In multiple sclerosis (MS), MRI is the most important paraclinical tool used to inform diagnosis and for monitoring disease evolution, either natural or modified by treatment. The increased availability of ultra-high-field magnets (7 Tesla or higher) gives rise to questions about the main benefits of and challenges for their use in patients with MS. The main advantages of ultra-high-field MRI are the improved signal-to-noise ratio, greater chemical shift dispersion, and improved contrast due to magnetic susceptibility variations, which lead to increased sensitivity to the heterogeneous pathological substrates of the disease. At present, ultra-high-field MRI is mainly used to improve our understanding of MS pathogenesis. This review discusses the main achievements that have so far come from the use of these scanners, which are: better visualisation of white matter lesions and their morphological characteristics; an improvement in the ability to visualise grey matter lesions and their exact location; the quantification of ‘novel’ metabolites which may have a role in axonal degeneration; and greater sensitivity to iron accumulation. The application of ultra-high-field systems in standard clinical practice is still some way off since their role in the diagnostic work-up of patients at presentation with clinically isolated syndromes, or in monitoring disease progression or treatment response in patients with definite MS, needs to be established. Additional challenges remain in the development of morphological, quantitative and functional imaging methods at these field strengths, techniques which may ultimately lead to novel biomarkers for monitoring disease evolution and treatment response.
- Multiple Sclerosis
Statistics from Altmetric.com
MRI provides clinicians and researchers with access to information about the structure and function of human brain non-invasively. The signal-to-noise ratio (SNR) in MRI increases strongly with the polarising magnetic field strength, and this extra SNR can be traded off to give improved spatial resolution. Contrast in brain MRI depends on regional differences in tissue water density as well as on the tissue relaxation properties. Differences in the longitudinal (T1) and transverse (T2) relaxation time values readily enable differentiation of tissues such as grey matter (GM) and white matter (WM) in vivo. This superior soft tissue contrast has made MRI the modality of choice for neurological imaging since the mid-1980s, with rapidly rising use in medical diagnostics.
While conventional MRI at 1.5 Tesla (T) has been the mainstay of diagnostic imaging for over 25 years, it has not yielded strong biomarkers correlating with disease progression in multiple sclerosis (MS). It is known that inflammation, demyelination, remyelination, oedema, gliosis and axonal loss are involved in MS, but 1.5 T MRI does not appear to offer measures that are capable of differentiating between these neuropathological substrates of the disease. Ultra-high-field MRI has the benefits of an increased SNR, greater chemical shift dispersion (CSD), and improved contrast due to magnetic susceptibility variations. Images acquired at 7 T and 8 T have shown that increased SNR can be used to visualise smaller structures or achieve faster imaging in echo planar imaging, useful for functional imaging applications, for example. Similarly, magnetic resonance spectroscopy (MRS) at ultra-high-field has benefitted from the greater CSD to produce high-resolution spectra from human brain in clinically acceptable acquisition times. N-acetylaspartate (NAA), creatine and choline, as well as many other metabolites (ie, , glutamate/glutamine (Glx) and gamma aminobutyric acid (GABA)) can be more readily detected in MRS at ultra-high-field.
After a summary of the main technical issues that should be taken into consideration when dealing with ultra-high-field MRI scanners, this review examines the improvements that MS pathologists would like to see by their use. Then, there is discussion of the findings from recent studies that have applied different ultra-high-field MR-based techniques for the assessment of focal and diffuse damage in the brains of patients with MS. Challenges and future developments that can be foreseen from the increased availability of these magnets will also be examined in the concluding section.
In an MRI scanner, the applied polarising magnetic field aligns the proton spin magnetic dipole moments and divides them in two groups of parallel and antiparallel populations. The antiparallel (+) group has more energy than the parallel (−) protons. This energy gap between these groups, ΔE=E+ − E−, increases linearly with the magnetic field strength (B0). Additionally, the difference between the number of protons in the two populations also increases with B0. These two factors together lead to an increase in the received MRI signal that, all other factors being equal, increases with the square of the polarising B0 field. While the relationship between the B0 field strength (and therefore resonance frequency) and the background noise level is more complicated, the SNR increases strongly with B0, which allows imaging with smaller voxel sizes, and hence, improved spatial resolution.
In spite of the aforementioned advantages, high magnetic fields have been perceived as posing a higher risk. A comprehensive review of the safety research was published in 1999 specifically addressing issues associated with exposure to static magnetic fields up to 10 T.1 This work convinced the scientific community that safe studies of human subjects at fields within that range were possible.1 Subsequent safety studies conducted at 7 T and 8 T confirmed those findings and paved the way for the development of ultra-high-field systems into viable medical imaging devices. It is generally accepted that the increase in SNR with B0 is better than linear, and that a gain in SNR of between 5 and 10 times is typically seen for 7 T scanners compared with the standard clinical scanners operating at 1.5 T. A 10-fold gain should allow a reduction of voxel dimensions by a factor of 2 in every dimension. Images acquired at 8 T with 2 mm slice thickness and 100 microns in-plane resolution support these theoretical predictions.2 The ability to depict biological structures, such as blood vessels, nerve bundles, microbleeds, microscopic lesions or infarcts, ischaemia and other cerebrovascular pathologies in a non-invasive way, will allow their roles in disease development and treatment to be explored. In MS, the ability to image at 100 μ resolution will open the door to the investigation of the role of vascular damage in the aetiology and progression of the disease.3 ,4 The series of images in online supplementary figure S1 shows how the improved SNR at ultra-high-field can be exploited to improve the quality of images in a number of MRI techniques.
Magnetic susceptibility variation within the patient is normally viewed as a disadvantage, since it gives rise to signal loss in tissues that are near to cavities (eg, the oral cavity, auditory canal, orbital sinuses, etc). However, this same effect can also be exploited, particularly at high field where the effect is greater, to enhance the contrast between veins and brain parenchyma. Such high-field amplification of susceptibility contrast can also be used to detect iron deposition in the brains of MS patients. Susceptibility-weighted imaging (SWI) is a technique that could serve as a surrogate biomarker for revealing MS lesion characteristics to which other MRI techniques are insensitive.5 The cause of higher iron content in deep GM nuclei, especially in basal ganglia and thalamus, could be investigated by SWI to explore whether the cause is degenerative breakdown of the blood-brain barrier or hindered iron outflow.
In MRS, the resonance line of each metabolite occupies a specific location in the frequency spectrum. The ability to distinguish between these peaks depends on their spacing, or spectral dispersion, which is proportional to B0. This dependence of the CSD on B0 leads to enhanced capability of metabolite detection and simplification of MRS spectra at higher fields, as overlapping peaks are disentangled and specific metabolites peaks are not obscured by higher concentration metabolites, as shown in online supplementary figure S2.
What an MS pathologist would like to see
Histopathology offers a unique view of the topography of pathological processes in MS and represents, for many, the ultimate proof of the diagnosis. It has the ability to visualise in exquisite detail, snapshots of the pathogenesis of the disease. What it lacks, is the ability to visualise the evolution of the pathological processes that lead to the state of the brain and the spinal cord when it presents as material for autopsy or biopsy. Another significant shortcoming of MS histopathology is that it relies, in most cases, on either autopsies of long-standing advanced cases, or on the evaluation of unusual lesions that require biopsy. With the diminishing number of new autopsies in most centres, it is imperative to gain more in-depth knowledge by high-field imaging. Ultra-high-field MRI can also help to answer questions that pathology cannot. In postmortem imaging, as well as imaging in vivo, the resolution that can be achieved compares well with traditional histopathological studies (figure 1). We consider three areas of major importance where pathology needs the help of ultra-high-field MRI: prelesional changes in the WM; evolution of GM lesions and pathogenesis of neurodegeneration. There have been interesting, but limited, pathological and radiological observations regarding the first steps of WM demyelination. However, there is still uncertainty about whether inflammation is the primary cause of demyelination, and if there are WM changes that could be detected prior to the development of perivenous demyelination. The possibility that oligodendrocyte pathology precedes inflammation has eloquently been made.6
It is fair to say that we have limited knowledge with respect to cortical lesions. We know that they exist, but we cannot assume that WM and GM lesions have the same pathogenesis. The subpial lesions that account for the majority of GM demyelination are unlikely to be caused by T-cell migration through venules, the presumed early phase of WM lesions. Both the presence of an early cortical inflammatory component7 and for most, the role that meningeal inflammation8 plays, are still debated. GM lesions appear to be more frequent in the progressive stage of the disease, but they have been detected even in very early MS. Do they simply correlate with, or cause disease progression? Remyelination of WM lesions has been shown by both pathology and imaging, but the same is not true for GM lesions.9 If a major therapeutic avenue in MS is remyelination, and MRI measures of the disease in GM seem to be correlated more with disability, we need to visualise and quantify GM remyelination.
Neither pathologists nor imaging experts can diagnose the subtype of the disease from the material they examine. Essentially, we still do not have a marker or a threshold of progressive disease. In most cases this is easier done clinically. The current hypothesis of neurodegeneration being the cause of progressive disease has been strengthened by the correlation of axonal loss with clinical measures of disability but, importantly, the causes of neurodegeneration in MS have not yet been uncovered.
As will be discussed in the next sections of this review, most of these questions are still unanswered. We believe that the best way to address such issues is likely to rely on the use of ultra-high-field MRI to longitudinally assess patients at the very early stages of the disease (ie, when they present with clinically isolated syndromes) in order to track lesion formation and evolution over time, and to investigate the relationship of lesion accrual with MRI measures of neurodegeneration and with changes of the patient clinical status.
Imaging WM lesions
Ultra-high-field MRI allows better definition of lesions located in the WM and GM, their morphology and their association with the vasculature4 ,10–12 at a resolution which is similar to that of pathological assessment. Several studies have shown that 7 T and 8 T systems detect a higher number of lesions within the brain WM in patients with established MS in comparison with 1.5 and 3 T scanners.3 ,13 This suggests that abnormalities detected using quantitative MR techniques in the normal-appearing WM (NAWM) are, at least in part, due to the presence of focal lesions which go undetected when using low-field magnets. Whether the assessment of lesion burden and distribution using ultra-high-field MRI scanners assists in making an earlier diagnosis of patients presenting with a clinically isolated syndrome suggestive of MS has not yet been evaluated. However, thanks to the morphological detail that can be seen with these scanners, several studies have contributed to the identification of some interesting lesion characteristics, which can aid the differential diagnosis between MS and other neurological conditions that can mimic the disease. Specifically, due to the better definition of the relationship between demyelinating lesions and the deep venous system, several studies3 ,10 ,12 ,14 ,15 have shown that MS plaques form around the microvasculature. This feature can help to distinguish WM lesions in MS patients from incidental WM lesions,15 and has been reinforced by an initial investigation of blood-brain barrier abnormalities in MS at 7 T, which showed that the majority of enhancing lesions are perivenular, and that the smallest lesions have a concentric pattern of enhancement, suggesting that they grow outward from a central vein.16 The presence of a central small vein and a rim of hypointensity on 7 T T2*-weighted magnitude images can also assist in the differentiation of WM lesions found in MS patients from those of patients with neuromyelitis optica spectrum disorders17 or Susac syndrome.18 In this latter condition, T1-hypointense lesions within the central part of the corpus callosum (CC), that are not commonly seen in MS, have also been detected.18
The use of iron-sensitive MRI sequences has provided additional insights into MS lesion characteristics by showing, for example, the presence of a peripheral ring of iron deposition around some acute and chronic MS lesions.10 A correlative MR/pathology study has shown that iron deposition occurs predominantly within oligodendrocytes in the NAWM, while in active MS lesions iron accumulates inside macrophages at the periphery of the plaques. Perivascular iron deposits do also occur and are most likely a reflection of impairment of vascular permeability.19 Another combined MR/histology study has suggested that the pooled analysis of R2* and phase images may contribute to a better characterisation of the pathological features of MS lesions, since R2* abnormalities were found to correspond to severe loss of iron and myelin, whereas negative phase shift abnormalities were associated with focal iron accumulation.20 A recent longitudinal study has shown that ring-phase lesions remained unchanged over a 2.5-year period in five relapsing-remitting MS patients, thus challenging the notion that such lesions reflect the presence of acute activated iron-rich macrophages.21
Imaging GM lesions
Cortical lesions have been imaged with improved spatial resolution both ex vivo3 and in vivo14 ,22 using ultra-high-field MRI systems (≥7 T), despite the challenges presented by the B0 and radiofrequency field inhomogeneities and the potential for higher RF deposition compared with lower fields. The use of T2*-weighted imaging at 7 T also improves GM/WM contrast, allowing better definition of the lesion territory.23 ,24 Additionally, the greater spatial resolution helps to minimise the partial volume between parenchyma and adjacent cerebrospinal fluid. Despite many comparative studies using ultra-high-field MRI being somewhat biased by the fact that they have simultaneously assessed two different parameters (ie, pulse sequences and field strengths), they have certainly contributed to provide important descriptive pieces of information. Several studies have tried to optimise 7 T imaging in order to improve the detection and classification of cortical MS lesions and to develop a clinical acquisition protocol.22 ,25 ,26 Sinneker et al22 showed that cortical lesions are hypointense on 3D magnetisation-prepared rapid acquisition gradient-echo scans, while Kilsdonk et al25 found that 3D fast fluid attenuated inversion recovery sequences detect a higher total number of GM lesions than 3D double inversion recovery (DIR) sequences.
The advantages of 7 T and multichannel receive technology have enabled the identification of different cortical lesion types in a small MS population, based on visual inspection of focal cortical hyperintensities on T2*-weighted fast low-angle shot (FLASH) and T2-weighted turbo spin echo (TSE) images.14 The frequency with which different lesion locations were observed in the cortical ribbon, including subpial lesions, conformed to previous descriptions of the neuropathology.27 The number of subpial lesions correlated with clinical disease severity measures, suggesting that ultra-high-field MRI is potentially a sensitive and specific marker of cortical pathology in MS. Interestingly, T2*-weighted images were the most sensitive for detecting cortical MS lesions, compared with phase, T1 and T2-TSE-weighted images.14
Postmortem MR examinations of MS brains have demonstrated an excellent retrospective sensitivity of ex vivo focal cortical lesion detection using T2*-weighted imaging at 7 T, validating preliminary in vivo findings.24 In a study assessing the sensitivity of 3D T2*-weighted gradient-echo and 3D inversion recovery WM attenuated turbo-field-echo (TFE) sequences at 7 T in formalin-fixed MS brain for detecting cortical demyelination, prospectively, 46% of cortical lesions were detected on T2*-weighted scans, and 42% on TFE images. These counts improved to 93% and 82%, respectively, with retrospective scoring, after comparison with histological sections. This technique has been recently applied in vivo in eight MS patients and showed a high sensitivity in the detection of Type I cortical lesions.28
In vivo data in a heterogeneous cohort of MS patients showed that the use of an optimised FLASH T2*-weighted sequence at 7 T MRI reveals about five to seven times the number of in vivo cortical lesions than does DIR imaging at 3 T,29 which has so far been the best MR tool for identifying cortical lesions in patients with MS, although detection of subpial lesions is suboptimal with DIR. Neuropathology studies report that subpial lesions may extend across multiple adjacent gyri, a phenomenon termed ‘general subpial demyelination’.30 In vivo observations with 7 T MRI revealed that in some MS cases, FLASH T2*-weighted magnitude images show, in addition to focal subpial lesions, band-like areas of signal hyperintensity that involve the outer cortical laminae, and may extend over an entire gyrus or multiple gyri, resulting in extensive involvement of the cortex (see online supplementary figure S3).14
The in vivo quantification of general subpial demyelination in MS presents a technical challenge. Advances in the study of diffuse subpial pathology in vivo can be achieved by combining T2*-weighted acquisition at 7 T with a surface-based analysis of the cortex.31 This type of analysis is based on the parametric reconstruction of the folded cortex from high-resolution anatomical scans, to identify the pial and WM/GM boundaries that define the cortical ribbon. It is then possible to selectively sample signal and contrast at various depths from the pial surface, and measure changes in tissue across the cortical width, hemispheres, gyri and sulci. The use of this analysis, by selectively sampling 7 T T2*-weighted signal at 50% depth from pial surface, demonstrated, in patients with established and late MS, distributed subpial T2*-weighted signal increases across the whole cortical mantle,31 which may reflect the diffuse subpial pathology described in postmortem studies. While subpial T2*-weighted signal increases were disseminated throughout the cortex, the correlation between subpial T2* changes and WM lesions involved only a few cortical areas, suggesting that subpial pathology is largely independent of WM damage, as observed in ex vivo studies.32 The surface-based analysis technique can be combined with quantitative indices of cortical tissue changes, including magnetisation transfer imaging33 and T2* relaxation decay,34 to measure diffuse tissue abnormalities that are otherwise difficult to quantify using signal intensity measurements alone.
Ultra-high-field MRI of cortical MS plaques can potentially provide useful information on the biophysical properties of such lesions. Correlations between histopathology and MRI of postmortem MS brains have evidenced hypointense rings on cortical lesions identified on 7 T T2*-weighted magnitude images. These areas correlate pathologically with iron-laden microglia present at the edge of chronic active lesions.24 Similar rings have been reported in WM lesions in patients with MS using T2*-weighted magnitude imaging and, more prominently, as areas of increased susceptibility using phase reconstruction techniques.10
Initial findings in a small patient dataset revealed that leukocortical lesions constitute the greatest fraction of MS cortical plaques with evidence of increased susceptibility effects on phase images,14 likely reflecting a greater degree of inflammation of this type of cortical lesion relative to subpial and intracortical types. Phase imaging at 7 T thus potentially represents a highly sensitive method for staging MS lesions by inflammatory activity in vivo.
Quantitative and metabolic techniques
One of the most promising research applications at ultra-high-field is MRS of brain metabolites with low concentrations (1–5 mM) that make their detection very challenging at lower field strengths.35 Glutathione (GSH) is an indicator of oxidative status in the human brain. In vivo detection and quantification of GSH at 7 T has been performed using proton MR spectroscopic imaging with a spectral editing scheme called band selective inversion with gradient dephasing.36 The application of this MRS technique to MS patients has shown that cortical GM and WM lesions are characterised by a significant reduction of GSH concentration in comparison with healthy controls, hinting at the potential of GSH to probe brain oxidative status.36
Increasing field strength also improves imaging and MR spectroscopy of nuclei other than hydrogen, such as sodium (23Na) and phosphorus (31P) that have lower MR sensitivity.37 23Na yields the second strongest nuclear magnetic resonance (NMR) signal among biologically relevant NMR-active nuclei. In most biological tissues, Na is distributed in two compartments: extracellular (∼140 mmol/L) and intracellular ([Na]in) (∼15 mmol/L).38 The use of multiple-quantum filters (MQFs) is considered to be the best method for monitoring changes in intracellular sodium non-invasively in the human brain. A preliminary study of MS patients using a novel triple-quantum filtered 23Na MRI sequence at 7 T has shown an increase of whole brain intracellular Na concentration in MS patients when compared with healthy controls (see online supplementary figure S4).39 Recent studies have suggested that intra-axonal Na accumulation contributes to axonal degeneration by reversing the action of the sodium/calcium exchanger, and thus inducing a lethal rise in intra-axonal calcium concentration.38 Therefore, a non-invasive technique able to quantify intracellular sodium concentration in the brain may help in the understanding of mechanisms underlying axonal degeneration, and may provide a marker of cellular viability.
Compounds with a high magnetic susceptibility, such as those containing iron, increase the local magnetic field. This provides a contrast mechanism which is more pronounced at ultra-high-fields. Using the phase of a gradient-recalled echo image and a newly developed postprocessing technique, a recent MRI study enabled high-resolution quantitative imaging of local magnetic field shifts in patients with MS.10 The phase images showed an increased local field in the caudate, putamen and globus pallidus of patients relative to control subjects, with contrast in 74% of WM lesions, and distinct peripheral rings in the larger lesions. This is consistent with the results of postmortem histological studies of MS showing pathological iron accumulation in both deep GM and WM plaques.40 Increased magnetic susceptibility (reflecting increased iron concentration) has been recently found in the deep GM of patients at presentation with clinically isolated syndromes suggestive of MS.41 An in vivo contrast mechanism sensitive and specific to the presence of iron may help in understanding the role of iron in neurodegenerative pathology and in developing biomarkers for disease progression.
Ultra-high-field MRI is improving our understanding of MS pathogenesis. The main advantages demonstrated so far by the use of these scanners are: a better visualisation of WM lesions and of some of their particular morphological characteristics; a net gain in the capability to visualise GM lesions and their location; the quantification of ‘novel’ metabolites which may have a role in axonal injury; and greater sensitivity to iron accumulation (see table 1 for a summary).
However, at present, the application of these magnets in standard clinical practice is still some way off, while their role in the diagnostic work-up of patients at presentation with a clinically isolated syndrome, or in monitoring disease progression, or treatment response in patients with definite MS, needs to be established. Due to the high cost of installation and maintenance, ultra-high-field MRI scanners are still relatively rare (at present, less than 50 worldwide), and therefore, it is not likely to become the ‘gold-standard’ for MS imaging. However, the information that ultra-high-field MRI can offer should guide the development of combined imaging strategies at lower field, contribute to prove or disprove pathological hypotheses and influence the development of novel therapeutic strategies.
Several challenges remain. An increased MRI field strength results in an increased field inhomogeneity, which leads to different tissue contrasts in different regions of the brain. Clearly, this might pose difficulties in lesion identification, and in the application of segmentation algorithms, and metabolic analysis. Patient safety and comfort issues, such as radiofrequency energy deposition, compatibility with metallic implants, and sensory symptoms experienced during scanning, must also be considered. Development of new biomarkers of brain function could allow measurement of metabolic pathways governing excitatory and inhibitory neurotransmission. In this context, a new approach has been developed to investigate the diffusion properties of intracellular, cell-type-specific metabolites; this provided preliminary data showing that it is possible to distinguish axonopathy from other processes, such as inflammation, oedema, demyelination and gliosis. This approach is called diffusion tensor spectroscopy and combines features of DT imaging and MRS. In a cross-sectional pilot study,42 the diffusion of the NAA was measured in the human normal-appearing CC on a 7.0 T MRI scanner. In MS, the NAA diffusivity parallel to the axonal fibre direction was lower in comparison with healthy controls, and inversely correlated with both water parallel diffusivity and clinical severity. Improvements in spatial resolution might result in better measurements of GM/WM volume, regional volume, deep GM structures, infratentorial regions, cortical surface area, curvature and thickness. Since cortical thickness can reveal neurodevelopmental trajectory in early life, these studies may help to assess the role and time-course of demyelination and cortical column dysfunction.43 It remains still to be seen whether such improvements will impact on the spinal cord and optic nerve imaging. The advantages offered by ultra-high-field MRI scanners for functional imaging techniques (active fMRI tasks or resting state analysis) increase the likelihood of fMRI being used in the assessment of clinical disability and cognitive function, as well as for monitoring various non-pharmaceutical interventions, such as cognitive therapies and rehabilitation. Determination of the contribution that different cognitive systems make to preserving brain plasticity in patients would have a significant impact on the selection of a care strategy by the neurologist. Since in MS, adaptive plasticity is preserved even when there is a high lesion load, the investigation of clinical approaches to aid recovery is warranted even in patients with advanced disease. Another functional index that can be measured using MRI is perfusion, which could prove useful for monitoring cortical function in MS patients. A recent study has shown reduced cortical perfusion that may be correlated with chronic WM injury.44 Finally, diffusion tensor imaging is also likely to show gains at ultra-high-fields, where the higher spatial resolution may help to probe WM fibres at a smaller scale than is currently possible.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online figures
Contributors MF: writing and organising the manuscript; revising the article for important intellectual content; and final approval of the version to be published; study supervision and coordination. NE: summarising the data on pathology; revising the article for important intellectual content; and final approval of the version to be published. AK: summarising the data on technical issues; revising the article for important intellectual content; and final approval of the version to be published. MI: summarising the data on quantitative and metabolic techniques; revising the article for important intellectual content; and final approval of the version to be published. CM: summarising the data on cortical lesions; revising the article for important intellectual content; and final approval of the version to be published. MAH: interpretation of the data; revising the article for important intellectual content; and final approval of the version to be published. MAR: review concept and design; summarising the data on WM lesions; revising the article for important intellectual content; and final approval of the version to be published.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.