Spinal cord markers in ALS – Diagnostic and biomarker considerations
REVIEW ARTICLE
Peter Bede,Arun L. W. Bokde, Susan Byrne, Marwa Elamin,Andrew J Fagan, Orla Hardiman
Trinity College Institute of Neuroscience, Dublin, Ireland
Centre for Advanced Medical Imaging (CAMI), St. James’s Hospital / Trinity College Dublin, Ireland
ABSTRACT
Background: Despite considerable involvement of the spinal cord in Amyotrophic Lateral Sclerosis (ALS), current biomarker research is primarily centred on brain imagingand CSF proteomics. In clinical practice, spinal cord imaging in ALS is performed primarily to rule out alternative conditions in the diagnostic phase of the disease. Quantitative spinal cord imaging has traditionally been regarded as challenging, as it requires high spatial resolution while minimizing partial volume effects, physiological motion and susceptibility distortions. In recent years however, as acquisition and post processing methods have been perfected,a number of exciting and promising quantitative spinal imaging and electrophysiology techniques have been developed.
Methods: We have performed a systematic review of the trends, methodologies, limitations and conclusions of recent spinal cord studies in ALS to explorethe diagnostic and prognostic potential of spinal markers.
Results: Novel corrective techniques for quantitative spinal cord imaging are systematically reviewed. Recent findings demonstrate that imaging techniques previously used in brain imaging, such as diffusion tensor, functional and metabolic imaging can now be successfully applied to the human spinal cord. Optimized electrophysiological approachesmake the non-invasive assessment of corticospinal pathways possible, and multimodal spinal techniques are likely to increase the specificity and sensitivity of proposed spinal markers.
Conclusions: Spinal cord imaging is an emerging area of ALS biomarker research. Novel quantitative spinal modalities have already been successfully used in ALS animal models andhave the potential for development into sensitive ALS biomarkers in humans.
INTRODUCTION
Biomarker research in Amyotrophic Lateral Sclerosis (ALS) is an active field of neurological research. Although no single such marker has been validated to date, serum[1],CSF, electrophysiological and neuroimaging[2] based markers are currently undergoing extensive investigation[3].
Because of the significant disease heterogeneity, sufficient patient numbers are required to augment statistical power.Structured international collaborations are now underway to share high quality multicentre imaging data in ALS.[4] In recent years novel quantitative MRI techniques, fast acquisition times and effective artefact correction methods have led to better clinical correlation and increased utility in spinal imaging in ALS both in humans and in animal models. However, despite these significant advances, no systematic analysis of spinal cord studies in ALS can be identified. Notwithstanding, the sensitivity of advanced spinal cord imaging techniques such as diffusion tensor imaging have been successfully demonstrated in several neurological conditions such as multiple sclerosis[5], spinal cord compression[6], cord injury[7], spinal artery occlusion[8], spinal tumours [9], syringiomyelia[10], inflammation[11] and arteriovenous malformations[12].
In addition to cross sectional spinal cord imaging studies, longitudinal studies have been published both in multiple sclerosis (MS)5 and ALS[13] and spinal imaging has been used to assess medication effect in multiple sclerosis[14].
SEARCH STRATEGY AND SELECTION CRITERIA
Publications were searched during a 4 month period between May2011 and August2011. Only articles published in English were reviewed. Further articles were identified through reviews of the references of these articles. Reports from meetings were also used if they presented relevant information.
The literature search was performed using Pub Med with the following keywords: Motor Neuron Disease (MND), Amyotrophic Lateral Sclerosis (ALS), Biomarker,Marker, Spinal cord, Diffusion tensor imaging, Magnetic resonance spectroscopy, Voxel based morphometry, Longitudinal, Cross-sectional, Upper motor neuron, Atrophy, Electrophysiology, Transcranial magnetic stimulation, ALS Animal Model, SOD1 Mice. The keywords Motor Neuron Disease (MND) and Amyotrophic Lateral Sclerosis (ALS) were separately combined with each of the other keywords as pairs of keywords.
RESULTS
Spinal markers in ALS Animal models
Magnetic Resonance Microscopy (MRM) of ALS animal models can achieve 41µm isotropic resolution on brain and 35 µm isotropic resolutionon spinal cord imaging[15]. Such high spatial resolution is comparable with traditional histological techniques without the tissue distorting effects of fixation, sectioning and staining.
The sensitivity of spinal DTI in SOD1 mouse model of ALS has been elegantly demonstrated recently[16] opening the opportunity for measuring therapeutic response in animal models.
Magnetic resonance spectroscopy (MRS) has also been applied to FALS mice spinal cord[17] showingdecreasedN-acetyl aspartate(NAA), N-acetylaspartylglutamate (NAAG) as well as increasedglutamate, taurine and inositol levels. Interestingly, in SOD1 mice the metabolic abnormalities are most pronounced in the spinal cord, followed by medulla and then the sensorimotor cortex. This observation may provide further rationale for the development and validation of spinal markers.
Pharmacologic MRI (phMRI) has been successfully used in a rat model of familial ALS[18] in which response to amphetamine was assessed, demonstrating upregulation of pre-motor areas and impaired activation of the primary motor cortex. These changes reflectresults of traditional motor task fMRI studies. This brain study indicateshow innovative pharmacological approaches can be applied to ALS imaging, and has the potential to open the way for future spinal applications.
Spinal markers in ALS – Human applications
Conventional MRI techniques
High signal changes along the corticospinal tracts (CST) in the brain on fluid attenuated inversion recovery (FLAIR), T1 and T2 weighted imaging have been long described in the literature, but have been shown to be poorly sensitive[19] and specific for ALS[20][21].
CST signs are also well described on spinal imaging[22], but their significance isuncertain.Terao et al.[23] reported that 9 out 13 patients had spinal CST signs. Thorpe et al.[24] reported that axial imaging revealed high signal in the lateral white matter in eight out of eleven patients. They highlight that two of their patients had only cord hyperintensities without brain signs, suggesting that spinal cord MRI may increase the sensitivity of detecting radiological signs of corticospinal pathway pathology. However, such signs are unlikely to be useful for diagnostic or biomarker purposes.
Simple measures of spinal cord atrophy such asspinal cord cross-sectional area (SCCA) have been successfully correlated with clinical function in Multiple Sclerosis (MS)[25] and in chronic incomplete spinal cord injury (SCI)[26]. Objective volumetric techniques used in brain imaging such as 3D-modified driven equilibrium Fourier transform (3D-MDEFT)based acquisition protocols have also been shown to quantify spinal cord cross-sectional areas reliably[27]. Spinal cord atrophy was originally considered to be a possible feature of ALS, and Agosta et al. found that cross-sectional area decreases over time on longitudinal spinal cord imaging in ALS.14 However, the specific methods by which cross sectional axial spinal cord areas are quantified vary considerably, and other imaging studies have failed to confirm this[28].
Diffusion Tensor Imaging (DTI)
Various studies use different DTI parameters such asFractional Anisotropy (FA), Mean Diffusivity (MD), Radial Diffusivity (RD), or Axial Diffusivity (AD) and have selecteddifferent spinal segments as their region of interest (ROI). Many studies have attempted clinical correlation with the intentionof validating the sensitivity of their imaging methods.
Valsasina et al.[29] conducted a study of 28 ALS patients and 20 healthy controls. They demonstrated that patients with ALS had significantly lower average fractional anisotropy of the cervicalcord and found significant correlation between cord average fractionalanisotropy and ALSFRS.
In a longitudinal spinal MRI study Agosta et al.[30] demonstrated a significant decrease in cord average FA and a significant increase in cord averagemean diffusivity (MD) in patients with ALSduring a mean follow-up of 9 months.
The majority of the studies have been of the superior cervical cord. Nair et al. conducted a multisegmental cervical spinal cord DTI study[31]and concluded that FA and RD differences between healthy subjects and ALS patients are greater in the more distal segments of the cervical cord.
A number of novel diffusion-weighted imaging techniques have been developed that are ideally suited to spinal applications and may be superior to standard DTI. Q-ball imaging (QBI) is a high angular resolution diffusion imaging (HARDI) method that allows the detection of crossing fibres. QBI offers significant additional benefit to conventional DTI by retrieving commissural, medio-lateral and dorso-ventral fibres in the spinal cord as well as the longitudinal fibres easily recovered by DTI[32].Contiguous-slice zonally orthogonal multislice (CO-ZOOM-EPI)[33] is another new technique that provides high resolution with reduced susceptibility distortions,making it an attractive spinal imaging method.
Similarly to brain studies, the usefulness and sensitivity of multi-parametric MRI has been demonstrated in the human spinal cord by Cohen-Adad et al[34], by combining high angular resolution diffusion imaging (HARDI), magnetization transfer(MT) and measures of cord atrophy, showing degeneration in normal appearing human spinal cord with good clinical correlation.
Magnetic resonance spectroscopy (MRS)
Magnetic resonance spectroscopy allows the non invasive measurement of different metabolites in a predefined voxel. The most frequently used metabolites include N-acetyl aspartate(NAA), Choline(Cho), Creatine(Cre), Myo-inositol(Myo), and Lactate. High absolute or relative (NAA/Cho, NAA/Cre+Cho)N-acetyl aspartatelevels are considered to be a marker of neural integrity. Creatine levels are regarded as a marker for metabolic activity, and choline a marker of membrane integrity. MRS studies can define their region of interest (ROI) in a single voxel – as if taking a virtual biopsy - or can use a grid of multiple volumes in multi-voxel techniques.
In the brain, MRS has been extensively used in ALS, in cross sectional, longitudinal[35], medication effect[36][37], motor cortex[38], brain stem[39][40], extra-motor cortex[41], single voxel[42] and multivoxel studies. The technique is generally regarded as a sensitive non invasive modality to identify and follow up early metabolite abnormalities in ALS. The validation of brain MRS studies in ALS is based on clinico-pathological correlations.
Despite the potential advantages of spinal cord MRS,such as measurement of medication effects and the inclusion of the anterior horns, very few successful spinal cord MRS studies can be identified in the literature. The technical challenges include relatively small volumes, small signal to noise ratio, longer imaging times and strong magnetic field inhomogeneities in the relevant regions.
A series of corrective imaging techniques, such as ECG triggering and higher-order shimming have been proposed to perform successful MRS in the cervical spinal region.[43]A single voxel cervical spinal MRS study[44] found that NAA/Myo and NAA/Cho correlated with the forced vital capacity (FVC).
One of the most interesting additions to the ALS spectroscopy literature is acervical spinal cord study of non-symptomatic people with SOD1 gene mutation. Carew et al.[45] found reduced neurometabolite radios in SOD1+ individuals compared to SOD1- healthy controls in advance of onset of symptoms. The predictive value of abnormal MRS in SOD1+ individuals in forecasting symptom onset is likely to be clarified by longitudinal studies. This study raises very important questions on presymptomatic biomarker development for asymptomatic people with known fALS mutations.
Spinal functional MRI (fMRI)
Functional brain MRI(fMRI) is used extensively in ALS research. Approaches include motor paradigms[46], experimental neuropsychology tasks[47][48] and resting state protocols[49].Despite technical difficulties (Table 1.) a number of spinal cord fMRI studies havenow been carried out both in animal models[50] and humans[51].
Spinal fMRI challenges / Correction techniquesSmall cross-sectional dimensions of the spinal cord require high resolution imaging to distinguish between gray and white matter and to avoid partial volume effects, however small voxels provide poor signal-to-noise ratio / Multiple methods have been developed:
1., transverse / axial slices with high in-plane resolution <2mm, with relatively large slice thickness ~10mm
2., Thin slice sagittal acquisition, followed by 3D volume reformatting and corrected by smoothing in the rostral-caudal direction[52]
Magnetic field inhomogeneities caused by spinal cord / vertebral body interface / spin-echo imaging techniques with short echo times
Using SEEP (signal enhancement by extravascular water protons) contrast (Stroman et al. 2001)
Motion artifacts due to cerebrospinal fluid, blood flow, respiration / Cardiac and respiratory gating
Breath-hold acquisition
Flow compensation gradients
Spatial saturation pulses
Retrospective image correction techniques based on cardio-respiratory physiological noise models (Brooks et al.[53])( Figley et al[54])
Independentcomponent analysis (ICA) to identify cardiaccomponents in spinal fMRI data (Piche et al)[55]
Poor spatial distribution of activationand reproducibility / gradient echo echo-planar imaging (GE-EPI)is superior to turbo spin echo(TSE) spinal fMRI[56]
Table 1. Technical challenges of spinal functional MRI
Early human spinal fMRI studies were primarily based on sensory stimulation, demonstrating dorsal horn activation in the relevant cord segments[57]. Motor task based human spinal fMRI studies have been successfully carried out in healthy individuals both with upper[58][59] and lower limb tasks[60].Clinical applications of spinal fMRI techniques includeMultiple Sclerosis (MS)[61], spinal cord injury[62]and pain medicine. While no spinal fMRI study has been published in ALS to date, this technique might be particularly applicable to ALS research. Activity of brain stem cranial nerve nuclei can be visualised with fMRI[63], making it a potentially attractive method in bulbar ALS patients.Spinal cord fMRI could also be investigated and validated as a potential non invasive marker in ALS.
Current clinical applications
In current clinical practice spinal MRI is only performed in selected atypical cases to exclude possible alternative diagnosis[64] (Table 2). However, recent innovations have addressed many of the technical limitations of spinal imaging(Table 3) and the superior cervical spinal cord can now be successfully scanned in humans without significant motion artifacts. The main ALS specific MRI challenges include dyspnoea, orthopnoea, sialorrhea and impaired communication between the radiographer and patients with bulbar involvement.
Hirayama disease[65]Space occupying lesions; Spinal tumours, abscesses or haematomas
Degenerative changes; cervical or lumbar radiculomyelopathy (spondylosis)
Syringomylelia
Anterior spinal cysts (Extra- or intradural)[66]
Transverse myelitis
Supeficial Sideoris [67]
Table 2. Possible ALS mimics requiring spinal MRI for exclusion
Source of imaging artifacts in spinal imaging / Techniques for CorrectionsCSF flow / Quick acquisition protocols
Small axial cross sectional area, partial volume errors (PVE) / High field scanners, high resolution, axial acquisitions
Low Signal to Noise Ratio in DTI sequences / Decreasing EPI factor, fast single-shot EPI with use of sensitivitiy encoding (SENSE)[68] , multishot echo-planar imaging[69]
Cardiac, arterial pulsation / Cardiac gating
Respiratory effects / Pulseoxymetric triggering, Quick acquisition, superior cervical imaging
Swallowing effects, especially in patients with sialorrhea / Rest slab or Saturation band placement to cover pharyngeal regions anterior to the spinal cord,
Quick acquisition, superior cervical imaging
Susceptibility artifacts and image distortion, Problematic regions: skull base, the spinal cordand the surrounding bony vertebral column / Parallel imaging technique using arrays of multiple receiver coils[70]
Shimming over the region, using smaller EPI factor, using thin slices, or short TE
Table 3.Technical challenges of spinal imaging
Pathological considerations
Although the spinal cord is heavily involved in ALS pathology and lower motor neuron involvement is a diagnostic requirement in ALS, the majority of MRI biomarker research studies in ALS are brain centred.
The “dying-back”[71] hypothesis is often used by imaging studies as an argument to study the spinal cord in ALS. The concept of “Dying back”of axons towards the cell body is based on early pathological observations[72] that corticospinal tract degeneration in the spinal cord tends to be more pronounced caudally than rostrally. In a detailed histological ALS case series Brownel et al.[73]indicated that in 14 out of 35 cases pyramidal degeneration could not be traced above a certain level. The ‘Dying back’ hypothesis was later supported by SOD1 animal models, where motor unit loss was noted to precede motor neuron death. SOD1 mutant ALS mouse model studies[74] suggest that end-plate denervation takes place before there is evidence of ventral root or cell body loss, and then ventral root axonal damage precedes motor neuron loss.However, mouse and human motor systems showsignificant differences, so patterns of spread are best studied on human case series. An elegant 3 dimensional model of disease spread was proposed by Ravits et al.[75] based on focal onsetand the differing LMN and UMN somatotopicanatomy.
There is now emerging evidence that neuroimmune processes[76] contribute to ALS pathology in the brain[77] and that significant microglia/macrophage activation also takes place in the spinal cord in ALS[78][79]. This immune response provides a further biomarker target in the spinal cord, especially that microglial activation in the spinal cord occurs early[80], in the preclinical stage of ALS in SOD1 animal models. Sites of microinflamation can be highlighted by magnetically labelled anti-CD4 antibodies a technique called immune MRI (iMRI), a method that has already been successfully applied to ALS.[81]
Disruption of the blood-brain barrier (BBB) and blood-spinal cord barrier (BSCB) have been proposed as important contributors to the microenvironment of ALS pathology[82]. BBB and BSCB permeability can be assessed by dynamic contract-enhanced magnetic resonance imaging[83](DCE-MRI)using Gd-DTPA[84] contrast or by the identification of hemosiderin deposition. Despite several human studies[85], most of the BBB and BSCB data in ALS is based on animal studies[86][87]. Recently, a high field 7 Tesla brain imaging study in humans has shown no evidence of BBB disruptionin ALS[88], but the study was confined to brain imaging. BBB and BSCB disruption in ALS offers another possible biomarker target that has not been fully explored to date.