Abstract

Neuropathological studies in multiple sclerosis have suggested that meningeal inflammation in the brain may be linked to disease progression. Inflammation in the spinal cord meninges has been associated with axonal loss, a pathological substrate for disability. Quantitative magnetic resonance imaging facilitates the investigation of spinal cord microstructure by approximating histopathological changes. We acquired structural and quantitative imaging of the cervical spinal cord from which we calculated magnetization transfer ratio in the outer spinal cord—an area corresponding to the expected location of the pia mater and subpial region—and in spinal cord white and grey matter. We studied 26 healthy controls, 22 people with a clinically isolated syndrome, 29 with relapsing-remitting, 28 with secondary-progressive and 28 with primary-progressive multiple sclerosis. Magnetization transfer ratio values in the outermost region of the spinal cord were higher than the white matter in controls and patients: controls (51.35 ± 1.29 versus 49.87 ± 1.45, P < 0.01), clinically isolated syndrome (50.46 ± 1.39 versus 49.13 ± 1.19, P < 0.01), relapsing-remitting (48.86 ± 2.89 versus 47.44 ± 2.70, P < 0.01), secondary-progressive (46.33 ± 2.84 versus 44.75 ± 3.10, P < 0.01) and primary-progressive multiple sclerosis (46.99 ± 3.78 versus 45.62 ± 3.40, P < 0.01). In linear regression models controlling for cord area and age, higher outer spinal cord magnetization transfer ratio values were seen in controls than all patient groups: clinically isolated syndrome (coefficient = −0.32, P = 0.03), relapsing-remitting (coefficient = −0.48, P < 0.01), secondary-progressive (coefficient = −0.51, P < 0.01) and primary-progressive multiple sclerosis (coefficient = −0.38, P < 0.01). In a regression analysis correcting for age and cord area, magnetization transfer ratio values in the outer cord were lower in relapsing-remitting multiple sclerosis compared with clinically isolated syndrome (coefficient = −0.28, P = 0.02), and both primary and secondary-progressive compared to relapsing-remitting multiple sclerosis (coefficients = −0.29 and −0.24, respectively, P = 0.02 for both). In the clinically isolated syndrome and relapsing-remitting multiple sclerosis groups, outer cord magnetization transfer ratio was decreased in the absence of significant cord atrophy. In a multivariate regression analysis an independent association was seen between outer cord magnetization transfer ratio and cord atrophy (coefficient = 0.40, P < 0.01). Our in vivo imaging observations suggest that abnormalities in a region involving the pia mater and subpial cord occur early in the course of multiple sclerosis and are more marked in those with a progressive course.

Introduction

Multiple sclerosis is an inflammatory disorder of the CNS, and is a common cause of disability among young adults (Confavreux and Vukusic, 2006). Although the exact causative mechanisms are unclear, it has been suggested from the findings in neuropathological studies that meningeal inflammation may play a role in the development of earlier disability in progressive forms of the disease (Magliozzi et al., 2007; Howell et al., 2011; Choi et al., 2012). It has also been reported that B cell follicles (as well as T cells) are present in the meninges of post-mortem brain specimens (Serafini et al., 2004; Lovato et al., 2011). However, Frischer et al. (2009) reported that B cells may only be found in the progressive forms of the disease. Neuropathology studies have also suggested that cortical demyelination may be a characteristic finding in multiple sclerosis (Dawson, 1916; Brownwell and Hughes, 1962). Another post-mortem study has suggested that cortical demyelination is more commonly found in progressive multiple sclerosis and inflammation is more diffuse, when compared directly to the relapsing-remitting form of the disease (Kutzelnigg et al., 2005).

Taking into account the above observations, a hypothesis has arisen that the pathological processes of meningeal inflammation and cortical demyelination may be interconnected and in turn play a role in the development of a progressive disease course. Magliozzi et al. (2010) reported the co-localization of areas of meningeal inflammation (composed of B cells and dendritic cells) and cortical demyelination. Furthermore, extensive subpial demyelination has been recorded in the cortex in multiple sclerosis (et al., 2003; et al., 2007). It has been proposed that cortical demyelination (especially subpial) may occur in the presence of meningeal inflammation due to the release of inflammatory cytokines in the subarachnoid space (Brown and Sawchenko, 2007; Ransohoff, 2009). In progressive forms of multiple sclerosis, where there seems to be long standing inflammation, this may create a milieu, which favours further retention of inflammatory cells (Krumbholz et al., 2005; Meinl et al., 2008). This retention of inflammatory cells in multiple sclerosis may result in the activation of microglia and subsequently demyelination (Magliozzi et al., 2010).

It has been reported that in relapsing-remitting multiple sclerosis, meningeal inflammation may be present, with inflammation involving T cells rather than B cells (Frischer et al., 2009). As neuropathology studies are predominantly composed of cases of progressive multiple sclerosis, an important insight was gained into the process of meningeal inflammation in a study examining biopsies of patients with early relapsing-remitting multiple sclerosis or a clinically isolated syndrome (Lucchinetti et al., 2011). In this paper it was found that subpial cortical demyelination may be associated with meningeal inflammation.

Imaging studies have also provided a means of studying patients when they present with a clinically isolated syndrome suggestive of multiple sclerosis, which is the earliest clinical manifestation of the condition in most cases. The use of a double inversion recovery sequence has demonstrated the presence of cortical lesions in patients with clinically isolated syndrome (Calabrese et al., 2007). Furthermore, a decrease in magnetization transfer ratio (MTR) of the cortex, which might reflect demyelination (Chen et al., 2013), has also been observed in subjects with clinically isolated syndrome (Audoin et al., 2005; Fernando et al., 2005). These imaging observations also suggest that cortical demyelination occurs early in the disease course.

The meninges of the spinal cord have also been reported to be inflamed in multiple sclerosis (Androdias et al., 2010; DeLuca et al., 2013), with the inflammatory cells predominantly composed of T cells in one study (Androdias et al., 2010). Meningeal inflammation in the spinal cord has been related to axonal loss (Androdias et al., 2010; DeLuca et al., 2013). As axonal loss has been hypothesized to be the pathological substrate for disability (Ganter et al., 1999; Evangelou et al., 2000; Lovas et al., 2000; Schirmer et al., 2011), and because much of the physical disability from multiple sclerosis arises from spinal cord involvement, a greater understanding of meningeal abnormalities in the spinal cord, and their relationship to disease course, would seem to be relevant.

However, to date the effect of multiple sclerosis on the spinal cord meninges has not been investigated in vivo, due to image resolution constraints and other technical challenges associated with MRI protocol optimization in the spinal cord (Dietrich et al., 2008; Stroman et al., 2014). MTR imaging in the spinal cord has been used for some time (Mezzapesa et al., 2004; Charil et al., 2006). The MTR of a tissue is related to its macromolecular structure and an important contribution in CNS tissue comes from myelin (Dousset et al., 1992). Recent MTR imaging studies in the spinal cord have provided insights into the mechanisms of disability in multiple sclerosis (Zackowski et al., 2009; Oh et al., 2013). With the implementation of higher resolution structural and MTR acquisitions in the spinal cord than those previously acquired (Yiannakas et al., 2012), the potential to investigate the effects of multiple sclerosis on the outer region of the spinal cord, that would be expected to include the pia mater of the meninges, could provide further insights into the pathophysiology of multiple sclerosis.

The aims of this paper are threefold:

  • To characterize the outermost region of the spinal cord, which is expected to include contributions from the pia mater and subpial region of the spinal cord, using high in-plane resolution, magnetization transfer-weighted images to measure the MTR in healthy control subejcts and in patients with multiple sclerosis or a clinically isolated syndrome;

  • To compare the outer spinal cord MTR measures of people with a clinically isolated syndrome or multiple sclerosis with those of healthy controls; and

  • To compare outer spinal cord MTR findings seen in different clinical subgroups: clinically isolated syndrome, relapsing-remitting, primary- and secondary-progressive multiple sclerosis, and explore the relationship of outer cord MTR with measures of both spinal cord atrophy and physical disability.

Materials and methods

Subjects

We recruited subjects with no prior neurological diseases (n = 26) and patients with either clinically isolated syndrome (n = 22) or multiple sclerosis: relapsing-remitting (n = 29), secondary-progressive (n = 28) and primary-progressive (n = 28). Multiple sclerosis was diagnosed using the 2010 McDonald criteria (Polman et al., 2011). The clinically isolated syndrome cohort was recruited following a single clinical episode consistent with demyelination and at least one lesion on a T2-weighted axial brain scan. The multiple sclerosis subgroups were classified using the Lublin-Reingold criteria (1996). Informed written consent was obtained from each participant before inclusion in the study.

All patients with clinically isolated syndrome or multiple sclerosis had Expanded Disability Status Scale (EDSS; Kurtzke, 1983) determined by a neurostatus certified assessor as well as multiple sclerosis functional composite (Fischer et al., 1999). Subsequently we calculated Z-scores for the 25-foot Timed Walk Test, 9-Hole Peg Test and 3 s Paced Auditory Serial Addition Test B (PASAT). Z-scores were calculated from normative values displayed in the National Multiple Sclerosis Society Task Force database (Fischer et al., 1999). We also recorded American Spinal Injury Association (ASIA) motor (m) and sensory (s) scores (Maynard et al., 1997) for all subjects with multiple sclerosis and clinically isolated syndrome. Assessment of physical function was performed immediately before the MRI.

None of the subjects had experienced a relapse or received a course of corticosteroids within 1 month before imaging.

Magnetic resonance imaging protocol

Subjects were scanned at 3 T using a Philips Achieva MRI system with radiofrequency (rf) multi-transmit technology (Philips Healthcare). A 16-channel receive-only neurovascular coil was used for spinal cord scanning and brain scanning was performed using the product 32-channel receive-only coil. A polystyrene-filled vacuum fixation bag was placed behind the neck of all participants to provide a head rest and neck restraint and to minimize motion artefacts. The following sequences were acquired:

  • For tissue-specific spinal cord segmentation into outer cord, grey matter and white matter, a 3D fat-suppressed fast field echo (3D-FFE) sequence was used with repetition time = 23 ms; echo time = 5 ms; flip angle = 7°; field of view = 240 × 180 mm2; voxel size = 0.5 × 0.5 × 5 mm3; 10 axial contiguous slices centred at the level of the C2–3 intervertebral disc;

  • MTR data were obtained by acquiring a 3D slab selective spoiled gradient echo sequence with two echoes using repetition time = 36 ms, echo time 1/echo time 2 = 3.5/5.9 ms, flip angle 9°, with and without Sinc-Gaussian shaped magnetization transfer saturating pulses of nominal α = 360°, offset frequency 1 kHz, duration 16 ms applied before the excitation pulse. Twenty-two 5-mm slices were acquired in an axial orientation, with field of view = 180 × 240 mm2 and acquisition matrix = 240 × 320 (voxel size 0.75 × 0.75 mm2, reconstructed to 0.5 × 0.5 mm2), with SENSE factor 2 in the foot/head direction and two number of excitations. The prescribed imaging volume for the MTR sequence was centred at the same level and geometrically matched to the volumetric FFE sequence;

  • For cord cross-sectional area measurements i.e. measurement of atrophy, a 3D-phase sensitive inversion recovery (3D-PSIR) sequence was acquired in the axial plane in the cervical cord centred at C2/3 intervertebral disc, with a voxel size of 0.5 × 0.5 × 3 mm3, repetition time = 8 ms; echo time = 3.7 ms; flip angle α = 5°; field of view = 256 × 256 mm; number of excitations = 1;

  • For brain lesion volume measurements, we acquired axial proton density/T2 images using a 2D turbo spin echo sequence (TSE) with 3-mm slice thickness; the following parameters were used: repetition time = 3500 ms; echo time = 19/85 ms; matrix 240 × 240 mm2; in plane resolution = 1 × 1 mm2; and

  • For brain volume measurements i.e. atrophy measurements, a 3D magnetization-prepared turbo field echo (3D-TFE) sequence was used with slice thickness = 1 mm; repetition time = 6.8 ms; echo time = 3.1 ms; matrix = 256 × 256 mm2; in plane resolution 1 × 1 mm2.

Image analysis

Magnetization transfer values

An inclusion criterion for all MRI images was the absence of motion artefacts. Using JIM 6.0 (Xinapse systems, http://www.xinapse.com) we extracted the central 10 slices of the MTR volume and the magnetization transfer-off and -on images (Fig. 1A and B) were registered independently to the volumetric 3D-FFE using linear registration in FSL (http://www.fmrib.ox.ac.uk/fsl/). Subsequently we created an MTR-map in the 3D-FFE space (Fig. 1C and D) (Yiannakas et al., 2012). All MTR maps were reviewed again by the same reader to confirm that no artefacts were introduced during the registration step.

Figure 1

(A) Axial magnetization transfer-off image (resolution 0.5 × 0.5 × 5mm3) and (B) axial magnetization transfer-on image (resolution 0.5 × 0.5 × 5mm3) before registration to the 3D-FFE image to create the MTR map. (C) Axial 3D-FFE image (resolution 0.5 × 0.5 × 5 mm3) through the C2–3 intervertebral disc. (D) Following independent linear registration of the magnetization transfer-on and magnetization transfer-off to the FFE the MTR map is created in this space (resolution 0.5 × 0.5 × 5 mm3). (E) Axial 3D-FFE image demonstrating the spinal cord outline created using the active surface model. (F) Axial 3D-FFE demonstrating grey matter region of interest outlined using the fuzzy connector.

Figure 1

(A) Axial magnetization transfer-off image (resolution 0.5 × 0.5 × 5mm3) and (B) axial magnetization transfer-on image (resolution 0.5 × 0.5 × 5mm3) before registration to the 3D-FFE image to create the MTR map. (C) Axial 3D-FFE image (resolution 0.5 × 0.5 × 5 mm3) through the C2–3 intervertebral disc. (D) Following independent linear registration of the magnetization transfer-on and magnetization transfer-off to the FFE the MTR map is created in this space (resolution 0.5 × 0.5 × 5 mm3). (E) Axial 3D-FFE image demonstrating the spinal cord outline created using the active surface model. (F) Axial 3D-FFE demonstrating grey matter region of interest outlined using the fuzzy connector.

We extracted three slices centred at the C2–3 intervertebral disc from the volumetric 3D-FFE image and then created two regions of interest: (i) spinal cord outline using an active surface model (Horsfield et al., 2010); and (ii) spinal cord grey matter using a semi-automated fuzzy connector method (Udupa and Samarasekera, 1996) (Fig. 1E and F). Using these two regions of interest we created a mask of the cord outline and grey matter from the MTR map. We then converted both masks to binary format using FSL tools (http://www.fmrib.ox.ac.uk/fsl/). This process enabled us to subtract the binary grey matter mask from the binary cord outline resulting in two masks: (i) spinal cord white matter MTR mask; and (ii) spinal cord grey matter MTR mask. We recorded the mean value of the spinal cord grey matter MTR mask for each participant.

We then further analysed the white matter MTR mask by eroding the image using iterations based on 4-connected neighbours written in Matlab 2012a (Mathworks). We discarded the outermost row of pixels of the image to avoid contamination by CSF. As the image was acquired at 0.75 × 0.75 mm2 resolution and subsequently reconstructed to 0.5 × 0.5 mm2, we considered the next voxel layer to be also contaminated by CSF due to the voxels’ interpolation and consequently discarded this voxel too. Therefore, the next voxel layer was designated as being the outermost voxel layer of the spinal cord and we recorded the mean MTR values of the voxels in this layer across the three selected slices for each subject. The classification of peripheral voxels in the image is illustrated in Fig. 2 and on MRI in Fig. 3; we considered this layer likely to include both pia mater and subpial spinal cord tissue. Following removal of the three outermost voxels we recorded the mean value from the spinal cord white matter-MTR mask for each subject and designated this as the white matter-MTR value. Finally, we recorded the mean grey matter MTR value across the three slices.

Figure 2

Graphical representation of voxel layers analysed in the outermost region of the spinal cord.

Figure 2

Graphical representation of voxel layers analysed in the outermost region of the spinal cord.

Figure 3

(A) MTR-map (resolution 0.5 × 0.5 × 5 mm3) demonstrating a square-shaped region of interest from which B was obtained in the periphery of the spinal cord. (B) Zoomed image of the periphery of the spinal cord demonstrating from right to left: CSF; (P) outer exclusion voxel layer–partial volume with CSF; (1) inner exclusion layer–interpolation of voxels; (2) outermost spinal cord voxel layer; and (WM) spinal cord white matter.

Figure 3

(A) MTR-map (resolution 0.5 × 0.5 × 5 mm3) demonstrating a square-shaped region of interest from which B was obtained in the periphery of the spinal cord. (B) Zoomed image of the periphery of the spinal cord demonstrating from right to left: CSF; (P) outer exclusion voxel layer–partial volume with CSF; (1) inner exclusion layer–interpolation of voxels; (2) outermost spinal cord voxel layer; and (WM) spinal cord white matter.

Spinal cord area

We measured spinal cord mean cross-sectional area from the 3D-PSIR image using the active surface model (Horsfield et al., 2010). To achieve this we extracted five 3-mm thick slices centred at C2–C3 and then recorded the area of each of the five slices (Kearney et al., 2014). Spinal cord area was normalized by the number of slices used i.e. we calculated the mean area of the five slices for each participant. This method of normalization was chosen as a previously published report has demonstrated that normalization of spinal cord volume by slice number is superior to the use of brain volume or other measures of body habitus (Healy et al., 2012). We present the spinal cord measurements as area normalized by slice number rather than as volume, as cord area is more commonly reported in clinical trials of multiple sclerosis (Kalkers et al., 2002; Leary et al., 2003; Lin et al., 2003; Montalban et al., 2009).

Brain parenchymal fraction and T2 lesion volume

Lesions on each subject’s 3D T1-weighted volume were marked and filled with values consistent with normal-appearing white matter signal intensity to prevent misclassification of tissue during segmentation (Chard et al., 2010). Segmentation of the lesion-filled image was then performed using SPM8 (statistical parametric mapping; Wellcome Trust centre for NeuroImaging, UCL Institute of Neurology, London). The brain parenchymal fraction (the sum of white and grey matter relative to total intracranial volume) was then recorded for each subject.

We outlined T2 hyperintense lesions in all participants (clinically isolated syndrome or multiple sclerosis) using the semi-automated edge finding tool in JIM 6 on the axially acquired T2-weighted images. We then recorded the volume of T2-weighted lesions in millilitres.

Statistical analysis

SPSS 21 (IBM) was used for statistical analysis.

Comparisons of magnetic resonance imaging measures in controls and patient subgroups

In healthy controls we evaluated the differences in mean MTR values between: (i) outer spinal cord and spinal cord white matter; (ii) outer spinal cord and spinal cord grey matter; and (iii) spinal cord white matter with grey matter, using a paired samples t-test. Subsequently we used the same tests in patients with clinically isolated syndrome and each subgroup of multiple sclerosis separately. We also tested differences in cord area and brain parenchymal fraction between patients with multiple sclerosis and clinically isolated syndrome and control subjects using an independent samples t-test.

To explore the hypothesis that MTR abnormalities in each tissue component of the spinal cord (outer spinal cord, white matter and grey matter) may differ between clinically isolated syndrome or each subtype of multiple sclerosis patients and controls, we constructed a linear regression model, with the case (i.e. multiple sclerosis subgroup, clinically isolated syndrome and control) set as the dependent variable and MTR added as an independent variable. To compare each patient group (clinically isolated syndrome or multiple sclerosis subgroup) with controls, or with each other, a separate regression model was constructed for each comparison of MTR values (outer spinal cord, white matter and grey matter) being performed. To investigate whether differences seen in MTR values between subject groups (dependent variable) were affected by cord atrophy in the multiple sclerosis subgroups, we included spinal cord area as an independent variable in every model and also adjusted for age and gender by adding these as independent variables. Adjustments for multiple comparisons were not performed due to the exploratory nature of this study (Bender and Lange, 2001).

Relationship between cord magnetization transfer ratio measures and cord area

To investigate the relationship between spinal cord MTR and cord area we firstly calculated univariate Pearson’s correlation coefficient of the MTR of the grey matter, white matter and outer spinal cord versus cord area in all multiple sclerosis patients combined. To identify the components of the spinal cord MTR (i.e. outer spinal cord, grey and white matter) which were associated with atrophy, independently from the others and from age and gender, a multivariate linear regression model was constructed. In this model, cord area was set as the dependent variable, and MTR values from each region of the spinal cord that had a significant univariate correlation with cord area were added as independent variables, in a forward stepwise manner, to determine those with independent associations with cord area. Independent variables retained in the final model with a P-value of <0.05 were considered to be independently associated with cord area (dependent variable).

Relationship between magnetic resonance imaging and disability measures

To explore the relationship of each MRI measure analysed (spinal cord area, spinal cord) MTR (outer cord, white matter and grey matter), brain parenchymal fraction and brain lesion load with disability, we firstly calculated univariate correlations between these variables, in all multiple sclerosis patients combined. We used Spearman’s rank correlation coefficient for the EDSS, as this scale is logarithmic. For all other disability scales (multiple sclerosis functional composite Z-scores, ASIAs and ASIAm) we calculated univariate correlations with the MRI parameters using Pearson’s correlation coefficient. For univariate associations, P < 0.01 was considered significant.

To investigate the relationship between the MRI parameters analysed and physical disability, we sought to determine independent associations between these two variables. To achieve this, we constructed a multivariate linear regression model with the disability measure of interest set as the dependent variable. A separate model was constructed for each disability measure used i.e. multiple comparisons were not performed. To refine the choice of independent variables added to the model, only MRI parameters that had a significant univariate correlation with the dependent variable (i.e. the disability measure of interest), were included in the regression analysis. In each model constructed, age and gender were also added as independent variables, to correct for any influence these may have on the association being tested. The MRI parameters were added to the model, as independent variables, in a forward stepwise manner and those variables retained in the final model, with a P-value of <0.05, were deemed to be independently associated with the disability scale (dependent variable).

Results

Subjects and their clinical and conventional magnetic resonance imaging characteristics

We studied a total of 133 subjects: 26 healthy controls, 22 subjects with clinically isolated syndrome, 29 with relapsing-remitting multiple sclerosis, 28 with secondary-progressive multiple sclerosis and 28 with primary-progressive multiple sclerosis. Characteristics of the group are shown in Table 1. In the clinically isolated syndrome cohort there were 18 cases of unilateral optic neuritis, one multifocal clinically isolated syndrome (consisting of optic neuritis, facial numbness and diplopia), one hemispheric presentation (unilateral hand weakness due to a lesion in the right motor cortex) and two cases of partial myelitis (one cervical and one thoracic).

Table 1

Demographics and conventional MRI parameters in healthy controls, and patients with clinically isolated syndrome and with multiple sclerosis

 Controls n = 26 Clinically isolated syndrome n = 22 Relapsing-remitting multiple sclerosis n = 29 Secondary-progressive multiple sclerosis n = 28 Primary-progressive multiple sclerosis n = 28 
Gender female:male 18:8 12:10 20:9 17:11 12:16 
Age (years) 40.6 ± 10.2 36.2 ± 9.3 38.1 ± 9.5 51.3 ± 9.4 50.5 ± 9.9 
Mean disease duration (years)  0.5 ± 0.4 6.1 ± 4.0 20.11 ± 11.89 10.9 ± 7.6 
Median EDSS (range)  1 (0–3) 2.5 (0–7) 6.5 (4–8.5) 6.0 (2–8) 
Brain parenchymal fraction 0.824 ± 0.015 0.822 ± 0.013 0.811 ± 0.017* 0.788 ± 0.022* 0.798 ± 0.014* 
T2 lesion volume (ml)  2.9 ± 3.8 13.1 ± 14.3 23.4 ± 17.0 16.4 ± 20.2 
Spinal cord cross-sectional area (mm280.3 ± 7.7 82.6 ± 7.3 78.1 ± 9.0 63.6 ± 9.8* 68.1 ± 9.7* 
 Controls n = 26 Clinically isolated syndrome n = 22 Relapsing-remitting multiple sclerosis n = 29 Secondary-progressive multiple sclerosis n = 28 Primary-progressive multiple sclerosis n = 28 
Gender female:male 18:8 12:10 20:9 17:11 12:16 
Age (years) 40.6 ± 10.2 36.2 ± 9.3 38.1 ± 9.5 51.3 ± 9.4 50.5 ± 9.9 
Mean disease duration (years)  0.5 ± 0.4 6.1 ± 4.0 20.11 ± 11.89 10.9 ± 7.6 
Median EDSS (range)  1 (0–3) 2.5 (0–7) 6.5 (4–8.5) 6.0 (2–8) 
Brain parenchymal fraction 0.824 ± 0.015 0.822 ± 0.013 0.811 ± 0.017* 0.788 ± 0.022* 0.798 ± 0.014* 
T2 lesion volume (ml)  2.9 ± 3.8 13.1 ± 14.3 23.4 ± 17.0 16.4 ± 20.2 
Spinal cord cross-sectional area (mm280.3 ± 7.7 82.6 ± 7.3 78.1 ± 9.0 63.6 ± 9.8* 68.1 ± 9.7* 

*Significant differences between MRI parameters from controls (i.e. P < 0.01).

With regard to the conventional MRI scans, in the clinically isolated syndrome cohort 15 cases fulfilled criteria for dissemination in space (Polman et al., 2011); the criteria for dissemination in time could not be evaluated on this single scan as a gadolinium-contrast agent was not administered. Asymptomatic T2-weighted brain lesions were seen in 18 of 22 cases and in one case the symptomatic lesion was seen (in the motor cortex). There were eight spinal cord lesions seen on cord PSIR and FFE scans, seven asymptomatic and one due to cervical cord partial myelitis.

Comparison of cord area and brain volume between groups

There were no significant differences in the cord area of healthy controls with either the clinically isolated syndrome (P = 0.06) or relapsing-remitting multiple sclerosis (P = 0.18) groups. The secondary-progressive (63.5 ± 10.0 mm2 versus 80.2 ± 6.8 mm2, P < 0.01) and primary-progressive multiple sclerosis (68.1 ± 9.7 mm2, P < 0.01) groups both had lower cord areas than healthy controls. Results for all groups are presented in Table 1.

No significant difference was seen in brain parenchymal fraction in the clinically isolated syndrome group compared to controls, but significant differences in brain parenchymal fraction were seen between controls and all subgroups of multiple sclerosis (P < 0.01), with smaller brain parenchymal fractions in the multiple sclerosis subgroups.

Comparison of outer spinal cord and spinal cord white matter magnetization transfer ratio values within each subject group

In each subject group, the MTR of the outer spinal cord was higher than the MTR of spinal cord white matter: controls (51.35 ± 1.29 versus 49.87 ± 1.45, P < 0.01), clinically isolated syndrome (50.46 ± 1.39 versus 49.13 ± 1.19, P < 0.01), relapsing-remitting multiple sclerosis (48.86 ± 2.89 versus 47.44 ± 2.70, P < 0.01), secondary-progressive multiple sclerosis (46.33 ± 2.84 versus 44.75 ± 3.10, P < 0.01), and primary-progressive multiple sclerosis (46.99 ± 3.78 versus 45.62 ± 3.40, P < 0.01) (Table 2).

Table 2

Mean MTR values (±SD) of spinal cord region subtypes (outer cord, white matter, grey matter) in the control group and in each patient group

 Controls CIS CIS (excluding two cases of myelitis) RRMS SPMS PPMS 
Outer spinal cord MTR 
    Comparison between outer spinal cord 51.35 ± 1.29 50.46 ± 1.39 49.9 ± 1.40 48.86 ± 2.89 46.33 ± 2.84 46.99 ± 3.78 
    and white matter MTR values P < 0.01 P < 0.01 P < 0.01 P < 0.01 P < 0.01 P < 0.01 
MTR spinal cord white matter 
    Comparison between spinal cord white 49.87 ± 1.45 49.13 ± 1.19 49.21 ± 1.31 47.44 ± 2.70 44.75 ± 3.10 45.62 ± 3.4 
    and grey matter MTR values P < 0.01 P < 0.01 P < 0.01 P < 0.01 p = 0.02 P < 0.01 
MTR spinal cord grey matter 48.23 ± 1.76 47.72 ± 1.23 47.8 ± 1.30 46.6 ± 2.43 43.88 ± 2.62 44.88 ± 3.09 
 Controls CIS CIS (excluding two cases of myelitis) RRMS SPMS PPMS 
Outer spinal cord MTR 
    Comparison between outer spinal cord 51.35 ± 1.29 50.46 ± 1.39 49.9 ± 1.40 48.86 ± 2.89 46.33 ± 2.84 46.99 ± 3.78 
    and white matter MTR values P < 0.01 P < 0.01 P < 0.01 P < 0.01 P < 0.01 P < 0.01 
MTR spinal cord white matter 
    Comparison between spinal cord white 49.87 ± 1.45 49.13 ± 1.19 49.21 ± 1.31 47.44 ± 2.70 44.75 ± 3.10 45.62 ± 3.4 
    and grey matter MTR values P < 0.01 P < 0.01 P < 0.01 P < 0.01 p = 0.02 P < 0.01 
MTR spinal cord grey matter 48.23 ± 1.76 47.72 ± 1.23 47.8 ± 1.30 46.6 ± 2.43 43.88 ± 2.62 44.88 ± 3.09 

CIS = clinically isolated syndrome; RRMS = relapsing-remitting multiple sclerosis; SPMS = secondary-progressive multiple sclerosis; PPMS = primary-progressive multiple sclerosis. The clinically isolated syndrome cohort is presented with and without the two cases of myelitis included.

Comparison of spinal cord white matter and grey matter magnetization transfer ratio values within each subject group

In each subject group, the spinal cord white matter had higher MTR values than spinal cord grey matter (Table 2).

Outer spinal cord magnetization transfer ratio in controls versus patient subgroups

The outer spinal cord MTR values were higher in controls than all patient groups: controls versus clinically isolated syndrome [coefficient = −0.32, P = 0.03, 95% confidence interval (CI) −0.22 to −0.01], controls versus relapsing-remitting multiple sclerosis (coefficient = −0.48, P < 0.01, 95% CI −0.28 to 0.09), controls versus secondary-progressive multiple sclerosis (coefficient = −0.51, P < 0.01, 95% CI −0.31 to −0.15), controls versus primary-progressive multiple sclerosis (coefficient = −0.38, P < 0.01, 95% CI −0.36 to −0.06). Cord area was not a significant covariate in any of the models used (Table 3).

Table 3

Comparison of MTR values between control groups and in each patient group.

MTR value Comparison groups SE Coefficient Significance (P-value) 95% CI (lower, upper) 
Outer spinal cord MTR CIS/control 0.05 −0.32 0.03 −0.22, −0.01 
RRMS/control 0.05 −0.48 <0.01 −0.28, −0.09 
SPMS control 0.04 −0.51 <0.01 −0.31, −0.15 
PPMS/control 0.07 −0.38 <0.01 −0.36, −0.06 
CIS RRMS 0.02 −0.28 0.02 −0.10, −0.01 
RRMS/SPMS 0.02 −0.24 0.02 −0.07, −0.01 
RRMS/PPMS 0.04 −0.29 0.02 −0.16, −0.01 
SPMS PPMS 0.02 −0.02 0.89 −0.05, 0.04 
MTR white matter CIS /control 0.39 −0.34 0.03 −1.69, −1.05 
RRMS/control 0.30 −0.56 <0.01 −1.95, −0.76 
SPMS/control 0.23 −0.98 <0.01 −2.74, −1.80 
PPMS /control 0.22 −0.80 <0.01 −1.79, −0.92 
MTR grey matter CIS /control 0.42 −0.17 0.26 −1.3, 0.37 
RRMS /control 0.29 −0.36 <0.01 −1.38, −0.21 
SPMS /control 0.20 −0.74 <0.01 −1.84, −1.05 
PPMS /control 0.19 −0.75 <0.01 −1.50, −0.73 
MTR value Comparison groups SE Coefficient Significance (P-value) 95% CI (lower, upper) 
Outer spinal cord MTR CIS/control 0.05 −0.32 0.03 −0.22, −0.01 
RRMS/control 0.05 −0.48 <0.01 −0.28, −0.09 
SPMS control 0.04 −0.51 <0.01 −0.31, −0.15 
PPMS/control 0.07 −0.38 <0.01 −0.36, −0.06 
CIS RRMS 0.02 −0.28 0.02 −0.10, −0.01 
RRMS/SPMS 0.02 −0.24 0.02 −0.07, −0.01 
RRMS/PPMS 0.04 −0.29 0.02 −0.16, −0.01 
SPMS PPMS 0.02 −0.02 0.89 −0.05, 0.04 
MTR white matter CIS /control 0.39 −0.34 0.03 −1.69, −1.05 
RRMS/control 0.30 −0.56 <0.01 −1.95, −0.76 
SPMS/control 0.23 −0.98 <0.01 −2.74, −1.80 
PPMS /control 0.22 −0.80 <0.01 −1.79, −0.92 
MTR grey matter CIS /control 0.42 −0.17 0.26 −1.3, 0.37 
RRMS /control 0.29 −0.36 <0.01 −1.38, −0.21 
SPMS /control 0.20 −0.74 <0.01 −1.84, −1.05 
PPMS /control 0.19 −0.75 <0.01 −1.50, −0.73 

CIS = clinically isolated syndrome; RRMS = relapsing-remitting multiple sclerosis; SPMS = secondary-progressive multiple sclerosis; PPMS = primary-progressive multiple sclerosis.

A linear regression model was used adjusted for age, gender and cord area (SE = standard error).

Spinal cord white matter and grey matter magnetization transfer ratio in controls versus patient subgroups

The MTR of spinal cord white matter was higher in controls than in all patient groups: controls versus CIS (P = 0.03), controls versus multiple sclerosis (P < 0.01). The MTR of spinal cord grey matter was not different between controls and the group with clinically isolated syndrome but was significantly higher in controls than in all three multiple sclerosis subgroups (P < 0.01) (Table 3).

Outer spinal cord magnetization transfer ratio: comparison between patient subgroups

The outer spinal cord MTR was lower in relapsing-remitting multiple sclerosis than clinically isolated syndrome (coefficient = −0.28, P = 0.02, 95% CI −0.10 to −0.01). Both secondary-progressive (coefficient −0.24, P = 0.02, 95% CI −0.07 to −0.01) and primary-progressive multiple sclerosis (coefficient = −0.29, P = 0.02, 95% CI −0.16 to −0.01) had lower outer spinal cord MTR values compared to relapsing-remitting multiple sclerosis. No significant difference was found between secondary- and primary-progressive multiple sclerosis (Table 3).

Univariate correlations and association between magnetization transfer ratio measures and cord area in all multiple sclerosis patients combined

The following MRI parameters were correlated with cord area: outer spinal cord MTR (r = 0.39, P < 0.01), MTR white matter (r = 0.36, P < 0.01), MTR grey matter (r = 0.36, P < 0.01) and brain parenchymal fraction (r = 0.27 P = 0.01). However, outer spinal cord MTR (coefficient = 0.40, P < 0.01, 95% CI 0.63 to 1.88) alone was found to be independently associated with cord area; no independent associations were seen between grey or white matter MTR and cord area.

Univariate correlations between magnetic resonance imaging parameters and disability in all multiple sclerosis patients combined

Significant univariate correlations between MRI parameters and disability are presented in Table 4. Although significant correlations were seen between outer spinal cord MTR and each disability measure used (apart from PASAT z-score), stronger correlations were seen between cord area and disability measures reflecting motor function. Brain T2 lesion volume was not found to be correlated with any disability measures except for PASAT z-score (Table 4).

Table 4

R-values for disability measures versus MRI parameter.

MRI parameter Disability measure
 
 EDSS* ASIAm ASIAs Z-score PASAT Z-score 9-Hole Peg Test Z-score 25-foot Timed Walk Test Z-score multiple sclerosis functional composite 
Outer spinal cord MTR −0.41 0.38 0.30 0.03 0.41 0.36 0.42 
MTR spinal cord white matter −0.32 0.23 0.30 0.01 0.42 0.40 0.43 
MTR spinal cord grey matter −0.34 0.27 0.29 0.05 0.38 0.37 0.40 
Spinal cord area −0.60 0.49 0.39 0.14 0.48 0.38 0.44 
Brain parenchymal fraction −0.40 0.22 0.06 0.35 0.48 0.29 0.40 
T2 lesion volume 0.19 −0.04 0.17 −0.36 −0.29 −0.08 −0.29 
MRI parameter Disability measure
 
 EDSS* ASIAm ASIAs Z-score PASAT Z-score 9-Hole Peg Test Z-score 25-foot Timed Walk Test Z-score multiple sclerosis functional composite 
Outer spinal cord MTR −0.41 0.38 0.30 0.03 0.41 0.36 0.42 
MTR spinal cord white matter −0.32 0.23 0.30 0.01 0.42 0.40 0.43 
MTR spinal cord grey matter −0.34 0.27 0.29 0.05 0.38 0.37 0.40 
Spinal cord area −0.60 0.49 0.39 0.14 0.48 0.38 0.44 
Brain parenchymal fraction −0.40 0.22 0.06 0.35 0.48 0.29 0.40 
T2 lesion volume 0.19 −0.04 0.17 −0.36 −0.29 −0.08 −0.29 

Significant (P < 0.01) univariate correlations between MRI parameters and disability measure in all patients with multiple sclerosis combined are indicated in bold font (*Spearman’s coefficient used for EDSS and Pearson’s coefficient for all other disability measures).

Independent associations between magnetic resonance imaging parameters and disability in all multiple sclerosis patients combined

Significant independent associations between MRI variables and disability measures are summarized in Table 5. The 9-Hole Peg Test z-score was associated with MTR in the outer spinal cord. Tests of motor function (EDSS and ASIAm) were associated with cord area. The 25-foot Timed Walk Test z-score was associated with spinal cord white matter MTR in multiple sclerosis. The PASAT z-score was associated with both brain T2 lesion volume and brain parenchymal fraction.

Table 5

Summary of MRI parameters significantly associated with disability measures from linear regression models using disability as dependent variable

Disability Measure MRI parameter associated with disability SE Coefficient Significance (P-value) 95% CI (lower, upper) 
EDSS Spinal cord area 0.02 −0.48 <0.01 −0.13, −0.06 
Brain parenchymal fraction 9.45 −0.29 <0.01 −48.78, −11.17 
ASIAm Spinal cord area 0.13 0.42 <0.01 0.32, 0.85 
ASIAs Spinal cord area 0.08 0.26 0.03 0.02, 0.34 
Z-score PASAT Brain parenchymal fraction 8.42 0.25 0.02 2.82, 36.32 
T2 lesion volume 0.01 −0.26 0.02 −0.04, −0.004 
Z-score 9-Hole Peg Test Outer spinal cord MTR 0.04 0.27 <0.01 0.04, 0.19 
Brain parenchymal fraction 0.41 0.41 <0.01 16.44, 14.99 
Z-score 25-foot Timed Walk Test MTR white matter 0.15 0.36 <0.01 0.30, 0.89 
Z-score multiple sclerosis functional composite MTR white matter 0.07 0.30 <0.01 0.07, 0.36 
Spinal cord area 0.023 0.22 0.04 0.001, 0.09 
Disability Measure MRI parameter associated with disability SE Coefficient Significance (P-value) 95% CI (lower, upper) 
EDSS Spinal cord area 0.02 −0.48 <0.01 −0.13, −0.06 
Brain parenchymal fraction 9.45 −0.29 <0.01 −48.78, −11.17 
ASIAm Spinal cord area 0.13 0.42 <0.01 0.32, 0.85 
ASIAs Spinal cord area 0.08 0.26 0.03 0.02, 0.34 
Z-score PASAT Brain parenchymal fraction 8.42 0.25 0.02 2.82, 36.32 
T2 lesion volume 0.01 −0.26 0.02 −0.04, −0.004 
Z-score 9-Hole Peg Test Outer spinal cord MTR 0.04 0.27 <0.01 0.04, 0.19 
Brain parenchymal fraction 0.41 0.41 <0.01 16.44, 14.99 
Z-score 25-foot Timed Walk Test MTR white matter 0.15 0.36 <0.01 0.30, 0.89 
Z-score multiple sclerosis functional composite MTR white matter 0.07 0.30 <0.01 0.07, 0.36 
Spinal cord area 0.023 0.22 0.04 0.001, 0.09 

Quality control

Our study procedures aimed to minimize the risk of imaging artefacts compromising the study findings. These included a neck immobilization procedure to prevent motion artefact, as already described in the ‘Materials and methods’ section. Further information is provided in online Supplementary material of how effects of potential partial volume, motion and other magnetic resonance-related artefacts were mitigated (see Supplementary Figs 1–6 and Supplementary Tables 1a–c and 2a and b).

Discussion

There are a number of novel findings in this study. First, we measured in vivo a quantitative imaging metric (MTR) in the outer spinal cord—a region likely to include the pia mater and subpial spinal cord. Second, in healthy control subjects, the outer cord MTR was higher than cord white matter MTR. Third, there was a reduction in outer cord MTR at an early stage of multiple sclerosis, and in the absence of significant cord atrophy, as reflected by the findings in clinically isolated syndrome and relapsing-remitting multiple sclerosis subgroups. Fourth, outer cord MTR abnormality was significantly and independently correlated with cord atrophy and was also greater in progressive multiple sclerosis subgroups when compared with relapsing-remitting multiple sclerosis. Finally, there were several independent associations between spinal cord MRI metrics and disability measures. These findings are discussed in turn.

The outer spinal cord magnetization transfer ratio measure and what it reflects anatomically

We quantified MTR in a voxel layer in the expected location of the pia mater and subpial region of the spinal cord. As the spinal cord is surrounded by CSF the outer voxels of the cord are susceptible to partial volume effects (Tench et al., 2005). However, by excluding the most peripheral voxel layers we are confident that this influence was negligible as the outer cord MTR values were higher than the remaining (deeper) spinal cord tissues, which will not be affected by CSF.

The pia mater is composed of two layers in the spinal cord; the intima pia and epipial layer, of which the epipial layer (not present over the cerebral hemispheres) is the thicker of the two due to its composition of connective tissue (Millen and Woollam, 1961). The combined thickness of these layers was 0.2 mm in a post-mortem study of the thoracic spinal cord with an increase in higher segments of the cord (Reina et al., 2004). The cord samples in this study were fixed with formaldehyde, and a reduction in tissue size of up to 19% can occur using a 10% formalin solution (Mouritzen Dam, 1979). It therefore seems likely that a significant proportion of the outer cord voxel layer of the upper cervical cord region that we studied (which has an in-plane resolution of 0.5 mm) will contain pial meningeal tissue, although it is also likely to include subpial spinal cord white matter.

Higher outer than white matter spinal cord magnetization transfer ratio in healthy control subjects

A question arising is why the MTR in the outer cord region should be higher than that of normally myelinated cord white matter. One possible explanation is that there is a tissue component other than myelin in the outer cord that has a relatively high MTR. In this regard, both the pial and subpial tissues in the spinal cord contain collagenous fibres (Reina et al., 2004) and in vitro data obtained in phantoms have shown that higher MTR is correlated with higher collagen concentration (Laurent et al., 2001). It seems therefore possible that collagen contributes to the higher outer cord MTR.

Outer spinal cord magnetization transfer ratio abnormalities in clinically isolated syndrome and relapsing-remitting multiple sclerosis in the absence of cord atrophy

The finding of reduced outer cord MTR without atrophy in the clinically isolated syndrome and relapsing-remitting multiple sclerosis groups supports the robustness of the outer cord measure: had partial volume effects of CSF caused a decrease in outer cord MTR, an abnormality would not be seen when there was no difference in cord area between patient and control groups.

In the clinically isolated syndrome cohort, 15 fulfilled the criteria for dissemination in space and 18 of 22 participants had at least one asymptomatic T2-weighted brain lesion evident on their brain MRI scan and are therefore at higher risk of conversion to multiple sclerosis (Fisniku et al., 2008). Seven of this group had asymptomatic spinal cord lesions, the presence of which also increases the risk for conversion to multiple sclerosis (Sombekke et al., 2013). The abnormalities detected in the outermost spinal cord in clinically isolated syndrome compared to control subjects indicates that changes occur at a very early stage in relapse-onset multiple sclerosis.

The relapsing-remitting multiple sclerosis cohort were also at a relatively early stage of disease, had little disability and no cord atrophy. As cord atrophy is related to axonal loss in neuropathology studies (McGavern et al., 2000; DeLuca et al., 2004; Evangelou et al., 2005), the decrease in MTR in the outer spinal cord seen at this stage of multiple sclerosis may be occurring in the absence of—and by implication preceding—significant axonal loss.

The pathological basis of the reduced outer cord MTR warrants further consideration. In an inflammatory animal model of multiple sclerosis (experimental allergic encephalomyelitis) areas of oedema, signifying inflammation, exhibited a mildly decreased MTR in the absence of demyelination (Dousset et al., 1992). A decrease in MTR also occurs with inflammation in spinal cord experimental allergic encephalomyelitis (Cook et al., 2004). A post-mortem study in multiple sclerosis has demonstrated reduced MTR corresponding with an increased number of inflammatory T cells (Moll et al., 2009). Other post-mortem studies in multiple sclerosis also report a reduction in MTR in regions of demyelination in the cerebral cortex (Schmierer et al., 2010; Chen et al., 2013), brain white matter (Schmierer et al., 2004) and spinal cord (Bot et al., 2004). As the outer spinal cord voxel layer in this study is likely to contain both the pia mater and subpial white matter tissue, a combination of inflammation in the former and demyelination in the latter may be responsible for the decrease seen in multiple sclerosis.

Although the limits of image resolution prevent a more specific interpretation of the MTR decrease that we saw in outer spinal cord region, it may nevertheless reflect a distinct pathogenic process in so far as co-localized subpial demyelination may occur secondary to meningeal inflammation. These pathological changes have been previously associated in brain biopsies of patients with early multiple sclerosis, where meningeal inflammation was found to have a 90% probability to be topographically associated with subpial demyelination (Lucchinetti et al., 2011). Further pathological studies will be needed to determine whether such changes are topographically related in the spinal cord.

Association of outer spinal cord magnetization transfer ratio with cord atrophy and progressive multiple sclerosis

In this study, outer spinal cord MTR was independently associated with cord atrophy. In a previous study of inflammation in the spinal cord meninges there was an association seen between meningeal inflammation and diffuse axonal loss in the spinal cord parenchyma (Androdias et al., 2010). In a similar study by DeLuca et al. (2013) it was found that small peripheral axons are preferentially lost. In these studies inflammatory cells and mediators were demonstrated to be present in the cord meninges. Thus, in both of these pathology studies, meningeal inflammation and axonal loss were evident. It is likely that axonal loss, which can be profound in multiple sclerosis spinal cord (Ganter et al., 1999; Lovas et al., 2000), is the major substrate of spinal cord atrophy. Thus, the link we observed between outer cord MTR and cord atrophy would appear concordant with pathological association of meningeal inflammation and axonal loss in the cord. In contrast, the lack of independent association between the inner spinal cord (grey and white matter) MTR (implying demyelination) and atrophy (consistent with neuroaxonal loss) is consistent with a dissociation between these pathological processes, in line with previous pathological reports in the spinal cord (DeLuca et al., 2006) and brain (Wegner et al., 2006).

Although at present it is not known how meningeal inflammation might be associated with cord pathology, including axonal loss, a possible anatomical connection may be via the epi-pial layer of the spinal cord, which contains branches of blood vessels that penetrate the spinal cord (Millen and Woollam, 1961). Furthermore, spinal cord lesions tend to occur around small veins (Oppenheimer et al., 1978), and therefore the small epi-pial vessels could potentially provide a route of entry for inflammatory cells from the meninges into the cord parenchyma.

The comparison of clinical subgroups showed that greater outer cord MTR abnormality in both primary and secondary multiple sclerosis groups compared with relapsing-remitting multiple sclerosis (Tables 2 and 3). In the comparison of the two progressive subtypes of multiple sclerosis no significant differences were found. These results suggest that the outer spinal cord (and by implication pial and/or subpial) abnormalities are greater in the progressive stage of multiple sclerosis.

The findings are consistent with previous neuropathology studies in the brain that have linked both meningeal inflammation and subpial demyelination with the progressive stage of multiple sclerosis (Magliozzi et al., 2007). There can be extensive cortical subpial demyelination in progressive multiple sclerosis (Peterson et al., 2001; et al., 2003; Kutzelnigg et al., 2005; Vercellino et al., 2005; Wenger et al., 2006). Although cortical subpial lesions are rarely visible on MRI (Geurts et al., 2005), post-mortem study has identified that regions of cortical demyelination have a reduced MTR (Chen et al., 2013), and a recent in vivo study reported a reduced MTR of the outer cortex in multiple sclerosis that was most marked in those with a secondary progressive course (Samson et al., 2014). These observations support use of MTR to reflect subpial demyelination in the cortex and suggest that subpial demyelination could contribute to the lower MTR we observed in the outermost region of the spinal cord.

Associations between spinal cord magnetic resonance imaging metrics and disability measures

Although outer spinal cord MTR had a univariate correlation with each disability measure used in this study, in a regression analysis it was found only to be associated with the 9-Hole Peg Test z-score. This limited independent association with function may reflect the small region of spinal cord included in the outer voxel layer; the purpose in studying the outer cord MTR was to investigate for abnormalities that reflect a process of pathogenic importance (i.e. meningeal and subpial pathology) and not for an association with disability.

Spinal cord white matter-MTR was independently associated with 25-foot Timed Walk Test and multiple sclerosis functional composite z-scores; this relationship plausibly reflects pathology in functionally important motor and sensory pathways of the spinal cord. Our results also confirm an earlier finding of a univariate correlation between spinal cord grey matter MTR and EDSS (Agosta et al., 2007), although grey matter MTR was not independently associated with EDSS in the subsequent regression analysis. Amongst the several MRI metrics studied, the measure of spinal cord cross-sectional area (atrophy) had a generally stronger univariate correlation with each of the disability measures used. Furthermore, cord area was found to be independently associated with a number of measures of disability. The strong relationship between cord atrophy and disability may be due to this measure reflecting axonal loss, thought to be the pathological substrate for disability (Evangelou et al., 2000; Lovas et al., 2000; Schirmer et al., 2011). A biologically coherent finding was that the PASAT z-score (a measure of cognition) was associated only with the two brain MRI metrics (brain parenchymal fraction and T2 lesion volume) and with none of the spinal cord measures.

Limitations and future directions

Limitations should be noted when considering the findings in this study. Firstly, we only studied the upper cervical spinal cord. However, at this level, a robust registration technique could be used and segmentation of the cord into grey matter and white matter regions was possible. Furthermore, spinal cord involvement by multiple sclerosis is most common in the cervical cord (Oppenheimer et al., 1978), and we believe that our approach was the most appropriate within the technical constraints of spinal cord MRI.

Second, this study was cross-sectional in nature and future longitudinal studies will be needed to elucidate outer spinal cord MTR changes over time and its relationship with evolution of disability and cord atrophy.

Third, the in-plane resolution of the axial images was constrained by time limitations necessitated in a clinical study. Future studies of the spinal cord at 7 T field strength (Zhao et al., 2013) may provide higher resolution images within an acceptable time frame.

Finally, our in vivo imaging study does not include MTR findings for histopathologically confirmed pia mater and subpial cord. To our knowledge no such findings have been published to date, either in ex vivo multiple sclerosis spinal cord or in animal models of the disease. A post-mortem MRI-histopathology correlation study is needed to consolidate the findings of our in vivo study.

In conclusion we have found MTR abnormalities in an area corresponding to the expected location of the pia mater and subpial region in the outer cervical spinal cord. We have found that these outer spinal cord abnormalities occur early in the course of multiple sclerosis before significant cord atrophy and that a greater reduction in MTR values is seen in progressive multiple sclerosis.

Acknowledgements

The authors would also like to thank Dr Daniel Altmann for his advice on statistical analysis of these data.

Funding

The NMR Research Unit at the Queen Square MS Centre is supported by the MS Society of Great Britain and Northern Ireland, and UCLH-UCL Biomedical Research Centre.

Supplementary material

Supplementary material is available at Brain online.

Abbreviations

    Abbreviations
  • EDSS

    Expanded Disability Status Scale

  • FFE

    fast field echo

  • MTR

    magnetization transfer ratio

  • PASAT

    Paced Auditory Serial Addition Test

References

Agosta
F
Pagani
E
Caputo
D
Filippi
M
Associations between cervical cord gray matter damage and disability in patients with multiple sclerosis
Arch Neurol
 , 
2007
, vol. 
64
 (pg. 
1302
-
5
)
Androdias
G
Reynolds
R
Chanal
M
Ritleng
C
Confavreux
C
Nataf
S
Meningeal T cells associate with diffuse axonal loss in multiple sclerosis spinal cords
Ann Neurol
 , 
2010
, vol. 
68
 (pg. 
465
-
76
)
Audoin
B
Au Duong
MV
Ranjeva
JP
Ibarrola
D
Malikova
I
Confort-Gouny
S
, et al.  . 
Magnetic resonance study of the influence of tissue damage and cortical reorganization on PASAT performance at the earliest stage of multiple sclerosis
Hum Brain Mapp
 , 
2005
, vol. 
24
 (pg. 
216
-
28
)
Barkhof
F
Calabresi
PA
Miller
DH
Reingold
SC
Imaging outcomes for neuroprotection and repair in multiple sclerosis trials
Nat Rev Neurol
 , 
2009
, vol. 
5
 (pg. 
256
-
66
)
Bender
R
Lange
S
Adjusting for multiple testing: when and how?
J Clin Epidemiol
 , 
2001
, vol. 
54
 (pg. 
343
-
349
)
Brown
DA
Sawchenko
PE
Time course and distribution of inflammatory and neurodegenerative events suggest structural bases for the pathogenesis of experimental autoimmune encephalomyelitis
J Comp Neurol
 , 
2007
, vol. 
502
 (pg. 
236
-
60
)
Brownell
B
Hughes
JT
The distribution of plaques in the cerebrum in multiple sclerosis
J Neurol Neurosurg Psychiatry
 , 
1962
, vol. 
25
 (pg. 
315
-
20
)
L
Geurts
JJ
van der Valk
P
Polman
C
Barkhof
F
Lack of correlation between cortical demyelination and white matter pathologic changes in multiple sclerosis
Arch Neurol
 , 
2007
, vol. 
64
 (pg. 
76
-
80
)
L
Vedeler
CA
Nyland
HI
Trapp
BD
Mørk
SJ
Subpial demyelination in the cerebral cortex of multiple sclerosis patients
J Neuropathol Exp Neurol
 , 
2003
, vol. 
62
 (pg. 
723
-
32
)
Charil
A
Caputo
D
Cavarretta
R
Sormani
MP
Ferrante
P
Filippi
M
Cervical cord magnetization transfer ratio and clinical changes over 18 months in patients with relapsing-remitting multiple sclerosis: a preliminary study
Mult Scler
 , 
2006
, vol. 
12
 (pg. 
662
-
5
)
Calabrese
M
De Stefano
N
Atzori
M
Bernardi
V
Mattisi
I
Barachino
L
, et al.  . 
Detection of cortical inflammatory lesions by double inversion recovery magnetic resonance imaging in patients with multiple sclerosis
Arch Neurol
 , 
2007
, vol. 
64
 (pg. 
1416
-
22
)
Chard
DT
Jackson
JS
Miller
DH
Wheeler-Kingshott
CA
Reducing the impact of white matter lesions on automated measures of brain gray and white matter volumes
J Magn Reson Imaging
 , 
2010
, vol. 
32
 (pg. 
223
-
8
)
Chen
JT
Easley
K
Schneider
C
Nakamura
K
Kidd
GJ
Chang
A
, et al.  . 
Clinically feasible MTR is sensitive to cortical demyelination in MS
Neurology
 , 
2013
, vol. 
80
 (pg. 
246
-
52
)
Choi
SR
Howell
OW
Carassiti
D
Magliozzi
R
Gveric
D
Muraro
PA
, et al.  . 
Meningeal inflammation plays a role in the pathology of primary progressive multiple sclerosis
Brain
 , 
2012
, vol. 
135
 (pg. 
2925
-
37
)
Confavreux
C
Vukusic
S
Accumulation of irreversible disability in multiple sclerosis: from epidemiology to treatment
Clin Neurol Neurosurg
 , 
2006
, vol. 
108
 (pg. 
327
-
32
)
Cook
LL
Foster
PJ
Mitchell
JR
Karlik
SJ
In vivo 4.0-T magnetic resonance investigation of spinal cord inflammation, demyelination, and axonal damage in chronic-progressive experimental allergic encephalomyelitis
J Magn Reson Imaging
 , 
2004
, vol. 
20
 (pg. 
563
-
71
)
Dawson
JW
The histology of multiple sclerosis
Trans R Soc Edinburgh
 , 
1916
, vol. 
50
 (pg. 
517
-
740
)
DeLuca
GC
Alterman
R
Martin
JL
Mittal
A
Blundell
S
Bird
S
, et al.  . 
Casting a light on multiple sclerosis heterogeneity: the role of HLA-DRB1 on spinal cord pathology
Brain
 , 
2013
, vol. 
136
 (pg. 
1025
-
34
)
DeLuca
GC
Ebers
GC
Esiri
MM
Axonal loss in multiple sclerosis: a pathological survey of the corticospinal and sensory tracts
Brain
 , 
2004
, vol. 
127
 (pg. 
1009
-
18
)
DeLuca
GC
Williams
K
Evangelou
N
Ebers
GC
Esiri
MM
The contribution of demyelination to axonal loss in multiple sclerosis
Brain
 , 
2006
, vol. 
129
 (pg. 
1507
-
16
)
Dietrich
O
Reiser
MF
Schoenberg
SO
Artifacts in 3-Tesla MRI: physical background and reduction strategies
Eur J Radiol
 , 
2008
, vol. 
65
 (pg. 
29
-
35
)
Dousset
V
Grossman
RI
Ramer
KN
Schnall
MD
Young
LH
Gonzalez-Scarano
F
, et al.  . 
Experimental allergic encephalomyelitis and multiple sclerosis:lesion characterization with magnetization transfer imaging
Radiology
 , 
1992
, vol. 
182
 (pg. 
483
-
91
)
Evangelou
N
DeLuca
GC
Owens
T
Esiri
MM
Pathological study of spinal cord atrophy in multiple sclerosis suggests limited role of local lesions
Brain
 , 
2005
, vol. 
128
 (pg. 
29
-
34
)
Evangelou
N
Esiri
MM
Smith
S
Palace
J
Matthews
PM
Quantitative pathological evidence for axonal loss in normal appearing white matter in multiple sclerosis
Ann Neurol
 , 
2000
, vol. 
47
 (pg. 
391
-
95
)
Fernando
KT
Tozer
DJ
Miszkiel
KA
Gordon
RM
Swanton
JK
Dalton
CM
Barker
GJ
Plant
GT
Thompson
AJ
Miller
DH
Magnetization transfer histograms in clinically isolated syndromes suggestive of multiple sclerosis
Brain
 , 
2005
, vol. 
128
 (pg. 
2911
-
25
)
Fischer
J
Rudick
RA
Cutter
G
Reingold
S
The Multiple Sclerosis Functional Composite Measure (MSFC): an integrated approach to MS clinical outcome assessment. National MS society clinical outcomes assessment task force
Mult Scler
 , 
1999
, vol. 
5
 (pg. 
244
-
50
)
Fisniku
LK
Brex
PA
Altmann
DR
Miszkiel
KA
Benton
CE
Lanyon
R
, et al.  . 
Disability and T2 MRI lesions: a 20-year follow-up of patients with relapse onset of multiple sclerosis
Brain
 , 
2008
, vol. 
131
 (pg. 
808
-
17
)
Frischer
JM
Bramow
S
Dal-Bianco
A
Lucchinetti
CF
Rauschka
H
Schmidbauer
M
, et al.  . 
The relation between inflammation and neurodegeneration in multiple sclerosis brains
Brain
 , 
2009
, vol. 
132
 (pg. 
1175
-
89
)
Ganter
P
Prince
C
Esiri
MM
Spinal cord axonal loss in multiple sclerosis: a post-mortem study
Neuropathol Appl Neurobiol
 , 
1999
, vol. 
25
 (pg. 
459
-
67
)
Geurts
JJ
L
Pouwels
PJ
Castelijns
JA
Polman
CH
Barkhof
F
Cortical lesions in multiple sclerosis: combined postmortem MR imaging and histopathology
AJNR Am J Neuroradiol
 , 
2005
, vol. 
26
 (pg. 
572
-
7
)
Healy
BC
Arora
A
Hayden
MA
Ceccarelli
A
Tauhid
SS
Neema
M
, et al.  . 
Approaches to normalization of spinal cord volume: application to multiple sclerosis
J Neuroimaging
 , 
2012
, vol. 
22
 (pg. 
12
-
9
)
Horsfield
MA
Sala
S
Neema
M
Absinta
M
Bakshi
A
Sormani
MP
, et al.  . 
Rapid semi-automatic segmentation of the spinal cord from magnetic resonance images: application in multiple sclerosis
Neuroimage
 , 
2010
, vol. 
50
 (pg. 
446
-
55
)
Howell
OW
Reeves
CA
Nicholas
R
Carassiti
D
Radotra
B
Gentleman
SM
, et al.  . 
Meningeal inflammation is widespread and linked to cortical pathology in multiple sclerosis
Brain
 , 
2011
, vol. 
134
 (pg. 
2755
-
71
)
Kalkers
NF
Barkhof
F
Bergers
E
van Schijndel
R
Polman
CH
The effect of the neuroprotective agent riluzole on MRI parameters in primary progressive multiple sclerosis: a pilot study
Mult. Scler
 , 
2002
, vol. 
8
 (pg. 
532
-
3
)
Kearney
H
Yiannakas
MC
Abdel-Aziz
K
Wheeler-Kingshott
CA
Altmann
DR
Ciccarelli
O
, et al.  . 
Improved MRI quantification of spinal cord atrophy in multiple sclerosis
J Magn Reson Imaging
 , 
2014
, vol. 
39
 (pg. 
617
-
23
)
Krumbholz
M
Theil
D
Derfuss
T
Rosenwald
A
Schrader
F
Monoranu
CM
, et al.  . 
BAFF is produced by astrocytes and up-regulated in multiple sclerosis lesions and primary central nervous system lymphoma
J Exp Med
 , 
2005
, vol. 
201
 (pg. 
195
-
200
)
Kurtzke
JF
Rating neurologic impairment in multiple sclerosis: an expanded disability status scale (EDSS)
Neurology
 , 
1983
, vol. 
33
 (pg. 
1444
-
52
)
Kutzelnigg
A
Lucchinetti
CF
Stadelmann
C
Brück
W
Rauschka
H
Bergmann
M
, et al.  . 
Cortical demyelination and diffuse white matter injury in multiple sclerosis
Brain
 , 
2005
, vol. 
128
 (pg. 
2705
-
12
)
Laurent
D
Wasvary
J
Yin
J
Rudin
M
Pellas
TC
O'Byrne
E
Quantitative and qualitative assessment of articular cartilage in the goat knee with magnetization transfer imaging
Magn Reson Imaging
 , 
2001
, vol. 
19
 (pg. 
1279
-
86
)
Leary
SM
Miller
DH
Stevenson
VL
Brex
PA
Chard
DT
Thompson
AJ
Interferon β-1a in primary progressive MS: an exploratory, randomized, controlled trial
Neurology
 , 
2003
, vol. 
60
 (pg. 
44
-
51
)
Lin
X
Tench
CR
Turner
B
Blumhardt
LD
Constantinescu
CS
Spinal cord atrophy and disability in multiple sclerosis over four years: application of a reproducible automated technique in monitoring disease progression in a cohort of the interferon β-1a (Rebif) treatment trial
J Neurol Neurosurg Psychiatry
 , 
2003
, vol. 
74
 (pg. 
1090
-
4
)
Lovas
G
Szilágyi
N
Majtényi
K
Palkovits
M
Komoly
S
Axonal changes in chronic demyelinated cervical spinal cord plaques
Brain
 , 
2000
, vol. 
123
 (pg. 
308
-
17
)
Lovato
L
Willis
SN
Rodig
SJ
Caron
T
Almendinger
SE
Howell
OW
, et al.  . 
Related B cell clones populate the meninges and parenchyma of patients with multiple sclerosis
Brain
 , 
2011
, vol. 
134
 (pg. 
534
-
41
)
Lublin
FD
Reingold
SC
Defining the clinical course of multiple sclerosis: results of an international survey. National Multiple Sclerosis Society (USA) advisory committee on clinical trials of new agents in multiple sclerosis
Neurology
 , 
1996
, vol. 
46
 (pg. 
907
-
911
)
Lucchinetti
CF
Popescu
BF
Bunyan
RF
Moll
NM
Roemer
SF
Lassmann
H
, et al.  . 
Inflammatory cortical demyelination in early multiple sclerosis
N Engl J Med
 , 
2011
, vol. 
365
 (pg. 
2188
-
97
)
Magliozzi
R
Howell
OW
Reeves
C
Roncaroli
F
Nicholas
R
Serafini
B
, et al.  . 
A gradient of neuronal loss and meningeal inflammation in multiple sclerosis
Ann Neurol
 , 
2010
, vol. 
68
 (pg. 
477
-
93
)
Magliozzi
R
Howell
O
Vora
A
Serafini
B
Nicholas
R
Puopolo
M
, et al.  . 
Meningeal B-cell follicle in secondary progressive multiple sclerosis associate with early onset of disease and severe cortical pathology
Brain
 , 
2007
, vol. 
130
 (pg. 
1089
-
104
)
Maynard
FM
Jr
Bracken
MB
Creasey
G
Ditunno
JF
Jr
Donovan
WH
Ducker
TB
, et al.  . 
International standards for neurological and functional classification of spinal cord injury. american spinal injury association
Spinal Cord
 , 
1997
, vol. 
35
 (pg. 
266
-
74
)
McGavern
DB
Murray
PD
Rivera-Quiñones
C
Schmelzer
JD
Low
PA
Rodriguez
M
Axonal loss results in spinal cord atrophy, electrophysiological abnormalities and neurological deficits following demyelination in a chronic inflammatory model of multiple sclerosis
Brain
 , 
2000
, vol. 
123
 (pg. 
519
-
31
)
Meinl
E
Krumbholz
M
Derfuss
T
Junker
A
Hohlfeld
R
Compartmentalization of inflammation in the CNS: a major mechan- ism driving progressive multiple sclerosis
J Neurol Sci
 , 
2008
, vol. 
274
 (pg. 
42
-
4
)
Mezzapesa
DM
Rocca
MA
Falini
A
Rodegher
ME
Ghezzi
A
Comi
G
Filippi
M
A preliminary diffusion tensor and magnetization transfer magnetic resonance imaging study of early-onset multiple sclerosis
Arch Neurol
 , 
2004
, vol. 
61
 (pg. 
366
-
8
)
Millen
JW
Woollam
DH
On the nature of the pia mater
Brain
 , 
1961
, vol. 
84
 (pg. 
514
-
20
)
Moll
NM
Cossoy
MB
Fisher
E
Staugaitis
SM
Tucky
BH
Rietsch
AM
, et al.  . 
Imaging correlates of leukocyte accumulation and CXCR4/CXCL12 in multiple sclerosis
Arch Neurol
 , 
2009
, vol. 
66
 (pg. 
44
-
53
)
Montalban
X
Sastre-Garriga
J
Tintoré
M
Brieva
L
Aymerich
FX
Río
J
, et al.  . 
A single-center, randomized, double-blind, placebo-controlled study of interferon beta-1b on primary progressive and transitional multiple sclerosis
Mult Scler
 , 
2009
, vol. 
15
 (pg. 
1195
-
205
)
Mouritzen Dam
A
Shrinkage of the brain during histological procedures with fixation in formaldehyde solutions of different concentrations
J Hirnforsch
 , 
1979
, vol. 
20
 (pg. 
115
-
19
)
Oh
J
Zackowski
K
Chen
M
Newsome
S
Saidha
S
Smith
SA
, et al.  . 
Multiparametric MRI correlates of sensorimotor function in the spinal cord in multiple sclerosis
Mult Scler
 , 
2013
, vol. 
19
 (pg. 
427
-
35
)
Oppenheimer
DR
The cervical cord in multiple sclerosis
Neuropathol Appl Neurobiol
 , 
1978
, vol. 
4
 (pg. 
151
-
62
)
Peterson
JW
L
Mörk
S
Chang
A
Trapp
BD
Transected neurites, apoptotic neurons, and reduced inflammation in cortical multiple sclerosis lesions
Ann Neurol
 , 
2001
, vol. 
50
 (pg. 
389
-
400
)
Polman
CH
Reingold
SC
Banwell
B
Clanet
M
Cohen
JA
Filippi
M
, et al.  . 
Diagnostic criteria for multiple sclerosis: 2010 revisions to the McDonald criteria
Ann Neurol
 , 
2011
, vol. 
69
 (pg. 
292
-
302
)
Ransohoff
RM
Immunology: in the beginning
Nature
 , 
2009
, vol. 
492
 (pg. 
41
-
2
)
Reina
MA
De León Casasola Ode
L
Villanueva
MC
López
A
Machés
F
De Andrés
JA
Ultrastructural findings in human spinal pia mater in relation to subarachnoid anesthesia
Anesth Analg
 , 
2004
, vol. 
98
 (pg. 
1479
-
85
)
Samson
RS
Cardoso
MJ
Muhlert
N
Sethi
V
Wheeler-Kingshott
CA
Ron
M
, et al.  . 
Investigation of outer cortical magnetisation transfer ratio abnormalities in multiple sclerosis clinical subgroups
Mult Scler
 , 
2014
 
[Epub ahead of print] PMID: 24552746
Schirmer
L
Antel
JP
Brück
W
Stadelmann
C
Axonal loss and neurofilament phosphorylation changes accompany lesion development and clinical progression in multiple sclerosis
Brain Pathol
 , 
2011
, vol. 
21
 (pg. 
428
-
40
)
Schmierer
K
Scaravilli
F
Altmann
DR
Barker
GJ
Miller
DH
Magnetization transfer ratio and myelin in postmortem multiple sclerosis brain
Ann Neurol
 , 
2004
, vol. 
56
 (pg. 
407
-
15
)
Schmierer
K
Parkes
HG
So
PW
An
SF
Brandner
S
Ordidge
RJ
, et al.  . 
High field (9.4 Tesla) magnetic resonance imaging of cortical grey matter lesions in multiple sclerosis
Brain
 , 
2010
, vol. 
133
 (pg. 
858
-
67
)
Serafini
B
Rosicarelli
B
Magliozzi
R
Stigliano
E
Aloisi
F
Detection of ectopic B-cell follicles with germinal centers in the meninges of patients with secondary progressive multiple sclerosis
Brain Pathol
 , 
2004
, vol. 
14
 (pg. 
164
-
74
)
Sombekke
MH
Wattjes
MP
Balk
LJ
Nielsen
JM
Vrenken
H
Uitdehaag
BM
, et al.  . 
Spinal cord lesions in patients with clinically isolated syndrome: a powerful tool in diagnosis and prognosis
Neurology
 , 
2013
, vol. 
80
 (pg. 
69
-
75
)
Stroman
PW
Wheeler-Kingshott
C
Bacon
M
Schwab
JM
Bosma
R
Brooks
J
, et al.  . 
The current state-of-the-art of spinal cord imaging: methods
Neuroimage
 , 
2014
, vol. 
84
 (pg. 
1070
-
81
)
Tench
CR
Morgan
PS
Constantinescu
CS
Measurement of cervical spinal cord cross-sectional area by MRI using edge detection and partial volume correction
J Magn Reson Imaging
 , 
2005
, vol. 
21
 (pg. 
197
-
203
)
Udupa
JK
Samarasekera
S
Fuzzy connectedness and object definition: theory, algorithms and applications in image segmentation
Graph Models Image Process
 , 
1996
, vol. 
58
 (pg. 
246
-
61
)
Wegner
C
Esiri
MM
Chance
SA
Palace
J
Matthews
PM
Neocortical neuronal, synaptic, and glial loss in multiple sclerosis
Neurology
 , 
2006
, vol. 
67
 (pg. 
960
-
7
)
Vercellino
M
Plano
F
Votta
B
Mutani
R
Giordana
MT
Cavalla
P
Grey matter pathology in multiple sclerosis
J Neuropathol Exp Neurol
 , 
2005
, vol. 
64
 (pg. 
1101
-
7
)
Yiannakas
MC
Kearney
H
Samson
RS
Chard
DT
Ciccarelli
O
Miller
DH
, et al.  . 
Feasibility of grey matter and white matter segmentation of the upper cervical cord in vivo: a pilot study with application to magnetisation transfer measurements
Neuroimage
 , 
2012
, vol. 
63
 (pg. 
1054
-
9
)
Zackowski
KM
Smith
SA
Reich
DS
Gordon-Lipkin
E
Chodkowski
BA
Sambandan
DR
, et al.  . 
Sensorimotor dysfunction in multiple sclerosis and column-specific magnetization transfer-imaging abnormalities in the spinal cord
Brain
 , 
2009
, vol. 
132
 (pg. 
1200
-
9
)
Zhao
W
Cohen-Adad
J
Polimeni
JR
Keil
B
Guerin
B
Setsompop
K
Serano
P
Mareyam
A
Hoecht
P
Wald
LL
Nineteen-channel receive array and four-channel transmit array coil for cervical spinal cord imaging at 7T
Magn Reson Med
 , 
2013
 
Advance Access published on August 20, doi: 10.1002/mrm.24911 [Epub ahead of print]