Advanced neuroimaging in ALS

Get Complete Project Material File(s) Now! »

Hints of spinal cord anatomy

The SC lies within the spinal canal surrounded by the cerebrospinal fluid (CSF) and by the bone and cartilaginous discs between the vertebral bodies. It 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., 2014) (figure 1). The CSF flows in the head–foot direction with each heart-beat with an amplitude that diminishes with greater distance from the head, thus inducing concomitant movements of the SC itself (Figley et al., 2008).
Figure 1: 7T cervical spinal cord MRI: sagittal T2-weighted image and T2*-weighted image (on the right side) of the C3-C4 axial slice with identification of main anatomical components (Barry et al., 2014).
The spinal cord has 31 segments (8 cervical, 12 thoracic, 5 lumbar, 5 sacral, and 1 coccygeal), each of which (except the first cervical segment, which has only a ventral root) has a pair of dorsal and ventral roots and a pair of spinal nerves. There are no sharp boundaries between the segments within the cord, but the cervical and lumbar enlargements which give rise to nerve roots for arms and legs, respectively, are clearly apparent. Each dorsal and ventral root join in the intervertebral foramina to form a spinal nerve (figure 2).
Figure 2: Anterior view of the cord showing dorsal and ventral roots and formation of the spinal nerves. Drawing by Frank Netter, MD. (Netter illustration from Elsevier, Inc. All rights reserved).
SC and vertebral column levels do not correspond (Cadotte et al., 2015) (figure 3): at the upper cervical level, the cord segment corresponds to the like-numbered vertebral body. From C5 to C8 the SC level is 1 level higher than the corresponding vertebral body. In the upper thoracic region, the vertebral spinal process is 2 segments above the corresponding cord segment. In the lower thoracic and upper lumbar regions, the difference between the vertebral and cord level is 2 or 3 segments while sacral cord levels correspond to vertebral T12-L1 levels (Bican et al., 2013).
Figure 3: Vertebral body and spinal cord segment location across 10 subjects. Vertebral bodies are represented for each subject by light-shaded bars, whereas spinal cord segments are represented by colored bars (Cadotte et al., 2015).
On the transversal plane, the SC is incompletely divided into two halves by a deep anterior median fissure and a posterior median sulcus. The GM of the SC is an H-shaped structure with 2 symmetric halves connected by a narrow bridge or commissure. The ratio between WM and GM is variable along the SC, with bigger GM surfaces observed at the cervical and lumbar enlargements. An imaginary coronal line through the central canal divides the GM into anterior and posterior horns. Alpha and gamma motor neurons are located in the anterior horn.
The central GM is surrounded by WM layers including descending and ascending fibers. The corticospinal tract (CST), arising from the precentral motor cortex, is the largest and most significant descending tract of the human SC. Almost 90% of the corticospinal tract fibers decussate in the lower medulla to form the lateral CST, whereas 8% of the non-decussating descending fibers form the anterior CST and 2% of them generate the uncrossed lateral CST.
On the other side, all afferent axons have their primary neurons in the dorsal root ganglia. The level of decussation varies among ascending systems. The dorsal column tract is responsible for the transmission of sensations of vibration, proprioception (position sense), and 2-point discrimination from the skin and joints (figure 4).
Figure 4: Transversal representation of the cervical SC. In dark grey: GM of anterior and posterior horns. The lateral corticospinal tract (CST) is showed in red.

Basic principles of spinal cord MRI

MRI makes use of the magnetic properties of certain atom nuclei present in body tissues, as for example the hydrogen one. By applying a strong and uniform magnetic field, the net magnetic moment (M) of hydrogen nuclei is induced to align (Andrew, 1992). This alignment is next perturbed by the introduction of brief radiofrequency pulses that tilt M, inducing its decay and return to the resting alignment through various relaxation processes, thus generating a current within the receiving coil. The decay between the excitation and the return to equilibrium of M is exploited to generate a contrast difference between tissues (Bitar et al., 2006). T1- and T2-weighted sequences characterized by different combinations of relaxation times can be used to give details about tissues specific features. The T1-time (longitudinal relaxation time) is the time constant which determines the rate at which excited protons return to equilibrium. The T2-time (transverse relaxation time) is the time constant which determines the rate at which excited protons reach equilibrium or go out of phase with each other (Edelman and Warach, 1993) (figure 5).
Figure 5: Physics of MRI. In the absence of a magnetic field, the magnetic axes of a group of protons are randomly oriented. (A) In the presence of a strong magnetic field, protons align both with and against the field. (B) A radiofrequency pulse applied at the resonance frequency will cause the protons to flip and align with the higher energy state, from which they relax back to their original alignment at a rate determined by T1 and T2 relaxation times (Edelman and Warach, 1993).
In the Central Nervous System (CNS), the T1-weighted sequence produces a high contrast difference between WM and GM, while the T2-weighted sequence is more sensitive to tissues’ water content and generates an excellent contrast between the CSF and the WM. T2*-weighted sequences are derived from T2 decay using gradient echoes. They are useful to accentuate local magnetic homogeneity effects generating high contrast difference between CSF, WM and GM (Martin et al., 2017).
Despite relevant technical improvements, MR imaging of the SC remains challenging and is frequently focused only on its cervical portion. The major imaging difficulties come from the fact that the SC 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 by organs that are prone to motion (Stroman et al., 2014, El Mendili et al., 2019).
Small cross-sectional dimensions: Considered the small dimensions of the SC, in order to effectively depict its anatomical details, high spatial resolution (at least 1 mm × 1 mm) and relatively thin imaging slices (1–2 mm) are required. Positioning the slices transverse to the SC anatomy is favorable to obtain the highest spatial resolution in the plane of the SC cross-section, where the anatomy is more varied, and the greatest resolution is needed (El Mendili et al., 2019). An unavoidable disadvantage of axial slices is that a large number of images is needed to view a sufficiently extended rostral–caudal part of the cord, thus prolonging the acquisition time (Stroman et al., 2014). Moreover, positioning slices transverse to the SC can increase the frequency of artifacts from surrounding tissues and organs (aliasing) whenever the field-of-view (FOV) is not extended enough to cover the whole SC cross-section. In this case, widening the FOV and the application of suppression pulses will be needed to improve the quality of the image (Hakky et al., 2013). An alternative is also the acquisition of sagittal slices to take advantage of the low curvature of the SC and of the smaller anterior-posterior dimensions of the chest.
Partial volume effect: Partial volume refers to the situation when different tissues contribute to the same voxel. In the SC, this occurs when a voxel is at the CSF/WM, WM/GM, and eventually CSF/WM/blood vessels interfaces, resulting in indistinct tissue-boundaries. Partial volume effects can be reduced by increasing the spatial resolution, but this in turn results in lower signal-to–noise and contrast-to-noise ratios (SNR and CNR). Higher magnet field strength (3T or 7T), higher number of phased-array coils with parallel imaging and multi-channel image acquisition (20, 32 or 64 channels) can improve spatial resolution, SNR and CNR reducing this kind of artefact (Zhao et al., 2014; Massire et al., 2018).
Inhomogeneous magnetic field: The spinal canal is surrounded by bones, ligaments, disks, arteries, and venous plexi. Its proximity to the esophagus, mediastinum and the lungs, each containing various amounts of air, create a challenging scanning environment. Adipose tissue, bone and air have different magnetic susceptibility profiles, which contribute to the inhomogeneity of the magnetic field around the SC, resulting in geometric distortions and signal intensity loss. To some extent, these artefacts can be counteracted with ‘shimming’. Shimming aims at compensating for field inhomogeneities by creating an auxiliary magnetic field via shim coils (Roméo and Hoult, 1984). While shimming improves overall field homogeneity, it is limited to smooth variations across large regions and cannot fully compensate for small and localized field inhomogeneities, such as those observed at cartilaginous discs between the vertebral bodies. Echo planar imaging sequences, such as diffusion tensor imaging (DTI), are particularly sensitive to geometric distortion around vertebral disks (Cohen-Adad et al., 2011; Rasoanandrianina et al., 2017). In addition to shimming, parallel imaging and careful slices positioning (i.e. slices centered in the middle of each vertebral body and perpendicular to the SC) may reduce magnetic field inhomogeneity. The image quality can be further optimized by a suitable choice of the pulse sequence. With few exceptions, MRI methods are either based on a gradient-echo or a spin echo pulse sequence. As echo time (TE) increases, the sequences become progressively T2*- and T2-weighted respectively. The key difference between them is that the spin-echo employs a refocusing pulse in order to reverse the effects of static field inhomogeneity for a brief instant of time. The MRI signal at the peak of the spin-echo is effectively free of the negative effects of the inhomogeneous magnetic field, and thus spin-echo imaging provides significant advantages to obtain high quality images of the SC (Stroman et al., 2014, El Mendili et al. 2019).
Physiological motions: Due to its proximity to the lungs and the heart, almost the entire SC undergoes periodical movements due to respiratory and cardiac pulsation (Kharbanda et al., 2006; Morozov et al., 2018). Spinal imaging is also affected by CSF flow and dynamic movements and is susceptible to movement artefacts from swallowing (Verma and Cohen-Adad, 2014). By ‘gating’ the acquisition, i.e. synchronizing it with the respiratory and cardiac cycles, the effect of periodical movements can be significantly reduced (Summers et al., 2006; Cohen-Adad et al., 2011; Massire et al., 2018). Motion artefacts can also be reduced using ‘saturation bands’ that cover the esophagus, chest and abdomen, attenuating signals from moving structures. Velocity compensating gradient sequences and signal averaging across multiple phases of motion can also be applied to minimize motion artefacts. Reducing acquisition time by using fast sequences and parallel imaging increasing acquisition speed effectively reduce both physiological and subject motion effects (Jaermann et al., 2004; Noebauer-Huhmann et al., 2007).
SC MRI has been extensively applied in the study of spinal injury, inflammatory and neurodegenerative diseases. Standard analyses focus both on morphological images through the study of SC atrophy and on WM structure through the use of diffusion sequences (Wheeler-Kingshott et al., 2014).
Cord morphometry: Gross axonal, WM and GM loss have traditionally been estimated by measuring SC cross-sectional areas at specific levels and interpreted as a proxy of atrophy in the context of reference normative values (Cohen-Adad et al., 2013a; Branco et al., 2014; El Mendili et al., 2015b; Paquin et al., 2018; El Mendili et al.; 2019). The ‘cross-sectional approach’ consists of estimating a mean cord cross-sectional area over a representative number of slices at a given vertebral level, which can be relatively easily calculated from conventional anatomical MRI sequences such as T1- or T2-weighted images. Segmentation of the SC is often performed in the cervical tract and has been mostly used to quantify SC atrophy in multiple sclerosis. Nevertheless, it can also be applied for co-registration and spatial normalization to a common coordinate space (i.e., template), which can be used to quantify morphometric changes or to perform atlas-based quantitative multiparametric analyses to investigate the structural and functional integrity of the SC (De Leener et al., 2016). Segmentation can be performed on 2D or 3D images and can be done entirely manually, semi-automatically (requiring only a few manual interventions) or fully automatically (El Mendili et al., 2015a; El Mendili et al., 2015b; Taso et al., 2015; De Leener et al., 2017; Dupont et al., 2017; Gros et al., 2018; Papinutto and Henry, 2019; Gros et al., 2019) (figure 6) with high reliability and precision. A variety of indexes, such as anterior-posterior dimension, left-right width, cord eccentricity and radial distance can be derived from different SC segmentation approaches. These measures reflect on different aspects of pathology, such as global versus regional, lateral versus anterior tissue loss, and are often related to predominant motor or sensory involvement (Lundell et al., 2011).
Figure 6: Example of automated cervical SC segmentation with identification of the global surface as well as of specific WM and GM surface for different vertebral levels (Fonov et al., 2014).
A new generation of MR scanners and coils as well as dedicated sequences (MEDIC, PSIR) (Massire et al., 2016; Martin et al., 2017; Olney et al., 2018) enables the production of high-resolution images that allows picturing WM and GM at specific SC levels (Dupont et al., 2017).
Recently, fully automated methods using multiparametric co-registration of different sequences and template-based analyses have been proposed, which have led to the development of dedicated software supporting the repeatability of analyses and the application in multi-centers studies (Taso et al., 2014; Taso et al., 2015; De Leener et al., 2016; De Leener et al., 2017; Dupont et al., 2017). An example of automate segmentation and co-registration of different imaging modalities is represented in figure 7.
Figure 7: Representation of automated segmentation, vertebral level identification, co-registration of different sequences and template-based analyses at the cervical SC level performed using the Spinal Cord Toolbox software (De Leener et al., 2017).
Most studies have been performed on 3-Tesla scanners but up-coming applications on 7-Testa machines seem to be more reliable and provide higher images quality and therefore better analyses precision (Cohen-Adad et al., 2013b). Moreover, technical improvement allows the application of new informatic tools, such as machine learning, to SC segmentation, further increasing the possibility of identifying modifications in tissue integrity leading to SC atrophy (Gros et al., 2019).
Diffusion weighted imaging: Diffusion weighted imaging (DWI) relies on the evaluation of water diffusion in CNS tissues and is primarily used to characterize the microscopic integrity of WM fibers (Duval et al., 2016). Diffusion results from the random movement of molecules in vivo (Brownian motion). Diffusion tensor imaging (DTI) consists of a mathematical computer-based model (a 3×3 symmetric matrix with six parameters) of diffusion gradients: in a spherical volume, the diffusion of water has no main direction (Le Bihan et al., 2001). This is the case of structures such as the CSF or the GM and is called isotropic (identical in all directions). In the WM, diffusion occurs preferentially along the axis of orientation of fibers bundles and is than called anisotropic (figure 8). By analyzing the diffusion of water in three main directions, four main parameters constituting the diffusion tensor model can be computed: fractional anisotropy (FA), radial and axial diffusivity (RD, AD), and mean diffusivity (MD). The first three parameters describe the spatial variation of water movements and are related to the orientation of the studied structures, giving an index of how strongly directional the water displacement is within tissue. In contrast, MD corresponds to the mean displacement of water molecules within the volume, independently from their orientation (Le Bihan et al., 2001).
Figure 8: Schematic representation of isotropic and anisotropic molecules diffusion.
Diffusion properties of tissues can be studied through an anisotropy map, which provides information about the microstructural organization of the WM (whole-brain voxel-wise methods); while tract-based spatial statistics (TBSS) and tractography methods allow the analysis of specific WM tracts. Specific DTI indices (AD, RD) have been associated with determined pathological processes, such as axonal versus myelin-related degeneration, but this interpretation is likely to be simplistic, as DTI measures are affected by axonal density, axonal diameter, myelin thickness and fiber orientation, fiber coherence and acquisition parameters. Intact WM is usually characterized by restricted diffusion parallel to the main fibers’ direction (leading to higher FA and lower MD), whereas damage to WM will cause diffusivity to be less restricted (i.e., lower FA and higher MD) (Andre and Bammer, 2010). Axial DTI of the SC is usually preferred to sagittal slices since it can reveal more information about specific WM bundles. Most DTI studies employ voxels that are elongated along the SC (e.g. 1 × 1 × 5 mm3). However, this is sub-optimal for tractography studies where isotropic voxels are preferred. Tractography-based quantification consists of reconstructing the fiber bundles via mathematical algorithms and subsequently quantifying DTI metrics within the tracts (Stroman et al., 2014). An alternative approach is to use region of interest (ROI)-based quantification to probe the integrity of specific tracts. Recent technical improvements allow automated template-based cervical SC tract identification and DTI computation (Lévy et al., 2015; De Leener et al., 2017; De Leener et al., 2018) (figure 9).
Figure 9: Template-based identification of the main cervical spinal cord white matter tracts (De Leener et al., 2017)
Novel high-directional approaches, such as high-angular resolution diffusion imaging (HARDI), q-ball imaging and diffusion kurtosis imaging are particularly well suited to reliably assess the integrity of crossing-fibers (Cohen-Adad et al., 2008, El Mendili et al., 2019). Emerging diffusion techniques such as NODDI help to estimate the microstructural complexity of dendrites and axons (Zhang et al., 2012).
Magnetization transfer imaging: Hydrogen nuclei linked to macromolecules such as the proteins and lipids of the myelin sheet have an extremely short T2 signal. While these macromolecules are not directly detectable by standard MRI sequences, magnetization transfer (MT) imaging enables their characterization. Macromolecular spins can be saturated using an off-resonance pulse, then the magnetization transfer between bound and free pools can be measured. MT occurs by mean of cross relaxation processes, such as dipole-dipole interactions and chemical exchange. Magnetization transfer ratio (MTR) is calculated as the percentage difference of MT images with macromolecules signal saturation and one without (M0) (Stroman et al., 2014, Combès et al., 2018). MTR enables inferences on myelin content, axonal count and density and has been used extensively to assess demyelination, remyelination and degeneration. However, in the SC, high resolution is necessary, and thus motion becomes a significant challenge when calculating the MTR (De Leener et al., 2018; El Mendili et al., 2019).
Inhomogeneous magnetization transfer imaging: Inhomogeneous magnetization transfer (ihMT) imaging is a novel method (Varma et al., 2015; Girard et al., 2017) which allows the unprecedented characterization of myelin integrity by isolating key myelin components from the broader macromolecular pool. ihMT shows unparalleled potential to detect and quantify demyelination and may be adapted to spinal applications (Taso et al., 2016; Van Obberghen et al., 2018).
MR spectroscopy: Magnetic resonance spectroscopy (MRS) is a well-established, non-invasive imaging tool which provides neurochemical insights on the studied tissue based on the concentration and relaxation profile of specific metabolites. The proton is the nucleus with the highest magnetic resonance sensitivity and natural abundance in living tissues (>99.9%). Proton magnetic resonance spectroscopy (1H-MRS) uses the slight differences in the magnetic field produced by electrons and nuclei in a localized region of tissue to resolve and quantify certain metabolites. It can distinguish N-acetyl aspartate (NAA; a marker of neuronal integrity), choline (Cho; a marker of membrane integrity), creatine (Cr; a chemical involved in energy metabolism) and myo-Inositol (mI). Furthermore, glutamate-related metabolites (glutamate and glutamine), as well as γ-amino butyric acid (GABA) can, depending on variables such as magnetic field strength, field homogeneity, and signal-to-noise ratio, be quantified separately or as a composite. Results are expressed by peaks.
Relatively few studies have used 1H-MRS to characterize metabolic changes at the spinal level, mainly focusing on multiple sclerosis patients (Marliani et al., 2010; Hock et al., 2013).
Functional MRI: Functional MRI (fMRI) detects local variations in blood oxygenation level-dependent (BOLD) MR signal at rest and during activation paradigms. Emerging SC fMRI studies in healthy controls recently provided proof of feasibility and high and super high field scanners, novel acquisition sequences and analyses techniques allow to overcome the presence of artefacts paving the way for applications in SC pathologies (Hock et al., 2013; Cohen-Adad, 2017; Powers et al., 2018).

READ  Evolution of the dust-to-oxygen ratio

Amyotrophic Lateral Sclerosis

Amyotrophic Lateral Sclerosis (ALS) is a relentless neurodegenerative disease leading to death after 3 to 5 years from symptom onset (Robberecht and Philips, 2013). ALS was first described at the end of the 19th century (1869) by Jean-Martin Charcot (Cleveland and Rothstein, 2001) and is characterized by the progressive degeneration of motor neurons in the brain, brainstem and anterior horn of the SC. Degeneration of upper motor neurons (UMN) in the motor cortex and in the SC determines the presence of spasticity, while the degeneration of lower motor neurons (LMN) induces the presence of muscle weakness and atrophy, fasciculations and cramps (figure 10) (Rowland and Shneider, 2001; Swinnen and Robberecht, 2014).
Figure 10: On the left side, the motor system composed of corticospinal (upper) motor neurons in the motor cortex and bulbar and spinal (lower) motor neurons, which innervate skeletal muscle (Rowland and Shneider, 2001). On the right side: main involvement sites and related symptoms in ALS (Swinnen and Robberecht, 2014).
ALS is a rare disease with an incidence of 2-3 individuals per 100.000 in the Caucasian population and is slightly more common in men than in women (Al-Chalabi and Hardiman, 2013). The estimated prevalence in Europe is 5/100.000 (Chiò et al., 2013). The role of genetic and environmental factors in the pathogenesis of the disease is still under debate (Al-Chalabi and Hardiman, 2013).
Similarly to other neurodegenerative diseases, ALS origins focally and spreads: symptoms start in the limbs or in the bulbar region and progress to paralysis of nearly all skeletal muscles (Ravits et al., 2007; Ravits and La Spada, 2009).

Clinical presentation

The clinical presentation of ALS is extremely heterogeneous concerning onset symptoms, disease progression rate and survival. Conventionally, the disease is classified according to the site of onset of the muscle weakness (Ravits et al., 2007; Ravits and La Spada, 2009), with identification of five main phenotypes:
Classic ALS: In most patients (65%), ALS starts with asymmetric weakness in a limb and is referred to as spinal-onset ALS. Muscle weakness is usually associated with atrophy, fasciculations, hyperreflexia and often with a mild to severe hypertonia with possible Babinski or Hoffmann sign. The classic form of ALS is more frequent in male than female subjects (ratio 1.65:1) (Chio et al., 2011), with a peak of incidence in the seventh decade in both genders. Mean survival time is 2.6 years.
Bulbar onset ALS: 30% of ALS patients present with bulbar onset of the symptoms manifesting as dysarthria, dysphagia and tongue fasciculations (Chio et al., 2011; Swinnen and Robberecht, 2014). A brisk jaw jerk is often found in these patients and pseudobulbar affect (uncontrolled crying or laughing) is sometimes present. The peak of onset is in the eighth decade of life (Chio et al., 2011) and the prognosis is worse than in patients with spinal onset, with a mean survival of 2 years and long-term survival (>10 years) of only 3%. The poor prognosis is mostly attributable to the higher probability of contracting aspiration pneumonia due to dysphagia, to the presence of nutritional impairment and to early respiratory failure.
Respiratory onset ALS: In about 3–5% of patients, ALS starts with respiratory involvement characterized by orthopnea or dyspnea and mild or even absent spinal or bulbar signs. Respiratory onset ALS has a male predominance (Shoesmith et al., 2007; Swinnen and Robberecht, 2014). The prognosis is poor, with a mean survival of 1.4 years, which is little improved by non-invasive ventilation (Gautier et al., 2010).
Flail leg syndrome: Some patients present with isolated, often asymmetric and frequently distal LMN involvement in the lower limbs for at least 12 months (Wijesekera et al., 2009). Subtle signs of UMN degeneration might be present as well and, subsequently, upper limb and bulbar muscles may be involved too (Kobayashi et al., 2010). This phenotype has a similar incidence in the two genders, a median survival time of 3.0 years and a 10 year survival rate of 12.8%, similar to that of classic ALS (Chio et al., 2011).
Flail arm syndrome: A minority of ALS patients present LMN involvement that remains limited to the upper limbs for at least 12 months (Hu et al., 1998; Gamez et al., 1999). This phenotype is rare and more common in men (male to female ratio of 4:1). Mean age at onset is around 60 years of age. Median survival time is 4 years with a 10-year survival rate of 17.4%. Respiratory impairment associated to cervical motor neurons degeneration is frequent (Czaplinski et al., 2004).
A second axe of classification of ALS is based on prevalent UMN or LMN involvement:
UMN-dominant ALS and primary lateral sclerosis (PLS): Patients presenting with prevalent UMN and only subtle LMN signs (Gordon et al., 2009) are considered to have UMN-dominant ALS (D’Amico et al., 2013). If the UMN signs remain isolated for 4 or more years, they are defined as having PLS findings (Gordon et al., 2006). PLS constitutes about 5% of all cases of MND. It is characterized by slow progression rate and sparing of respiratory function with long survival (up to 20 years) (Le Forestier et al., 2001). Patients with UMN-dominant ALS have a relatively slow progression rate even if conversion to classic ALS is possible (Tartaglia et al., 2007). Cognitive impairment with frontotemporal degeneration is retained to be more frequent in patients with prevalent UMN involvement (Agarwal et al., 2018; de Vries et al., 2018).
Progressive muscular atrophy (PMA): The term PMA is used to describe an adult onset form of MND affecting only the LMN with asymmetric and relatively slowly progressive limb muscles weakness. It represents about 5% of all MND cases with a prognosis that is slightly better than that of the classic ALS form (Rowland, 2010; Garg et al., 2017).
All the ALS phenotypes are considered to be part of a disease spectrum. A correlation between some specific phenotypes and age of symptom onset has been observed, with classic ALS, PLS and PMA incidence decreasing with age and that of the bulbar phenotype increasing in older subjects (Chio et al., 2011). Survival is also different according to the phenotype, with PLS and UMN-dominant ALS having the longer survival and classic and bulbar ALS having the worse prognosis (Swinnen and Robberecht, 2014; Calvo et al., 2017).

Extra-motor involvement in ALS

ALS has long been retained a pure motor syndrome, even if early evidence of cognitive and behavioral abnormalities was already acknowledged at the beginning of the 20th century (Mitsuyama, 1984). Such evidence has been confirmed and extended through the last 25 years and sustained by pathological, genetic and neuroimaging data describing a clinical continuum between ALS and Frontotemporal dementia (FTD) (Goldstein and Abrahams, 2013). Nowadays, it is retained that almost 50% of ALS patients have some cognitive impairment, while up to 15% of them reach the criteria for FTD (Ringholz et al., 2005; Witgert et al., 2010; Gordon et al., 2011). Cognitive impairment in ALS may manifest as behavioral (ALS-bi) or as executive and language dysfunction (ALSci) (Phukan et al., 2007), which can extend to full-blown FTD according to the Neary criteria (Phukan et al., 2012; Goldstein and Abrahams, 2013). Interestingly, it has been demonstrated that about 15% of FTD patients develop an associated MND during the course of the disease and that 30-40% of them have subclinical signs of motor neuron degeneration, further supporting the hypothesis that ALS and FTD are part of the same disease spectrum (Josephs et al., 2006; Burrell et al., 2011; Burrell et al., 2016). Cognitive impairment is more frequent in bulbar onset ALS (up to 60% of the patients) (Chio et al., 2011) and is a negative prognostic factor (Olney et al., 2005; Elamin et al., 2013). It may be present before the onset of motor symptoms and can be suggestive of peculiar genetic mutations such as those in the c9orf72 gene. Cognitive impairment can be progressive and seems to be associated with the pathological spread of degeneration (Lulé et al., 2018a).
Recent studies have demonstrated that a distinct neuropsychological profile specific to ALS is detectable also in non-demented patients and that it has peculiar features when compared to pure FTD (Abrahams et al., 1995; Lulé et al., 2018b). It can be resumed as follows (Barulli et al., 2015; Christidi et al., 2018a):
Executive dysfunction: It is one of the most studied and common features of the ALS cognitive profile (Abrahams et al., 2000). Executive functions are a complex group of higher-order brain processes related to the frontal and prefrontal cortex encompassing the capacity of planning and initiating actions but also behavioral regulation, situation-appropriated decision-making, complex problem solving and verbal fluency. Reduced phonemic fluency and semantic dysfunction are frequent in demented and non-demented ALS patients (Abrahams et al., 2000) and can be detected already at the beginning of the disease using dedicated neuropsychological tests such as the Edinburgh Cognitive and Behavioral ALS Screening (ECAS) (Lulé et al., 2015; Niven et al., 2015; Crockford et al., 2018). ALS patients may also present difficulties in concepts formation and reduced mental flexibility (Evans et al., 2015) as well as disability in tasks implicating working memory (Goldstein and Abrahams, 2013).
Language deficits: Language dysfunction is recognized as a key element of cognitive impairment in ALS (Pinto-Grau et al., 2018) and has been consistently depicted also in subjects without executive dysfunction (Taylor et al., 2013). ALS patients show impaired syntactic processing, deficits in verb naming and action verb management (Ash et al., 2015).
Social cognition deficits: Deficits in social behavior, social perception and empathy are acknowledged as a core feature of the ALS neuropsychological profile (Girardi et al., 2011; Andrews et al., 2017) and are associated with impairment in processes related to the theory of mind, a complex cognitive function localized in the frontal lobe and accounting for the recognition of others affects and intentions (van der Hulst et al., 2015; Keller et al., 2018). Patients with ALS may also exhibit impaired emotional processing and reduced ability to interpret emotional facial expressions (Aho-Özhan et al., 2016). Such symptoms are correlated to the presence of cortical atrophy of the right orbitofrontal, superior temporal, occipital and posterior cingulate regions (Beeldman et al., 2018).
Behavioral deficits: The most frequent behavioral modifications in ALS are perseveration, apathy and disinhibition followed by loss of disease insight, indifference and irritability (Chiò et al., 2010; Raaphorst et al., 2012; Radakovic et al., 2016; Christidi et al., 2018a).
Memory deficits: Episodic memory is usually intact in ALS and ALS-FTD patients. Nevertheless, semantic memory may be affected independently from the presence of executive dysfunction and is related to the presence of mesial temporal lobe degeneration (Hervieu-Bègue et al., 2016; Christidi et al., 2018b).
Psychiatric symptoms are infrequent in classic ALS, but delusions and psychosis may be part of a distinct syndrome related to c9orf72 gene mutations.
Extrapyramidal involvement: The association between ALS and extrapyramidal signs and symptoms is possible, with atypical parkinsonian signs being more common than overt Parkinson’s Disease (Desai and Swash, 1999). Mild parkinsonism is present in 5–15% of patients with ALS and is typically characterized by postural instability and backward falls (Feron et al., 2018). In such patients, abnormalities in the nigrostriatal system can be seen on dopamine transporter imaging or at autopsy (Park et al., 2011).
Sensitive involvement: Clinically evident signs and symptoms of sensory involvement are infrequent in classic ALS. Nevertheless, sub-clinical degeneration of the sensory pathways has been described both through neurophysiological and imaging techniques (Cohen-Adad et al., 2013a; Iglesias et al., 2015), suggesting that the involvement of ascending spinal tracts is associated with impaired neural processing of sensory inputs at cortical and subcortical level (Lulé et al., 2010). These data support the hypothesis that, in ALS, alterations of connectivity, excitability and integrative properties extend from motor to non-motor areas (Sangari et al., 2018).


The diagnosis of ALS is mainly clinical and is based on the identification of neurological signs of both UMN and LMN degeneration. It should be anyway kept in mind that the clinical presentation can be blurred and the rate of disease progression extremely heterogeneous. To homogenize the diagnostic definition of ALS, the El Escorial criteria have been approved in 1994 by the World Federation of Neurology (Brooks, 1994) and further modified and renamed in the Airlie House criteria in 2008. They are resumed in table 1 (Hardiman O et al., 2011).
Nowadays, the El Escorial criteria are used especially for patients stratification in clinical trials but they are not considered exhaustive in a clinical setting (Beghi et al., 2002). Atypical phenotypes still represent a diagnostic challenge and a long diagnostic delay (in average 12 months) is frequent (Nzwalo et al., 2014; Paganoni et al., 2014). The diagnosis of ALS remains an exclusion one and needs to be sustained by several complementary tests.
Laboratory investigations: They should include standard biochemical screening, erythrocyte sedimentation rate, thyroid function tests and cerebrospinal fluid analysis. A heavy metal screen should be performed in individuals with a potential history of exposure. Serum creatinine-kinase may be moderately elevated. To exclude symptoms secondary to systemic diseases, screening for autoimmunity and for paraneoplastic syndromes should be carried out as well (Hardiman et al.,  Electrodiagnostic studies are the most relevant complementary test in the investigation of ALS. Electromyography can identify loss of LMN, which is transposed in fasciculations, spontaneous denervation discharges (fibrillation potentials and positive sharp waves) and polyphasic units indicative of reinnervation (Eisen and Swash, 2001). More recent neurophysiological techniques able to quantify the number and the dimension of motor units (MUNE and MUNIX) have been proved reliable and sensitive in the early detection of LMN loss and of reinnervation (Gooch and Shefner, 2004; Escorcio-Bezerra et al., 2016).

Table of contents :

1. Background
1.1 Hints on spinal cord anatomy
1.2 Basic principles of MRI
1.3 Amyotrophic Lateral Sclerosis
1.3.1 Clinical presentation
1.3.2 Extra-motor involvement in ALS
1.3.3 Diagnosis
1.3.4 Clinical genetics of ALS
1.3.5 Pathology and pathogenesis in ALS
1.3.6 Pathological staging of ALS
1.3.7 Advanced neuroimaging in ALS
1.3.8 Brain imaging in ALS
1.3.9 Spinal cord imaging in ALS
1.3.10 Biomarkers in ALS
1.4 5q-Spinal Muscular Atrophy (SMA)
1.4.1 Clinical presentation of adult 5q-SMA.
1.4.2 Genetic background
1.4.3 Diagnosis
1.4.4 Outcome measures.
1.4.5 Upcoming therapies
2. Objective of the study.
3. Results
Study 1: Spinal cord multi-parametric magnetic resonance imaging for survival prediction in amyotrophic lateral sclerosis
Study 2: Multi-modal spinal cord MRI offers accurate diagnostic classification in ALS
Study 3: Presymptomatic longitudinal cord pathology in c9orf72 mutation carriers: longitudinal neuroimaging study
Study 4: The spinal and cerebral profile of adult spinal-muscular atrophy: a multimodal imaging study
Study 5: The motor neuron number index (MUNIX) profile of patients with adult Spinal Muscular Atrophy
Study 6: Biomarkers definition for adult spinal muscular atrophy (SMA): experience from a longitudinal study
4. General discussion
4.1 Pathological implications of spinal cord MRI studies
4.2 Development of diagnostic and prognostic biomarkers
4.3 Future perspectives


Related Posts