [101]P.Valsasina,M.A.Horsfeld, M. A. Rocca, M. Absinta, G. Comi, with conventional single-shot echo-planar imaging,” American and M. Filippi, “Spatial normalization and regional assessment Journal of Neuroradiology,vol.32,no.5,pp.813–820, 2011. of cord atrophy: voxel-based analysis of cervical cord 3DT1- [86]D.A.PorterandR.M.Heidemann,“Highresolutiondifusion- weighted images,” American Journal of Neuroradiology,vol.33, weighted imaging using readout-segmented echo-planar imag- no. 11,pp.2195–2200, 2012. ing, parallel imaging and a two-dimensional navigator-based [102] M. M. El Mendili, R. Chen, B. Tiret et al., “Validation of a reacquisition,” Magnetic Resonance in Medicine,vol.62,no.2, semiautomated spinal cord segmentation method,” Journal of pp. 468–475, 2009. Magnetic Resonance Imaging, 2014. [87]D.C.vanEssenandK.Ugurbil,“Te future of the human [103] B. B. Avants, N. J. Tustison, G. Song, P. A. Cook, A. Klein, and connectome,” NeuroImage,vol.62,no.4,pp.2222–2231, 2012. J. C. Gee, “Areproducible evaluation of ANTs similarity metric [88]J.A.McNab,B.L.Edlow,T.Witzeletal.,“Tehuman performance in brain image registration,” NeuroImage,vol.54, connectome project and beyond: initial applications of 300 no. 3,pp.2033–2044, 2011. mT/m gradients,” NeuroImage,vol.80,pp.234–245, 2013. [104] C. Konrad, H. Henningsen, J. Bremer et al., “Pattern of cor- [89] U. Ferizi, T. Schneider, E. Panagiotaki et al., “A ranking of tical reorganization in amyotrophic lateral sclerosis: a func- difusion MRI compartment models with in vivo human brain tional magnetic resonance imaging study,” Experimental Brain data,” Magnetic Resonance in Medicine, 2013. Research,vol.143,no.1,pp.51–56, 2002. [90] D. C. Alexander, P. L. Hubbard, M. G. Hall et al., “Orienta- [105] C. Tempelmann, C. Gaul, G. R. Kuhnel¨ et al., “Functional tionally invariant indices of axon diameter and density from motor compensation in amyotrophic lateral sclerosis,” Journal difusion MRI,” NeuroImage,vol.52,no.4,pp.1374–1389, 2010. of Neurology,vol.252,no.8,pp.944–952, 2005. [91] J. R. Korzan, M. Gorassini, D. Emery, Z. A. Taher, and C. [106] A. Poujois, F. C. Schneider, I. Faillenot et al., “Brain plasticity Beaulieu, “In vivo magnetic resonance imaging of the human in the motor network is correlated with disease progression in cervical spinal cord at 3 Tesla,” Journal of Magnetic Resonance amyotrophic lateral sclerosis,” Human Brain Mapping,vol.34, Imaging,vol.16,no.1,pp.21–27, 2002. no. 10,pp.2391–2401, 2012. BioMed Research International 9

[107] M. D. Greicius, G. Srivastava, A. L. Reiss, and V. Menon, risk for familial ALS,” Neurology,vol.77,no.14,pp.1370–1375, “Default-mode network activity distinguishes Alzheimer’s dis- 2011. ease from healthy aging: evidence from functional MRI,” [123] H. Zhu, R. A. E. Edden, R. Ouwerkerk, and P. B. Barker, “High Proceedings of the National Academy of Sciences of the United resolution spectroscopic imaging of GABA at 3 Tesla,” Magnetic States of America,vol.101,no.13,pp.4637–4642, 2004. Resonance in Medicine,vol.65,no.3,pp.603–609, 2011. [108] F. Agosta, P. Valsasina, M. Absinta et al., “Sensorimotor func- [124] B. R. Foerster, B. C. Callaghan, M. Petrou, R. A. E. Edden, T. tional connectivity changes in amyotrophic lateral sclerosis,” L. Chenevert, and E. L. Feldman, “Decreased motor cortex - Cerebral Cortex,vol.21,no.10,pp.2291–2298, 2011. aminobutyric acid in amyotrophic lateral sclerosis,” Neurology, [109] G. Douaud, N. Filippini, S. Knight, K. Talbot, and M. R. Turner, vol. 78,no.20,pp.1596–1600, 2012. $ “Integration of structural and functional magnetic resonance [125] M. Waragai, Y.Takaya, and M. Hayashi, “Serial MRI and SPECT imaging in amyotrophic lateral sclerosis,” Brain,vol.134,no.12, in amyotrophic lateral sclerosis: a case report,” Journal of the pp. 3470–3479, 2011. Neurological Sciences,vol.148,no.1,pp.117–120, 1997. [110] E. Verstraete, M. P. van den Heuvel, J. H. Veldink et al., [126] G. Waldemar, S. Vorstrup, T. Staehelin Jensen, A. Johnsen, “Motor network degeneration in amyotrophic lateral sclerosis: and G. Boysen, “Focal reductions of cerebral blood fow in a structural and functional connectivity study,” PLoS ONE,vol. amyotrophic lateral sclerosis: a [99mTc]-d,l-HMPAO SPECT 5,no.10, Article ID e13664, 2010. study,” Journal of the Neurological Sciences,vol.107,no.1,pp. [111] L. M. Jelsone-Swain, B. W. Fling, R. D. Seidler, R. Hovatter, K. 19–28, 1991. Gruis, and R. C. Welsh, “Reduced interhemispheric functional [127] K. Abe, S. Yorifuji, and Y. Nishikawa, “Reduced isotope uptake connectivity in the motor cortex during rest in limb-onset restricted to the motor area in patients with amyotrophic lateral amyotrophic lateral sclerosis,” Frontiers in Systems Neuroscience, sclerosis,” Neuroradiology,vol.35,no.6,pp.410–411, 1993. vol. 4, article 158, 2010. [128] S. Kalra and D. Arnold, “Neuroimaging in amyotrophic lateral [112]F.Agosta,E.Canu,P.Valsasinaetal.,“Divergentbrainnetwork sclerosis,” Amyotrophic Lateral Sclerosis,vol.4,no.4,pp.243– connectivity in amyotrophic lateral sclerosis,” Neurobiology of 248, 2003. Aging,vol.34,no.2,pp.419–427, 2013. [129] M.-O. Habert, L. Lacomblez, P. Maksud, G. E. L. Fakhri, [113]K.Kollewe,S.Korner,¨ R. Dengler, S. Petri, and B. Mohammadi, J.-F. Pradat, and V. Meininger, “Brain perfusion imaging in “Magnetic resonance imaging in amyotrophic lateral sclerosis,” amyotrophic lateral sclerosis: extent of cortical changes accord- Neurology Research International,vol.2012, Article ID 608501, ing to the severity and topography of motor impairment,” 9 pages, 2012. Amyotrophic Lateral Sclerosis,vol.8,no.1,pp.9–15, 2007. [114] C. Luo, Q. Chen, R. Huang et al., “Patterns of spontaneous brain [130] J. M. Hofman, J. C. Mazziotta, T. C. Hawk, and R. Sumida, activity in amyotrophic lateral sclerosis: a resting-state fMRI “Cerebral glucose utilization in motor neuron disease,” Archives study,” PloS ONE,vol.7,no.9, Article ID e45470, 2012. of Neurology,vol.49,no.8,pp.849–854, 1992. [131]J.Hatazawa,R.A.Brooks,M.C.Dalakas,L.Mansi,andG. [115]C.Pohl,W.Block,J.Karitzkyetal.,“Protonmagneticreso- Di Chiro, “Cortical motor-sensory hypometabolism in amy- nance spectroscopy of the motor cortex in 70 patients with otrophic lateral sclerosis: a PET study,” Journal of Computer amyotrophic lateral sclerosis,” Archives of Neurology,vol.58,no. Assisted Tomography,vol.12,no.4,pp.630–636, 1988. 5,pp.729–735, 2001. [132]J.J.M.Kew,P.N.Leigh,E.D.Playfordetal.,“Cortical [116] P. Sarchielli, G. P. Pelliccioli, R. Tarducci et al., “Magnetic function in amyotrophic lateral sclerosis. A positron emission resonance imaging and 1H-magnetic resonance spectroscopy in tomography study,” Brain,vol.116,no.3,pp.655–680, 1993. amyotrophic lateral sclerosis,” Neuroradiology,vol.43,no.3,pp. 189–197, 2001. [133]A.Cistaro,M.C.Valentini,A.Chio` et al., “Brain hyperme- tabolism in amyotrophic lateral sclerosis: a FDG PET study in [117] J. Suhy, R. G. Miller, R. Rule et al., “Early detection and ALS of spinal and bulbar onset,” European Journal of Nuclear longitudinal changes in amyotrophic lateral sclerosis by 1H Medicine and Molecular Imaging,vol.39,no.2,pp.251–259, 2012. MRSI,” Neurology,vol.58,no.5,pp.773–779, 2002. [134]A.Cistaro,M.Pagani,A.Montuschietal.,“Temetabolic [118]O.Gredal,S.Rosenbaum,S.Topp,M.Karlsborg,P.Strange, signature of C9ORF72-related ALS: FDG PET comparison with and L. Werdelin, “Quantifcation of brain metabolites in amy- nonmutated patients,” European Journal of Nuclear Medicine otrophic lateral sclerosis by localized proton magnetic reso- and Molecular Imaging,pp.1–9, 2014. nance spectroscopy,” Neurology,vol.48,no.4,pp.878–881, 1997. [135] C. M. Lloyd, M. P. Richardson, D. J. Brooks, A. Al-Chalabi, and [119] E. P. Pioro, J. P. Antel, N. R. Cashman, and D. L. Arnold, P. N. Leigh, “Extramotor involvement in ALS: PET studies with “Detection of cortical neuron loss in motor neuron disease the GABA(A) ligand [11C]fumazenil,” Brain,vol.123,no.11,pp. by proton magnetic resonance spectroscopic imaging in vivo,” 2289–2296, 2000. Neurology,vol.44,no.10,pp.1933–1938, 1994. [136] M. R. Turner, A. Hammers, A. Al-Chalabi et al., “Distinct [120]S.Kalra,C.C.Hanstock,W.R.W.Martin,P.S.Allen,and cerebral lesions in sporadic and ’D90A’ S O D 1 ALS: studies with W. S. Johnston, “Detection of cerebral degeneration in amy- [11C]fumazenil PET,” Brain,vol.128,no.6,pp.1323–1329, 2005. otrophic lateral sclerosis using high-feld magnetic resonance [137] S. Przedborski, V. Dhawan, D. M. Donaldson et al., “Nigros- spectroscopy,” Archives of Neurology,vol.63,no.8,pp.1144– triatal dopaminergic function in familial amyotrophic lateral 1148, 2006. sclerosis patients with and without copper/zinc superoxide [121] C. J. Stagg, S. Knight, K. Talbot, M. Jenkinson, A. A. Maudsley, dismutase mutations,” Neurology,vol.47,no.6,pp.1546–1551, and M. R. Turner, “Whole-brain magnetic resonance spec- 1996. troscopic imaging measures are related to disability in ALS,” [138] G. D. Borasio, R. Linke, J. Schwarz et al., “Dopaminergic defcit Neurology,vol.80,no.7,pp.610–615, 2013. in amyotrophic lateral sclerosis assessed with [I-123]IPTsingle [122] J. D. Carew, G. Nair, P. M. Andersen et al., “Presymptomatic photon emission computed tomography,” Journal of Neurology spinal cord neurometabolic fndings in SOD1-positive people at Neurosurgery & Psychiatry,vol.65,no.2,pp.263–265, 1998. 10 BioMed Research International

[139] M. R. Turner, E. A. Rabiner, A. Hammers et al., “[11C]- WAY100635 PET demonstrates marked 5-HT1Areceptor changes in sporadic ALS,” Brain,vol.128,no.4,pp.896–905, 2005. [140]F.Chauveau,N.vanCamp,F.Dolleetal.,“Comparative´ evaluation of the translocator protein radioligands 11C-DPA- 713, 18F-DPA-714,and11C-PK11195 in a rat model of acute neuroinfammation,” Journal of Nuclear Medicine,vol.50,no. 3,pp.468–476, 2009. [141] P. Corcia, C. Tauber, J. Vercoullie et al., “Molecular imaging of microglial activation in amyotrophic lateral sclerosis,” PLoS ONE,vol.7,no.12, Article ID e52941, 2012.

Annex B

Grolez G, Moreau C, Danel-Brunaud V, Delmaire C, Lopes R, Pradat PF, El Mendili MM, Defebvre L and Devos D

2015

The value of magnetic resonance imaging as a biomarker for amyotrophic lateral sclerosis: a systematic review

BMC Neurology

- 124 - Grolez et al. BMC Neurology (2016) 16:155 DOI 10.1186/s12883-016-0672-6

RESEARCHARTICLE Open Access The value of magnetic resonance imaging as a biomarker for amyotrophic lateral sclerosis: a systematic review G. Grolez1,2 , C. Moreau1,2, V. Danel-Brunaud1,2, C. Delmaire2,3, R. Lopes2,3, P. F. Pradat4,5, M. M. El Mendili4, L. Defebvre1,2 and D. Devos1,2,6*

Abstract Background: Amyotrophic lateral sclerosis (ALS) is a fatal, rapidly progressive neurodegenerative disease that mainly affects the motor system. A number of potentially neuroprotective and neurorestorative disease-modifying drugs are currently in clinical development. At present, the evaluation of a drug’s clinical efficacy in ALS is based on the ALS Functional Rating Scale Revised, motor tests and survival. However, these endpoints are general, variable and late-stage measures of the ALS disease process and thus require the long-term assessment of large cohorts. Hence, there is a need for more sensitive radiological biomarkers. Various sequences for magnetic resonance imaging (MRI) of the brain and spinal cord have may have value as surrogate biomarkers for use in future clinical trials. Here, we review the MRI findings in ALS, their clinical correlations, and their limitations and potential role as biomarkers. Methods: The PubMed database was screened to identify studies using MRI in ALS. We included general MRI studies with a control group and an ALS group and longitudinal studies even if a control group was lacking. Results: A total of 116 studies were analysed with MRI data and clinical correlations. The most disease-sensitive MRI patterns are in motor regions but the brain is more broadly affected. Conclusion: Despite the existing MRI biomarkers, there is a need for large cohorts with long term MRI and clinical follow-up. MRI assessment could be improved by standardized MRI protocols with multicentre studies. Keywords: Amyotrophic lateral sclerosis, Magnetic resonance imaging, Morphometry, Diffusion tensor imaging, Magnetic resonance spectroscopy, Spinal cord, Biomarkers Abbreviations: AD, Axial diffusivity; ADC, Apparent diffusion coefficient; ALFF, Amplitude of low frequency fluctuations; ALS, Amyotrophic lateral sclerosis; ALSFRS, Amyotrophic lateral sclerosis functional rating scale; ALSFRS- R, Amyotrophic lateral sclerosis functional rating scale revised; CAFS, Combined assessment of function and survival; Cho, Choline; Cr, Creatine; CSF, Cerebro spinal fluid; CST(s), Cortico-spinal tract(s); DTI, Diffusion tensor imaging; FA, Fractional anisotropy; fMRI, Functional magnetic resonance imaging; GABA, Gamma aminobutyric acid; Gln, Glutamine; Glu, Glutamate; Glx, Glutamine and glutamate; Ins, myo-inositol; MD, Mean diffusivity; MRI, Magnetic resonance imaging; MRS, Magnetic resonance spectroscopy; NAA, N-acetyl-aspartate; NiALS, Neurimaging symposium in amyotrophic lateral sclerosis; Pcr, Phosphocreatine; PD, Parkinson’s disease; PLIC, Posterior limb of internal capsule; QSM, Quantitative susceptibility mapping; RD, Radial diffusivity; SWI, Susceptibility weighted imaging; VBM, Voxel-based morphometry

* Correspondence: [email protected] 1Department of Movement Disorders and Neurology, Lille University Hospital, Faculty of Medicine, University of Lille, Lille, France 2INSERM U1171, Lille University Hospital, Faculty of Medicine, University of Lille, Lille, France Full list of author information is available at the end of the article

© 2016 The Author(s). Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Grolez et al. BMC Neurology (2016) 16:155 Page 2 of 17

Background biomarkers in ALS is facilitated by the hypothesis that the Amyotrophic lateral sclerosis (ALS) is a neurodegenera- pathological changes in ALS radiate out from the initial tive disease that mainly affects the motor system. At spinal and brain areas in which the most severe upper and present, the only drug to have produced an increase in lower motor neuron dysfunction respectively occur (from patient survival in a controlled clinical trial is riluzole symptom onset onwards) with variability in the predomin- [1]. The condition is always fatal; the median survival ance of upper and lower motor neuron dysfunction [13]. time after onset is 36 [2], although there are great indi- The neurodegeneration then progresses along the motor vidual variations [3]. Although ALS is a clinically neurons of the ventral horn of the spinal cord and into recognizable condition (in terms of the pattern of pro- the corresponding areas of the motor cortex [13]. Despite gressive upper and lower motor neuron degeneration), this modelization, the pattern of onset and progression re- the clinical presentation and progression are heteroge- mains complex [13]. Hence, MRI of the spinal cord should neous [4]. Moreover, there are now strong reasons for be an early, sensitive marker of subtle degeneration- considering a broader aetiological and pathogenic related changes. However, MRI of the spinal cord is com- spectrum [5]. This clinical heterogeneity is observed in plicated by a number of technical difficulties (relative to other neurodegenerative diseases, such as Parkinson’s MRI of the brain). Firstly, the magnetic field is in- disease (PD). Nevertheless, it has been found that all PD homogeneous because the spinal cord is close to patients display degeneration and iron overload of the bone, soft tissues and air. Secondly, the spine has a substantia nigra, which might become a surrogate bio- small cross-sectional dimension, which thus requires marker [6]. In ALS, iron overload in the motor cortex high-resolution data acquisition. Thirdly, the spine is was confirmed [7] after being suspected in mouse subject to motion of the CSF [14] and physiological models [8]. Ongoing drug trials in the field of neurode- motion due to breathing [15]. generative disease are mainly seeking to establish neuro- Many different MRI techniques and sequences – in- protection and neurorestoration. Since neuroprotection cluding voxel-based morphometry (VBM), diffusion ten- can never be unambiguously demonstrated in a given sor imaging (DTI), magnetic resonance spectroscopy patient, the concept of disease-modifying drugs (with a (MRS), iron-sensitive sequences (T2*, R2* and SWI) and slowing of the disease progression) has arisen. At functional MRI (fMRI) - are available for studying ALS- present, the evaluation of a drug’s clinical efficacy in related changes in the brain or spinal cord. Hence, we ALS is based on the ALS Functional Rating Scale Re- reviewed the correlations between MRI findings and vised, motor tests, survival or a combination of these clinical scores in ALS, and sought to establish whether measures (such as the Combined Assessment of Func- particular sequences may have value as surrogate bio- tion and Survival (CAFS)) [9, 10]. However, these end- markers in this disease. points are general, variable and late-stage measures of the ALS disease process and thus require the long-term Methods assessment of large cohorts. These assessments are risky We searched the PubMed database with the following and expensive, which considerably limits the number of keywords: “ALS AND MRI OR amyotrophic lateral scler- trials being conducted. A sensitive surrogate biomarker osis AND magnetic resonance imaging”, “ALS AND VBM might help to (i) reduce the sample size in pilot studies, OR amyotrophic lateral sclerosis AND magnetic reson- (ii) better define the sample size required in Phase III ance imaging ”, “ALS AND MR spectroscopy OR amyo- clinical trials (by comparison with clinical scales) and trophic lateral sclerosis AND magnetic resonance (iii) define endophenotypes for separate assessment in spectroscopy”, “amyotrophic lateral sclerosis AND diffu- clinical trials. sion tensor imaging OR ALS AND DTI”, “amyotrophic Biomarkers are typically divided into “wet biomarkers” lateral sclerosis AND functional magnetic resonance im- and “dry biomarkers”. Wet biomarkers are biological sub- aging OR ALS AND fMRI”. There was no limitation on stances that are measured in a body fluid (such as whole the publication date but only publications in French or blood, serum, plasma, saliva or cerebrospinal fluid (CSF)). English were considered. We included (i) general MRI “Dry biomarkers” are based on functional scales, task per- studies with a control group and an ALS group and (ii) formance, electrophysiology or imaging. Magnetic reson- longitudinal follow-up studies (even if a control group was ance imaging is a widely available and non-invasive lacking). We excluded therapeutic trials articles, reviews, technique, which makes it an appropriate candidate for studies focused only on particular subgroups of ALS biomarkers. Although early reports stated that ALS starts patients such as cognitive impairment, genetic mutations. in the spinal cord, the initial site of neurodegeneration has not been unambiguously identified; disease may start in Results the spinal cord, in the motor cortex or at both sites simul- A total of 1336 articles were identified using database taneously [11, 12]. The development of radiological searching, 651 were recorded after duplicates removal. Grolez et al. BMC Neurology (2016) 16:155 Page 3 of 17

405 were excluded (not ALS, not MRI or, therapeutic or ALSFRS-R score, divided by the disease duration or clinical trials, not English or French, reviews) and 130 the time interval between two clinical assessments), the articles were finally excluded (studies focused only on disease duration and the survival time. Lower ALSFRS particulars subgroups of patients, other MRI sequences or ALSFRS-R scores were associated with larger ventri- or lacking of a control group). We analysed 116 articles. cles and greater volume loss in the basal ganglia [38], and volume changes in several parts of the frontal lobe Brain imaging (such as Brodman area 10) [18]. In another study, a Structural magnetic resonance imaging (Table 1) higher ALSFRS-R progression rate was associated with Three methods based on a three-dimensional, T1- MRI changes in the left sensorimotor area; cortical weighted sequence have been applied to ALS: VBM, vol- thickness was lower in patients with predominantly ume analysis and the measurement of cortical thickness. upper motor neuron impairment (especially in the pre- As a volume-based analysis, VBM involves the automatic central gyrus and the left paracentral lobule), whereas segmentation of grey matter and inter-subject compari- patients with predominantly lower motor neuron im- son of the local grey matter density [16, 17]. After all the pairment showed less difference relative to controls [20]. data have been transformed in the same stereotaxic Grey matter density was lower in the cingulate and right space, the images are partitioned and corrected for the inferior frontal gyrus in patients with impaired emo- separation between white matter, grey matter and CSF tional empathy [36]. Poor survival in ALS is correlated as a function of voxel intensity [17]. Measurement of with volume loss in the basal ganglia and limbic struc- cortical thickness is a surface-based technique analysis tures [38]. The initial side of limb weakness in ALS in which the white matter and pial surfaces are automat- patients was not correlated with contralateral brain cor- ically extracted by applying image analysis packages such tical loss of volume [27]. Although atrophy was predom- as Freesurfer (http://surfer.nmr.mgh.harvard.edu/). inant in the left sensorimotor cortex [27, 28], it was not Most of these techniques have evidenced atrophy in the related to the side of onset [27]. A subgroup analysis (by precentral gyri [18–32]. However, some DTI studies used side of onset) found more widespread atrophy in the VBM to show that ALS vs. control differences in the frac- contralateral brain cortex than in the group of ALS pa- tional anisotropy (FA) of the precentral gyri were not due tients as a whole [28]. Precentral cortical thickness was to atrophy [33, 34]. In other studies, cortical atrophy was lower in patients with upper motor neuron clinical not limited to the precentral gyri but extended to other re- features than in those with lower motor neuron clinical gions of the frontal lobe [18–24, 26–28, 32, 35] - especially features and in the ALS group as a whole [29]. in apathetic patients with involvement of the cingulate In a study with a post mortem T1 sequence and histo- cortex [27, 36]. Other regions involved include several logical assessment, an abnormally small difference in in- parts of the temporal cortex [19–24, 32, 35], the hippo- tensity between cortical grey matter and subcortical campus [24, 37–39], the parietal cortex (mostly in the white matter was observed not only in the motor cortex post-central cortex [18, 21, 23, 28]) and the insula [24]. (including the precentral gyrus, supplemental motor area Occipital grey matter atrophy [19, 20, 24, 35] and cerebel- and premotor cortex) but also in somatosensory areas lar atrophy were less common [24]. When considering the and the primary visual cortex. These signal abnormalities basal ganglia, a low volume of the thalamus [22, 24, 37] were linked to neuron loss and an elevated number of and the caudate nucleus [37] have been observed. Few astrocytes [41]. studies have evidenced a relative reduction in the volume of white matter (mainly in the frontal and temporal lobes) Diffusion tensor imaging (Table 2) [23]. There were no marked differences in the corpus cal- DTI (also known as magnetic resonance tractography) is losum [40]. Whole-brain analyses have evidenced low vol- based on the random diffusion (Brownian motion) of umes in the neocortex, grey matter, white matter [28] and molecules. In a spherical volume, the diffusion of water brain parenchyma [28, 32]. A longitudinal study showed has no main direction and its diffusion in the three di- (i) a volume reduction in the left subiculum, the right rections (λ1, λ2 and λ3) is equally likely. However, this is hippocampal area CA4 and the right dentatus gyrus and no longer the case in an ellipsoid or cylindrical volume (ii) an enlargement of the ventricles in 39 of the 112 ALS and the diffusion is anisotropic. λ1 is the volume’s main patients [38]. axis (axial diffusivity, AD) and λ2 and λ3 are the minor axes (radial diffusivity, RD). By analysing the diffusion of Clinical correlations Specific MRI abnormalities have water in three directions, four parameters can be de- been correlated with clinical scores, the disease stage (as fined: FA, RD, AD and mean diffusivity (MD, which is estimated with the original or revised ALS Functional the average of diffusion in the λ1, λ2 and λ3 axis, also Rating Scale (ALSFRS and ALSFRS-R, respectively), the referred to as the apparent diffusion coefficient (ADC)). speed of disease progression (i.e. the change in ALSFRS The first three parameters describe the spatial variation Grolez et al. BMC Neurology (2016) 16:155 Page 4 of 17

Table 1 The main results of volumetric studies Publication Numbers of Main results Main clinical correlations patients/controls Abdulla et al., 58/29 Low volume in the right hippocampus Verbal memory test performance is correlated [39, 80] with the left hippocampal volume. Agosta et al., 25/18 GM volume reduction in right precentral, left Negative with ALSFRS-R and DD. [21] inferior frontal cortex and temporal superior cortex. Agosta et al., 44/26 Reduction in cortical thickness in the precentral, DPR is correlated with the mean cortical thickeness [35] frontal, limbic, parietal, temporal and occipital lobes. of left sensorimotor cortex. Bede et al., 39/44 GM volume reduction in the thalamus, caudate Not available. [37] nucleus, hippocampus, left putamen. Canu et al., 23/24 GM volume reduction in the precentral, right Not available. [23] opercular and angular cortex, WM volume reduction in the frontal lobe (especially the subcortical motor areas) and the temporal lobe. Cerami et al., 14/20 Low GM density in the anterior cingulate and Not available. [36] right inferior frontal gyri. Chang et al., 20 (10 ALS and GM volume reduction in the precentral, frontal and Not available. [22] 10 ALS-FTD)/22 temporal cortex and left posterior thalamus. Chapman et al., 25/22 No intergroup difference in the corpus Not available. [40] callosum area Devine et al., 30/17 GM volume reduction in the left precentral cortex, Increased GM volume reduction in the dominant [27] left frontal gyrus, medial frontal gyri and anterior (left) motor cortex irrespective to the side of limb cingulate gyri. onset. Asymmetric GM reduction of the left somatosensory cortex and the temporal gyri in in right limb onset ALS Ellis et al., [26] 16 (8 bulbar onset GM volume reduction in the superior, middle and Compared to limb onset ALS, bulbar onset ALS and 8 limb onset)/8 medial frontal cortex. WM volume reduction in the showed GM volume reduction in the brainstem, the right frontal cortex. cerebellum and in the fusiform gyri (BA 34) and WM volume reduction along the left corticospinal tract. Grosskreutz et al., 17/17 GM volume reduction in the precentral, frontal ALSFRS-R correlated positively with GM volume [18] and parietal cortex. No WM atrophy. reduction in the right medial frontal gyrus (BA 10). Kassubek et al., 22/22 GM volume reduction in the right primary motor Negative with ALSFRS-R and DD. [32] cortex and the left medial gyrus and bilateral inferior temporal gyrus. Meadcroft et al., 8/6 Post mortem study; loss of signal difference Not available. 2014 between WM and GM (mainly in the primary motor cortex). Mezzapesa et al., 16/9 GM volume reduction in the precentral, frontal Not available. [19] and parietal cortex. Mezzapesa et al., 29/20 GM volume reduction in the precentral, frontal, Different patterns of volume reduction in ALS [20] right temporal and right occipital cortex. group, high UMN burden, spinal onset and faster progression compared to controls. Sach et al., [33] 15/12 No difference -study not designed to voxel-based Not available. morphometry. Sage et al., [34] 28/26 No difference -study not designed to voxel-based Not available. morphometry. Schuster et al., 93 (60 ALS, 17 with Correlations between the site of onset and upper Different patterns of volume reduction in ALS [25] dominant UMN, 16 vs. lower motor neuron involvement, with focal subgroups: bulbar and spinal UMN signs, only with dominant reductions in cortical thickness in the precentral spinal UMN signs, classical ALS, UMN ALS-variants, LMN)/67 cortex. LMN ALS-variants or sites of onset compared to controls. GM reduction is most important in the left precentral cortex in bulbar onset vs. limb onset. No correlations of MRI with DPR, DD and ALSFRS-R. Thivard et al., 15/25 GM volume reduction in the precentral, frontal Not available. [24] and temporal cortex (especially the hippocampus), the parietal and occipital cortex, the thalamus and the cerebellum. No difference in WM. Verstraete et al., 12/12 GM reduction in the right and left precentral cortex. Not available. [30] Grolez et al. BMC Neurology (2016) 16:155 Page 5 of 17

Table 1 The main results of volumetric studies (Continued) Walhout et al., 153 (112 ALS, 19 GM reduction in the right and (predominantly) left Bulbar and arms ALSFRS-R subscores are correlated [29] UMN, 19 LMN)/60 precentral cortex and the right paracentral cortex. with cortical thickness of corresponding precentral cortex areas of the motor homunculus. DPR is correlated with the thickness of the right inferior temporal cortex, the postcentral cortex and the right paracentral cortex. Westenberg 112/60 GM volume reduction in the hippocampus and Larger ventricles are correlated to a lower ALSFRS-R et al., 2014 left subiculum. score. Smaller basal ganglia, smaller limbic structures and larger ventricles are associated with shorter survival. Zhang et al., [28] 43/43 GM reduction in the left precentral cortex, left GM volume reduction is most important in the motor supplemental area and left postcentral gyrus. cortex of the contralateral hemisphere of the limb of Reduction in neocortex volume, GM and WM onset. ALSFRS is positively correlated with GM density volume and brain parenchymal fraction. in the left postcentral cortex and DPR is negatively correlated with GM density in the right precentral cortex Zhu et al., 22/22 GM reduction in the right and left precentral cortex. Negative with DD, DPR and all other clinical [31, 87] parameters. GM grey matter, WM white matter, BA Brodman area, DD disease duration, DPR Disease progression rate, UMN upper motor neuron of water movement and are related to the orientation of staging system [72], suggesting that the disease process ex- the studied structures. In contrast, MD corresponds to tended in the following sequence: the corticospinal tract the mean displacement of the water molecules within (stage 1), the corticorubral and corticopontine tracts the volume. The main application of DTI is the analysis (stage 2), the corticostriatal tract (stage 3) and the prox- of the anatomical bundles that compose the white mat- imal portion of the perforant pathway (stage 4) [73]. The ter of the brain and the spinal cord [42–44]. Whereas researchers suggested that this MRI-based staging system DTI corresponds to the analysis of the above-mentioned could be used as a surrogate marker of disease progression parameters in a defined volume, tractography corre- in clinical studies of ALS. sponds to the analysis of a complete white matter tract. Diffusivity in ALS The most common observation in Fractional anisotropy in ALS Low FA along entire cor- patients with ALS is a relative increase in MD, reflecting ticospinal tracts (CSTs) is the most common observation changes in white matter tracts. The MD was elevated in patients with ALS [34, 45–58]. However, some studies along entire CSTs [51, 53, 69] and in parts of CSTs, in- have only reported low FA in one or several parts of a cluding subcortical white matter in the precentral gyrus CST: the subcortical white matter of the precentral gyrus [23, 34, 58, 60, 61], the corona radiata [34, 60], the PLIC [23, 24, 30, 33, 59–65], the corona radiata [24, 60–62, [23, 34, 60, 74], the cerebral peduncles and the pons 66–68], the posterior limb of the internal capsule (PLIC) [34]. Diffusivity was also elevated in several parts of the [24, 33, 60, 62, 63], the cerebral peduncles [21, 59, 61] frontal lobe [23, 24, 52, 58, 60, 61], the temporal lobe and the pons [61]. Lastly, only one study failed to ob- [23, 24, 52, 61], the hippocampal region [24, 34, 52], serve a difference in FA along a CST in ALS patients by the parahippocampal region [60], the parietal lobe region of interest approach whereas tract based spatial [23, 58, 61], the insula [23, 24, 52], the cingulum [53, statistics (TBSS) showed decreased FA in corona radiate 60], the occipital lobe [23], the cerebellum [23, 60] and corpus callosum [68]. (especially in C9ORF72-positive patients) [75], the Low FA has also been observed in other regions: the corpus callosum [23, 52, 60, 69] and the thalamus [53]. frontal lobe (excluding the CST) [21, 24, 33, 34, 45, 52, 56, Conversely, two studies did not detect any differences in 57, 61, 64], the cingulum [60], the corpus callosum [21, diffusivity in ALS patients vs. controls [21, 63]. 33, 45, 50, 52, 58, 61, 67–69], the parietal lobe [24, 34, 45], the temporal lobe [21], parahippocampal areas [61], the hippocampus [34], the insula [34, 52], the cerebellum [61] Radial and axial diffusivity in ALS Several studies have and the thalamus [24, 33, 53]. In a whole-brain compari- found elevated RD and AD values in the right CST, the son, many other small areas of ALS-modified white matter right cingulum and the left anterior thalamic radiations tracts were observed [70]. (on the basis of changes in FA or MD) [53]. In the PLIC, A tractography study revealed that the CST volumes of AD is low [63] and RD is elevated [67] - especially in pa- ALS patients are lower than in controls [71]. A DTI study tients presenting with T2 hyperintensities in CSTs [63]. using a tract of interest-based fibre tracking found that RD is elevated in white matter connected to the primary the radiological results mirrored the neuropathological motor cortex, the premotor cortex [56] and the corpus Table 2 The main results of DTI studies Grolez Publication Numbers of Fractional anisotropy Mean diffusivity or apparent Radial and axial diffusivity Main clinical correlations patients/controls diffusion coefficient Neurology BMC al. et Abe et al., [64] 7/11 Low in parts of the right frontal Not available lobe and in the left subcortical precentral area Agosta et al., [21] 25/18 Low in parts of CSTs, the CC, No difference Negative parts of the frontal lobe and parts of the temporal lobe (2016) 16:155 Bede et al., [75] 27/42 Low along CSTs, the CC, RD was low in the Not available cerebellum, brainstem, occipital cerebellum, brainstem, lobe and opercular and insular occipital lobe and opercular regions and insular regions Canu et al., [23] 23/14 Low in subcortical parts of CSTs Elevated in parts of the CSTs, left Correlation between MD in bilateral and other parts of the frontal postcentral gyrus, left insula, parts of the orbitofrontal region and DD lobe left temporal lobe, right angular cortex, parts of the frontal lobe, CC, parts of the occipital lobe and parts of the cerebellum Chapman et al., [67] 21/21 Low in the CC, corona radiata RD was low in the CC, Correlation between FA in parts of CC and and PLIC corona radiate and PLIC. AD ALSFRS-R and negative correlation with DD, was elevated in the left correlation between RD in parts of CC and corona radiata and the DD internal and external capsules Ciccarelli et al., [45] 26/41 Low along CSTs, the CC, anterior Correlation between DPR and FA in left limb of the internal capsule, cerebral peduncle, right PLIC, right corona external capsule, parts of frontal radiate right WM adjascent to the lobe WM and postcentral gyri precentral gyrus and CC Ding et al., [74]10/10 ElevatedinthePLIC Notavailable Ellis et al., [51] 22/20 Low along CSTs Elevated along CSTs Correlation between mean diffusivity in CST and DD and between FA along CSTs and ALS severity scale and spasticity scales. Filippini et al., [56] 24/24 Low along CSTs and in the CC Elevated RD in WM linked Inverse correlation between FA and UMN to the primary motor and score along CSTs, correlation between FA premotor cortex and the CC and ALSFRS-R and DD along CSTs Foerster et al., [55] 29/30 Low along CSTs Correlation between FA and ALSFRS-R along CSTs Iwata et al., [49] 31/31 Low along CSTs Negative Iwata et al., [50] 18/19 Low along CSTs and in the Elevated along CSTs Correlation between DD and FA along motor part of the CC CSTs. Inverse correlation between FA in CSTs and global and localized UMN

impairment score and between FA along 17 of 6 Page CST and UMN rapidity index. Kassubek et al., [12, 73] 111/74 Low along CSTs Not available Table 2 The main results of DTI studies (Continued) Grolez

Keil et al., [61] 24/24 Low in parts of CSTs, Elevated in motor areas, parts of CSTs, parts Correlation between FA along CSTs and Neurology BMC al. et supplementary motor area, CC, of the frontal and temporal lobe, and the ALSFRS-R. Correlation between executives parts of the frontal lobe and the postcentral gyrus functions (sfs36 score) and FA in cerebel- parahippocampal area lum. At 6 months, negative correlation between FA along CSTs and DD correlation between FA along CSTs and frontal WM and ALSFRS-R. Correlation between FA in CST at brainstem level and executives func-

tions, negative correlation between ADC in (2016) 16:155 the cerebellum and parahippocampal gyri and executives functions (sfs36) Keller et al., [68] 33/30 Low in the corona radiata and Negative CC Liu et al., 2014 19/13 Low along CSTs Correlation between FA along left CST and ALSFRS-R Menke et al., [77] 21/0 (follow-up Progressive reduction in the Correlation between FA at baseline and DPR study) PLIC Metwalli et al., [69]12/19LowalongCSTs,theCCElevatedalongCSTs,theCC ADandRDwereelevatedNegative along CSTs, RD was elevated in the CC, parts of frontal and parietal lobe Muller et al., 2011 19/19 Low in parts of CSTs, the Correlation between FA in parts of CSTs parahippocampal area, insula and ALSFRS-R and brainstem Nickerson et al., [46]2/0(follow-upAlinearreductionalongCSTs Not available study) during one year follow-up Poujois et al., [65] 19/21 Low in CSTs from the left Muscular strength is lower on the right side corona radiate to the precentral corresponding to the lower FA in the left gyrus and in both cerebral CST peduncles Prell et al., [60] 17/17 Low in parts of CSTs and the Elevated parts of CSTs, parts of frontal lobe, Correlation between FA in internal capsule cingulate gyrus cingulate gyrus, parahippocampal region, and contralateral strength of the lower limb. CC, cerebellum. Different patterns in FA and ADC of bulbar and limb onset compared to controls. Prudlo et al., [70] 22/21 Low throughout the CSTs, the Correlation between FA in many voxels of a anterior limb of the internal whole DTI brain analyses with ALSFRS-R capsule, thalamic radiations, the CC, association fibres and the middle cerebellar peduncle Pyra et al., [59] 14/14 Low in left precentral gyrus Correlation between spasticity and ADC in contralateral precentral gyrus ae7o 17 of 7 Page Rajagopalan et al., [63] 47/10 Low in the left subcortical motor AD was low in the PLIC. RD Not available area and right PLIC was elevated in the PLIC of patients with T2 hyperintensities in the CSTs Table 2 The main results of DTI studies (Continued) Grolez

Rosskopf et al., [57] 100/93 Low along CSTs Correlation between FA along CSTs and Neurology BMC al. et ALSFRS-R Sach et al., [33] 15/12 Low in parts of CSTs, premotor Low FA in patients without UMN signs in areas, CC and right thalamus parts of CST, CC and right thalamus. No correlation available with clinical score Sage et al., [34] 28/26 Low along CSTs and the right Elevated in parts of CSTs Correlation between FA and ALSFRS in postcentral gyrus several parts of CSTs and in prefrontal lobe (2016) 16:155 Sage et al., [52] 28/26 Low in parts of CSTs, parts of Elevated along CSTs, hippocampus, insula, Correlation between FA along CSTs, in the frontal lobe, insula, parts of the temporal and frontal lobe and CC prefrontal area and ALSFRS. Negative hippocampus, cerebral correlation between MD along CST, peduncles and CC hippocampus, cerebellum, parietal and temporal lobe and ALSFRS. Sarica et al., [53] 14/14 Low in right CSTs and left Elevated in right CSTs, cingulum and left RD elevated in right CSTs Negative (p ≤ 0.05) anterior thalamic radiations anterior thalamic radiations and left anterior thalamic radiations. AD elevated in the right cingulum Schirimrigt et al., 2007 10/20 Low along CSTs (nb: study with Correlation between FA along CSTs and DD a technical objective) Stagg et al., [47] 13/14 Low along CSTs Elevated along CSTs Negative Tang et al., [58] 69/23 Low along CSTs, in frontal WM Elevated in the centrum semi-ovale and Not available and the genu of the CC frontal and parietal WM Thivard et al., [24] 15/25 Low along CSTs, premotor Elevated in the motor cortex, premotor Correlation between FA along CST, insula, cortex, right thalamus, insula, cortex, insula, hippocampus, and right premotor cortex, cingulum, precuneus, CC parts of parietal lobe superior temporal gyrus and ALSFRS-R, negative correlation between FA in CC and centrum semiovale and DD Verstraete et al., [30] 12/12 Low in the rostral part of CSTs Not available and the CC Wang et al., [71] 16/17 Low CST volume in DTI Negative Yin et al., [48] 8/12 Low along CSTs Not available Zhang et al., [62] 17/19 Low in parts of CSTs Elevated in parts of the CSTs Correlation between FA in right superior CST and ALSFRS-R and motor subscore of ALSFRS-R FA fractional anisotropy, MD mean diffusivity, ADC apparent diffusion coefficient, RD radial diffusivity, AD axial diffusivity, CST corticospinal tract, PLIC posterior limb of the internal capsule, CC corpus callosum, DD disease duration, DPR disease progression rate, UMN upper motor neuron, LMN lower motor neuron ae8o 17 of 8 Page Grolez et al. BMC Neurology (2016) 16:155 Page 9 of 17

callosum [56, 67, 69]. Lastly, RD and AD are elevated in Magnetic resonance spectroscopy (Table 3) the cerebellum of C9ORF72-positive patients [75]. MRS is a non-invasive means of assessing brain metabolism, based on the chemical properties of molecular Longitudinal studies In longitudinal analyses, low FA is structures. Results are expressed as peaks. The main observed in the right superior CST [62], precentral sub- brain metabolites monitored with MRS include N- cortical regions, mesencephalic CSTs and parts of cere- acetyl-aspartate (NAA), choline (Cho, a cell membrane bellum [61]. FA was found to fall in a progressive, linear marker), creatine (Cr) and phosphocreatine (Pcr). The manner along the CSTs [34, 46, 76], and the diffusivity Cr + Pcr peak is a marker of energy metabolism; the level of the external and internal capsule was elevated [61]. of Cr is assumed to be stable and is used to calculate the AD is elevated in CSTs [77]. However, a recent longitu- metabolite ratio. At a field strength of 1.5 T, the peaks dinal study found that white matter defects (as assessed for glutamine (Gln), glutamate (Glu, an excitatory by DTI) progressed much slowly that in the grey matter neurotransmitter) and gamma aminobutyric acid defects - notably in the basal ganglia [78]. (GABA, an inhibitory neurotransmitter) overlap to form a complex referred to as “Glx”; higher field strengths are Clinical correlations Many studies have found that dis- required for more accurate quantification of this me- ease progression (as defined by the change over time in tabolites. The simple sugar myo-inositol (Ins) is ab- the ALSFRS or ALSFRS-R score) is correlated with the sent from neurons but present in glial cells. Hence, changes in FA in CSTs [24, 34, 45, 61, 62, 77, 79]. Further- elevated brain levels of Ins are associated with glial more, the disease duration is correlated with FA in entire proliferation, whereas low levels are associated with CSTs or parts of CSTs [24, 50, 54, 56, 61, 77]. Conversely, glial destruction [82–84]. the disease duration was negatively correlated with FA in The main results for the CST concern the ratio between the cerebellum [61], the subcortical white matter of insula, the NAA peak on one hand and the Cho and Cr + Pcr the ventrolateral premotor cortex, the cingulum, the pre- peaks on the other. In ALS patients, low levels of NAA cuneus and the splenium of the corpus callosum [24]. have been observed in the precentral corticosubcortical Lower-limb muscle strength was correlated with the region [85–91], in the PLIC [85] and in entire CSTs [55, contralateral FA in CSTs [60]. Diffusivity values in CSTs 59, 86, 92]. Glu and Glu + Gln were found to be elevated and in the parietal lobe, temporal lobe and cerebel- in the precentral corticosubcortical region and in the PLIC lum were also correlated with disease progression [52, [85], whereas Cho was found to be elevated in the precen- 79]. Diffusivity in CSTs was correlated with disease tral region [88, 93, 94] and along the CSTs [47, 59, 92]. duration [79], and contralateral diffusivity in CSTs Levels of the inhibitory neurotransmitter GABA were low was correlated with spasticity [59]. in the precentral cortex [55, 95], and levels of Ins were Whereas both limb-onset patients and bulbar-onset low in the precentral corticosubcortical region [89]. patients differed from controls in terms of FA along Whereas the above-mentioned studies used defined CSTs, the two patient groups had similar values [60]. voxels for MRS acquisition, whole-brain spectroscopy The CSTs were more impaired in bulbar-onset patients has revealed low NAA/Cho or NAA/Cr ratios in many than in limb-onset patients when considering FA, MD other regions [93]. and RD but not AD [80]. Tractography revealed a large Follow-up studies have shown a decrease in the NAA number of differences between controls and patients de- peak in the precentral cortex (with no differences for fined as definite/probable according to the El Escorial other metabolites) [96], a decrease in the NAA/Cho peak criteria, whereas there were fewer differences between and an increase in the Cho/Cr peak [91]. controls and a possible/suspected group [79].

Neurophysiological correlations FA in CSTs was re- Clinical correlations Patients with definite or probable lated to the central conduction time in a transcranial ALS (according to the El Escorial criteria) had low levels magnetic stimulation study [33, 49] (Iwata et al., [49]; of NAA and high levels of Ins and Cho in the precentral Sach et al., [33]). cortex. In patients with possible or suspected ALS, only Cho and (on the left side) Ins were elevated. In another Diagnostic accuracy An individual patient data meta- study of suspected ALS patients, only right-side Ins and analysis of CST studies concluded that DTI lacks suffi- right-side Cho were elevated [79]. Low NAA and high cient diagnostic discrimination [81]. The pooled sensitiv- Cho and Ins levels are associated with clinical disease se- ity and specificity were only 0.68 and 0.73, respectively. verity [94]. Lower NAA levels in the precentral gyrus, Researchers have also suggested that a multimodal ap- disease duration and disease progression were intercor- proach (combining DTI with methods such as MRS) related [59], and a low NAA/Cho ratio was associated may be more promising [55]. with poor survival [88]. Grolez et al. BMC Neurology (2016) 16:155 Page 10 of 17

Table 3 main results of magnetic resonance spectroscopy studies Publication Numbers of Main results Main clinical correlations patients/controls Bowen et al., [94] 18/12 Cho and ins were elevated in the MC. NAA and Cr Correlation between ins in MC and UMN disability, were correlated in left MC negative correlation between Naa in MC and UMN disability, higher Cho in sever UMN disability group Cervo et al., [90] 84/28 NAA/(Cho + Cr) was low in the MC Negative Foerster et al., [95] 10/9 Low gamma aminobutyric acid in the MC but not in Negative WM Foerster et al., [55] 29/30 NAA and gamma aminobutyric acid were low and Negative correlation between gamma aminobutyric ins was elevated in the left MC acid in MC and DD, correlation between Naa peak in MC and ALSFRS-R Govind et al., [92] 38/70 NAA was low and Cho was elevated in most parts of Negative correlation between Cho/Naa in the left CST, and Cho/NAA was elevated in all parts of the entire CST and forced vital capacity, negative CSTs correlation between Cho/Naa in the left CST and right and left finger tap rate, negative correlation between Cho/Naa in left MC and semiovale centrum and right finger tape or forced vital capacity Han et al., [85] 15/15 NAA/Cr peak was low in the MC and PLIC, Glu/Cr Negative correlation between Glu + Gln/Cr with and Glu + Gln/Cr peaks were low in the MC and Norris score PLIC Kalra et al., [88, 89] 63/18 NAA/Cho and NAA/Cr was low in the MC, and Cho/ Relation between decreased Naa/Cho in the MC and Cr was elevated in the MC reduced survival. Kalra et al., [88, 89] 17/15 NAA/Ins, NAA/Cr and NAA/Cho were low in the MC, Negative (p ≤ 0.05) and Ins/Cr was elevated in the MC Liu et al., [87] 19/13 NAA/Cr was low in the MC Negative Lombardo et al., [79] 32/19 NAA/Cr was low and ins/Cr and Cho/Cr were Abnormalities were correlated with the El Escorial elevated in the MC. score Pohl et al., [91] 70/48 NAA, Pcr + Cr, NAA/Cho and NAA/(Pcr + Cr) were Not available low in the MC. At 12 months, NAA/Cho was low and Cho/(Pcr + Cr) was elevated Pyra et al., [59] 14/14 NAA/Cho and NAA/Cr were low in the MC and Correlation between Naa/Cho peak in MC and DD, corona radiata, and Cho/Cr was elevated in the MC negative correlation between Naa/Cho in MC and corona radiata and DPR Rooney et al., [86] 10/9 NAA/(Cho + Cr) was low in the MC and CST but not Correlation between Naa/(Cho + Cr) in the MC and in other regions maximum finger tape rate Stagg et al., [47] 13/14 NAA was low along the CSTs Correlation between Naa peak along CSTs and ALSFRS-R Unrath et al., [96] 8/0 Progressive decrease over time in the NAA peak Progressive decrease of Naa/(Cr + Cho) in the less throughout the MC affected hemisphere. Correlation between Naa and the more or less affected side (ALSFRS subscore). Correlation between Naa/(Cr + Cho) and the less affected side (ALSFRS subscore) Verma et al., [93] 21/10 NAA/Cho was low in the right lingual gyrus, parts of Not available the occipital lobe, left supramarginal gyrus and left caudate. NAA/Cr was low in the right MC, left frontal inferior operculum, right cuneus, parts of the occipital lobe, left caudate and left Heschl gyrus MC motor cortex, CST corticospinal tract, PLIC posterior limb of the internal capsule, WM white matter, NAA N-acetylaspartate, Cho choline, Gln glutamine, Glu glutamate, PCr phosphocreatine, Cr creatine or creatine + Pcr (the distinction is not relevant for interpretation), DD disease duration, DPR disease progression rate

Diagnostic accuracy In a recent study, the combination considered alone, the sensitivity and specificity values did of MRS and DTI yielded a sensitivity of 0.93 and a specifi- not exceed 0.71 and 0.75, respectively [90]. city of 0.85, whereas the DTI data alone gave values of 0.86 and 0.70, respectively [55]. A criterion combining the Iron imaging MRS data, hypointensity in the precentral gyri and hyper- Iron overload appears to be involved in the physiopa- intensity in the CSTs yielded a sensitivity of 0.78 and a thology of ALS [8, 97]. However, very few studies have specificity of 0.82; when each of the three parameters was analysed the iron content in the brain of ALS patients. A Grolez et al. BMC Neurology (2016) 16:155 Page 11 of 17

variety of techniques are available, including relaxometry segments of the cervical spinal cord [114]. Atrophy of (T2, T2* or R2*) and the susceptibility-weighted imaging the cervical and upper thoracic spinal cord progressed [98, 99] (SWI, which is probably more sensitive than over time, whereas the DTI changes were relatively relaxometry [100]). Multi-echo T2* and R2* sequences stable [115]. This observation agrees with the results of are sensitive to magnetic field inhomogeneities. The a longitudinal brain MRI study showing that grey matter value of R2* is measured by voxel-by-voxel modelling of defects (but not white matter defects) progressed over the exponential decrease in signal [101, 102]. In post- time [78]. Sensory pathways were also involved, with mortem studies, R2* values have been found to be corre- adecreasedFAandanelevatedMDandelevatedRD lated with the iron content [103]. in the posterior part of the cervical spinal cord [115, 116]. MRS of the cervical spinal cord revealed low Susceptibility-weighted imaging and quantitative NAA/Cr+Pcr and NAA/Ins peaks and an elevated susceptibility mapping (QSM) Ins/Cr peak [117]. SWI is derived from T2* sequences and can be described as a flow-compensated, high-spatial-resolution T2* Clinical correlations Cervical and upper thoracic spinal weighted sequence. The main advantage over conven- cord atrophy was correlated with upper arm muscle tional T2* sequence is better contrast [101]. QSM is strength, as measured by manual testing [114]. FA in the based on the analysis of phase information. The main cervical spinal cord was correlated with the ALSFRS sources of phase shifts in biological tissues are the iron, score. The rate of atrophy in a follow-up study was cor- calcium, lipid and myelin contents [104]. The value of related with the rates of changes of an arm ALSFRS-R QSM in ALS was confirmed in a post-mortem study, in score and an arm strength score (in a manual test) [115]. which the iron content of grey matter was assessed more The NAA/Cr + Pcr and NAA/Ins peaks were positively accurately than that of white matter [105]. correlated with the ALSFRS-R score and forced vital The initial studies in the field noted a shorter T2 time in capacity but negatively correlated with disease progres- the motor cortex of most ALS patients (vs. controls) [106], sion [117]. although this was not confirmed a few years later [107]. Later studies found a correlation being T2* shortening and Functional MRI iron deposition in the microglia of the motor cortex (ac- Brain fMRI is a recent technique based on blood flow cording to a post-mortem examination) [7, 106, 108]. Iron and blood-oxygen-level dependent (BOLD) contrast, overload in the motor cortex was also described in SWI which in turn are based on the neurovascular conse- studies [109] and qualitative studies [100]; in the later study, quences of neuronal activation [118]. The paramagnetic there were no differences between ALS patients and con- properties of deoxyhaemoglobin lead to signal reduction trols in terms of T2 or T2*. A few quantitative studies have in gradient echo sequences [101]. Brain fMRI sequences looked at white matter or deep grey matter [109, 110]. are acquired in the resting state or during performance There was a trend towards iron accumulation in CST white of a motor, auditory, cognitive, or visual task [119–121]. matter in a relaxometry study [110]. Iron overload was also The resting state sequences are quite similar to T2* se- present in the red nucleus, substantia nigra, globus pallidus, quences [101]. putamen [109] and caudate nucleus [110]. The use of SWI In the resting state, functional connectivity is based on sequences revealed widespread iron overload and myelin fluctuations of the BOLD signal. In general, seven “de- defects in the white matter of all brain lobes and corpus fault” networks can be identified: default mode, execu- callosum [111]. A retrospective study of iron deposition in tive control, visual salience, sensorimotor, dorsal motor cortex found that quantitative susceptibility mapping attention and auditory networks [118, 122]. Functional was more accurate than T2*, T2 or FLAIR sequences [112]. connectivity between specific brain regions and/or net- A follow-up study has highlighted the progression of a works and the rest of the brain can then be analysed T2 hypointense area in the motor cortex after 6 months, [118]. Resting-state analysis can also provide information which was strongly and negative correlated with the about local activation (by analysing the amplitude of ALSFRS score [108]. Lastly, changes in SWI sequences low-frequency fluctuations (ALFF), for example). in the corpus callosum are correlated with the ALSFRS- R score [111]. The resting state Resting-state studies have evidenced the widespread reorganization of functional connectivity Imaging of the spinal cord in ALS patients. This reorganization can be observed as The few studies to have analysed the spinal cord re- either elevated functional connectivity or low functional vealed atrophy in the cervical and upper thoracic regions connectivity. Functional connectivity was found to be [113–115]. A DTI study reported low FA, elevated RD low in several parts of the frontal lobe (the right orbito- and a high magnetization transfer ratio in lateral frontal cortex, the left inferior frontal cortex [123] and Grolez et al. BMC Neurology (2016) 16:155 Page 12 of 17

especially the motor cortex [124]) but was elevated in [135]. Using another type of stimulus, activation of the the parietal lobe (the left precuneus and the right angu- hippocampus also fell after 3 months [135]. In a theory- lar gyrus [123]) and the left frontoparietal network [123]. of-mind task designed to evaluation the mirror neuron This agrees with the elevated sensorimotor connectivity system, ALS patients displayed abnormally strong activa- of the left sensorimotor cortex with several regions of tion of the right anterior cortical areas [136]. A link the right hemisphere (mainly the cingulate cortex, para- between the severity of motor weakness and the pattern of hippocampal gyrus and cerebellum crus II) [125]. When activation has never been observed, although the compared with alterations of the CSTs (according to FA ALSFRS-R score was correlated with the signal change in measurements), there were also changes in right sensori- the sensorimotor cortex in one study [129]. When consid- motor connectivity; the connectivity was elevated if the ering the clinical phenotype, only a bulbar-onset subgroup CST was normal in a DTI analysis but was low if the showed low activation of the motor cortex during tongue CST was abnormal [125]. A network-by-network analysis movements [131]. Shifts in cerebellar or hippocampal acti- revealed low activity in the default mode network [126] vation were correlated with the ALSFRS-R score. and elevated activity in the sensorimotor network (both of which include parts of the frontal lobe [126, 127], Discussion although not the same ones). Thalamic connectivity is also When considering the data as a whole, the most disease- elevated [127]. sensitive MRI patterns are located in motor regions (and Resting-state activity (as measured by the ALFF) especially along the CST from the cortex to the spinal was elevated in the frontal lobe (the left anterior cin- cord). However, ALS also affects the brain more broadly; gulate, right superior frontal cortex [31] left medial other parts of the frontal lobe, the temporal lobe, the frontal gyrus and right inferior frontal gyrus [128]), hippocampus, the parietal lobe, the cingulum and the in- the temporal lobe (the right parahippocampal cortex sula are commonly involved, whereas the occipital lobe and the left inferior temporal cortex) and the occipital is less frequently involved. For most MRI parameters, lobe (the middle occipital cortex) [31] but low in the patient vs. control differences are mostly apparent in the right fusiform gyrus, the visual cortex and right post- left (dominant) hemisphere. The various MRI patterns central gyrus [128]. appear to reflect disease severity, progression and dur- The resting-state connectivity in the frontal cortex is ation. With fMRI, abnormally high levels of activation in correlated with disease progression and duration [128]. the motor cortex are probably related to the disease- The connectivity between the left sensorimotor cortex related loss of inhibition. Excessive activation appears to and the right parahippocampal gyri and fourth cerebellar be associated with compensation for the loss of function lobule is correlated with the ALSFRS-R [125]. Connect- caused by neuron loss. ivity of the dorsal part of the precentral gyri was correlated with hand strength [124]. Activity in the right Limitations of neuroimaging as a biomarker for ALS parahippocampal gyri was correlated with disease pro- The low average number of participants represents the gression, whereas activity in the left anterior cingulate main limitation: very few published studies have featured and left temporal gyrus was correlated with cognitive pa- more than 30 subjects per group (i.e. ALS and control rameters [31]. groups). Together with the differences in the proportions of the various patient phenotypes, this explains the Activation in fMRI In motor tasks, activation of the high variability of the results - particularly outside the motor cortex was elevated [129–131] and spread widely motor cortex and the CSTs [137]. across the supplementary motor area, the premotor cortex Only a small proportion of morphometric studies in- [132] and the parietal somatosensory cortex [129, 132]. In- cluded a whole-brain analysis. The proportion is higher hibition was low [130]. Activation of the ipsilateral sen- for DTI studies and low (only one study) for MRS. The sorimotor cortex was possibly linked to a compensatory selection of regions of interest may over-emphasize mechanism in ALS [133]. A visual task was associated changes in the CSTs and the motor system and under- with poor activation of the secondary visual cortex and emphasize changes outside these areas. strong activation of associative areas, whereas an auditory Very few studies had a longitudinal design, which is task produced delayed activation in the secondary audi- nevertheless required for the full interpretation of abnor- tory cortex, and somatosensory stimulation produced pro- malities. The main difficulty in setting up longitudinal longed activation of the right inferior frontal gyrus and studies relates to the rapid disease progression and the right posterior insular cortex [134]. In a modified “go”/ appearance of respiratory failure, which prevents pa- “no-go” task, activation of the primary motor cortex, other tients from lying in the MRI for any length of time. In motor areas, the frontal lobe, the insula and the cerebel- studies longer than 6 months, the drop-out rate is usu- lum was found to decrease over a three-month period ally over 50 %. Grolez et al. BMC Neurology (2016) 16:155 Page 13 of 17

fMRI studies have highlighted both elevated activity AcomprehensiveanalysisofMRIparametersinvery and low activity in several parts of the brain. The main large cohorts of ALS patients might reveal a radiological physiopathological explanations relate to a physiological surrogate marker for use in clinical trials. Several points compensation response or elevated activity due to the should be considered for large scale studies: in 2010, the loss of inhibitory mechanisms [127]. first Neuroimaging Symposium in ALS (NiALS) had the Spinal cord imaging is a rarely applied imaging mo- aim of establishing consensus about the various applica- dality in ALS, despite the pivotal involvement of the an- tions of MRI to the study of ALS and the possibility of mul- terior horn of the spinal cord in ALS [137]. Spinal cord ticentre collaboration. Consensus criteria for the main MRI imaging appears to be most commonly used in studies sequences (VBM, DTI, fMRI and spectroscopy) and the of multiple sclerosis, since all studies in this disease con- main clinical dataset were established [144]. The aim of this text found differences (relative to controls) in the cer- consensus was to lead large multicentre and longitudinal vical spinal cord. To the best of our knowledge, no studies in imaging research or therapeutic trials [144].

Author details 20. Mezzapesa DM, D’Errico E, Tortelli R, Distaso E, Cortese R, Tursi M, et al. 1Department of Movement Disorders and Neurology, Lille University Cortical thinning and clinical heterogeneity in amyotrophic lateral sclerosis. Hospital, Faculty of Medicine, University of Lille, Lille, France. 2INSERM U1171, PLoS One. 2013;8(11):e80748. Lille University Hospital, Faculty of Medicine, University of Lille, Lille, France. 21. Agosta F, Pagani E, Rocca MA, Caputo D, Perini M, Salvi F, et al. Voxel-based 3Department of Neuroradiology, Lille University Hospital, Faculty of Medicine, morphometry study of brain volumetry and diffusivity in amyotrophic University of Lille, Lille, France. 4Laboratoire d’Imagerie Biomédicale, lateral sclerosis patients with mild disability. Hum Brain Mapp. Sorbonne Universités, UPMC Univ Paris 06, CNRS, INSERM, Paris, France. 2007;28(12):1430–8. 5Département des Maladies du Système Nerveux, Groupe Hospitalier 22. Chang JL, Lomen-Hoerth C, Murphy J, Henry RG, Kramer JH, Miller BL, et al. Pitié-Salpêtrière, APHP, Paris, France. 6Department of Medical Pharmacology, A voxel-based morphometry study of patterns of brain atrophy in ALS and Lille University Hospital, Faculty of Medicine, University of Lille, Lille, France. ALS/FTLD. Neurology. 2005;65(1):75–80. 23. Canu E, Agosta F, Riva N, Sala S, Prelle A, Caputo D, et al. The Received: 4 March 2016 Accepted: 10 August 2016 topography of brain microstructural damage in amyotrophic lateral sclerosis assessed using diffusion tensor MR imaging. AJNR Am J Neuroradiol. 2011;32(7):1307–14. 24. Thivard L, Pradat P-F, Lehéricy S, Lacomblez L, Dormont D, Chiras J, et al. References Diffusion tensor imaging and voxel based morphometry study in 1. Bensimon G, Lacomblez L, Meininger V. A controlled trial of riluzole in amyotrophic lateral sclerosis: relationships with motor disability. J Neurol amyotrophic lateral sclerosis. ALS/Riluzole Study Group. N Engl J Med. 1994; Neurosurg Psychiatry. 2007;78(8):889–92. 330(9):585–91. 25. Schuster C, Kasper E, Machts J, Bittner D, Kaufmann J, Benecke R, et al. Focal 2. Soriani M-H, Desnuelle C. Épidémiologie de la SLA. Rev Neurol (Paris). 2009; thinning of the motor cortex mirrors clinical features of amyotrophic lateral 165(8–9):627–40. sclerosis and their phenotypes: a neuroimaging study. J Neurol. 2013; 3. Beghi E, Logroscino G, Chiò A, Hardiman O, Mitchell D, Swingler R, et al. 260(11):2856–64. The epidemiology of ALS and the role of population-based registries. 26. Ellis CM, Suckling J, Amaro E, Bullmore ET, Simmons A, Williams SC, et al. Biochim Biophys Acta (BBA) - Mol Basis Dis. 2006;1762(11–12):1150–7. Volumetric analysis reveals corticospinal tract degeneration and extramotor 4. Mitchell JD, Borasio GD. Amyotrophic lateral sclerosis. Lancet. 2007; involvement in ALS. Neurology. 2001;57(9):1571–8. 369(9578):2031–41. 27. Devine MS, Pannek K, Coulthard A, Mccombe PA, Rose SE, Henderson RD. 5. Turner MR, Bowser R, Bruijn L, Dupuis L, Ludolph A, Mcgrath M, et al. Exposing asymmetric gray matter vulnerability in amyotrophic lateral Mechanisms, models and biomarkers in amyotrophic lateral sclerosis. sclerosis. NeuroImage Clin. 2015;7:782–7. Amyotroph Lateral Scler Front Degener. 2013;14(S1):19–32. 6. Devos D, Moreau C, Devedjian JC, Kluza J, Petrault M, Laloux C, et al. 28. Zhang Q, Mao C, Jin J, Niu C, Bai L, Dang J, et al. Side of Limb-Onset Targeting Chelatable Iron as a Therapeutic Modality in Parkinson’s Disease. Predicts Laterality of Gray Matter Loss in Amyotrophic Lateral Sclerosis. – Antioxid Redox Signal. 2014;21(2):195–210. BioMed Res Int. 2014;2014:1 11. 7. Kwan JY, Jeong SY, Van Gelderen P, Deng H-X, Quezado MM, Danielian LE, 29. Walhout R, Westeneng H-J, Verstraete E, Hendrikse J, Veldink JH, van den et al. Iron Accumulation in Deep Cortical Layers Accounts for MRI Signal Heuvel MP, et al. Cortical thickness in ALS: towards a marker for upper Abnormalities in ALS: Correlating 7 Tesla MRI and Pathology. Ashizawa T, motor neuron involvement. J Neurol Neurosurg Psychiatry. editor. PLoS ONE. 2012 Apr 17;7(4):e35241. 2015;86(3):288–94. 8. Jeong SY, Rathore KI, Schulz K, Ponka P, Arosio P, David S. 30. Verstraete E, van den Heuvel MP, Veldink JH, Blanken N, Mandl RC, Hulshoff Dysregulation of Iron Homeostasis in the CNS Contributes to Disease Pol HE, et al. Motor Network Degeneration in Amyotrophic Lateral Sclerosis: Progression in a Mouse Model of Amyotrophic Lateral Sclerosis. J A Structural and Functional Connectivity Study. Zhan W, editor. PLoS ONE. Neurosci. 2009;29(3):610–9. 2010 Oct 27;5(10):e13664. 9. Kollewe K, Mauss U, Krampfl K, Petri S, Dengler R, Mohammadi B. ALSFRS-R 31. Zhu W, Fu X, Cui F, Yang F, Ren Y, Zhang X, et al. ALFF Value in Right score and its ratio: A useful predictor for ALS-progression. J Neurol Sci. 2008; Parahippocampal Gyrus Acts as a Potential Marker Monitoring Amyotrophic 275(1–2):69–73. Lateral Sclerosis Progression: a Neuropsychological, Voxel-Based 10. Rudnicki SA, Berry JD, Ingersoll E, Archibald D, Cudkowicz ME, Kerr DA, et al. Morphometry, and Resting-State Functional MRI Study. J Mol Neurosci Dexpramipexole effects on functional decline and survival in subjects with [Internet]. 2015 Jun 4 [cited 2015 Jul 29]; Available from: http://link.springer. amyotrophic lateral sclerosis in a Phase II study: Subgroup analysis of com/10.1007/s12031-015-0583-9 demographic and clinical characteristics. Amyotroph Lateral Scler Front 32. Kassubek J, Unrath A, Huppertz H, Lulé D, Ethofer T, Sperfeld A, et al. Global Degener. 2013;14(1):44–51. brain atrophy and corticospinal tract alterations in ALS, as investigated by 11. Braak H, Brettschneider J, Ludolph AC, Lee VM, Trojanowski JQ, Tredici KD. voxel‐based morphometry of 3‐D MRI. Amyotroph Lateral Scler. Amyotrophic lateral sclerosis—a model of corticofugal axonal spread. Nat 2005;6(4):213–20. Rev Neurol. 2013;9(12):708–14. 33. Sach M, Winkler G, Glauche V, Liepert J, Heimbach B, Koch MA, et al. 12. The ALS Schwaben Register Group, Rosenbohm A, Kassubek J, Weydt P, Diffusion tensor MRI of early upper motor neuron involvement in Marroquin N, Volk AE, et al. Can lesions to the motor cortex induce amyotrophic lateral sclerosis. Brain J Neurol. 2004;127(Pt 2):340–50. amyotrophic lateral sclerosis? J Neurol. 2014;261(2):283–90. 34. Sage CA, Peeters RR, Görner A, Robberecht W, Sunaert S. Quantitative 13. Ravits JM, La Spada AR. ALS motor phenotype heterogeneity, focality, and spread diffusion tensor imaging in amyotrophic lateral sclerosis. Neuroimage. 2007; Deconstructing motor neuron degeneration. Neurology. 2009;73(10):805–11. 34(2):486–99. 14. Stroman PW, Wheeler-Kingshott C, Bacon M, Schwab JM, Bosma R, Brooks J, 35. Agosta F, Valsasina P, Riva N, Copetti M, Messina MJ, Prelle A, et al. The cortical et al. The current state-of-the-art of spinal cord imaging: Methods. signature of amyotrophic lateral sclerosis. PLoS One. 2012;7(8):e42816. Neuroimage. 2014;84:1070–81. 36. Cerami C, Dodich A, Canessa N, Crespi C, Iannaccone S, Corbo M, et al. 15. Verma T, Cohen-Adad J. Effect of respiration on the B 0 field in the human Emotional empathy in amyotrophic lateral sclerosis: a behavioural and spinal cord at 3T: Effect of Respiration on B 0 Field in Human Spinal Cord. voxel-based morphometry study. Amyotroph Lateral Scler Front Degener. Magn Reson Med. 2014;72(6):1629–36. 2014;15(1–2):21–9. 16. Bandettini PA. What’s New in Neuroimaging Methods? Ann N Y Acad Sci. 37. Bede P, Bokde ALW, Byrne S, Elamin M, McLaughlin RL, Kenna K, et al. 2009;1156(1):260–93. Multiparametric MRI study of ALS stratified for the C9orf72 genotype. 17. Ashburner J, Friston KJ. Voxel-based morphometry–the methods. Neurology. 2013;81(4):361–9. Neuroimage. 2000;11(6 Pt 1):805–21. 38. Westeneng H-J, Verstraete E, Walhout R, Schmidt R, Hendrikse J, Veldink JH, 18. Grosskreutz J, Kaufmann J, Frädrich J, Dengler R, Heinze H-J, Peschel T. et al. Subcortical structures in amyotrophic lateral sclerosis. Neurobiol Aging Widespread sensorimotor and frontal cortical atrophy in amyotrophic lateral [Internet]. 2014 Sep [cited 2014 Nov 6]; Available from: http://linkinghub. sclerosis. BMC Neurol. 2006;6(1):17. elsevier.com/retrieve/pii/S0197458014005971 19. Mezzapesa DM, Ceccarelli A, Dicuonzo F, Carella A, De Caro MF, Lopez M, 39. Abdulla S, Machts J, Kaufmann J, Patrick K, Kollewe K, Dengler R, et al. et al. Whole-brain and regional brain atrophy in amyotrophic lateral Hippocampal degeneration in patients with amyotrophic lateral sclerosis. sclerosis. AJNR Am J Neuroradiol. 2007;28(2):255–9. Neurobiol Aging. 2014;35(11):2639–45. Grolez et al. BMC Neurology (2016) 16:155 Page 15 of 17

40. Chapman MC, Jelsone-Swain L, Fling BW, Johnson TD, Gruis K, Welsh RC. 62. Zhang Y, Schuff N, Woolley SC, Chiang GC, Boreta L, Laxamana J, et al. Corpus callosum area in amyotrophic lateral sclerosis. Amyotroph Lateral Progression of white matter degeneration in amyotrophic lateral sclerosis: A Scler. 2012;13(6):589–91. diffusion tensor imaging study. Amyotroph Lateral Scler. 2011;12(6):421–9. 41. Meadowcroft MD, Mutic NJ, Bigler DC, Wang J, Simmons Z, Connor JR, et al. 63. Rajagopalan V, Allexandre D, Yue GH, Pioro EP. Diffusion Tensor Imaging Histological-MRI correlation in the primary motor cortex of patients with Evaluation of Corticospinal Tract Hyperintensity in Upper Motor Neuron- amyotrophic lateral sclerosis: MRI and Histological Analysis of the PMC in Predominant ALS Patients. J Aging Res. 2011;2011:481745. ALS. J Magn Reson Imaging. 2015;41(3):665–75. 64. Abe O, Yamada H, Masutani Y, Aoki S, Kunimatsu A, Yamasue H, et al. 42. Le Bihan D, Mangin JF, Poupon C, Clark CA, Pappata S, Molko N, et al. Amyotrophic lateral sclerosis: diffusion tensor tractography and voxel-based Diffusion tensor imaging: concepts and applications. J Magn Reson Imaging analysis. NMR Biomed. 2004;17(6):411–6. JMRI. 2001;13(4):534–46. 65. Poujois A, Schneider FC, Faillenot I, Camdessanché J-P, Vandenberghe N, 43. Habas C. Basic principles of diffusion tensor MR tractography. J Radiol. 2004; Thomas-Antérion C, et al. Brain plasticity in the motor network is correlated 85(3):281–6. with disease progression in amyotrophic lateral sclerosis: Functional 44. Yamada K, Sakai K, Akazawa K, Yuen S, Nishimura T. MR tractography: a Cerebral Reorganization, Disease Progression, and ALS. Hum Brain Mapp. review of its clinical applications. Magn Reson Med Sci MRMS Off J Jpn Soc 2013;34(10):2391–401. Magn Reson Med. 2009;8(4):165–74. 66. Müller H-P, Lulé D, Unrath A, Ludolph AC, Riecker A, Kassubek J. 45. Ciccarelli O, Behrens TE, Johansen-Berg H, Talbot K, Orrell RW, Howard RS, Complementary image analysis of diffusion tensor imaging and 3- et al. Investigation of white matter pathology in ALS and PLS using tract- dimensional t1-weighted imaging: white matter analysis in amyotrophic based spatial statistics. Hum Brain Mapp. 2009;30(2):615–24. lateral sclerosis. J Neuroimaging Off J Am Soc Neuroimaging. 46. Nickerson JP, Koski CJ, Boyer AC, Burbank HN, Tandan R, Filippi CG. Linear 2011;21(1):24–33. longitudinal decline in fractional anisotropy in patients with amyotrophic 67. Chapman MC, Jelsone-Swain L, Johnson TD, Gruis KL, Welsh RC. Diffusion lateral sclerosis: preliminary results. Klin Neuroradiol. 2009;19(2):129–34. tensor MRI of the corpus callosum in amyotrophic lateral sclerosis: DTI of 47. Stagg CJ, Knight S, Talbot K, Jenkinson M, Maudsley AA, Turner MR. Whole- the Corpus Callosum in ALS. J Magn Reson Imaging. 2014;39(3):641–7. brain magnetic resonance spectroscopic imaging measures are related to 68. Keller J, Vymazal J, Ridzoň P, Rusina R, Kulišt’ák P, Malíková H, et al. disability in ALS. Neurology. 2013;80(7):610–5. Quantitative brain MR imaging in amyotrophic lateral sclerosis. Magma N Y 48. Yin H, Cheng SHT, Zhang J, Ma L, Gao Y, Li D, et al. Corticospinal tract N. 2011;24(2):67–76. degeneration in amyotrophic lateral sclerosis: a diffusion tensor imaging 69. Metwalli NS, Benatar M, Nair G, Usher S, Hu X, Carew JD. Utility of axial and and fibre tractography study. Ann Acad Med Singapore. 2008;37(5):411–5. radial diffusivity from diffusion tensor MRI as markers of neurodegeneration 49. Iwata NK, Aoki S, Okabe S, Arai N, Terao Y, Kwak S, et al. Evaluation of in amyotrophic lateral sclerosis. Brain Res. 2010;1348:156–64. corticospinal tracts in ALS with diffusion tensor MRI and brainstem 70. Prudlo J, Bißbort C, Glass A, Grossmann A, Hauenstein K, Benecke R, et al. stimulation. Neurology. 2008;70(7):528–32. White matter pathology in ALS and lower motor neuron ALS variants: a 50. Iwata NK, Kwan JY, Danielian LE, Butman JA, Tovar-Moll F, Bayat E, et al. diffusion tensor imaging study using tract-based spatial statistics. J Neurol. White matter alterations differ in primary lateral sclerosis and amyotrophic 2012;259(9):1848–59. lateral sclerosis. Brain. 2011;134(9):2642–55. 71. Wang S, Poptani H, Bilello M, Wu X, Woo JH, Elman LB, et al. Diffusion 51. Ellis CM, Simmons A, Jones DK, Bland J, Dawson JM, Horsfield MA, et al. tensor imaging in amyotrophic lateral sclerosis: volumetric analysis of the Diffusion tensor MRI assesses corticospinal tract damage in ALS. Neurology. corticospinal tract. Am J Neuroradiol. 2006;27(6):1234–8. 1999;53(5):1051–8. 72. Brettschneider J, Del Tredici K, Toledo JB, Robinson JL, Irwin DJ, Grossman 52. Sage CA, Van Hecke W, Peeters R, Sijbers J, Robberecht W, Parizel P, et al. M, et al. Stages of pTDP-43 pathology in amyotrophic lateral sclerosis: ALS Quantitative diffusion tensor imaging in amyotrophic lateral sclerosis: Stages. Ann Neurol. 2013;74(1):20–38. revisited. Hum Brain Mapp. 2009;30(11):3657–75. 73. Kassubek J, Muller H-P, Del Tredici K, Brettschneider J, Pinkhardt EH, Lule D, 53. Sarica A, Cerasa A, Vasta R, Perrotta P, Valentino P, Mangone G, et al. et al. Diffusion tensor imaging analysis of sequential spreading of disease in Tractography in amyotrophic lateral sclerosis using a novel probabilistic amyotrophic lateral sclerosis confirms patterns of TDP-43 pathology. Brain. tool: a study with tract-based reconstruction compared to voxel-based 2014;137(6):1733–40. approach. J Neurosci Methods. 2014;224:79–87. 74. Ding X-Q, Kollewe K, Blum K, Körner S, Kehbel S, Dengler R, et al. Value of 54. Schimrigk SK, Bellenberg B, Schlüter M, Stieltjes B, Drescher R, Rexilius J, quantitative analysis of routine clinical MRI sequences in ALS. Amyotroph et al. Diffusion tensor imaging-based fractional anisotropy quantification Lateral Scler. 2011;12(6):406–13. in the corticospinal tract of patients with amyotrophic lateral sclerosis 75. Bede P, Elamin M, Byrne S, McLaughlin RL, Kenna K, Vajda A, et al. Patterns using a probabilistic mixture model. AJNR Am J Neuroradiol. of cerebral and cerebellar white matter degeneration in ALS. J Neurol 2007;28(4):724–30. Neurosurg Psychiatry. 2015;86(4):468–70. 55. Foerster BR, Carlos RC, Dwamena BA, Callaghan BC, Petrou M, Edden RAE, 76. Jacob S, Finsterbusch J, Weishaupt JH, Khorram-Sefat D, Frahm J, Ehrenreich et al. Multimodal MRI as a diagnostic biomarker for amyotrophic lateral H. Diffusion tensor imaging for long-term follow-up of corticospinal tract sclerosis. Ann Clin Transl Neurol. 2014;1(2):107–14. degeneration in amyotrophic lateral sclerosis. Neuroradiology. 56. Filippini N, Douaud G, MacKay CE, Knight S, Talbot K, Turner MR. Corpus 2003;45(9):598–600. callosum involvement is a consistent feature of amyotrophic lateral 77. Menke RAL, Abraham I, Thiel CS, Filippini N, Knight S, Talbot K, et al. sclerosis. Neurology. 2010;75(18):1645–52. Fractional Anisotropy in the Posterior Limb of the Internal Capsule and 57. Rosskopf J, Müller H-P, Dreyhaupt J, Gorges M, Ludolph AC, Kassubek J. Ex Prognosis in Amyotrophic Lateral Sclerosis. Arch Neurol. 2012;69(11):1493. post facto assessment of diffusion tensor imaging metrics from different 78. Menke RAL, Korner S, Filippini N, Douaud G, Knight S, Talbot K, et al. Widespread MRI protocols: Preparing for multicentre studies in ALS. Amyotroph Lateral grey matter pathology dominates the longitudinal cerebral MRI and clinical Scler Front Degener. 2015;9:1–10. landscape of amyotrophic lateral sclerosis. Brain. 2014;137(9):2546–55. 58. Tang M, Chen X, Zhou Q, Liu B, Liu Y, Liu S, et al. Quantitative assessment 79. Lombardo F, Frijia F, Bongioanni P, Canapicchi R, Minichilli F, Bianchi F, et al. of amyotrophic lateral sclerosis with diffusion tensor imaging in 3.0 T Diffusion tensor MRI and MR spectroscopy in long lasting upper motor magnetic resonance. Int J Clin Exp Med. 2015;8(5):8295–303. neuron involvement in amyotrophic lateral sclerosis. Arch Ital Biol. 59. Pyra T, Hui B, Hanstock C, Concha L, Wong JCT, Beaulieu C, et al. 2009;147(3):69–82. Combined structural and neurochemical evaluation of the corticospinal 80. Cardenas-Blanco A, Machts J, Acosta-Cabronero J, Kaufmann J, Abdulla S, tract in amyotrophic lateral sclerosis. Amyotroph Lateral Scler. 2010; Kollewe K, et al. Central white matter degeneration in bulbar- and limb- 11(1–2):157–65. onset amyotrophic lateral sclerosis. J Neurol. 2014;261(10):1961–7. 60. Prell T, Peschel T, Hartung V, Kaufmann J, Klauschies R, Bodammer N, et al. 81. Foerster BR, Dwamena BA, Petrou M, Carlos RC, Callaghan BC, Churchill CL, Diffusion tensor imaging patterns differ in bulbar and limb onset et al. Diagnostic accuracy of diffusion tensor imaging in amyotrophic lateral amyotrophic lateral sclerosis. Clin Neurol Neurosurg. 2013;115(8):1281–7. sclerosis: a systematic review and individual patient data meta-analysis. 61. Keil C, Prell T, Peschel T, Hartung V, Dengler R, Grosskreutz J. Longitudinal Acad Radiol. 2013;20(9):1099–106. diffusion tensor imaging in amyotrophic lateral sclerosis. BMC Neurosci. 82. Galanaud D, Nicoli F, Confort-Gouny S, Le Fur Y, Dormont D, Girard N, et al. 2012;13(1):141. Brain magnetic resonance spectroscopy. J Radiol. 2007;88(3 Pt 2):483–96. Grolez et al. BMC Neurology (2016) 16:155 Page 16 of 17

83. Jissendi Tchofo P, Balériaux D. Brain (1)H-MR spectroscopy in clinical 106. Oba H, Araki T, Ohtomo K, Monzawa S, Uchiyama G, Koizumi K, et al. neuroimaging at 3T. J Neuroradiol. 2009;36(1):24–40. Amyotrophic lateral sclerosis: T2 shortening in motor cortex at MR imaging. 84. Soares DP, Law M. Magnetic resonance spectroscopy of the brain: review of Radiology. 1993;189(3):843–6. metabolites and clinical applications. Clin Radiol. 2009;64(1):12–21. 107. Hecht MJ, Fellner C, Schmid A, Neundörfer B, Fellner FA. Cortical T2 signal 85. Han J, Ma L. Study of the features of proton MR spectroscopy ((1)H-MRS) on shortening in amyotrophic lateral sclerosis is not due to iron deposits. amyotrophic lateral sclerosis. J Magn Reson Imaging JMRI. 2010;31(2):305–8. Neuroradiology. 2005;47(11):805–8. 86. Rooney WD, Miller RG, Gelinas D, Schuff N, Maudsley AA, Weiner MW. 108. Ignjatović A, Stević Z, Lavrnić S, Daković M, Bačić G. Brain iron MRI: A Decreased N-acetylaspartate in motor cortex and corticospinal tract in ALS. biomarker for amyotrophic lateral sclerosis: Brain Iron MRI: A Biomarker for Neurology. 1998;50(6):1800–5. ALS. J Magn Reson Imaging. 2013;38(6):1472–9. 87. Liu C, Jiang R, Yi X, Zhu W, Bu B. Role of diffusion tensor imaging or magnetic 109. Yu J, Qi F, Wang N, Gao P, Dai S, Lu Y, et al. Increased iron level in motor resonance spectroscopy in the diagnosis and disability assessment of cortex of amyotrophic lateral sclerosis patients: An in vivo MR study. amyotrophic lateral sclerosis. J Neurol Sci. 2015;348(1–2):206–10. Amyotroph Lateral Scler Front Degener. 2014;15(5–6):357–61. 88. Kalra S, Vitale A, Cashman NR, Genge A, Arnold DL. Cerebral degeneration 110. Langkammer C, Enzinger C, Quasthoff S, Grafenauer P, Soellinger M, Fazekas predicts survival in amyotrophic lateral sclerosis. J Neurol Neurosurg F, et al. Mapping of iron deposition in conjunction with assessment of Psychiatry. 2006;77(11):1253–5. nerve fiber tract integrity in amyotrophic lateral sclerosis. J Magn Reson 89. Kalra S, Hanstock CC, Martin WW, Allen PS, Johnston WS. Detection of Imaging JMRI. 2010;31(6):1339–45. cerebral degeneration in amyotrophic lateral sclerosis using high-field 111. Prell T, Hartung V, Tietz F, Penzlin S, Ilse B, Schweser F, et al. Susceptibility- magnetic resonance spectroscopy. Arch Neurol. 2006;63(8):1144–8. Weighted Imaging Provides Insight into White Matter Damage in 90. Cervo A, Cocozza S, Saccà F, Giorgio SMDA, Morra VB, Tedeschi E, et al. The Amyotrophic Lateral Sclerosis. Kassubek J, editor. PLOS ONE. 2015 Jun 25; combined use of conventional MRI and MR spectroscopic imaging 10(6):e0131114. increases the diagnostic accuracy in amyotrophic lateral sclerosis. Eur J 112. Schweitzer AD, Liu T, Gupta A, Zheng K, Seedial S, Shtilbans A, et al. Quantitative Radiol. 2015;84(1):151–7. Susceptibility Mapping of the Motor Cortex in Amyotrophic Lateral Sclerosis and 91. Pohl C, Block W, Träber F, Schmidt S, Pels H, Grothe C, et al. Proton Primary Lateral Sclerosis. Am J Roentgenol. 2015;204(5):1086–92. magnetic resonance spectroscopy and transcranial magnetic stimulation for 113. Branco LMT, De Albuquerque M, De Andrade HMT, Bergo FPG, Nucci A, the detection of upper motor neuron degeneration in ALS patients. J França MC. Spinal cord atrophy correlates with disease duration and Neurol Sci. 2001;190(1–2):21–7. severity in amyotrophic lateral sclerosis. Amyotroph Lateral Scler Front 92. Govind V, Sharma KR, Maudsley AA, Arheart KL, Saigal G, Sheriff S. Degener. 2014;15(1–2):93–7. Comprehensive evaluation of corticospinal tract metabolites in amyotrophic 114. Cohen-Adad J, Mendili M-ME, Morizot-Koutlidis R, Lehéricy S, Meininger V, lateral sclerosis using whole-brain 1H MR spectroscopy. PLoS One. Blancho S, et al. Involvement of spinal sensory pathway in ALS and 2012;7(4):e35607. specificity of cord atrophy to lower motor neuron degeneration. 93. Verma G, Woo JH, Chawla S, Wang S, Sheriff S, Elman LB, et al. Whole-brain Amyotroph Lateral Scler Front Degener. 2013;14(1):30–8. analysis of amyotrophic lateral sclerosis by using echo-planar spectroscopic 115. El Mendili M-M, Cohen-Adad J, Pelegrini-Issac M, Rossignol S, Morizot- imaging. Radiology. 2013;267(3):851–7. Koutlidis R, Marchand-Pauvert V, et al. Multi-parametric spinal cord MRI as 94. Bowen BC, Pattany PM, Bradley WG, Murdoch JB, Rotta F, Younis AA, et al. potential progression marker in amyotrophic lateral sclerosis. PLoS One. MR imaging and localized proton spectroscopy of the precentral gyrus in 2014;9(4):e95516. amyotrophic lateral sclerosis. Am J Neuroradiol. 2000;21(4):647–58. 116. Iglesias C, Sangari S, El Mendili M-M, Benali H, Marchand-Pauvert V, Pradat 95. Foerster BR, Callaghan BC, Petrou M, Edden RAE, Chenevert TL, Feldman EL. P-F. Electrophysiological and spinal imaging evidences for sensory Decreased motor cortex γ-aminobutyric acid in amyotrophic lateral dysfunction in amyotrophic lateral sclerosis. BMJ Open. 2015;5(2):e007659. sclerosis. Neurology. 2012;78(20):1596–600. 117. Ikeda K, Murata K, Kawase Y, Kawabe K, Kano O, Yoshii Y, et al. Relationship 96. Unrath A, Ludolph AC, Kassubek J. Brain metabolites in definite between cervical cord 1H-magnetic resonance spectroscopy and clinoco- amyotrophic lateral sclerosis: A longitudinal proton magnetic resonance electromyographic profile in amyotrophic lateral sclerosis. Muscle Nerve. spectroscopy study. J Neurol. 2007;254(8):1099–106. 2013;47(1):61–7. 97. Wang Q, Zhang X, Chen S, Zhang X, Zhang S, Youdium M, et al. Prevention of 118. Barkhof F, Haller S, Rombouts SARB. Resting-state functional MR imaging: a Motor Neuron Degeneration by Novel Iron Chelators in SOD1(G93A) Transgenic new window to the brain. Radiology. 2014;272(1):29–49. Mice of Amyotrophic Lateral Sclerosis. Neurodegener Dis. 2011;8(5):310–21. 119. Habas C. Physiological basis of functional MRI. J Radiol. 2002;83(11):1737–41. 98. Deistung A, Schäfer A, Schweser F, Biedermann U, Turner R, Reichenbach JR. 120. Delmaire C, Krainik A, Lethuc V, Reyns N, Duffau H, Capelle L, et al. Toward in vivo histology: A comparison of quantitative susceptibility Functional magnetic resonance imaging: physiopathology, techniques and mapping (QSM) with magnitude-, phase-, and R2*-imaging at ultra-high applications. J Radiol. 2007;88(3 Pt 2):497–509. magnetic field strength. Neuroimage. 2013;65:299–314. 121. Gore JC. Principles and practice of functional MRI of the human brain. J Clin 99. Haacke EM, Cheng NYC, House MJ, Liu Q, Neelavalli J, Ogg RJ, et al. Imaging Invest. 2003;112(1):4–9. iron stores in the brain using magnetic resonance imaging. Magn Reson 122. Raichle ME. The Restless Brain. Brain Connect. 2011;1(1):3–12. Imaging. 2005;23(1):1–25. 123. Agosta F, Canu E, Valsasina P, Riva N, Prelle A, Comi G, et al. Divergent brain 100. Adachi Y, Sato N, Saito Y, Kimura Y, Nakata Y, Ito K, et al. Usefulness of SWI for network connectivity in amyotrophic lateral sclerosis. Neurobiol Aging. the Detection of Iron in the Motor Cortex in Amyotrophic Lateral Sclerosis: 2013;34(2):419–27. Usefulness of SWI for the Diagnoses of ALS. J Neuroimaging. 2015;25(3):443–51. 124. Jelsone-Swain LM, Fling BW, Seidler RD, Hovatter R, Gruis K, Welsh RC. 101. Chavhan GB, Babyn PS, Thomas B, Shroff MM, Haacke EM. Principles, Reduced Interhemispheric Functional Connectivity in the Motor Cortex Techniques, and Applications of T2*-based MR Imaging and Its Special during Rest in Limb-Onset Amyotrophic Lateral Sclerosis. Front Syst Applications1. RadioGraphics. 2009;29(5):1433–49. Neurosci [Internet]. 2010 [cited 2015 Jul 15];4. Available from: http://journal. 102. Haacke EM, Miao Y, Liu M, Habib CA, Katkuri Y, Liu T, et al. Correlation of frontiersin.org/article/10.3389/fnsys.2010.00158/abstract putative iron content as represented by changes in R2* and phase with age 125. Agosta F, Valsasina P, Absinta M, Riva N, Sala S, Prelle A, et al. Sensorimotor in deep gray matter of healthy adults. J Magn Reson Imaging JMRI. 2010; Functional Connectivity Changes in Amyotrophic Lateral Sclerosis. Cereb 32(3):561–76. Cortex. 2011;21(10):2291–8. 103. Langkammer C, Krebs N, Goessler W, Scheurer E, Ebner F, Yen K, et al. 126. Mohammadi B, Kollewe K, Samii A, Krampfl K, Dengler R, Münte TF. Changes Quantitative MR imaging of brain iron: a postmortem validation study. of resting state brain networks in amyotrophic lateral sclerosis. Exp Neurol. Radiology. 2010;257(2):455–62. 2009;217(1):147–53. 104. Haacke EM, Liu S, Buch S, Zheng W, Wu D, Ye Y. Quantitative susceptibility 127. Douaud G, Filippini N, Knight S, Talbot K, Turner MR. Integration of mapping: current status and future directions. Magn Reson Imaging. 2015; structural and functional magnetic resonance imaging in amyotrophic 33(1):1–25. lateral sclerosis. Brain. 2011;134(12):3470–9. 105. Langkammer C, Schweser F, Krebs N, Deistung A, Goessler W, Scheurer E, 128. Luo C, Chen Q, Huang R, Chen X, Chen K, Huang X, et al. Patterns of et al. Quantitative susceptibility mapping (QSM) as a means to measure Spontaneous Brain Activity in Amyotrophic Lateral Sclerosis: A Resting-State brain iron? A post mortem validation study. Neuroimage. 2012;62(3):1593–9. fMRI Study. Duce JA, editor. PLoS ONE. 2012 Sep 20;7(9):e45470. Grolez et al. BMC Neurology (2016) 16:155 Page 17 of 17

129. Mohammadi B, Kollewe K, Samii A, Dengler R, Münte TF. Functional neuroimaging at different disease stages reveals distinct phases of neuroplastic changes in amyotrophic lateral sclerosis. Hum Brain Mapp. 2011;32(5):750–8. 130. Mohammadi B, Kollewe K, Cole DM, Fellbrich A, Heldmann M, Samii A, et al. Amyotrophic lateral sclerosis affects cortical and subcortical activity underlying motor inhibition and action monitoring: ALS and Motor Inhibition. Hum Brain Mapp. 2015;36(8):2878–89. 131. Kollewe K, Münte TF, Samii A, Dengler R, Petri S, Mohammadi B. Patterns of cortical activity differ in ALS patients with limb and/or bulbar involvement depending on motor tasks. J Neurol. 2011;258(5):804–10. 132. Konrad C, Henningsen H, Bremer J, Mock B, Deppe M, Buchinger C, et al. Pattern of cortical reorganization in amyotrophic lateral sclerosis: a functional magnetic resonance imaging study. Exp Brain Res. 2002;143(1):51–6. 133. Schoenfeld MA, Tempelmann C, Gaul C, Kühnel GR, Düzel E, Hopf J-M, et al. Functional motor compensation in amyotrophic lateral sclerosis. J Neurol. 2005;252(8):944–52. 134. Lule D, Diekmann V, Muller H-P, Kassubek J, Ludolph AC, Birbaumer N. Neuroimaging of multimodal sensory stimulation in amyotrophic lateral sclerosis. J Neurol Neurosurg Psychiatry. 2010;81(8):899–906. 135. Stoppel CM, Vielhaber S, Eckart C, Machts J, Kaufmann J, Heinze H-J, et al. Structural and functional hallmarks of amyotrophic lateral sclerosis progression in motor- and memory-related brain regions. NeuroImage Clin. 2014;5:277–90. 136. Jelsone-Swain L, Persad C, Burkard D, Welsh RC. Action Processing and Mirror Neuron Function in Patients with Amyotrophic Lateral Sclerosis: An fMRI Study. Iacoboni M, editor. PLOS ONE. 2015 Apr 17;10(4):e0119862. 137. Bede P, Hardiman O. Lessons of ALS imaging: Pitfalls and future directions - A critical review. NeuroImage Clin. 2014;4:436–43. 138. Brooks BR, Miller RG, Swash M, Munsat TL, World Federation of Neurology Research Group on Motor Neuron Diseases. El Escorial revisited: revised criteria for the diagnosis of amyotrophic lateral sclerosis. Amyotroph Lateral Scler Mot Neuron Disord Off Publ World Fed Neurol Res Group Mot Neuron Dis. 2000;1(5):293–9. 139. Andersen PM, Borasio GD, Dengler R, Hardiman O, Kollewe K, Leigh PN, et al. EFNS task force on management of amyotrophic lateral sclerosis: guidelines for diagnosing and clinical care of patients and relatives. Eur J Neurol. 2005;12(12):921–38. 140. Turner MR, Kiernan MC, Leigh PN, Talbot K. Biomarkers in amyotrophic lateral sclerosis. Lancet Neurol. 2009;8(1):94–109. 141. Pradat P-F, Bruneteau G. [Clinical characteristics of amyotrophic lateral sclerosis subsets]. Rev Neurol (Paris). 2006 Jun;162 Spec No 2:4S29-24S33. 142. Zoccolella S, Beghi E, Palagano G, Fraddosio A, Samarelli V, Lamberti P, et al. Predictors of delay in the diagnosis and clinical trial entry of amyotrophic lateral sclerosis patients: a population-based study. J Neurol Sci. 2006;250(1– 2):45–9. 143. Pradat P-F, El Mendili M-M. Neuroimaging to investigate multisystem involvement and provide biomarkers in amyotrophic lateral sclerosis. BioMed Res Int. 2014;2014:467560. 144. Turner MR, Grosskreutz J, Kassubek J, Abrahams S, Agosta F, Benatar M, et al. Towards a neuroimaging biomarker for amyotrophic lateral sclerosis. Lancet Neurol. 2011;10(5):400–3. 145. Filippi M, Agosta F, Grosskreutz J, Benatar M, Kassubek J, Verstraete E, et al. Progress towards a neuroimaging biomarker for amyotrophic lateral sclerosis. Lancet Neurol. 2015;14(8):786–8. 146. Müller H-P, Turner MR, Grosskreutz J, Abrahams S, Bede P, Govind V, et al. A large-scale multicentre cerebral diffusion tensor imaging study in amyotrophic lateral sclerosis. J Neurol Neurosurg Psychiatry. 2016;87(6):570–9. Submit your next manuscript to BioMed Central and we will help you at every step: • We accept pre-submission inquiries • Our selector tool helps you to ﬁnd the most relevant journal • We provide round the clock customer support • Convenient online submission • Thorough peer review • Inclusion in PubMed and all major indexing services • Maximum visibility for your research

Submit your manuscript at www.biomedcentral.com/submit

References

Agosta F, Chiò A, Cosottini M, De Stefano N, Falini A, Mascalchi M, Rocca MA, Silani V, Tedeschi G, Filippi M. The present and the future of neuroimaging in amyotrophic lateral sclerosis. AJNR Am J Neuroradiol. 2010 Nov;31(10):1769-77.

Agosta F, Gorno-Tempini ML, Pagani E, Sala S, Caputo D, Perini M, Bartolomei I, Fruguglietti ME, Filippi M. Longitudinal assessment of grey matter contraction in amyotrophic lateral sclerosis: A tensor based morphometry study. Amyotroph Lateral Scler. 2009 Jun;10(3):168-74.

Agosta F, Rocca MA, Valsasina P, Sala S, Caputo D, Perini M, Salvi F, Prelle A, Filippi M. A longitudinal diffusion tensor MRI study of the cervical cord and brain in amyotrophic lateral sclerosis patients. J Neurol Neurosurg Psychiatry. 2009 Jan;80(1):53-5.

Alcaz S, Jarebinski M, Pekmezovic T, Stevic-Marinkovic Z, Pavlovic S, Apostolski S. Epidemiological and clinical characteristics of ALS in Belgrade, Yugoslavia. Acta Neurol Scand. 1996 Oct;94(4):264-8.

Ambikairajah A, Devenney E, Flanagan E, Yew B, Mioshi E, Kiernan MC, Hodges JR, Hornberger M. A visual MRI atrophy rating scale for the amyotrophic lateral sclerosis-frontotemporal dementia continuum. Amyotroph Lateral Scler Frontotemporal Degener. 2014 Jun;15(3-4):226-34.

Andersen PM, Al-Chalabi A. Clinical genetics of amyotrophic lateral sclerosis: what do we really know? Nat Rev Neurol. 2011 Oct 11;7(11):603-15.

Anderson WA. Electrical current shims for correcting magnetic fields. Review of Scientific Instruments 1961;32(3):241–250.

Atlas SW. Magnetic Resonance Imaging of the Brain and Spine. Lippincott Williams & Wilkins, 2009.

Averback P, Crocker P. Regular involvement of Clarke's nucleus in sporadic amyotrophic lateral sclerosis. Arch Neurol. 1982 Mar;39(3):155-6.

Bali T, Self W, Liu J, Siddique T, Wang LH, Bird TD, Ratti E, Atassi N, Boylan KB, Glass JD, Maragakis NJ, Caress JB, McCluskey LF, Appel SH, Wymer JP, Gibson S, Zinman L, Mozaffar T,

Callaghan B, McVey AL, Jockel-Balsarotti J, Allred P, Fisher ER, Lopate G, Pestronk A, Cudkowicz

- 142 -

ME, Miller TM. Defining SOD1 ALS natural history to guide therapeutic clinical trial design. J Neurol Neurosurg Psychiatry. 2016 Jun 3.

Beaulieu C. The basis of anisotropic water diffusion in the nervous system - a technical review. NMR Biomed. 2002 Nov-Dec;15(7-8):435-55.

Bede P, Iyer PM, Schuster C, Elamin M, Mclaughlin RL, Kenna K, Hardiman O. The selective anatomical vulnerability of ALS: 'disease-defining' and 'disease-defying' brain regions. Amyotroph Lateral Scler Frontotemporal Degener. 2016 Apr 18:1-10.

Beers DR, Henkel JS, Xiao Q, Zhao W, Wang J, Yen AA, Siklos L, McKercher SR, Appel SH. Wild- type microglia extend survival in PU.1 knockout mice with familial amyotrophic lateral sclerosis. Proc Natl Acad Sci U S A. 2006 Oct 24;103(43):16021-6.

Bérard C, Payan C, Hodgkinson I, Fermanian J; MFM Collaborative Study Group. A motor function measure for neuromuscular diseases. Construction and validation study. Neuromuscul Disord. 2005 Jul;15(7):463-70.

Boyer JG, Deguise MO, Murray LM, Yazdani A, De Repentigny Y, Boudreau-Larivière C, Kothary R. Myogenic program dysregulation is contributory to disease pathogenesis in spinal muscular atrophy. Hum Mol Genet. 2014 Aug 15;23(16):4249-59.

Bradley WG, Good P, Rasool CG, Adelman LS. Morphometric and biochemical studies of peripheral nerves in amyotrophic lateral sclerosis. Ann Neurol. 1983 Sep;14(3):267-77.

Branco LM, De Albuquerque M, De Andrade HM, Bergo FP, Nucci A, França MC Jr. Spinal cord atrophy correlates with disease duration and severity in amyotrophic lateral sclerosis. Amyotroph Lateral Scler Frontotemporal Degener. 2014 Mar;15(1-2):93-7.

Bruneteau G, Simonet T, Bauché S, Mandjee N, Malfatti E, Girard E, Tanguy ML, Behin A, Khiami F, Sariali E, Hell-Remy C, Salachas F, Pradat PF, Fournier E, Lacomblez L, Koenig J, Romero NB, Fontaine B, Meininger V, Schaeffer L, Hantaï D. Muscle histone deacetylase 4 upregulation in amyotrophic lateral sclerosis: potential role in reinnervation ability and disease progression. Brain. 2013 Aug;136(Pt 8):2359-68.

Budde MD, Frank JA. Neurite beading is sufficient to decrease the apparent diffusion coefficient after

- 143 -

ischemic stroke. Proc Natl Acad Sci U S A. 2010 Aug 10;107(32):14472-7.

Budde MD, Kim JH, Liang HF, Schmidt RE, Russell JH, Cross AH, Song SK. Toward accurate diagnosis of white matter pathology using diffusion tensor imaging. Magn Reson Med. 2007 Apr;57(4):688-95.

Byrne S, Walsh C, Lynch C, Bede P, Elamin M, Kenna K, McLaughlin R, Hardiman O. Rate of familial amyotrophic lateral sclerosis: a systematic review and meta-analysis. J Neurol Neurosurg Psychiatry. 2011 Jun;82(6):623-7.

Byun MS, Kim SE, Park J, Yi D, Choe YM, Sohn BK, Choi HJ, Baek H, Han JY, Woo JI, Lee DY; Alzheimer’s Disease Neuroimaging Initiative. Heterogeneity of Regional Brain Atrophy Patterns Associated with Distinct Progression Rates in Alzheimer's Disease. PLoS One. 2015 Nov 30;10(11):e0142756.

Campbell JS, Siddiqi K, Rymar VV, Sadikot AF, Pike GB. Flow-based fiber tracking with diffusion tensor and q-ball data: validation and comparison to principal diffusion direction techniques. Neuroimage. 2005 Oct 1;27(4):725-36.

Camu W, Khoris J, Moulard B, Salachas F, Briolotti V, Rouleau GA, Meininger V. Genetics of familial ALS and consequences for diagnosis. French ALS Research Group. J Neurol Sci. 1999 Jun;165 Suppl 1:S21-6.

Carbonell-Caballero J, Manjón JV, Martí-Bonmatí L, Olalla JR, Casanova B, de la Iglesia-Vayá M, Coret F, Robles M. Accurate quantification methods to evaluate cervical cord atrophy in multiple sclerosis patients. MAGMA. 2006 Nov;19(5):237-46.

Carr PA, Blanchard S, Doeden L, Al-Marsoummi S. Interneuron loss in a mouse model of amyotrophic lateral sclerosis. 2010. Program No.82.8. Neuroscience Meeting Planner. San Diego, CA: Society for Neuroscience, Online.

Carrara G, Carapelli C, Venturi F, Ferraris MM, Lequio L, Chiò A, Calvo A, Sirgiovanni S, Cistaro A, Valentini MC. A distinct MR imaging phenotype in amyotrophic lateral sclerosis: correlation between T1 magnetization transfer contrast hyperintensity along the corticospinal tract and diffusion tensor imaging analysis. AJNR Am J Neuroradiol. 2012 Apr;33(4):733-9.

- 144 -

Carter GT, Abresch RT, Fowler WM Jr, Johnson ER, Kilmer DD, McDonald CM. Profiles of neuromuscular diseases. Spinal muscular atrophy. Am J Phys Med Rehabil. 1995 Sep-Oct;74(5 Suppl):S150-9.

Chang Q, Martin LJ. Glycine receptor channels in spinal motoneurons are abnormal in a transgenic mouse model of amyotrophic lateral sclerosis. J Neurosci 2011; 31(8):2815-2827.

Cheah BC, Vucic S, Krishnan AV, Kiernan MC. Riluzole, neuroprotection and amyotrophic lateral sclerosis. Curr Med Chem. 2010;17(18):1942-199.

Cheung G, Gawel MJ, Cooper PW, Farb RI, Ang LC, Gawal MJ. Amyotrophic lateral sclerosis: correlation of clinical and MR imaging findings. Radiology 1995;194(1):263-270.

Chevis CF, da Silva CB, D'Abreu A, Lopes-Cendes I, Cendes F, Bergo FP, França MC Jr. Spinal cord atrophy correlates with disability in Friedreich's ataxia. Cerebellum. 2013 Feb;12(1):43-7.

Chiò A, Calvo A, Moglia C, Mazzini L, Mora G; PARALS study group. Phenotypic heterogeneity of amyotrophic lateral sclerosis: a population based study. J Neurol Neurosurg Psychiatry. 2011 Jul;82(7):740-6.

Chiò A, Logroscino G, Traynor BJ, Collins J, Simeone JC, Goldstein LA, White LA. Global epidemiology of amyotrophic lateral sclerosis: a systematic review of the published literature. Neuroepidemiology. 2013;41(2):118-30.

Chiò A, Mora G, Calvo A, Mazzini L, Bottacchi E, Mutani R; PARALS. Epidemiology of ALS in Italy: a 10-year prospective population-based study. Neurology. 2009 Feb 24;72(8):725-31.

Chiò A, Mora G, Leone M, Mazzini L, Cocito D, Giordana MT, Bottacchi E, Mutani R; Piemonte and Valle d'Aosta Register for ALS (PARALS). Early symptom progression rate is related to ALS outcome: a prospective population-based study. Neurology. 2002 Jul 9;59(1):99-103.

Chiò A, Pagani M, Agosta F, Calvo A, Cistaro A, Filippi M. Neuroimaging in amyotrophic lateral sclerosis: insights into structural and functional changes. Lancet Neurol. 2014 Dec;13(12):1228-40.

Choe AS, Stepniewska I, Colvin DC, Ding Z, Anderson AW. Validation of diffusion tensor MRI in the central nervous system using light microscopy: quantitative comparison of fiber properties. NMR

- 145 -

Biomed. 2012 Jul;25(7):900-8.

Clarke JL, Jackson JH. On a Case of Muscular Atrophy, with Disease of the Spinal Cord and Medulla Oblongata. Med Chir Trans. 1867;50:489-498.1.

Clement AM, Nguyen MD, Roberts EA, Garcia ML, Boillée S, Rule M, McMahon AP, Doucette W, Siwek D, Ferrante RJ, Brown RH Jr, Julien JP, Goldstein LS, Cleveland DW. Wild-type nonneuronal cells extend survival of SOD1 mutant motor neurons in ALS mice. Science. 2003 Oct 3;302(5642):113- 7. Erratum in: Science. 2003 Oct 24;302(5645):568.

Cohen-Adad J, Benali H, Rossignol S. Methodology for MR diffusion tensor imaging of the cat spinal cord. Conf Proc IEEE Eng Med Biol Soc. 2007;2007:323-6.

Cohen-Adad J, Descoteaux M, Rossignol S, Hoge RD, Deriche R, Benali H. Detection of multiple pathways in the spinal cord using q-ball imaging. Neuroimage. 2008 Aug 15;42(2):739-49.

Cohen-Adad J, El Mendili MM, Lehéricy S, Pradat PF, Blancho S, Rossignol S, Benali H. Demyelination and degeneration in the injured human spinal cord detected with diffusion and magnetization transfer MRI. Neuroimage. 2011 Apr 1;55(3):1024-33.

Cohen-Adad J, El Mendili MM, Morizot-Koutlidis R, Lehéricy S, Meininger V, Blancho S, Rossignol S, Benali H, Pradat PF. Involvement of spinal sensory pathway in ALS and specificity of cord atrophy to lower motor neuron degeneration. Amyotroph Lateral Scler Frontotemporal Degener. 2013 Jan;14(1):30-8.

Cohen-Adad J, Wheeler-Kingshott CAM. Quantitative MRI of the Spinal Cord. Elsevier, 2014.

Comi G, Rovaris M, Leocani L. Neuroimaging in amyotrophic lateral sclerosis. European Journal of Neurology 1999;6(6):629-637.

Concha L, Gross DW, Wheatley BM, Beaulieu C. Diffusion tensor imaging of time-dependent axonal and myelin degradation after corpus callosotomy in epilepsy patients. Neuroimage. 2006 Sep;32(3):1090-9.

Concha L, Livy DJ, Beaulieu C, Wheatley BM, Gross DW. In vivo diffusion tensor imaging and histopathology of the fimbria-fornix in temporal lobe epilepsy. J Neurosci. 2010 Jan 20;30(3):996- 1002.

- 146 -

Concha L. A macroscopic view of microstructure: using diffusion-weighted images to infer damage, repair, and plasticity of white matter. Neuroscience. 2014 Sep 12;276:14-28.

Constantinovici A. Abnormal somatosensory evoked potentials in amyotrophic lateral sclerosis. Rom J Neurol Psychiatry. 1993 Jul-Dec;31(3-4):273-8.

Cosi V, Poloni M, Mazzini L, Callieco R. Somatosensory evoked potentials in amyotrophic lateral sclerosis. J Neurol Neurosurg Psychiatry. 1984 Aug;47(8):857-61.

Coulon O, Hickman SJ, Parker GJ, Barker GJ, Miller DH, Arridge SR. Quantification of spinal cord atrophy from magnetic resonance images via a B-spline active surface model. Magn Reson Med. 2002 Jun;47(6):1176-85.

Cruz-Sánchez FF, Moral A, Tolosa E, de Belleroche J, Rossi ML. Evaluation of neuronal loss, astrocytosis and abnormalities of cytoskeletal components of large motor neurons in the human anterior horn in aging. J Neural Transm (Vienna). 1998;105(6-7):689-701.

D'Amico A, Mercuri E, Tiziano FD, Bertini E. Spinal muscular atrophy. Orphanet J Rare Dis. 2011 Nov 2;6:71.

D'Amico E, Pasmantier M, Lee YW, Weimer L, Mitsumoto H. Clinical evolution of pure upper motor neuron disease/dysfunction (PUMMD). Muscle Nerve 2013; 47(1):28-32. d'Errico P, Boido M, Piras A, Valsecchi V, De Amicis E, Locatelli D, Capra S, Vagni F, Vercelli A, Battaglia G. Selective vulnerability of spinal and cortical motor neuron subpopulations in delta7 SMA mice. PLoS One. 2013 Dec 6;8(12):e82654.

Davison C, Goodhart SP, Lander J. Multiple sclerosis and amyotrophies. Arch Neurol Psychiatry 1934;31:270 –289. de Carvalho M, Dengler R, Eisen A, England JD, Kaji R, Kimura J, Mills K, Mitsumoto H, Nodera H, Shefner J, Swash M. Electrodiagnostic criteria for diagnosis of ALS. Clin Neurophysiol. 2008 Mar;119(3):497-503. de Carvalho M, Matias T, Coelho F, Evangelista T, Pinto A, Luis ML. Motor neuron disease presenting

- 147 -

with respiratory failure. J Neurol Sci 1996; 139 Suppl:117-122.

De Leener B, Lévy S, Dupont SM, Fonov VS, Stikov N, Louis Collins D, Callot V, Cohen-Adad J. SCT: Spinal Cord Toolbox, an open-source software for processing spinal cord MRI data. Neuroimage. 2016 Oct 5.

De Leener B, Taso M, Cohen-Adad J, Callot V. Segmentation of the human spinal cord. MAGMA. 2016 Apr;29(2):125-53.

Defelipe J, Fields RD, Hof PR, Höistad M, Kostovic I, Meyer G, Rockland KS. Cortical white matter: beyond the pale remarks, main conclusions and discussion. Front Neuroanat. 2010 Mar 24;4:4. del Aguila MA, Longstreth WT Jr, McGuire V, Koepsell TD, van Belle G. Prognosis in amyotrophic lateral sclerosis: a population-based study. Neurology. 2003 Mar 11;60(5):813-9.

Dubowitz V. Infantile muscular atrophy--a broad spectrum. Clin Proc Child Hosp Dist Columbia. 1967 Sep;23(8):223-39.

Dubowitz V. Infantile muscular atrophy. A prospective study with particular reference to a slowly progressive variety. Brain. 1964 Dec;87:707-18.

Dubowitz V. Ramblings in the history of spinal muscular atrophy. Neuromuscul Disord. 2009 Jan;19(1):69-73.

Dupuis L, Gonzalez de Aguilar JL, di Scala F, Rene F, de Tapia M, Pradat PF, Lacomblez L, Seihlan D, Prinjha R, Walsh FS, Meininger V, Loeffler JP. Nogo provides a molecular marker for diagnosis of amyotrophic lateral sclerosis. Neurobiol Dis. 2002 Aug;10(3):358-65.

Ellingson BM, Salamon N, Holly LT. Advances in MR Imaging for Cervical Spondylotic Myelopathy. European spine journal: official publication of the European Spine Society, the European Spinal Deformity Society, and the European Section of the Cervical Spine Research Society. 2015;24(Suppl 2):197-208.

Fahl CN, Branco LM, Bergo FP, D'Abreu A, Lopes-Cendes I, França MC Jr. Spinal cord damage in Machado-Joseph disease. Cerebellum. 2015 Apr;14(2):128-32.

- 148 -

Figley CR, Stroman PW. Investigation of human cervical and upper thoracic spinal cord motion: implications for imaging spinal cord structure and function. Magn Reson Med. 2007 Jul;58(1):185-9.

Filippi M, Agosta F. Magnetization transfer MRI in multiple sclerosis. J Neuroimaging. 2007 Apr;17 Suppl 1:22S-26S.

Filippi M, Agosta F, Abrahams S, Fazekas F, Grosskreutz J, Kalra S, Kassubek J, Silani V, Turner MR, Masdeu JC; European Federation of Neurological Societies. EFNS guidelines on the use of neuroimaging in the management of motor neuron diseases. Eur J Neurol. 2010 Apr;17(4):526-e20.

Filippi M, Rocca MA. Magnetization transfer magnetic resonance imaging of the brain, spinal cord, and optic nerve. Neurotherapeutics. 2007 Jul;4(3):401-13.

Fischer LR, Culver DG, Tennant P, Davis AA, Wang M, Castellano-Sanchez A, Khan J, Polak MA, Glass JD. Amyotrophic lateral sclerosis is a distal axonopathy: evidence in mice and man. Exp Neurol. 2004 Feb;185(2):232-40.

Foerster BR, Callaghan BC, Petrou M, Edden RA, Chenevert TL, Feldman EL. Decreased motor cortex gamma-aminobutyric acid in amyotrophic lateral sclerosis. Neurology 2012; 78(20):1596-1600.

Fonov VS, Le Troter A, Taso M, De Leener B, Lévêque G, Benhamou M, Sdika M, Benali H, Pradat PF, Collins DL, Callot V, Cohen-Adad J. Framework for integrated MRI average of the spinal cord white and gray matter: the MNI-Poly-AMU template. Neuroimage. 2014 Nov 15;102 Pt 2:817-27.

Forbes RB, Colville S, Swingler RJ; Scottish ALS/MND Register. The epidemiology of amyotrophic lateral sclerosis (ALS/MND) in people aged 80 or over. Age Ageing. 2004 Mar;33(2):131-4.

Fradet L, Arnoux PJ, Ranjeva JP, Petit Y, Callot V. Morphometrics of the entire human spinal cord and spinal canal measured from in vivo high-resolution anatomical magnetic resonance imaging. Spine (Phila Pa 1976). 2014 Feb 15;39(4):E262-9.

Freund P, Weiskopf N, Ashburner J, Wolf K, Sutter R, Altmann DR, Friston K, Thompson A, Curt A. MRI investigation of the sensorimotor cortex and the corticospinal tract after acute spinal cord injury: a prospective longitudinal study. Lancet Neurol. 2013 Sep;12(9):873-81.

Fruehwald-Pallamar J, Szomolanyi P, Fakhrai N, Lunzer A, Weber M, Thurnher MM, Pallamar M,

- 149 -

Trattnig S, Prayer D, Noebauer-Huhmann IM. Parallel imaging of the cervical spine at 3T: optimized trade-off between speed and image quality. AJNR Am J Neuroradiol. 2012 Nov;33(10):1867-74.

Gardner-Medwin D, Hudgson P, Walton JN. Benign spinal muscular atrophy arising in childhood and adolescence. J Neurol Sci. 1967 Jul-Aug;5(1):121-58.

Gil J, Funalot B, Verschueren A, Danel-Brunaud V, Camu W, Vandenberghe N, Desnuelle C, Guy N, Camdessanche JP, Cintas P, Carluer L, Pittion S, Nicolas G, Corcia P, Fleury MC, Maugras C, Besson G, Le Masson G, Couratier P. Causes of death amongst French patients with amyotrophic lateral sclerosis: a prospective study. Eur J Neurol. 2008 Nov;15(11):1245-51.

Glockner JF, Hu HH, Stanley DW, Angelos L, King K. Parallel MR imaging: a user's guide. Radiographics. 2005 Sep-Oct;25(5):1279-97.

Goetz CG. Amyotrophic lateral sclerosis: early contributions of Jean-Martin Charcot. Muscle Nerve. 2000 Mar;23(3):336-43.

Goodin DS, Rowley HA, Olney RK. Magnetic resonance imaging in amyotrophic lateral sclerosis. Ann Neurol. 1988 Apr;23(4):418-20.

Gordon PH, Cheng B, Katz IB, Pinto M, Hays AP, Mitsumoto H, Rowland LP. The natural history of primary lateral sclerosis. Neurology. 2006 Mar 14;66(5):647-53.

Gordon PH, Salachas F, Bruneteau G, Pradat PF, Lacomblez L, Gonzalez-Bermejo J, Morelot-Panzini C, Similowski T, Elbaz A, Meininger V. Improving survival in a large French ALS center cohort. J Neurol. 2012 Sep;259(9):1788-92.

Goto N, Otsuka N. Development and anatomy of the spinal cord. Neuropathology 1997; 17: 25-31.

Graham JM, Papadakis N, Evans J, Widjaja E, Romanowski CA, Paley MN, Wallis LI, Wilkinson ID, Shaw PJ, Griffiths PD. Diffusion tensor imaging for the assessment of upper motor neuron integrity in ALS. Neurology. 2004 Dec 14;63(11):2111-9.

Gregory R, Mills K, Donaghy M. Progressive sensory nerve dysfunction in amyotrophic lateral sclerosis: a prospective clinical and neurophysiological study. J Neurol. 1993 May;240(5):309-14.

- 150 -

Guo YS, Wu DX, Wu HR, Wu SY, Yang C, Li B, Bu H, Zhang YS, Li CY. Sensory involvement in the SOD1-G93A mouse model of amyotrophic lateral sclerosis. Exp Mol Med. 2009 Mar 31;41(3):140-50.

Hamilton G, Gillingwater TH. Spinal muscular atrophy: going beyond the motor neuron. Trends Mol Med. 2013 Jan;19(1):40-50.

Haralick RM, Shapiro LG. Computer and robot vision: vol. 1. Bos- ton: Addison-Wesley. 1991.

Hausmanowa-Petrusewicz I, Fidzianska-Dolot A. Clinical features of infantile and juvenile spinal muscular atrophy. In: Gamstorp I, Sarnat HB, editors. Progressive spinal muscular atrophies. New York: Raven Press, 1984:31–42.

Hecht MJ, Fellner F, Fellner C, Hilz MJ, Heuss D, Neundörfer B. MRI-FLAIR images of the head show corticospinal tract alterations in ALS patients more frequently than T2-, T1- and proton-density- weighted images. J Neurol Sci. 2001 May 1;186(1-2):37-44.

Hickman SJ, Hadjiprocopis A, Coulon O, Miller DH, Barker GJ. Cervical spinal cord MTR histogram analysis in multiple sclerosis using a 3D acquisition and a B-spline active surface segmentation technique. Magn Reson Imaging. 2004 Jul;22(6):891-5.

Hofmann E, Ochs G, Pelzl A, Warmuth-Metz M. The corticospinal tract in amyotrophic lateral sclerosis: an MRI study. Neuroradiology. 1998 Feb;40(2):71-5.

Horsfield MA, Sala S, Neema M, Absinta M, Bakshi A, Sormani MP, Rocca MA, Bakshi R, Filippi M. Rapid semi-automatic segmentation of the spinal cord from magnetic resonance images: application in multiple sclerosis. Neuroimage. 2010 Apr 1;50(2):446-55.

Howard RS, Wiles CM, Loh L. Respiratory complications and their management in motor neuron disease. Brain 1989; 112 ( Pt 5):1155-1170.

Hua X, Ching CR, Mezher A, Gutman BA, Hibar DP, Bhatt P, Leow AD, Jack CR Jr, Bernstein MA, Weiner MW, Thompson PM; Alzheimer's Disease Neuroimaging Initiative. MRI-based brain atrophy

- 151 -

rates in ADNI phase 2: acceleration and enrichment considerations for clinical trials. Neurobiol Aging. 2016 Jan;37:26-37.

Iglesias C, Sangari S, El Mendili MM, Benali H, Marchand-Pauvert V, Pradat PF. Electrophysiological and spinal imaging evidences for sensory dysfunction in amyotrophic lateral sclerosis. BMJ Open 2015; 5(2):e007659.

Ilieva H, Polymenidou M, Cleveland DW. Non-cell autonomous toxicity in neurodegenerative disorders: ALS and beyond. J Cell Biol. 2009 Dec 14;187(6):761-72.

Ille R, Schäfer A, Scharmüller W, Enzinger C, Schöggl H, Kapfhammer HP, Schienle A. Emotion recognition and experience in Huntington disease: a voxel-based morphometry study. J Psychiatry Neurosci. 2011 Nov;36(6):383-90.

Ince PG, Evans J, Knopp M, Forster G, Hamdalla HH, Wharton SB, Shaw PJ. Corticospinal tract degeneration in the progressive muscular atrophy variant of ALS. Neurology. 2003 Apr 22;60(8):1252- 8.

Ishiura H, Tsuji S. Epidemiology and molecular mechanism of frontotemporal lobar degeneration/amyotrophic lateral sclerosis with repeat expansion mutation in C9orf72. J Neurogenet. 2015;29(2-3):85-94.

Jaermann T, Crelier G, Pruessmann KP, Golay X, Netsch T, van Muiswinkel AM, Mori S, van Zijl PC, Valavanis A, Kollias S, Boesiger P. SENSE-DTI at 3 T. Magn Reson Med. 2004 Feb;51(2):230-6.

Joensen P. Incidence of amyotrophic lateral sclerosis in the Faroe Islands. Acta Neurol Scand. 2012 Jul;126(1):62-6.

Jones C, Oskoui M, Zielinski D, Vinikoor L, Farwell W. Systematic review of incidence and prevalence of spinal muscular atrophy (SMA). Eur J Paediatr Neurol. 2016;19s(2015):S64-S65.

Jucker M, Walker LC. Self-propagation of pathogenic protein aggregates in neurodegenerative diseases. Nature 2013; 501(7465):45-51.

Kariya S, Park GH, Maeno-Hikichi Y, Leykekhman O, Lutz C, Arkovitz MS, Landmesser LT, Monani UR. Reduced SMN protein impairs maturation of the neuromuscular junctions in mouse models of

- 152 -

spinal muscular atrophy. Hum Mol Genet. 2008 Aug 15;17(16):2552-69.

Kato Y, Matsumura K, Kinosada Y, Narita Y, Kuzuhara S, Nakagawa T. Detection of pyramidal tract lesions in amyotrophic lateral sclerosis with magnetization-transfer measurements. AJNR Am J Neuroradiol. 1997 Sep;18(8):1541-7.

Katsuno M, Banno H, Suzuki K, Adachi H, Tanaka F, Sobue G. Clinical features and molecular mechanisms of spinal and bulbar muscular atrophy (SBMA). Adv Exp Med Biol. 2010;685:64-74.

Katz JS, Dimachkie MM, Barohn RJ. Amyotrophic Lateral Sclerosis: A Historical Perspective. Neurol Clin. 2015 Nov;33(4):727-34.

Kawamura Y, Dyck PJ, Shimono M, Okazaki H, Tateishi J, Doi H. Morphometric comparison of the vulnerability of peripheral motor and sensory neurons in amyotrophic lateral sclerosis. J Neuropathol Exp Neurol. 1981 Nov;40(6):667-75.

Kearney H, Miller DH, Ciccarelli O. Spinal cord MRI in multiple sclerosis--diagnostic, prognostic and clinical value. Nat Rev Neurol. 2015 Jun;11(6):327-38.

Kharbanda HS, Alsop DC, Anderson AW, Filardo G, Hackney DB. Effects of cord motion on diffusion imaging of the spinal cord. Magn Reson Med. 2006 Aug;56(2):334-9.

Kim WK, Liu X, Sandner J, Pasmantier M, Andrews J, Rowland LP, Mitsumoto H. Study of 962 patients indicates progressive muscular atrophy is a form of ALS. Neurology. 2009 Nov 17;73(20):1686-92.

King AE, Blizzard CA, Southam KA, Vickers JC, Dickson TC. Degeneration of axons in spinal white matter in G93A mSOD1 mouse characterized by NFL and α-internexin immunoreactivity. Brain Res. 2012 Jul 17;1465:90-100.

King AE, Woodhouse A, Kirkcaldie MT, Vickers JC. Excitotoxicity in ALS: Overstimulation, or overreaction? Exp Neurol. 2016 Jan;275 Pt 1:162-71.

Kipps CM, Duggins AJ, Mahant N, Gomes L, Ashburner J, McCusker EA. Progression of structural neuropathology in preclinical Huntington's disease: a tensor based morphometry study. J Neurol Neurosurg Psychiatry. 2005 May;76(5):650-5.

- 153 -

Klein JP, Arora A, Neema M, Healy BC, Tauhid S, Goldberg-Zimring D, Chavarro-Nieto C, Stankiewicz JM, Cohen AB, Buckle GJ, Houtchens MK, Ceccarelli A, Dell'Oglio E, Guttmann CR, Alsop DC, Hackney DB, Bakshi R. A 3T MR imaging investigation of the topography of whole spinal cord atrophy in multiple sclerosis. AJNR Am J Neuroradiol. 2011 Jun-Jul;32(6):1138-42.

Kolb SJ, Kissel JT. Spinal Muscular Atrophy. Neurol Clin. 2015 Nov;33(4):831-46.

Kong L, Wang X, Choe DW, Polley M, Burnett BG, Bosch-Marcé M, Griffin JW, Rich MM, Sumner CJ. Impaired synaptic vesicle release and immaturity of neuromuscular junctions in spinal muscular atrophy mice. J Neurosci. 2009 Jan 21;29(3):842-51.

Kulgelberg E, Welander L. Heredofamilial juvenile muscular atrophy simulating muscular dystrophy. AMA Arch Neurol Psychiatry. 1956 May;75(5):500-9.

Kumar V, Sami N, Kashav T, Islam A, Ahmad F, Hassan MI. Protein aggregation and neurodegenerative diseases: From theory to therapy. Eur J Med Chem. 2016 Jul 26.

Lacomblez L, Bensimon G, Leigh PN, Guillet P, Powe L, Durrleman S, Delumeau JC, Meininger V. A confirmatory dose-ranging study of riluzole in ALS. ALS/Riluzole Study Group-II. Neurology. 1996 Dec;47(6 Suppl 4):S242-50.

Larriva-Sahd J, Condés Lara M, Martínez-Cabrera G, Varela-Echavarria A. Histological and ultrastructural characterization of interfascicular neurons in the rat anterior commissure. Brain Res. 2002 Mar 22;931(1):81-91.

Le Bihan D, Mangin JF, Poupon C, Clark CA, Pappata S, Molko N, Chabriat H. Diffusion tensor imaging: concepts and applications. J Magn Reson Imaging. 2001 Apr;13(4):534-46.

Leary SM, Parker GJ, Stevenson VL, Barker GJ, Miller DH, Thompson AJ. Reproducibility of magnetic resonance imaging measurements of spinal cord atrophy: the role of quality assurance. Magn Reson Imaging. 1999 Jun;17(5):773-6.

Lebouteux MV, Franques J, Guillevin R, Delmont E, Lenglet T, Bede P, Desnuelle C, Pouget J, Pascal- Mousselard H, Pradat PF. Revisiting the spectrum of lower motor neuron diseases with snake eyes appearance on magnetic resonance imaging. Eur J Neurol. 2014 Sep;21(9):1233-41.

- 154 -

Lefebvre S, Bürglen L, Reboullet S, Clermont O, Burlet P, Viollet L, Benichou B, Cruaud C, Millasseau P, Zeviani M, et al. Identification and characterization of a spinal muscular atrophy-determining gene. Cell. 1995 Jan 13;80(1):155-65.

Lefebvre S, Burlet P, Liu Q, Bertrandy S, Clermont O, Munnich A, Dreyfuss G, Melki J. Correlation between severity and SMN protein level in spinal muscular atrophy. Nat Genet. 1997 Jul;16(3):265-9.

Leung D, Hays AP, Karlikaya G, Del Bene ML, Rowland LP. Diagnosis of ALS: clinicopathologic analysis of 76 autopsies. Neurology. 1999; 52(suppl 2):A164.

Ling KK, Lin MY, Zingg B, Feng Z, Ko CP. Synaptic defects in the spinal and neuromuscular circuitry in a mouse model of spinal muscular atrophy. PLoS One. 2010 Nov 11;5(11):e15457.

Liu M, Gross DW, Wheatley BM, Concha L, Beaulieu C. The acute phase of Wallerian degeneration: longitudinal diffusion tensor imaging of the fornix following temporal lobe surgery. Neuroimage. 2013 Jul 1;74:128-39.

Lloyd CM, Richardson MP, Brooks DJ, al Chalabi A, Leigh PN. Extramotor involvement in ALS: PET studies with the GABA(A) ligand [(11)C]flumazenil. Brain 2000; 123 ( Pt 11):2289-2296.

Logroscino G, Traynor BJ, Hardiman O, Chiò A, Mitchell D, Swingler RJ, Millul A, Benn E, Beghi E; EURALS. Incidence of amyotrophic lateral sclerosis in Europe. J Neurol Neurosurg Psychiatry. 2010 Apr;81(4):385-90.

Losseff NA, Webb SL, O'Riordan JI, Page R, Wang L, Barker GJ, Tofts PS, McDonald WI, Miller DH, Thompson AJ. Spinal cord atrophy and disability in multiple sclerosis. A new reproducible and sensitive MRI method with potential to monitor disease progression. Brain. 1996 Jun;119 ( Pt 3):701-8.

Ludolph A, Drory V, Hardiman O, Nakano I, Ravits J, Robberecht W, Shefner J; WFN Research Group On ALS/MND. A revision of the El Escorial criteria - 2015. Amyotroph Lateral Scler Frontotemporal Degener. 2015;16(5-6):291-2.

Lukas C, Hahn HK, Bellenberg B, Hellwig K, Globas C, Schimrigk SK, Köster O, Schöls L. Spinal cord atrophy in spinocerebellar ataxia type 3 and 6 : impact on clinical disability. J Neurol. 2008 Aug;255(8):1244-9.

- 155 -

Lundell H, Barthelemy D, Skimminge A, Dyrby TB, Biering-Sørensen F, Nielsen JB. Independent spinal cord atrophy measures correlate to motor and sensory deficits in individuals with spinal cord injury. Spinal Cord. 2011 Jan;49(1):70-5.

Mackenzie IR, Bigio EH, Ince PG, Geser F, Neumann M, Cairns NJ, Kwong LK, Forman MS, Ravits J, Stewart H, Eisen A, McClusky L, Kretzschmar HA, Monoranu CM, Highley JR, Kirby J, Siddique T, Shaw PJ, Lee VM, Trojanowski JQ. Pathological TDP-43 distinguishes sporadic amyotrophic lateral sclerosis from amyotrophic lateral sclerosis with SOD1 mutations. Ann Neurol. 2007 May;61(5):427- 34.

Mailman MD, Heinz JW, Papp AC, Snyder PJ, Sedra MS, Wirth B, Burghes AH, Prior TW. Molecular analysis of spinal muscular atrophy and modification of the phenotype by SMN2. Genet Med. 2002 Jan- Feb;4(1):20-6.

Malaspina A, Ngoh SF, Ward RE, Hall JC, Tai FW, Yip PK, Jones C, Jokic N, Averill SA, Michael-Titus AT, Priestley JV. Activation transcription factor-3 activation and the development of spinal cord degeneration in a rat model of amyotrophic lateral sclerosis. Neuroscience. 2010 Aug 25;169(2):812- 27.

Manjaly ZR, Scott KM, Abhinav K, Wijesekera L, Ganesalingam J, Goldstein LH, Janssen A, Dougherty A, Willey E, Stanton BR, Turner MR, Ampong MA, Sakel M, Orrell RW, Howard R, Shaw CE, Leigh PN, Al-Chalabi A. The sex ratio in amyotrophic lateral sclerosis: A population based study. Amyotroph Lateral Scler. 2010 Oct;11(5):439-42.

Marin B, Boumédiene F, Logroscino G, Couratier P, Babron MC, Leutenegger AL, Copetti M, Preux PM, Beghi E. Variation in worldwide incidence of amyotrophic lateral sclerosis: a meta-analysis. Int J Epidemiol. 2016 May 16.

Mark Hallett. Transcranial Magnetic Stimulation: A Primer. Neuron 2007; 55(2):187-199.

Martin AR, Aleksanderek I, Cohen-Adad J, Tarmohamed Z, Tetreault L, Smith N, Cadotte DW, Crawley A, Ginsberg H, Mikulis DJ, Fehlings MG. Translating state-of-the-art spinal cord MRI techniques to clinical use: A systematic review of clinical studies utilizing DTI, MT, MWF, MRS, and fMRI. Neuroimage Clin. 2015 Dec 4;10:192-238.

- 156 -

Mazziotta J, Toga A, Evans A, Fox P, Lancaster J, Zilles K, Woods R, Paus T, Simpson G, Pike B, Holmes C, Collins L, Thompson P, MacDonald D, Iacoboni M, Schormann T, Amunts K, Palomero- Gallagher N, Geyer S, Parsons L, Narr K, Kabani N, Le Goualher G, Boomsma D, Cannon T, Kawashima R, Mazoyer B. A probabilistic atlas and reference system for the human brain: International Consortium for Brain Mapping (ICBM). Philos Trans R Soc Lond B Biol Sci. 2001 Aug 29;356(1412):1293-322.

McAndrew PE, Parsons DW, Simard LR, Rochette C, Ray PN, Mendell JR, Prior TW, Burghes AH. Identification of proximal spinal muscular atrophy carriers and patients by analysis of SMNT and SMNC gene copy number. Am J Hum Genet. 1997 Jun;60(6):1411-22.

McCombe PA, Henderson RD. Effects of gender in amyotrophic lateral sclerosis. Gend Med. 2010 Dec;7(6):557-70.

McGown A, McDearmid JR, Panagiotaki N, Tong H, Al Mashhadi S, Redhead N, Lyon AN, Beattie CE, Shaw PJ, Ramesh TM. Early interneuron dysfunction in ALS: insights from a mutant sod1 zebrafish model. Ann Neurol. 2013 Feb;73(2):246-58.

McIntosh C, Hamarneh G, Toom M, Tam RC. Spinal cord segmentation for volume estimation in healthy and multiple sclerosis subjects using crawlers and minimal paths. Proceedings of the First IEEE International Conference on Healthcare Informatics, Imaging and Systems Biology (HISB). 2011;25-31.

Meadows JC, Marsden CD, Harriman DGF. Chronic spinal muscular atrophy in adults. Part 1. The Kugelberg-Welander syndrome. J Neurol Sci. 1969;9:527–550.

Menke RA, Körner S, Filippini N, Douaud G, Knight S, Talbot K, Turner MR. Widespread grey matter pathology dominates the longitudinal cerebral MRI and clinical landscape of amyotrophic lateral sclerosis. Brain. 2014 Sep;137(Pt 9):2546-55.

Mirowitz S, Sartor K, Gado M, Torack R. Focal signal-intensity variations in the posterior internal capsule: normal MR findings and distinction from pathologic findings. Radiology. 1989 Aug;172(2):535-9.

- 157 -

Mitsumoto H, Chad DA, Pioro EP. Clinical features: signs and symptoms. In: Mitsumoto H, Chad DA, Pioro EP, editors. Amyotrophic Lateral Sclerosis. Philadelphie: FA Davis Company, 1998: 47-64.

Mohammadi S, Freund P, Feiweier T, Curt A, Weiskopf N. The impact of post-processing on spinal cord diffusion tensor imaging. Neuroimage. 2013 Apr 15;70:377-85.

Monani UR, De Vivo DC. Neurodegeneration in spinal muscular atrophy: from disease phenotype and animal models to therapeutic strategies and beyond. Future Neurol. 2014 Jan 1;9(1):49-65.

Mori S, Zhang J. Principles of diffusion tensor imaging and its applications to basic neuroscience research. Neuron. 2006 Sep 7;51(5):527-39.

Morra JH, Tu Z, Apostolova LG, Green AE, Avedissian C, Madsen SK, Parikshak N, Toga AW, Jack CR Jr, Schuff N, Weiner MW, Thompson PM; Alzheimer's Disease Neuroimaging Initiative. Automated mapping of hippocampal atrophy in 1-year repeat MRI data from 490 subjects with Alzheimer's disease, mild cognitive impairment, and elderly controls. Neuroimage. 2009 Mar;45(1 Suppl):S3-15.

Morrison BM, Janssen WG, Gordon JW, Morrison JH. Time course of neuropathology in the spinal cord of G86R superoxide dismutase transgenic mice. J Comp Neurol. 1998 Feb 2;391(1):64-77.

Morrison LR. The Effect of Advancing Age upon the Human Spinal Cord. Cam- bridge, Massachusetts: Harvard University Press; 1959.

Mottershead JP, Schmierer K, Clemence M, Thornton JS, Scaravilli F, Barker GJ, Tofts PS, Newcombe J, Cuzner ML, Ordidge RJ, McDonald WI, Miller DH. High field MRI correlates of myelin content and axonal density in multiple sclerosis--a post-mortem study of the spinal cord. J Neurol. 2003 Nov;250(11):1293-301.

Mühlau M, Engl C, Boucard CC, Schmidt P, Biberacher V, Görsch I, Sorg C, Wohlschläger A, Zimmer

- 158 -

C, Hemmer B, Weindl A. Spinal cord atrophy in early Huntington's disease. Annals of Clinical and Translational Neurology. 2014 Mar;1(4), 302-306.

Mukherjee P, Hess CP, Xu D, Han ET, Kelley DA, Vigneron DB. Development and initial evaluation of 7-T q-ball imaging of the human brain. Magn Reson Imaging. 2008 Feb;26(2):171-80.

Munsat TL, Davies KE. International SMA consortium meeting. (26-28 June 1992, Bonn, Germany). Neuromuscul Disord. 1992;2(5-6):423-8.

Murray LM, Comley LH, Thomson D, Parkinson N, Talbot K, Gillingwater TH. Selective vulnerability of motor neurons and dissociation of pre- and post-synaptic pathology at the neuromuscular junction in mouse models of spinal muscular atrophy. Hum Mol Genet. 2008 Apr 1;17(7):949-62.

Nair G, Carew JD, Usher S, Lu D, Hu XP, Benatar M. Diffusion tensor imaging reveals regional differences in the cervical spinal cord in amyotrophic lateral sclerosis. Neuroimage. 2010 Nov 1;53(2):576-83.

Noebauer-Huhmann IM, Glaser C, Dietrich O, Wallner CP, Klinger W, Imhof H, Schoenberg SO. MR imaging of the cervical spine: assessment of image quality with parallel imaging compared to non- accelerated MR measurements. Eur Radiol. 2007 May;17(5):1147-55.

O'Callaghan C, Hornberger M, Balsters JH, Halliday GM, Lewis SJ, Shine JM. Cerebellar atrophy in Parkinson's disease and its implication for network connectivity. Brain. 2016 Mar;139(Pt 3):845-55.

Oba H, Araki T, Ohtomo K, Monzawa S, Uchiyama G, Koizumi K, Nogata Y, Kachi K, Shiozawa Z, Kobayashi M. Amyotrophic lateral sclerosis: T2 shortening in motor cortex at MR imaging. Radiology. 1993 Dec;189(3):843-6.

Ozdinler PH, Benn S, Yamamoto TH, Güzel M, Brown RH Jr, Macklis JD. Corticospinal motor neurons and related subcerebral projection neurons undergo early and specific neurodegeneration in hSOD1G93A transgenic ALS mice. J Neurosci. 2011 Mar 16;31(11):4166-77.

Piepers S, van den Berg LH, Brugman F, Scheffer H, Ruiterkamp-Versteeg M, van Engelen BG, Faber CG, de Visser M, van der Pol WL, Wokke JH. A natural history study of late onset spinal muscular atrophy types 3b and 4. J Neurol. 2008 Sep;255(9):1400-4.

- 159 -

Pierpaoli C, Basser PJ. Toward a quantitative assessment of diffusion anisotropy. Magn Reson Med. 1996 Dec;36(6):893-906. Erratum in: Magn Reson Med 1997 Jun;37(6):972.

Pradat PF, Barani A, Wanschitz J, Dubourg O, Lombès A, Bigot A, Mouly V, Bruneteau G, Salachas F, Lenglet T, Meininger V, Butler-Browne G. Abnormalities of satellite cells function in amyotrophic lateral sclerosis. Amyotroph Lateral Scler. 2011 Jul;12(4):264-71.

Pradat PF, Bruneteau G, Gonzalez de Aguilar JL, Dupuis L, Jokic N, Salachas F, Le Forestier N, Echaniz-Laguna A, Dubourg O, Hauw JJ, Tranchant C, Loeffler JP, Meininger V. Muscle Nogo-A expression is a prognostic marker in lower motor neuron syndromes. Ann Neurol. 2007 Jul;62(1):15- 20.

Pradat PF, Bruneteau G, Munerati E, Salachas F, Le Forestier N, Lacomblez L, Lenglet T, Meininger V. Extrapyramidal stiffness in patients with amyotrophic lateral sclerosis. Mov Disord. 2009 Oct 30;24(14):2143-8.

Pradat PF, Dupuis L. Neuroprotective strategies in amyotrophic lateral sclerosis: modulation of neurotransmittory and neurotrophic input to motoneurons. In: Kostrzewa RM, editor. Handbook of Neurotoxicity. New-York: Springer, 2014: 1417-1434.

Pradat PF, Kabashi E, Desnuelle C. Deciphering spreading mechanisms in amyotrophic lateral sclerosis: clinical evidence and potential molecular processes. Curr Opin Neurol 2015; 28(5):455-461.

Pradat PF. Atrophie bulbo-spinale liée à l'X : une maladie rare mais un diagnostic différentiel fréquent de la sclérose latérale amyotrophique. Presse Med 2014; 43(5):580-586.

Pradat PF. Lower motor neuron syndrome with snake eyes appearance on MRI is a benign differential diagnosis of ALS and dysimmune neuropathies. Clin Neurol Neurosurg. 2016 Feb;141:130.

Prell T, Grosskreutz J. The involvement of the cerebellum in amyotrophic lateral sclerosis. Amyotroph Lateral Scler Frontotemporal Degener. 2013 Dec;14(7-8):507-15.

Pupillo E, Messina P, Logroscino G, Beghi E; SLALOM Group. Long-term survival in amyotrophic lateral sclerosis: a population-based study. Ann Neurol. 2014 Feb;75(2):287-97.

Qin W, Zhang M, Piao Y, et al. Wallerian Degeneration in Central Nervous System: Dynamic

- 160 -

Associations between Diffusion Indices and Their Underlying Pathology. Yang Y, ed. PLoS ONE. 2012;7(7):e41441. doi:10.1371/journal.pone.0041441.

Radcliffe CB, Lockhart Clarke J. An important case of paralysis and muscular atrophy with disease of the nervous centres. Brit & Foreign Medico-Chirurgical Rev. 1862;30:215–25.

Rajagopalan V, Pioro EP. Distinct patterns of cortical atrophy in ALS patients with or without dementia: an MRI VBM study. Amyotroph Lateral Scler Frontotemporal Degener. 2014 Jun;15(3-4):216-25.

Renton AE, Chiò A, Traynor BJ. State of play in amyotrophic lateral sclerosis genetics. Nat Neurosci. 2014 Jan;17(1):17-23.

Robberecht W, Philips T. The changing scene of amyotrophic lateral sclerosis. Nat Rev Neurosci 2013; 14(4):248-264.

Rocca MA, Valsasina P, Damjanovic D, Horsfield MA, Mesaros S, Stosic-Opincal T, Drulovic J, Filippi M. Voxel-wise mapping of cervical cord damage in multiple sclerosis patients with different clinical phenotypes. J Neurol Neurosurg Psychiatry. 2013 Jan;84(1):35-41.

Rocha AJ, Maia Júnior AC. Is magnetic resonance imaging a plausible biomarker for upper motor neuron degeneration in amyotrophic lateral sclerosis/primary lateral sclerosis or merely a useful paraclinical tool to exclude mimic syndromes? A critical review of imaging applicability in clinical routine. Arq Neuropsiquiatr. 2012 Jul;70(7):532-9.

Rodegerdts EA, Boss A, Riemarzik K, Lichy M, Schick F, Claussen CD, Schlemmer HP. 3D imaging of the whole spine at 3T compared to 1.5T: initial experiences. Acta Radiol. 2006 Jun;47(5):488-93.

Rodrigues NR, Owen N, Talbot K, Ignatius J, Dubowitz V, Davies KE. Deletions in the survival motor neuron gene on 5q13 in autosomal recessive spinal muscular atrophy. Hum Mol Genet. 1995 Apr;4(4):631-4.

Roméo F, Hoult DI. Magnet field profiling: analysis and correcting coil design. Magn Reson Med. 1984 Mar;1(1):44-65.

Ronen I, Ercan E, Webb A. Axonal and glial microstructural information obtained with diffusion- weighted magnetic resonance spectroscopy at 7T. Front Integr Neurosci. 2013 Mar 13;7:13.

- 161 -

Rosen DR, Siddique T, Patterson D, Figlewicz DA, Sapp P, Hentati A, Donaldson D, Goto J, O'Regan JP, Deng HX, et al. Mutations in Cu/Zn superoxide dismutase gene are associated with familial amyotrophic lateral sclerosis. Nature. 1993 Mar 4;362(6415):59-62. Erratum in: Nature. 1993 Jul 22;364(6435):362.

Ross CA, Poirier MA. Protein aggregation and neurodegenerative disease. Nat Med. 2004 Jul;10 Suppl:S10-7.

Sábado J, Casanovas A, Tarabal O, Hereu M, Piedrafita L, Calderó J, Esquerda JE. Accumulation of misfolded SOD1 in dorsal root ganglion degenerating proprioceptive sensory neurons of transgenic mice with amyotrophic lateral sclerosis. Biomed Res Int. 2014;2014:852163.

Samson RS, Ciccarelli O, Kachramanoglou C, Brightman L, Lutti A, Thomas DL, Weiskopf N, Wheeler-Kingshott CA. Tissue- and column-specific measurements from multi-parameter mapping of the human cervical spinal cord at 3 T. NMR Biomed. 2013 Dec;26(12):1823-30.

Sangari S, Iglesias C, El Mendili MM, Benali H, Pradat PF, Marchand-Pauvert V. Impairment of sensory-motor integration at spinal level in amyotrophic lateral sclerosis. Clin Neurophysiol. 2016 Apr;127(4):1968-77.

Sasaki S. Sporadic lower motor neuron disease with a snake eyes appearance on the cervical anterior horns by MRI. Clin Neurol Neurosurg. 2015 Sep;136:122-31.

Schlaeger R, Papinutto N, Panara V, Bevan C, Lobach IV, Bucci M, Caverzasi E, Gelfand JM, Green AJ, Jordan KM, Stern WA, von Büdingen HC, Waubant E, Zhu AH, Goodin DS, Cree BA, Hauser SL, Henry RG. Spinal cord gray matter atrophy correlates with multiple sclerosis disability. Ann Neurol. 2014 Oct;76(4):568-80.

Schlaeger R, Papinutto N, Zhu AH, Lobach IV, Bevan CJ, Bucci M, Castellano A, Gelfand JM, Graves JS, Green AJ, Jordan KM, Keshavan A, Panara V, Stern WA, von Büdingen HC, Waubant E, Goodin DS, Cree BA, Hauser SL, Henry RG. Association Between Thoracic Spinal Cord Gray Matter Atrophy and Disability in Multiple Sclerosis. JAMA Neurol. 2015 Aug;72(8):897-904.

Schmid A, DiDonato CJ. Animal models of spinal muscular atrophy. J Child Neurol. 2007 Aug;22(8):1004-12.

- 162 -

Schmierer K, Scaravilli F, Altmann DR, Barker GJ, Miller DH. Magnetization transfer ratio and myelin in postmortem multiple sclerosis brain. Ann Neurol. 2004 Sep;56(3):407-15.

Schmierer K, Tozer DJ, Scaravilli F, Altmann DR, Barker GJ, Tofts PS, Miller DH. Quantitative magnetization transfer imaging in postmortem multiple sclerosis brain. J Magn Reson Imaging. 2007 Jul;26(1):41-51.

Shefner JM, Logigian EL. The mixed nerve silent period in normal subjects and patients with amyotrophic lateral sclerosis. Electromyogr Clin Neurophysiol. 1998 Dec;38(8):505-10.

Shen D, Cui L, Cui B, Fang J, Li D, Ma J. A Systematic Review and Meta-Analysis of the Functional MRI Investigation of Motor Neuron Disease. Front Neurol. 2015 Nov 24;6:246.

Simone C, Ramirez A, Bucchia M, Rinchetti P, Rideout H, Papadimitriou D, Re DB, Corti S. Is spinal muscular atrophy a disease of the motor neurons only: pathogenesis and therapeutic implications? Cell Mol Life Sci. 2016 Mar;73(5):1003-20.

Smith SA, Edden RA, Farrell JA, Barker PB, Van Zijl PC. Measurement of T1 and T2 in the cervical spinal cord at 3 tesla. Magn Reson Med. 2008 Jul;60(1):213-9.

Smith SA, Farrell JA, Jones CK, Reich DS, Calabresi PA, van Zijl PC. Pulsed magnetization transfer imaging with body coil transmission at 3 Tesla: feasibility and application. Magn Reson Med. 2006 Oct;56(4):866-75.

Song SK, Sun SW, Ju WK, Lin SJ, Cross AH, Neufeld AH. Diffusion tensor imaging detects and differentiates axon and myelin degeneration in mouse optic nerve after retinal ischemia. Neuroimage. 2003 Nov;20(3):1714-22.

- 163 -

Spreux-Varoquaux O, Bensimon G, Lacomblez L, Salachas F, Pradat PF, Le Forestier N, Marouan A, Dib M, Meininger V. Glutamate levels in cerebrospinal fluid in amyotrophic lateral sclerosis: a reappraisal using a new HPLC method with coulometric detection in a large cohort of patients. J Neurol Sci. 2002 Jan 15;193(2):73-8.

Sreedharan J, Blair IP, Tripathi VB, Hu X, Vance C, Rogelj B, Ackerley S, Durnall JC, Williams KL, Buratti E, Baralle F, de Belleroche J, Mitchell JD, Leigh PN, Al-Chalabi A, Miller CC, Nicholson G, Shaw CE. TDP-43 mutations in familial and sporadic amyotrophic lateral sclerosis. Science. 2008 Mar 21;319(5870):1668-72

Stroman PW, Figley CR, Cahill CM. Spatial normalization, bulk motion correction and coregistration for functional magnetic resonance imaging of the human cervical spinal cord and brainstem. Magn Reson Imaging. 2008 Jul;26(6):809-14.

Stroman PW, Wheeler-Kingshott C, Bacon M, Schwab JM, Bosma R, Brooks J, Cadotte D, Carlstedt T, Ciccarelli O, Cohen-Adad J, Curt A, Evangelou N, Fehlings MG, Filippi M, Kelley BJ, Kollias S, Mackay A, Porro CA, Smith S, Strittmatter SM, Summers P, Tracey I. The current state-of-the-art of spinal cord imaging: methods. Neuroimage. 2014 Jan 1;84:1070-81.

Stroman PW. Discrimination of errors from neuronal activity in functional MRI of the human spinal cord by means of general linear model analysis. Magn Reson Med. 2006 Aug;56(2):452-6.

Suárez-Solá ML, González-Delgado FJ, Pueyo-Morlans M, Medina-Bolívar OC, Hernández-Acosta NC, González-Gómez M, Meyer G. Neurons in the white matter of the adult human neocortex. Front Neuroanat. 2009 Jun 9;3:7.

Sun SW, Liang HF, Cross AH, Song SK. Evolving Wallerian degeneration after transient retinal ischemia in mice characterized by diffusion tensor imaging. Neuroimage. 2008 Mar 1;40(1):1-10.

Sun SW, Liang HF, Trinkaus K, Cross AH, Armstrong RC, Song SK. Noninvasive detection of cuprizone induced axonal damage and demyelination in the mouse corpus callosum. Magn Reson Med. 2006 Feb;55(2):302-8.

Swash M, Scholtz CL, Vowles G, Ingram DA. Selective and asymmetric vulnerability of corticospinal and spinocerebellar tracts in motor neuron disease. J Neurol Neurosurg Psychiatry. 1988

- 164 -

Jun;51(6):785-9.

Swinnen B, Robberecht W. The phenotypic variability of amyotrophic lateral sclerosis. Nat Rev Neurol. 2014 Nov;10(11):661-70.

Swoboda KJ, Prior TW, Scott CB, McNaught TP, Wride MC, Reyna SP, Bromberg MB. Natural history of denervation in SMA: relation to age, SMN2 copy number, and function. Ann Neurol. 2005 May;57(5):704-12.

Takahashi M, Hackney DB, Zhang G, Wehrli SL, Wright AC, O'Brien WT, Uematsu H, Wehrli FW, Selzer ME. Magnetic resonance microimaging of intraaxonal water diffusion in live excised lamprey spinal cord. Proc Natl Acad Sci U S A. 2002 Dec 10;99(25):16192-6.

Tanabe JL, Vermathen M, Miller R, Gelinas D, Weiner MW, Rooney WD. Reduced MTR in the corticospinal tract and normal T2 in amyotrophic lateral sclerosis. Magn Reson Imaging. 1998 Dec;16(10):1163-9.

Taso M, Le Troter A, Sdika M, Cohen-Adad J, Arnoux PJ, Guye M, Ranjeva JP, Callot V. A reliable spatially normalized template of the human spinal cord--Applications to automated white matter/gray matter segmentation and tensor-based morphometry (TBM) mapping of gray matter alterations occurring with age. Neuroimage. 2015 Aug 15;117:20-8.

Taso M, Le Troter A, Sdika M, Ranjeva JP, Guye M, Bernard M, Callot V. Construction of an in vivo human spinal cord atlas based on high-resolution MR images at cervical and thoracic levels: preliminary results. MAGMA. 2014 Jun;27(3):257-67.

Tavazzi E, Laganà MM, Bergsland N, Tortorella P, Pinardi G, Lunetta C, Corbo M, Rovaris M. Grey matter damage in progressive multiple sclerosis versus amyotrophic lateral sclerosis: a voxel-based morphometry MRI study. Neurol Sci. 2015 Mar;36(3):371-7.

Tench CR, Morgan PS, Constantinescu CS. Measurement of cervical spinal cord cross-sectional area by MRI using edge detection and partial volume correction. J Magn Reson Imaging. 2005 Mar;21(3):197-

- 165 -

203.

Tessa C, Lucetti C, Giannelli M, Diciotti S, Poletti M, Danti S, Baldacci F, Vignali C, Bonuccelli U, Mascalchi M, Toschi N. Progression of brain atrophy in the early stages of Parkinson's disease: a longitudinal tensor-based morphometry study in de novo patients without cognitive impairment. Hum Brain Mapp. 2014 Aug;35(8):3932-44.

Testa D, Lovati R, Ferrarini M, Salmoiraghi F, Filippini G. Survival of 793 patients with amyotrophic lateral sclerosis diagnosed over a 28-year period. Amyotroph Lateral Scler Other Motor Neuron Disord. 2004 Dec;5(4):208-12.

Thivard L, Pradat PF, Lehéricy S, Lacomblez L, Dormont D, Chiras J, Benali H, Meininger V. Diffusion tensor imaging and voxel based morphometry study in amyotrophic lateral sclerosis: relationships with motor disability. J Neurol Neurosurg Psychiatry. 2007 Aug;78(8):889-92.

Thorpe JW, Moseley IF, Hawkes CH, MacManus DG, McDonald WI, Miller DH. Brain and spinal cord MRI in motor neuron disease. Journal of Neurology, Neurosurgery, and Psychiatry. 1996;61(3):314- 317.

Tuch DS. Q-ball imaging. Magn Reson Med. 2004 Dec;52(6):1358-72.

Turner MR, Swash M, Ebers GC. Lockhart Clarke's contribution to the description of amyotrophic lateral sclerosis. Brain. 2010 Nov;133(11):3470-9.

Turner MR, Verstraete E. What does imaging reveal about the pathology of amyotrophic lateral sclerosis? Curr Neurol Neurosci Rep. 2015 Jul;15(7):45.

Unrath A, Müller HP, Riecker A, Ludolph AC, Sperfeld AD, Kassubek J. Whole brain-based analysis of regional white matter tract alterations in rare motor neuron diseases by diffusion tensor imaging. Hum Brain Mapp. 2010 Nov;31(11):1727-40.

Vahdat S, Lungu O, Cohen-Adad J, Marchand-Pauvert V, Benali H, Doyon J. Simultaneous Brain- Cervical Cord fMRI Reveals Intrinsic Spinal Cord Plasticity during Motor Sequence Learning. PLoS Biol. 2015 Jun 30;13(6):e1002186.

Valsasina P, Agosta F, Benedetti B, Caputo D, Perini M, Salvi F, Prelle A, Filippi M. Diffusion anisotropy of the cervical cord is strictly associated with disability in amyotrophic lateral sclerosis. J

- 166 -

Neurol Neurosurg Psychiatry. 2007 May;78(5):480-4.

Valsasina P, Horsfield MA, Rocca MA, Absinta M, Comi G, Filippi M. Spatial normalization and regional assessment of cord atrophy: voxel-based analysis of cervical cord 3D T1-weighted images. AJNR Am J Neuroradiol. 2012 Dec;33(11):2195-200.

Valsasina P, Rocca MA, Horsfield MA, Absinta M, Messina R, Caputo D, Comi G, Filippi M. Regional cervical cord atrophy and disability in multiple sclerosis: a voxel-based analysis. Radiology. 2013 Mar;266(3):853-61.

Van Den Berg-Vos RM, Van Den Berg LH, Visser J, de Visser M, Franssen H, Wokke JH. The spectrum of lower motor neuron syndromes. J Neurol. 2003 Nov;250(11):1279-92. van der Graaff MM, de Jong JM, Baas F, de Visser M. Upper motor neuron and extra-motor neuron involvement in amyotrophic lateral sclerosis: a clinical and brain imaging review. Neuromuscul Disord. 2009 Jan;19(1):53-8. van der Graaff MM, Sage CA, Caan MW, Akkerman EM, Lavini C, Majoie CB, Nederveen AJ, Zwinderman AH, Vos F, Brugman F, van den Berg LH, de Rijk MC, van Doorn PA, Van Hecke W, Peeters RR, Robberecht W, Sunaert S, de Visser M. Upper and extra-motoneuron involvement in early motoneuron disease: a diffusion tensor imaging study. Brain. 2011 Apr;134(Pt 4):1211-28.

Verma A. Altered RNA metabolism and amyotrophic lateral sclerosis. Annals of Indian Academy of Neurology 2011;14(4):239-244.

Verma T, Cohen-Adad J. Effect of respiration on the B0 field in the human spinal cord at 3T. Magn Reson Med. 2014 Dec;72(6):1629-36.

Vogt J, Paul F, Aktas O, Müller-Wielsch K, Dörr J, Dörr S, Bharathi BS, Glumm R, Schmitz C, Steinbusch H, Raine CS, Tsokos M, Nitsch R, Zipp F. Lower motor neuron loss in multiple sclerosis and experimental autoimmune encephalomyelitis. Ann Neurol. 2009 Sep;66(3):310-22.

Vucic S, Lin CS, Cheah BC, Murray J, Menon P, Krishnan AV, Kiernan MC. Riluzole exerts central and peripheral modulating effects in amyotrophic lateral sclerosis. Brain. 2013 May;136(Pt 5):1361-70.

Wadman RI, Vrancken AF, van den Berg LH, van der Pol WL. Dysfunction of the neuromuscular

- 167 -

junction in spinal muscular atrophy types 2 and 3. Neurology. 2012 Nov 13;79(20):2050-5.

Wang CH, Finkel RS, Bertini ES, Schroth M, Simonds A, Wong B, Aloysius A, Morrison L, Main M, Crawford TO, Trela A; Participants of the International Conference on SMA Standard of Care. Consensus statement for standard of care in spinal muscular atrophy. J Child Neurol. 2007 Aug;22(8):1027-49.

Wang Y, Liu L, Ma L, Huang X, Lou X, Wang Y, Wu N, Liu T, Guo X. Preliminary study on cervical spinal cord in patients with amyotrophic lateral sclerosis using MR diffusion tensor imaging. Acad Radiol. 2014 May;21(5):590-6.

Wang Y, Song Y, Rajagopalan P, An T, Liu K, Chou YY, Gutman B, Toga AW, Thompson PM; Alzheimer's Disease Neuroimaging Initiative. Surface-based TBM boosts power to detect disease effects on the brain: an N=804 ADNI study. Neuroimage. 2011 Jun 15;56(4):1993-2010.

Wang Y, Zhang J, Gutman B, Chan TF, Becker JT, Aizenstein HJ, Lopez OL, Tamburo RJ, Toga AW, Thompson PM. Multivariate tensor-based morphometry on surfaces: application to mapping ventricular abnormalities in HIV/AIDS. Neuroimage. 2010 Feb 1;49(3):2141-57.

Waragai M. MRI and clinical features in amyotrophic lateral sclerosis. Neuroradiology 1997;39(12):847-851.

Werlauff U, Vissing J, Steffensen BF. Change in muscle strength over time in spinal muscular atrophy types II and III. A long-term follow-up study. Neuromuscul Disord. 2012 Dec;22(12):1069-74.

Wijesekera LC, Leigh PN. Amyotrophic lateral sclerosis. Orphanet J Rare Dis. 2009 Feb 3;4:3.

Winhammar JM, Rowe DB, Henderson RD, Kiernan MC. Assessment of disease progression in motor neuron disease. Lancet Neurol. 2005 Apr;4(4):229-38.

Winner B, Marchetto MC, Winkler J, Gage FH. Human-induced pluripotent stem cells pave the road for a better understanding of motor neuron disease. Hum Mol Genet. 2014 Sep 15;23(R1):R27-34.

Wolff SD, Balaban RS. Magnetization transfer contrast (MTC) and tissue water proton relaxation in vivo. Magn Reson Med. 1989 Apr;10(1):135-44.

- 168 -

Wyndaele M, Wyndaele JJ. Incidence, prevalence and epidemiology of spinal cord injury: what learns a worldwide literature survey? Spinal Cord. 2006 Sep;44(9):523-9.

Xiao Q, Zhao W, Beers DR, Yen AA, Xie W, Henkel JS, Appel SH. Mutant SOD1(G93A) microglia are more neurotoxic relative to wild-type microglia. J Neurochem. 2007 Sep;102(6):2008-19.

Yamashita S, Ando Y. Genotype-phenotype relationship in hereditary amyotrophic lateral sclerosis. Transl Neurodegener. 2015 Jul 24;4:13.

Yang E, Nucifora PG, Melhem ER. Diffusion MR imaging: basic principles. Neuroimaging Clin N Am. 2011 Feb;21(1):1-25, vii.

Yiannakas MC, Kearney H, Samson RS, Chard DT, Ciccarelli O, Miller DH, Wheeler-Kingshott CA. Feasibility of grey matter and white matter segmentation of the upper cervical cord in vivo: a pilot study with application to magnetisation transfer measurements. Neuroimage. 2012 Nov 15;63(3):1054-9.

Zarkov M, Stojadinović A, Sekulić S, Barjaktarović I, Perić S, Keković G, Drasković B, Stević Z. Association between the SMN2 gene copy number and clinical characteristics of patients with spinal muscular atrophy with homozygous deletion of exon 7 of the SMN1 gene. Vojnosanit Pregl. 2015 Oct;72(10):859-63.

Zeng H, Constable RT. Image distortion correction in EPI: comparison of field mapping with point spread function mapping. Magn Reson Med. 2002 Jul;48(1):137-46.

Zerres K, Rudnik-Schöneborn S, Forrest E, Lusakowska A, Borkowska J, Hausmanowa-Petrusewicz I. A collaborative study on the natural history of childhood and juvenile onset proximal spinal muscular atrophy (type II and III SMA): 569 patients. J Neurol Sci. 1997 Feb 27;146(1):67-72.

Zerres K, Rudnik-Schöneborn S. Natural history in proximal spinal muscular atrophy. Clinical analysis of 445 patients and suggestions for a modification of existing classifications. Arch Neurol. 1995 May;52(5):518-23.

Zhan W, Yang Y. How accurately can the diffusion profiles indicate multiple fiber orientations? A study on general fiber crossings in diffusion MRI. J Magn Reson. 2006 Dec;183(2):193-202.

Zhang J, Jones M, DeBoy CA, et al. Diffusion Tensor Magnetic Resonance Imaging of Wallerian

- 169 -

Degeneration in Rat Spinal Cord after Dorsal Root Axotomy. The Journal of neuroscience : the official journal of the Society for Neuroscience. 2009;29(10):3160-3171.

Zhao W, Cohen-Adad J, Polimeni JR, Keil B, Guerin B, Setsompop K, Serano P, Mareyam A, Hoecht P, Wald LL. Nineteen-channel receive array and four-channel transmit array coil for cervical spinal cord imaging at 7T. Magn Reson Med. 2014 Jul;72(1):291-300.

Zivadinov R, Banas AC, Yella V, Abdelrahman N, Weinstock-Guttman B, Dwyer MG. Comparison of three different methods for measurement of cervical cord atrophy in multiple sclerosis. AJNR Am J Neuroradiol. 2008 Feb;29(2):319-25.

- 170 -