Chapter 8 Discussion
Spinal cord lesions in multiple sclerosis are common (occurring in ~80-90% of patients) and have a major contribution to disability accrual. These lesions can impact ambulation and cause motor, sensory deficits as well as bladder/bowel/sexual dysfunction. Yet, we know a lot less about the pathophysiology of spinal cord disease in MS than for the brain. Studies including longitudinal spinal cord follow-up are scarce and radiological spinal cord outcomes are rarely used as an outcome in drug trials. As a result, knowledge is limited regarding silent spinal cord disease activity and the effect of therapy in inhibiting disease activity in the cord. In clinical practice, this raises a few important questions for the clinician:
- What is the best strategy to follow-up the spinal cord with imaging for my patient?
- How can I best treat patients with more spinal cord involvement?
To make a step in the direction of ultimately being able to answer these clinical questions, in this thesis, we explored the following themes:
- Factors driving differences between brain and cord involvement in MS (Chapter 2)
- Imaging tools to further aid exploration of cord lesions in vivo, in particular ultra-high field MRI. (Chapter 2 & 3)
- The effect of disease-modifying therapies in preventing cord lesions. (Chapter 4 & 5)
- Biomarkers that can support stratification of patients at higher risk of more spinal cord involvement. (Chapter 6)
These themes will be further discussed based on several questions:
8.1 How is MS pathology in the spinal cord different to the brain?
That a difference exists between brain and spinal cord as CNS regions in MS pathophysiology starts with a few observations from clinical practice. For example, there are MS patients who at presentation have no cord lesions at all but extensive brain lesions and others with much more cord than brain lesions. Disease progression can occur in one CNS site without progression in the other, and the degree of new disease activity can be completely different between the CNS sites. This is also highlighted in studies, where spinal cord and brain abnormalities correlate poorly, with regard to both lesion formation and atrophy [55, 56]. Also, in our local cohort (see chapter 4), there was no brain radiological disease activity in 37% of cases where the follow-up spinal cord MRIs showed new cord lesions. Furthermore, it is interesting that the spinal cord is considerably more affected in primary-progressive compared to the relapsing-remitting MS phenotype (McDonald criteria 2010 & 2017). In pathological studies, spinal cord lesions are considerably less often mixed active/inactive (smouldering lesions) than brain lesions and more often inactive [80–82]. Also, some studies show considerably less remyelinated plaques in the spinal cord compared to the brain [80]. Lastly, as we have seen in chapter 5, there seems to be a difference in effectiveness of higher-efficacy disease-modifying treatment in preventing lesions in the spinal cord compared to the brain (more about this later).
To try explain these observations we explored the literature (see chapter 2) for possible important factors that distinguish the cord and brain. There are the obvious anatomical differences that the cord has to the brain, in volume (~20ml vs. 1500ml), white/grey matter organization (white matter at the subpial surface vs. the grey matter) and vascularization. Then the blood-brain and blood-spinal cord barrier each have their own morphology and features, with the interesting observation that there is regional variation in permeability, also for cytokines [69]. This might have a role in the regional differences in disease activity but may also be relevant for DMT drug distribution [72].
Data on regional immunological differences between spinal cord and brain are limited. A recent transcriptomic study [303], using human CNS tissue samples from MS patients and controls, found strong glial subtype diversity within the different CNS regions (cerebrum, cerebellum, spinal cord) when evaluated in controls. In MS patients, however, the pattern of transcriptomic changes are shared across CNS regions and converge to specific pathways, foremost those regulating cellular stress and immune activation. When it comes to the adaptive immune system, it is suggested that the localization of lesions within the brain vs. spinal cord is associated with different effector T cell responses to myelin proteins [75]. Also, immunogenetic factors seem to play a role in regional variation; HLA-DRB1*1501 is associated with more extensive spinal cord pathology [77–79]. HLA-DRB1 encodes for the beta chain of the major histocompatibility complex class II, whose ligands are the CD4 receptors of helper T
Collectively, this makes it arguable that the spinal cord is a distinct entity from the brain in MS pathophysiology. Combined with the possible regional differences in treatment response, described in chapter 5, this suggests that findings from MS drug trials on the impact on progression in the brain cannot be simply extrapolated to the spinal cord.
What do we know from other autoinflammatory diseases where the spinal cord is involved?
In neuromyelitis optica spectrum disorder (NMOSD) auto-antibodies target aquaporin-4 (AQP4) on astrocytes, where lesion mainly occur in the spinal cord and optic nerves. Here, a theory on what explains the distribution of lesions is that this is due to that AQP4 is expressed in astrocyte processes in regions of the CNS without a blood-brain barrier, including the pre‑laminar optic nerve head, the circumventricular organs (midline structures around the third and fourth ventricles) and the root entry zones in the spinal cord [304, 305]. In systemic lupus erythematosus (SLE), the CNS can be involved in up to 45% patients [306]. Here, two processes seem to be at play resulting in the diffuse character of nervous system involvement: One ischemic in nature due to the vasculitis/vasculopathy component of the disease and another autoimmune component mediated by antibodies crossing dysfunctioning blood-brain barrier (possibly as a result of the vasculitis and vasculopathy) [306]. For myelin oligodendrocyte glycoprotein antibody disease (MOGAD) it is unclear why a predilection for the spinal cord and optic nerve exists and this relates to MOG expression level in the different CNS regions [307]. Also, in the case of neurosarcoïdosis the mechanisms of cord involvement are unknown.
8.2 What tools can aid us in further exploring these differences?
Pathological studies can provide detailed insight into MS pathophysiological processes within the CNS. However, post-mortem samples are inherently biased, as they mostly originate from patients with a long disease duration and in the progressive phase. Moreover, post-mortem studies provide only a snapshot in time and give limited longitudinal information. Thus, in chapter 2 and 3 we explored what tools can aid further the investigation into differences in cord and brain MS pathology in vivo, in particular if ultra-high field MRI will prove to be useful for the spinal cord.
Over the years, different advanced MRI techniques have contributed to better understanding spinal cord pathology in MS, especially outside focal lesions. Some examples are: (i) Magnetization transfer imaging, which measures the ratio of free protons and protons bound to macromolecules, the ratio decreases with demyelination. (ii) Myelin water imaging that quantifies the myelin water pool from multicomponent T2 relaxometry and has a high correlation with myelin density in histopathological studies [113]. (iii) Diffusion-tensor imaging measuring directional water diffusion which changes with demyelination [308]. These techniques helped in showing that there is diffuse demyelination outside focal lesions in the normal-appearing spinal cord [99, 101].
Additionally, techniques exist that are able to image blood-brain barrier integrity (e.g. dynamic contrast-enhanced MRI) which have been used in different MS studies [126, 309, 310] and perhaps it is also feasible to use these techniques for investigating the blood-spinal cord barrier in MS in vivo [241, 311]. It would be interesting to see if these techniques can show regional differences in permeability in the CNS in humans. Finally, positron-emission tomography (PET) could also be a potential tool to collect insights into regional immunological differences in vivo, though challenging to apply for the spinal cord. For example, using radioligands specifically targeted at microglia and macrophages, this taught us for the brain that in MS activated macrophages/microglia are present in the normal-appearing white matter, correlating with disability accumulation over time [120, 121].
Aside from advanced techniques, there are also challenges and opportunities for conventional MRI. The cords small diameter and the distortions from physiological motion (swallowing, CSF pulsation, respiratory and cardiac action) still makes imaging of the cord artifact-prone and time-consuming. Improvements in coils, sequence design and motion correction techniques are already paying off. However, while for brain imaging, increasing MRI field strength lead to considerable improvements in sensitivity and characterization of MS lesions in the brain, going from 1.5T to 3T for the spinal cord imaging in MS did not result in any gain [267]. In case of the spinal cord, the increase in signal that comes with the higher field strength is offset by the higher sensitivity for magnetic field inhomogeneities and motion, which are a larger problem in the cord region than for the brain. Could then going beyond that, beyond 3T, to ultra-high field strengths have any benefit for spinal cord imaging?
In chapter 3 a literature review was conducted to evaluate what kind of research was performed in the MS field, adapting ultra-high field (UHF) imaging for the cord. Here, we saw that there were research groups applying UHF spinal cord imaging successfully. For example, Ouellette and colleagues [55] elegantly employed the capabilities of 7T cervical cord MRI in MS patients to investigate the lesion distribution of spinal white and grey matter lesions. They acquired axial and sagittal T2* sequences of the whole cervical spine at a high in-plane resolution of 0.4 x 0.4 mm and 3 mm slice thickness within an acquisition time of less than 5 minutes. This beautifully demonstrates the potential of UHF imaging for the cord, and that with the current state of improvements in coil and sequence design, it is feasible at 7T to achieve whole cervical spine coverage at high resolutions within reasonable scan time. But with all the technical challenges, costs and expertise necessary to adapt spinal UHF MRI (discussed in chapter 3), for the time being it remains a tool for research and will not yet benefit MS imaging in daily clinical practice in the coming years.
Since the literature review, only very few new studies have been published adapting ultra-high field MRI for the spinal cord. This shows that the aforementioned challenges (and costs) cause that, even for research, it is currently only used to a limited extent. However, there is one recent study applying 7T susceptibility-weighted imaging (imaging focused on enhancing conspicuity of microvasculature and (para)magnetic substances) on MS spinal cords, showing for the first time that paramagnetic rims (MRI marker for chronically active lesions) can also occur in spinal cord lesions and also confirming the presence of central veins in a proportion of spinal cord MS lesions [65, 312].
In conclusion, albeit with technical challenges, advanced imaging techniques, improvements in conventional MRI and UHF imaging provide opportunities in growing our understanding of differences between the spinal- and brain disease process in MS, how this is affected by treatment and how repair mechanisms vary between CNS regions.
8.3 Are there regional differences in treatment effect within the CNS in MS?
If regional differences between brain and spinal cord exist in how DMTs inhibit disease activity, this has potential implications in how we personalize treatment and follow-up in MS patients. Do the therapies that we have work as well for inhibiting MS disease activity in the cord as is the case for the brain? Or are we potentially undertreating patients with a higher degree of spinal cord involvement? Ultimately, the question is whether patients with different topographical distributions of disease activity warrant a different treatment approach.
Evidence on the effectiveness of treatment in preventing cord lesions remains limited; the only randomized controlled trial explicitly including new spinal cord lesions as an outcome measure remains the APREGGIO trial. This trial investigated laquinimod in PPMS (negative trial, also on spinal cord endpoints) [131]. There are observational studies that investigated ‘no evidence of disease activity’ outcomes for different DMTs that included spinal cord imaging at baseline and follow-up, but these studies did not separately report to what degree there were new spinal cord lesions [154, 218, 220, 222]. Therefore, for the current DMTs used in clinical practice we know almost nothing on this subject from prospective studies.
In the local retrospective study from chapter 4, using a matching approach, we aimed to investigate whether intermediate- (iDMT) and high-efficacy DMTs (hDMT) have a protective effect with regard to spinal cord inflammatory activity. The main finding was that patients on i/hDMTs had a reduced risk (HR 0.29, p=0.01) of new cord lesions at follow-up, compared to a group that received lDMTs and/or no treatment for ≥90% of the follow-up period. This suggests that i/hDMTs also exert their effect in the spinal cord region. But with the data from that cohort it was not possible to demonstrate whether i/hDMT treatment also resulted in a larger risk reduction for new cord lesions when compared to low-efficacy DMTs (lDMT; so when patients receiving no DMT were excluded). To address this question we selected a subpopulation from the international MSBase observational dataset that started a DMT, having at least a spinal cord MRI around treatment initiation and one at follow-up (minimum interval 6 months). Here matched groups that received purely hDMTs or lDMTs for ≥90% of the follow-up period were compared on spinal cord lesions at follow-up, but also brain lesions and relapses (see chapter 5). The latter two outcomes were included as validation, as for those we already know that hDMTs are more effective from randomized controlled trials. For spinal cord lesions there was absolutely no difference between the hDMT and lDMT group (HR 0.99, p=0.97), in contrast to the other outcomes; brain lesions (HR 0.22, p<0.001) and relapses (HR 0.45, p=0.004) that occurred significantly less in the hDMT-group. This suggests that a potential discrepancy exists in the effect of hDMTs over lDMTs in preventing spinal cord lesions versus brain lesions and relapses.
But at present, we cannot state with full confidence that this discrepancy exists, due to the limitations of these two retrospective studies. The most important one being the potential selection bias toward patients with more cord involvement due to the selection of patients with spinal cord MRI imaging at follow-up, which in general is not part of the routine follow-up regimen. Despite the limitations, to date, it is best evidence we have on the effect of DMTs with regard to spinal cord lesions.
In both studies, we conducted an exploratory analysis to see whether there were groups on specific individual DMTs with notably few or many spinal cord lesions at follow-up. In the local study, none of the patients on anti-CD20 B cell depleting therapies (ocrelizumab and ofatumumab) and alemtuzumab (which depletes CD52-positive B and T cells) developed new cord lesions (see figure 4.3). Although, this was only in a small group of 34 patients using hDMT ≥90% of follow-up, of which 6 had cord lesions at follow-up. This finding was not replicated in the study in the MSBase dataset. Here, if we look at all therapies used during follow-up, by patients that had spinal cord MRI follow-up (see supplement 5 of chapter 5), the distribution of used therapies in the group with (n=442) compared to without (n=1219) new cord lesions is not considerably different (this was also the case for the matched dataset, see figure 5.2). Based on follow-up data from these two studies, there is currently no convincing evidence that maybe a specific DMT or DMT-group is better in preventing specifically spinal cord lesions.
So to answer the question heading this subchapter: There seems to be a positive effect of DMT in general in inhibiting spinal cord disease activity, however, based on current evidence the benefit of higher efficacy agents is possibly limited or more modest for the spinal cord compared to the brain. This is an interesting and important finding, but needs cautious interpretation because of the study limitations. It should be considered a plea to work towards more and a higher level of evidence to either confirm or reject this finding.
8.4 How can we best follow-up the spinal cord in MS patients?
Central to the question how follow-up of the spinal cord is best organised in MS patients is another question: What are the chances you miss disease activity at follow-up if you do not routinely scan the spinal cord and only scan if there are cord-related symptoms? In other words: How often do asymptomatic lesions in the spinal cord occur while there isn’t any disease activity in the brain?
Figures in literature on isolated asymptomatic cord lesions range from 2% to 16% and are summarized in this paragraph: In an observational cohort followed from first clinical attack (CIS suggestive of multiple sclerosis), the spinal cord MRI showed one or more asymptomatic lesions in 58 out of 178 (35%) patients [235]. When looking over time, a 7-year follow-up study in a general MS cohort on varying DMTs (n=219, spinal cord data available in 74%, mean disease duration at first visit 6.6 years), 8% to 12% each year had disease activity solely based on new spinal cord lesions on MRI [34]. In a smaller study (n=115; mean baseline disease duration 7.3 years) on a first-line DMTs this figure was 8% [291]. Another retrospective study (413 screened, 103 included, median disease duration at baseline 5.0 years, median cord MRI follow-up 17 months) showed during follow-up that 25.2% had asymptomatic cord lesions and 9.8% had asymptomatic cord lesions with no concomitant brain activity [32]. A comparable study (n=168, mean disease duration at baseline 9.8 years) reported asymptomatic cord lesions occurring in 15% of a clinically stable MS cohort over a median period of 14 months [313]. Granella and colleagues retrospectively looked at all spinal cord MRI scans with radiological activity in their local cohort (340 scans from 230 patients; baseline median disease duration 5.2 years) and found that 31.2% of cord lesions were asymptomatic and 12.1% had neither clinical activity nor brain radiological activity [33]. A study including all patients from a single center cohort, who had at minimum another spinal cord MRI with contrast sometime at follow-up (830 out of 1335 patients, median follow-up 7 years; median baseline disease duration 12 years) showed that 16.1% of patients had asymptomatic isolated spinal cord activity (here defined as new contrast enhancing lesions) [314]. Finally, Lim et al only found 2% of patients in a clinically stable cohort to have isolated spinal cord activity (median follow-up 14 months; disease duration at baseline not mentioned) [315].
Apart from the study by Lim et al, the mentioned studies all concluded (in different degrees) that there is a place for routine spinal cord monitoring at follow-up [33, 34, 313, 314], but “In what group?”, “For how long?”, “Should it include only cervical coverage or also thoracic?”, “Only sagittal or also axial sequences?” and whether it is cost-efficient are questions that remain.
To that end, in the MSpine study (currently recruiting; see chapter 7) the aim is to answer a part of these questions by following a cohort of prospectively included early MS patients initiating their first DMT with a yearly spinal cord MRI (whole cord, including axial coverage) added to the routine clinical care during 3 years. Based on the data from the local cohort as well as the MSBase study, most new spinal cord lesions in patients initiating a DMT seem to occur in the first 2-3 years. Furthermore, the number of spinal cord lesions during these first 3 years is also prognostic for future disability accrual [235]. Therefore, in clinical practice, it seems reasonable to implement yearly routine spinal cord follow-up in the first three years. However, more evidence is needed to further substantiate this recommendation, to which we hope that the MSpine can contribute.
As an initial step towards identifying patient subgroups at higher risk of more cord involvement, in chapter 6, we retrospectively investigated in our local cohort whether intrathecal immunoglobulin production was associated with more cord pathology at follow-up, since a recent study showed a cross-sectional relationship between mainly IgM intrathecal synthesis and spinal cord lesions [127]. While in our study there also was a cross-sectional relation between IgM intrathecal synthesis and cord lesions, intrathecal synthesis did not predict future cord lesions and thus did the study not support a role for intrathecal immunoglobulin production to be used to stratify patients that need more routine follow-up imaging of the spinal cord. We will see if this finding will be reproduced or rejected when prospectively evaluated in the MSpine study, where it will be more broadly investigated which patient groups are predisposed (based on demographic, blood immunological and CSF markers) to developing new spinal cord lesions during follow-up in early disease.
8.5 What are future directions regarding spinal cord disease in MS?
8.5.1 Pathophysiological differences between brain and cord involvement in MS
By better understanding the differences in regional pathophysiological processes, we can probably in turn better grasp how therapies affect this process and explain why there is the possible discrepancy between the effect of hDMTs on inhibiting lesion formation in the brain compared to the cord. There is still a lot unknown about what determines the degree of spinal cord involvement in MS and progress to be made in how we best image cord pathology, especially beyond focal lesions.
Further study into what degree genetics influence the involvement of each CNS region could shed some light on pathways involved in the heterogeneity of lesion distribution. Additionally, it would be interesting to compare characteristics of cord versus brain lesions in different stages of the lesions using advanced imaging techniques like for example: diffusion-based MRI techniques, myelin-water imaging, spectroscopy MR, TSPO PET, ultra-high field (susceptibility-weighted) MRI. These techniques have vastly improved over recent years providing the opportunity to adapt them to the spinal cord.
8.5.2 Spinal cord follow-up imaging
As discussed under How can we best follow-up the spinal cord in MS patients?, to develop a strategy for efficient spinal cord MRI follow-up in MS patients, future research needs to address the following subjects: (i) the incidence of asymptomatic cord lesions independent of brain disease activity, (ii) which subgroups of patients are at higher risk of more spinal cord involvement and (iii) protocol/sequence development for follow-up spinal cord MRI. Points (i) and (ii) we aim to address in the MSpine study. Regarding subject (iii) the MSpine study can hopefully give some insight to whether only sagittal or sagittal plus axial imaging is best for follow-up, as well as cervical cord only versus whole spinal cord. But future studies specifically dedicated to developing and testing MRI protocols for time/cost-efficient spinal cord imaging at follow-up will be necessary to eventually come to an optimal scanning strategy.
8.5.3 Effect of DMTs on spinal cord lesion formation
Evaluating the effect of an individual DMT compared to others on a specific outcome like spinal cord lesions is challenging; there are numerous different DMTs being used in clinical practice and there is the large variation in how MS manifests in every patient.
Ideally, a large group of patients from multiple centers starting a DMT is followed prospectively with routine spinal cord imaging for at least 2-3 years (preferably longer). Then, with a matching procedure (or marginal structural models), DMTs grouped by their mechanism of action can be compared against each other (e.g. S1P inhibitors [fingolimod, siponimod, ozanimod, ponesimod], B-cell depleting agents [ocrelizumab, ofatumumab, rituximab], broad lymphocyte depleting agents [alemtuzumab, cladribine], etc.). However, as conducting such a study would be costly and we already know that these therapies have a positive effect on clinical outcomes, it is not a very attractive study to set up. As an alternative, hopefully, in the future, there will be better registry data including routine cord imaging data from patients on different DMTs, as spinal cord imaging protocols becoming increasingly time-efficient and are applied at routine imaging more often than in the past. Furthermore, results from this thesis should provide an argument for future trials of new DMTs to also include spinal cord outcome measures.
Finally, the MSpine study (chapter 7) aims at including 155 patients, but as comparison of DMTs on spinal cord outcomes is not the goal of that study, there will probably be insufficient power to be able to compare treatment groups. However, maybe this data can be pooled with other prospective cohorts that include cord imaging, to possibly identify treatment groups more efficacious in inhibiting spinal cord lesion formation.
8.6 Summary
In this thesis we started with exploring factors driving differences between brain and cord involvement (theme 1) in MS by reviewing literature (chapter 2) on how these CNS regions are different and how MS pathology varies in these regions on (advanced) imaging and in pathological investigations. This includes differences anatomically (volume, white/grey matter configuration), in vascularization and in blood-brain and blood-spinal cord barrier function. When comparing lesions in the brain and cord in pathological studies, brain lesions are more often smouldering and a larger proportion of cord lesions are inactive [80–82]. Immunologically, spinal cord–predominant MS is associated with different effector T cell responses to myelin antigens compared to patients with brain-predominant MS [75] and the genetic variant HLA-DRB1*1501 is associated with more extensive spinal cord pathology [77–79].
To come to know more about cord lesions and the differences to brain lesions in vivo, we are mostly dependent on imaging tools (theme 2). In chapter 2 different studies were discussed applying techniques like magnetization transfer imaging, myelin water imaging, diffusion-tensor imaging, MR spectrography, PET and ultra-high field MRI that gave us new fragments on how spinal cord pathology manifests in MS. For the latter, ultra-high field MRI, we reviewed in more depth (chapter 3) what the benefits and challenges are for spinal cord imaging in MS. Benefits include increased signal- and contrast-to-noise, enabling imaging at higher spatial resolutions, which can improve MS lesion sensitivity in both the spinal white matter as well as grey matter. Additionally, this can aid imaging of microstructural abnormalities (also outside lesions) in the spinal cord in MS using advanced MRI techniques like functional imaging, MR spectroscopy and diffusion-based techniques. Technical challenges include increased magnetic field inhomogeneities and distortions from physiological motion. Due to the size, diameter and the presence of various sources of physiological motion in the spinal cord area. This makes ultra-high field imaging for the cord more difficult than for the brain. With these technical challenges, costs and expertise necessary to adapt spinal UHF MRI, for the time being it remains a tool for research and will not yet benefit MS imaging in daily clinical practice in the coming years.
With the differences between brain and spinal cord, a large part of this thesis was dedicated to exploring whether this also results in differences in the inhibitory effect of disease-modifying treatments on lesion formation in these CNS regions (theme 3). In the local retrospective study (chapter 4), we found that patients on i/hDMTs had a reduced risk of new cord lesions at follow-up, compared to a group that received lDMTs and/or no treatment, suggesting that i/hDMTs also exert their effect in the spinal cord region. With the limited sample size, it was not possible to demonstrate whether i/hDMT treatment also resulted in a larger reduction compared to low-efficacy DMTs (i.e., excluding patients [partially] receiving no DMT from the reference group). With data from patients with spinal cord follow-up from the international MSBase observational database (chapter 5), matched groups that received purely hDMTs or lDMTs for ≥90% of the follow-up period were compared on spinal cord lesions, brain lesions and relapses at follow-up. With regard to new spinal cord lesions at follow-up there was no difference between the hDMT and lDMT group, while brain lesions and relapses occurred significantly less in the hDMT-group. This raises the possibility of a discrepancy in the effect of hDMTs over lDMTs in preventing spinal cord lesions versus brain lesions and relapses.
Finally, it remains uncertain what an appropriate strategy is for follow-up imaging of the spinal cord - as there is a lack of data on the incidence of asymptomatic spinal cord lesions and factors that are prognostic for more cord involvement in the disease course (theme 4). We investigated whether intrathecal immunoglobulin could predict more cord involvement at follow-up (chapter 6). Especially, intrathecal IgM production was of interest as an earlier study showed a cross-sectional relationship between cord lesions and intrathecal IgM [127]. In our study we also found that cross-sectional relation, but intrathecal synthesis of IgM or IgG did not predict future cord lesions and thus did the study not support a role for intrathecal immunoglobulin production to be used in stratifying patients that need more routine follow-up imaging of the spinal cord. Currently, a multicenter prospective study is currently ongoing to investigate the incidence of asymptomatic cord lesions (i.e., lesions you would miss without routine imaging) and markers predicting more spinal cord disease in early MS (chapter 7). The goal is to provide us with knowledge to achieve personalized efficient spinal cord imaging follow-up.
In conclusion, the main findings from this thesis are: (i) variation in brain and spinal cord involvement are probably driven by a combination of anatomical, immunological, genetic and blood-brain/spinal cord barrier differences, (ii) ultra-high field MRI for spinal imaging in MS with its technical challenges, costs and expertise necessary, currently remains a tool for research and is not yet useful for MS imaging for clinical purposes, (iii) a possible discrepancy exists in the effect of high-efficacy DMTs over low-efficacy DMTs in preventing spinal cord lesions compared to brain lesions and relapses, and (iv) intrathecal synthesis of immunoglobulin M and G does not predict future cord lesions and is thus not useful for stratifying patients for routine spinal cord follow-up imaging.