Spinal cord spectroscopy and diffusion-based tractography to assess acute disability in multiple sclerosis
1Department of Brain Repair and Rehabilitation, Institute of Neurology, University College London, London, 2MRI Unit, National Society for Epilepsy, Chalfont St Peter and 3Department of Neuroinflammation, Institute of Neurology, University College London, London, UK
Correspondence to: Dr Olga Ciccarelli, Department of Brain Repair and Rehabilitation, Institute of Neurology, University College London, Queen Square, London WC1N 3BG, UK E-mail: o.ciccarelli{at}ion.ucl.ac.uk
 |
Summary |
|---|
 Top Summary IntroductionMethodsResultsDiscussionReferences |
|---|
There is a need to assess spinal cord involvement in multiple sclerosis with new imaging techniques in order to understand better the underlying pathology. We aimed to evaluate whether quantitative MRI measures, obtained using single-voxel 1H-MR spectroscopy of the cervical cord and diffusion-based tractography of the major spinal cord pathways, in patients with a cervical cord relapse, differed from controls and correlated with acute disability.
Fourteen patients at the onset of a cervical cord relapse with at least one lesion between C1 and C3 were imaged on a 1.5 T scanner and clinically assessed on the Expanded Disability Status Scale (EDSS), 9-hole peg test (HPT) and timed 25-foot walk test. Thirteen age- and gender-matched control subjects were also scanned. Metabolite concentrations, including total N-acetyl-aspartate (tNAA), choline-containing compounds (Cho), creatine plus phosphocreatine (Cr) and myo-Inositol (m-Ins), were quantified at C1–C3. Probabilistic tractography was performed at C1–C3 to track the lateral cortico-spinal tracts in the lateral columns, the anterior cortico-spinal tracts and the anterior spino-thalamic fasciculi in the anterior columns, and the bilateral fasciculus gracilis and cuneatus in the posterior columns. Diffusion- and tractography-derived measures of these tracts, including fractional anisotropy and voxel-based connectivity, which reflect fibre integrity, were obtained. These MRI measures were compared between patients and controls using the Mann–Whitney test. Univariate correlations between MRI measures and disability were assessed using the Spearman's rho correlation coefficient. Multiple regression analyses were performed to investigate which MRI measures independently correlated with the clinical scores, adjusting also for cross-sectional cord area, age and gender.
Patients showed lower tNAA of the cervical cord, lower connectivity and lower fractional anisotropy of the lateral cortico-spinal tracts and posterior tracts, than controls. In patients, there were significant correlations between: (i) EDSS and m-Ins, Cho, Cr and radial diffusivity of the lateral cortico-spinal tracts; (ii) HPT and Cr, radial diffusivity of the lateral cortico-spinal tracts, connectivity and fractional anisotropy of the posterior tracts, and connectivity of the anterior tracts. M-Ins was independently associated with the EDSS, while Cr, tNAA and connectivity of the posterior tracts were independently associated with the HPT.
MR spectroscopy and diffusion-based tractography of the cervical cord provide measures that are sensitive to the tissue damage occurring in this area in patients with a cervical cord relapse. These measures were found to correlate with acute disability. Our findings suggest that it would be worthwhile performing longitudinal studies and extending these novel techniques to other neurological diseases affecting the spinal cord.
Key Words: multiple sclerosis; spinal cord; fibre tracking; MR spectroscopy
Abbreviations: Cho, choline-containing compounds; Cr, creatine plus phosphocreatine; CST, cortico-spinal tract; EDSS, Expanded Disability Status Scale; FA, fractional anisotropy; m-Ins, myo-Inositol; MS, multiple sclerosis; tNAA, total N-acetyl-aspartate; HPT, 9-hole peg test
Received May 16, 2007. Revised June 5, 2007. Accepted June 6, 2007.
|
Introduction |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
Imaging of the spinal cord is challenging and needs to overcome several technical limitations, including the small size of the cord, the susceptibility artefacts due to tissue-bone interfaces, the motion artefacts arising from respiratory and cardiac activities and CSF pulsations. However, spinal cord imaging has particular importance in the management of patients with multiple sclerosis (MS), since patients’ disability often depends on the degree of spinal cord involvement. The relationship between conventional MRI findings of the spinal cord and clinical disability in MS is weak (Kidd et al., 1993
; Lycklama et al., 2003), underlying the need to develop MRI measures that are more clinically sensitive. These measures need to be pathologically specific, so that they can be used to evaluate axonal degeneration and repair mechanisms in MS, and can be extended to other neurological diseases, including intra-spinal cord tumours and traumatic spinal cord injury. Furthermore, new spinal cord imaging will be essential to monitor treatments that aim to enhance the repair mechanisms after spinal cord lesions (Freund et al., 2006), and have already been shown to be promising in human spinal cord injury (Feron et al., 2005).
Spinal cord spectroscopy and diffusion tensor imaging are two techniques with the potential to assess the extent and degree of spinal cord involvement in diseases such as MS. A recent study has performed cervical cord spectroscopy using a 3 T scanner in a group of healthy subjects (Marliani et al., 2007), suggesting that this technique is feasible on high magnetic fields and allows metabolite quantifications. One attempt to develop spinal cord spectroscopy on a 1.5 T scanner was previously made by our group with limited success in identifying the major metabolites in healthy volunteers (Gomez-Anson et al., 2000). To date, the only study that has applied spinal cord spectroscopy to patients with MS (Kendi et al., 2004) has reported a reduced N-acetyl-aspartate (NAA) in the normal-appearing white matter (NAWM) of the spinal cord, although the metabolite concentrations or the NAA/Cr ratios were not provided. Reduced NAA has been shown to be associated with axonal metabolic dysfunction (Narayana et al., 1998) and reduced axonal density (Bjartmar et al., 2000).
The potential of diffusion tensor imaging (DTI) to detect abnormalities in the spinal cord of patients with MS has been suggested in recent studies (Valsasina et al., 2005; Hesseltine et al., 2006). These studies, although using different sequences and methodologies, have reported reduced fractional anisotropy (FA) (Hesseltine et al., 2006), suggesting axonal degeneration and myelin breakdown (Beaulieu et al., 1996; Pierpaoli et al., 2001), and an association between the average cord FA and disability (Valsasina et al., 2005). Using the information contained in the DTI images, tractography methods can track in vivo white matter pathways (Mori and van Zijl, 2002). Tractography of the spinal cord has not been widely investigated, and few papers have been published that investigated either controls or patients with cord tumours using deterministic tractography (Ducreux et al., 2006; Voss et al., 2006).
Disability in MS can be divided into acute (potentially reversible) and chronic. Acute disability is associated with relapses, which can be described as acute deterioration of pre-existing symptoms or the development of new symptoms, while chronic disability relates more to the slow progression of symptoms that occurs independently of relapses. In our paper we focused on the first type of disability, and in particular, on the motor disability that is associated with an acute cervical cord relapse and the presence of at least one lesion in the cervical cord. The presence of acute motor and sensory limb signs during a cord relapse suggests the involvement of the lateral and anterior cortico-spinal tracts that run in the lateral and anterior columns of the spinal cord respectively, and of the anterior spinothalamic and dorsal column pathways, that run in the anterior and posterior columns respectively. These are the clinically relevant tracts which were examined in this study.
In this work we improved current MRI techniques and applied them to MS patients with a cervical cord relapse in order to address the three following questions: (1) Are MR spectroscopy and diffusion-based tractography feasible on a clinical scanner? (2) Do metabolite concentrations and diffusion- and tractography-derived measures of the spinal cord pathways differ from healthy subjects? (3) Do these new MRI measures correlate with acute disability?
|
Methods |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
Subjects
We recruited the first 14 patients with a diagnosis of MS (McDonald et al., 2001
), who attended the Outpatient Department at the National Hospital for Neurology and Neurosurgery in London and fulfill our criteria. The inclusion criteria were: (1) Acute development of motor signs, which were considered to be due to a lesion between C1 and C3; (2) Onset within 4 weeks of visit; (3) Presence of at least one lesion in the spine between C1 and C3 documented by conventional MRI scan.
On the day of the scanning all patients were clinically assessed and were scored on the Expanded Disability Status Scale (EDSS) (Kurtzke, 1983), Timed 25-foot Walk Test (TWT) (Cutter et al., 1999), 9-Hole Peg Test (HPT) (Goodkin et al., 1988) and Multiple Sclerosis Walking Scale-12 (MSWS-12) (Hobart et al., 2003). The average of two trials for the TWT and the average of four trials of the HPT (averaged as reciprocals of the mean times from two trials for each hand) (Fischer et al., 1999) were calculated. Six months after enrolment into the study, patients were again scored on the EDSS. Where patients had not reached the 6-month visit, the EDSS was obtained at 3 months.
Thirteen age- and gender-matched healthy subjects [mean age 40.9 y (SD 12.6), eight women and five men] were also studied.
All subjects gave informed, written consent before the study, which was approved by the joint ethics committee of the Institute of Neurology and the National Hospital for Neurology and Neurosurgery.
MRI protocol
All imaging was performed on a 1.5 T GE scanner.
Conventional imaging
Patients underwent spinal cord T2 and PD sequences at baseline and after 6 months (or 3 months in two cases) (TR 3300 ms/TE 110 ms and TR 3000 ms/TE 8.9 ms, respectively, echo train length 33 and 11, respectively, FOV 240 x 240 mm2, matrix 256 x 224 interpolated to 512 x 512, 12 contiguous sagittal-oblique and coronal-oblique slices, 3-mm slice thickness). An expert neuro-radiologist reviewed all the scans to identify the cord lesions and cord swelling, defined as a bulge in the contour of the spinal cord when compared to the adjacent levels on the sagittal images.
Single voxel spectroscopy
A volume of interest (VOI) with dimensions of about 6 x 8 x 50 mm3 (2.4 ml) was placed along the main axis of the cord between C1 and C3 on T2 images (Fig. 1). The volume of the VOI varied between subjects (because of inter-subject variability in the size and shape of the cord), but it did not differ between patients and controls [patients: mean 2.45 ml (SD 0.62), versus controls: mean 2.59 ml (SD 0.40), Mann–Whitney U test].
|
1H-MRS was obtained with the PROBE (PRESS) sequence (TE 30 ms, 192 averages) with CHESS water suppression, cardiac gating [TR 3RR (
3 s)] and spatial saturation pulses. Soon after the acquisition of each subject's spectrum (while the examination on the scanner was still ‘active’), a 2.7 L phantom with physiological concentrations of brain metabolites (18 cm diameter MRS HD sphere, model 2152220; General Electric, Milwaukee, WI) was positioned in the scanner such that its centre coincided with the centre of the subject's voxel. The spectroscopy series was then copied, maintaining the voxel prescription of the subject, and the acquisition was repeated with TR 2000 ms and TE 30 ms. The TR was shorter than in vivo since the phantom contained 1 g/l of gadolinium (to give physiological T2 values) and thus had short T1. Automated pre-scanning was repeated.
Diffusion tensor imaging
A CO-ZOOM-EPI technique (contiguous zonally orthogonal multislice echo-planar imaging) (Wheeler-Kingshott et al., 2002; Dowell et al., 2007) was acquired [FOV 70 x 47 mm2, matrix 48 x 32, acquired in-plane resolution 1.5 x 1.5 mm2 (reconstructed to 1 x 1 mm2), 30 contiguous axial slices, 5-mm slice thickness, TE 96 ms, cardiac gating (TR 15RR (15 s)), 6 b 0 s/mm2 images, diffusion gradients applied along 31 optimized diffusion directions (Cook et al., 2005), with max gradient amplitude of 33 x 10–3 Tm–1, maximum b factor of 1000 s/mm2 optimized for white matter (Jones et al., 1999)].
Cord atrophy
Spinal cord atrophy is commonly found in MS (Tench et al., 2005) and it correlates with the accumulation of clinical disability (Losseff et al., 1996; Sastre-Garriga et al., 2005). Therefore, it was included in our study.
A volume-acquired, inversion-prepared, fast spoiled-gradient recalled (FSPGR) sequence of the spinal cord was collected (TR 13.2 ms, TE 4.2 ms, TI 450 ms, flip angle 20°, FOV 250 x 250 mm2, matrix 256 x 256, 60 contiguous sagittal slices, 1-mm slice thickness) and used to calculate the cross-sectional cord area at C2–C3.
MRI processing
Single voxel spectroscopy
Analysis of the spectra was performed using LCModel (user-independent frequency-domain fitting software) (Provencher, 1993) and SAGE-IDL 2005.3 (GE, Milwaukee). First, LCModel was used to estimate the concentration of NAA in the phantom that was then divided by the known concentration of NAA in order to obtain a subject- and position-specific calibration factor. This calibration factor was then used to estimate the concentrations (in mM/l) and the quantification error (as a percentage SD) of the following metabolites: total N-acetyl-aspartate (tNAA) (equal to NAA + NAA-glutamate), choline-containing compounds (Cho), creatine plus phosphocreatine (Cr) and myo-Inositol (m-Ins). LCmodel estimates were retained if the percentage SD was lower than 20 for at least one of the metabolites of interest. One control spectrum was discarded because it did not meet this criterion. Therefore, the spectroscopic data came from 14 patients and 12 controls [mean age 41.7 years (SD 12.8), eight women and four men)]. All percentage SDs of the metabolites were less than 50 except in two patients and in one control. These metabolite concentrations, in these specific cases, were excluded from the analysis. Two typical post-processed spectra from a patient and a control are shown in Fig. 2.
|
Diffusion-based tractography
The diffusion data were processed to determine the diffusion behaviour on a voxel-by-voxel basis using Camino (http://www.cs.ucl.ac.uk/research/medic/camino/) (Cook et al., 2006
), and since one patient's images were discarded because of movement artefact, the patients’ data came from 13 subjects [mean age 36 years (SD 8), eight women and five men]. The information contained in the diffusion maps was used by the probabilistic tractography algorithm, called Probabilistic index of Connectivity (PICo) (Parker et al., 2003), to produce a connectivity map for each ‘seed’ voxel. In each subject, four seed voxels were placed on the middle slice of the axial b0 to reconstruct the following four tracts: (1) the left lateral cortico-spinal tract (CST) in the left lateral columns, (2) the right lateral CST in the right lateral column, (3) the anterior CSTs together with the anterior spino-thalamic fasciculi in the anterior columns and (4) the bilateral fasciculus gracilis and fasciculus cuneatus in the posterior columns (Fig. 3). Each connectivity map (between C1 and C3) was then thresholded at 0.1 [as thresholding has been largely used in brain studies (Ciccarelli et al., 2006; Powell et al., 2006; Thottakara et al., 2006)], and the mean connectivity value was calculated for each tract in each subject. These thresholded maps were then transformed into binary images, which were used to segment the diffusion maps to obtain their mean value (i.e. FA, MD, the first eigenvalue, and the mean of the second and third eigenvalue). The first eigenvalue was referred to as axial diffusivity (i.e. diffusivity parallel to the axonal fibres), whilst the average of the second and third eigenvalue was referred to as radial diffusivity (e.g. diffusivity perpendicular to the axonal fibres) (Song et al., 2002). Since there was no significant difference in all the diffusion- and tractography-derived measures between the right and left lateral tract using the Wilcoxon Signed Ranks Test, in both patient and control groups (results not shown), the mean values of the right and left lateral CSTs were used in the statistical analysis.
|
To calculate the intra-observer coefficient of variation (CV) for the diffusion- and tractography-derived measures of the lateral CST, anterior and posterior tracts, a new tractography experiment was run after 1 month on the data from eight subjects (four patients and four controls), including choosing the seed voxels. The observer was blind to the results of the first experiment. The CV was calculated using the following formula: CV = (SD/mean) x 100.
Cord atrophy
Cross-sectional cord area was measured using a semiatomated method previously described (Losseff et al., 1996).
Statistical analysis
Changes in EDSS between baseline and follow-up were assessed using the Wilcoxon Signed rank test.
Differences between groups
Statistical analyses were performed using SPSS 11.5 for Windows. To investigate differences in metabolite concentrations, voxel-based connectivity, FA, MD, eigenvalues of the tractography-derived lateral, posterior and anterior tracts and cross-sectional cord area between patients and controls, the Mann–Whitney U test was used. The same test was performed to assess differences between patients with and without cord swelling and controls.
Associations between spinal cord MRI and disability
We investigated, in the patient group, the association between the disability scores (e.g. EDSS, HPT, TWT and MSWS-12) and the MRI parameters (e.g. tNAA, m-Ins, Cho, Cr, FA, MD, connectivity and cross-sectional cord area) as follows. First, we tested for univariate correlations using Spearman's rho correlation coefficient. P-values <0.05 were considered to be significant. Secondly, in order to identify which MRI parameter was associated with disability independently from the others and from age and gender, a multiple regression analysis was repeated for each clinical score. In the case of EDSS, an ordinal logistic multiple regression was performed using the EDSS ‘categorized’ in four groups as a dependent variable, and all the other MRI variables, together with age and gender, as covariates. This analysis was chosen because the EDSS is not a linear scale and was not normally distributed, and a multiple linear regression would have been incorrect. The EDSS was categorized on the basis of disability as follows: (i) EDSS 
3.5 = group 1 (four patients with minimal disability and fully ambulatory); (ii) EDSS between 4.0 and 4.5 = group 2 (four patients with moderate disability and restricted ambulation); (iii) EDSS between 5.0 and 6.0 = group 3 (three patients with severe disability and restricted ambulation or who required unilateral assistance); and (iv) EDSS 
6.5 = group 4 (three patients who required constant bilateral assistance). For the HPT, TWT and MSWS-12, linear multiple regression analyses were performed using the clinical score as a dependent variable, and all the MRI measures, together with age and gender, as independent variables. In particular, we used the inverse of the mean values of HPT and the inverse of TWT to make the scores normally distributed, while for the MSWS-12 we used the original score, since this is an ordinal scale and was normally distributed.
|
Results |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
Patients’ characteristics
Fourteen patients (mean age 35.3 years (SD 8.16), nine women and five men, 13 relapsing-remitting and 1 secondary-progressive (Lublin and Reingold, 1996
), mean disease duration 9.2 years, SD 5.1] were recruited. Their median EDSS was 4 (range 2.5–6.5), the mean TWT was 9 s (SD 3.2), the mean HPT was 36.8 s (SD 37.6) and the MSWS-12 was 47 (SD 7). The mean delay from onset of the acute symptoms to the MRI scan was 23.8 days (SD 11.7). Although in three patients the delay from onset of symptoms was longer than 4 weeks (Table 1), these patients developed motor symptoms 2 weeks before the MRI scan. All patients, except one, complained of sensory symptoms in the upper limb (but not above the neck) and none had optic neuritis. All patients were receiving treatment with steroids and physiotherapy input around the time of the scanning (three patients had all three doses of 1 g of I.V. methylprednisolone in the days before the MRI, and the remainder were in mid-course at the time of scanning). All patients were on treatment with disease-modifying drugs. More details on patients’ characteristics, including spinal cord MRI findings, are summarized in Table 1.
|
Thirteen patients came back for a follow-up visit, which was at 6 months in all except two cases, who were seen at 3 months. The median EDSS at follow-up was 3.5 (range 1–6.5). There was a statistical significance in the EDSS score changes between baseline and follow-up [median EDSS change –1 (range 0, –2.5), P = 0.02]. If four patients (31%) who had a further relapse during the follow-up period are excluded (since this may contribute to the final EDSS), seven patients (78%) out of the remaining nine improved by at least 1 step in the EDSS, while two (22%) showed no change (Table 1). The MRI findings at follow-up showed a resolution of the swelling or a reduction of the cervical cord demyelination in nine patients, no change in three patients, and an increased lesion size in one patient (Table 1).
Differences in MRI measures between patients and controls
Patients showed significantly lower cervical cord tNAA than controls (P < 0.0001) and an increase in the concentration of m-Ins, but this did not reach statistical significance (Table 2).
|
Patients showed lower connectivity (P = 0.03) and lower FA (P = 0.006) of the tractography-derived lateral CST, and lower connectivity (P = 0.03) and lower FA (P = 0.02) of the posterior tracts compared with controls (Table 2).
There was no significant difference in the cross-sectional spinal cord area between patients and controls. Five patients showed swelling of the cervical cord (mean cord area 80.5 mm2, SD 5.6), and if they were excluded from the patient group, a significant reduction in the spinal cord area between the remainder of patients and controls was found (P = 0.01) (Table 2).
When patients with and without cord swelling were considered separately, differences in the MRI measures between controls and patients were similar to those obtained when all patients were considered together except for (1) the mean connectivity of the lateral CST and of the posterior tract of patients without cord swelling; (2) the FA of the posterior tract in patients with cord swelling, neither of which differed from controls.
The intra-observer CV was remarkably low: 1.1, 0.7 and 1.6% for connectivity, FA and MD of the lateral CST, 0.6, 0.8 and 0.3% for connectivity, FA and MD of the posterior tracts, and 0% for connectivity, FA and MD of the anterior tracts.
Association between spinal cord MRI and disability
The univariate correlation analysis showed that EDSS correlated significantly with m-Ins (r = 0.64, P = 0.02), Cho (r = 0.65, P = 0.01), Cr (r = 0.75, P = 0.003) (Fig. 4) and radial diffusivity of the lateral CST (r = 0.57, P = 0.04). There were significant correlations between HPT and Cr (r = 0.57, P = 0.04), radial diffusivity of the lateral CST (r = 0.67, P = 0.01), connectivity (r = –0.82, P = 0.001) and FA (r = –0.57, P = 0.04) of the posterior tracts (Fig. 4), and connectivity of the anterior tracts (r = –0.58, P = 0.04). MSWS-12 significantly correlated with Cr (r = 0.74, P = 0.004) (Fig. 4) and radial diffusivity of the lateral CST (r = 0.74, P = 0.004).
|
The multiple regression analyses showed that m-Ins was independently associated with the EDSS [the odds ratio of being in a category of greater EDSS was 1.7 for each unit increment of m-Ins, P = 0.02, 95% confidence interval (CI) 1.1, 2.5]. We also found that tNAA (P = 0.04, coeff. 0.008, 95% CI 0.001, 0.015), Cr (P = 0.005, coeff. –0.006, 95% CI –0.01, –0.002), and connectivity of the posterior tracts (P = 0.009, coeff. 0.20, 95% CI 0.06, 0.33) were independently associated with the HPT. Cr was independently associated with the MSWS-12 (P = 0.02, coeff. 3.84, 95% CI 0.86, 6.82). Finally, the regression model including all the MRI variables (including cord cross-sectional area), together with age and gender, as predictors of the inverse TWT, did not show any significant association between any MRI measure and TWT.
|
Discussion |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
We have demonstrated that MR spectroscopy of the cervical cord and diffusion-based tractography of the major spinal pathways are possible on a 1.5 T scanner. These cutting-edge techniques provide new measures, some of which detected differences between patients and controls and correlated with acute disability.
Spinal cord spectroscopy
Quantification of metabolites
Our concentrations of metabolites calculated in normal controls are similar to those recently reported (Marliani et al., 2007), with the exception of m-Ins, which appears to be lower in our study. However, given the different scanners and different acquisitions, incomplete agreement is unsurprising.
Comparisons between groups
We found that patients showed a significantly lower tNAA (a neuronal marker) than controls, suggesting the presence of axonal metabolic dysfunction (Narayana et al., 1998), or loss, (Bitsch et al., 1999) or both, in the cervical cord containing the acute lesion. This result mirrors the findings of acute brain lesions (Davie et al., 1994) and brain NAWM (van Walderveen et al., 1999; Chard et al., 2002). This finding extends the results of a previous report that described lower NAA in the NAWM of the cord in patients with MS (Kendi et al., 2004), although without metabolite quantification.
The other metabolites were not significantly different between patients and controls, although patients showed a trend towards higher m-Ins. Increased m-Ins has been found in both acute (Davie et al., 1994) and chronic (Brex et al., 2000) brain lesions and in the brain NAWM (Chard et al., 2002; Fernando et al., 2004), suggesting an increase in the activity or number of glial and inflammatory cells. We did not find any group differences for Cr and Cho, which have been reported to be elevated in demyelinating lesions (Brex et al., 1999; Suhy et al., 2000), although a decrease in Cr in acute MS lesions has also been described (Davies et al., 1995).
Association between metabolites and disability
We found that: (i) m-Ins was independently associated with EDSS, and, in particular, it was higher in patients with greater disability, (ii) Cr was independently associated with the upper limb function, as measured with the HPT, and with the MSWS-12, which is a scale developed to measure the impact of MS on walking; in particular, Cr was higher in patients with greater disability, and (iii) tNAA was independently associated with the upper limb function, being lower in patients with greater disability on the HPT. Interestingly, the latter association was not significant when performing the univariate analysis, that is, without adjustment for Cr concentration; the relationship between lower tNAA and higher HPT only became significant in patients with similar Cr, as assessed by the multiple regressions including Cr as independent variable. This appears to be due to negative confounding by the Cr concentration: there was a positive relationship between tNAA and Cr (patients with lower NAA showed lower Cr), and between Cr and HPT (patients with lower Cr showed lower HPT).
We found that NAA did not correlate with either EDSS or TWT whether using the univariate or multivariate analysis, which is consistent with a significant component of the tNAA decrease being due to potentially reversible metabolic dysfunction, rather than irreversible structural changes (Narayana et al., 1998; Cader et al., 2007). Therefore, as previously demonstrated in acute MS brain lesions (Davie et al., 1994; De Stefano et al., 1995), it is possible that the spinal cord NAA will increase over time after an acute event, without reaching the normal level, and that the spinal cord NAA level in the chronic phase will reflect irreversible axonal loss and thus will correlate more strongly with the residual disability.
Spinal cord diffusion tensor imaging
Reconstruction of spinal cord pathways using probabilistic tractography
This is the first time a probabilistic algorithm (Parker et al., 2003) has been applied to track the clinically relevant spinal cord pathways in patients with MS. The probability maps of the spinal pathways were used not only to obtain a voxel-based connectivity index that was used in the assessment of spinal cord damage, but also to calculate the mean of the diffusion indices along the tracts. The values of mean FA in controls were slightly lower than those previously reported using the ROIs at C2–C3 level (Hesseltine et al., 2006). These differences might be due to the different methodologies (ROIs versus tractography) and to the different imaging protocols, including a smaller number of diffusion encoding directions (6 versus 31) used by Helletine and colleagues., which can result in a biased measure of FA (Alexander and Barker, 2005). We do not think that this difference in FA is likely to be due to different degree of partial volume, since we also measured a similar value of MD for the posterior tracts and a lower value of MD for the anterior tracts.
Comparisons between groups
Patients showed a significantly lower mean connectivity and FA in the tractography-derived CSTs and posterior tracts when compared to controls. In particular, we found that the differences in connectivity of the CSTs and posterior tracts between patients and controls were mainly due to the contribution of patients with cord swelling, while changes in FA of the posterior tract were driven essentially by patients without cord swelling. These findings confirm that FA and connectivity provide complementary information and they might reflect different aspects of the pathological processes. There is evidence that a reduction in FA is consistent with either axonal fibre degeneration and myelin breakdown (Beaulieu et al., 1996; Pierpaoli et al., 2001), or myelin loss per se (Schmierer et al., 2007), whilst the pathological correlate of a reduction in connectivity still needs to be clarified by pathological studies. Our analysis of diffusivities suggests that the reduction in FA detected in the present patient cohort is due to an increase in radial diffusivity (diffusivity perpendicular to the axonal fibres). This suggests that the most relevant pathological mechanism occurring in the spinal tracts of these patients was probably demyelination, rather than axonal loss. Indeed, animal studies have reported that demyelination causes an increase in radial diffusivity without changing axial diffusivity (Song et al., 2002, 2005). However, the presence of a constant axial diffusivity could still be compatible with axonal loss, if the larger diameter fibres are preserved and the fibres of small diameter (<3 µm2) of the CST are preferentially lost (DeLuca et al., 2004).
In our study, we have found a smaller spinal cord cross-sectional area in patients without cord swelling when compared with controls. The reduction of cord cross-sectional area has been related to both axonal loss (Ganter et al., 1999) and loss of myelin (Bot et al., 2004).
We did not find any associations between the location of the lesions on the MRI scans and the tractography-derived measures (results not shown), but the number of lesions for each group (anterior, posterior, whole spine, for example) was very small, and future studies need to investigate further this relationship.
Association between diffusion- and tractography-derived measures and disability
We found that connectivity of the posterior column pathways was independently associated with HPT, and, in particular, it was lower in patients with greater upper limb dysfunction. We tracked the posterior column pathways because they mediate important sensory functions, such as touch, form, movement and position sense. It is likely that abnormalities in these pathways, which were reflected by lower values of voxel-based connectivity and related to the presence of cervical cord lesions, can affect the performance on the HPT. The univariate analysis showed more correlations, including those between HPT and the FA of the posterior tracts and the radial diffusivity of the lateral CST, but these relationships were not independent of the other variables, since they disappeared in the multiple regression model.
Limitations and future developments
A limitation of the study is that we did not perform gadolinium-enhanced scans, but rather recruited patients at the onset of a cervical cord relapse who had at least one lesion in the cervical cord which was thought to be responsible for their clinical signs. Therefore, we do not know whether one (or more) cervical lesion included in the spectroscopic voxel, or crossed by the tractography-derived pathways, were completely new, or a reactivation of inflammation in old plaques. We did not use gadolinium because: (i) the enhancement would have not clarified this issue, (ii) patients were receiving steroids around the time of the scanning and (iii) it would have prolonged the scanning time. However, there was clear evidence that the events were acute and related to the cervical cord lesions in all but two cases (patient no. 6—who dropped out from the study, and no. 11, Table 1). The evidence came from the clinical and radiological findings at follow-up (Table 1), together with (i) upper limb dysfunction, with no new symptoms or signs above the cord in all patients at baseline, (ii) cord swelling, which is a sign of acute inflammation, (iii) the normal cord MRI before the study in three cases. Furthermore, when the whole analysis was repeated without the two patients for whom acute changes could not be demonstrated, the results did not change substantially (results not shown).
Another limitation of this study is that, although the spectroscopic volume included the acute lesion, it also encompassed part of the normal-appearing white matter (NAWM), at least in those cases that showed a spinal cord lesion smaller than the voxel size. Therefore, it is possible that some of the MRI measure changes were due to the spinal cord NAWM abnormalities around the lesion. Similarly, the tractography-reconstructed tracts passed through one or more lesions, but the b = 0 and FA image resolution did not allow us to distinguish between the portion of the tracts that passed through the lesion from the tracts outside the lesion. Therefore we produced only mean values of diffusion parameters for the whole tracts.
A further limitation is that we performed a large number of statistical tests (about 100) without formal correction for multiple comparisons. Nevertheless, for an 
level of 0.05, one would expect on average 5 out of 100 false positive results. However, we reported over 20 significant results for P-value <0.05, making it very unlikely that type I error played a role. About half of these results had P-values <0.01, making them even more robust. We feel that this informal assessment of potential type I error is a sensible way to deal with the multiple comparisons issue in the context of an exploratory study, and the significant results presented here will help to inform the hypotheses for future studies in this field.
Finally, a more general observation is that the exact relationship between changes in the spinal cord cross-sectional area, metabolite concentrations, diffusion- and tractography-derived measures, and histopathological processes remains to be clarified in radiological and pathological correlation studies. A technical consideration, that is evident when discussing spinal cord imaging, is that higher magnetic field scanners with parallel imaging coils will provide an increased signal-to-noise ratio with advantages for the detection of metabolites and the quality of the spinal cord DTI images.
In conclusion, we have shown promising results from the application of advanced spinal cord imaging to MS. Our findings suggest that the spectroscopic and tractography-derived measures are clinically relevant, since they are associated with disability even in this small cohort of well-selected patients. We are currently investigating whether the diffusion and spectroscopic changes in the cervical cord at the onset of a cervical cord relapse predict the recovery process or the development of irreversible disability by carrying out a follow-up study of this cohort.
|
Acknowledgements |
|---|
This study is supported by the Wellcome Trust, grant 074618/Z/04. O. Ciccarelli is a Wellcome Trust Advanced Clinical Fellow. C. Wheeler-Kingshott is supported by the same grant. The authors thank D. Altmann for his advice on the statistics, C. Benton, L. Thirkell and K. Chappell for technical assistance with the MRI scans, T. Jenkins, F. Manfredonia and Z. Khaleeli for their assessment of MS patients, J. Chataway and the MS Relapse Team at the NHNN for recruiting patients, K. Miszkiel for reviewing the conventional scans, and the subjects for kindly agreeing to take part in this study. The MS NMR Research Unit is supported by the MS Society of Great Britain and Northern Ireland. This work was undertaken at UCLH/UCL who received a proportion of funding from the Department of Health's NIHR Biomedical Research Centres funding scheme. Funding to pay the Open Access publication charges for this article was provided by the Wellcome Trust.
|
References |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
Alexander DC, Barker GJ. Optimal imaging parameters for fiber-orientation estimation in diffusion MRI. Neuroimage (2005) 27:357–67.[CrossRef][ISI][Medline]
Beaulieu C, Does MD, Snyder RE, Allen PS. Changes in water diffusion due to Wallerian degeneration in peripheral nerve. Magn Reson Med (1996) 36:627–31.[ISI][Medline]
Bitsch A, Bruhn H, Vougioukas V, et al. Inflammatory CNS demyelination: histopathologic correlation with in vivo quantitative proton MR spectroscopy. AJNR Am J Neuroradiol (1999) 20:1619–27.
Bjartmar C, Kidd G, Mork S, Rudick R, Trapp BD. Neurological disability correlates with spinal cord axonal loss and reduced N-acetyl aspartate in chronic multiple sclerosis patients. Ann Neurol (2000) 48:893–901.[CrossRef][ISI][Medline]
Bot JC, Blezer EL, Kamphorst W, et al. The spinal cord in multiple sclerosis: relationship of high-spatial-resolution quantitative MR imaging findings to histopathologic results. Radiology (2004) 233:531–40.
Brex PA, Gomez-Anson B, Parker GJ, et al. Proton MR spectroscopy in clinically isolated syndromes suggestive of multiple sclerosis. J Neurol Sci (1999) 166:16–22.[CrossRef][ISI][Medline]
Brex PA, Parker GJ, Leary SM, et al. Lesion heterogeneity in multiple sclerosis: a study of the relations between appearances on T1 weighted images, T1 relaxation times, and metabolite concentrations. J Neurol Neurosurg Psychiatry (2000) 68:627–32.
Cader S, Johansen-Berg H, Wylezinska M, et al. Discordant white matter N-acetylasparate and diffusion MRI measures suggest that chronic metabolic dysfunction contributes to axonal pathology in multiple sclerosis. Neuroimage (2007) 36:19–27.[CrossRef][ISI][Medline]
Chard DT, Griffin CM, McLean MA, et al. Brain metabolite changes in cortical grey and normal-appearing white matter in clinically early relapsing-remitting multiple sclerosis. Brain (2002) 125:2342–52.
Ciccarelli O, Behrens TE, Altmann DR, et al. Probabilistic diffusion tractography: a potential tool to assess the rate of disease progression in amyotrophic lateral sclerosis. Brain (2006) 129:1859–71.
Cook PA, Bai Y, Nedjati-Gilani S, Seunarine KK, Hall MG, Parker GJ, Alexander DC. Camino: open-cource diffusion-MRI reconstruction and processing. Proc Int Soc Mag Reson Med (2006) 14. Abstract.
Cook PA, Boulby PA, Symms MR, Alexander DC. Optimal acquisition order of diffusion-weighted measurements on a sphere. Proc Int Soc Mag Reson Med (2005) 13. Abstract.
Cutter GR, Baier ML, Rudick RA, et al. Development of a multiple sclerosis functional composite as a clinical trial outcome measure. Brain (1999) 122:871–82.
Davie CA, Hawkins CP, Barker GJ, et al. Serial proton magnetic resonance spectroscopy in acute multiple sclerosis lesions. Brain (1994) 117(Pt 1):49–58.
Davies SE, Newcombe J, Williams SR, McDonald WI, Clark JB. High resolution proton NMR spectroscopy of multiple sclerosis lesions. J Neurochem (1995) 64:742–48.[ISI][Medline]
De Stefano N, Matthews PM, Arnold DL. Reversible decreases in N-acetylaspartate after acute brain injury. Magn Reson Med (1995) 34:721–7.[ISI][Medline]
DeLuca GC, Ebers GC, Esiri MM. Axonal loss in multiple sclerosis: a pathological survey of the corticospinal and sensory tracts. Brain (2004) 127(Pt 5):1009–18.
Dowell NG, Miller DH, Jenkins T, Wheeler-Kingshott CA. Contiguous-slice diffusion tensor imaging of the optic nerve with CSF suppressed IR CO-ZOOM. Proc Int Soc Magn Res Med (2007) Abstract.
Ducreux D, Lepeintre JF, Fillard P, Loureiro C, Tadie M, Lasjaunias P. MR diffusion tensor imaging and fiber tracking in 5 spinal cord astrocytomas. AJNR Am J Neuroradiol (2006) 27:214–6.
Fernando KT, McLean MA, Chard DT, et al. Elevated white matter myo-inositol in clinically isolated syndromes suggestive of multiple sclerosis. Brain (2004) 127:1361–9.
Feron F, Perry C, Cochrane J, et al. Autologous olfactory ensheathing cell transplantation in human spinal cord injury. Brain (2005) 128:2951–60.
Fischer JS, Jak AJ, Kniker JE, Rudick RA, Cutter G. Administration and scoring manual for the multiple sclerosis functional composite meaure (MSFC). (1999) New York: Demos.
Freund P, Schmidlin E, Wannier T, et al. Nogo-A-specific antibody treatment enhances sprouting and functional recovery after cervical lesion in adult primates. Nat Med (2006) 12:790–2.[CrossRef][ISI][Medline]
Ganter P, Prince C, Esiri MM. Spinal cord axonal loss in multiple sclerosis: a post-mortem study. Neuropathol Appl Neurobiol (1999) 25:459–67.[CrossRef][ISI][Medline]
Gomez-Anson B, MacManus DG, Parker GJ, et al. In vivo 1H-magnetic resonance spectroscopy of the spinal cord in humans. Neuroradiology (2000) 42:515–7.[CrossRef][ISI][Medline]
Goodkin DE, Hertsgaard D, Seminary J. Upper extremity function in multiple sclerosis: improving assessment sensitivity with box-and-block and nine-hole peg tests. Arch Phys Med Rehabil (1988) 69:850–4.[ISI][Medline]
Hesseltine SM, Law M, Babb J, et al. Diffusion tensor imaging in multiple sclerosis: assessment of regional differences in the axial plane within normal-appearing cervical spinal cord. AJNR Am J Neuroradiol (2006) 27:1189–93.
Hobart JC, Riazi A, Lamping DL, Fitzpatrick R, Thompson AJ. Measuring the impact of MS on walking ability: the 12-item MS Walking Scale (MSWS-12). Neurology (2003) 60:31–6.
Jones DK, Horsfield MA, Simmons A. Optimal strategies for measuring diffusion in anisotropic systems by magnetic resonance imaging. Magn Reson Med (1999) 42:515–25.[CrossRef][ISI][Medline]
Kendi AT, Tan FU, Kendi M, Yilmaz S, Huvaj S, Tellioglu S. MR spectroscopy of cervical spinal cord in patients with multiple sclerosis. Neuroradiology (2004) 46:764–9.[CrossRef][ISI][Medline]
Kidd D, Thorpe JW, Thompson AJ, et al. Spinal cord MRI using multi-array coils and fast spin echo. II. Findings in multiple sclerosis. Neurology (1993) 43:2632–7.
Kurtzke JF. Rating neurologic impairment in multiple sclerosis: an expanded disability status scale (EDSS). Neurology (1983) 33:1444–52.
Losseff NA, Webb SL, O'Riordan JI, et al. Spinal cord atrophy and disability in multiple sclerosis. A new reproducible and sensitive MRI method with potential to monitor disease progression. Brain (1996) 119(Pt 3):701–8.
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) 46:907–11.
Lycklama G, Thompson A, Filippi M, et al. Spinal-cord MRI in multiple sclerosis. Lancet Neurol (2003) 2:555–62.[CrossRef][ISI][Medline]
Marliani AF, Clementi V, Albini-Riccioli L, Agati R, Leonardi M. Quantitative proton magnetic resonance spectroscopy of the human cervical spinal cord at 3 tesla. Magn Reson Med (2007) 57:160–3.[CrossRef][ISI][Medline]
McDonald WI, Compston A, Edan G, et al. Recommended diagnostic criteria for multiple sclerosis: guidelines from the International Panel on the diagnosis of multiple sclerosis. Ann Neurol (2001) 50:121–7.[CrossRef][ISI][Medline]
Mori S, van Zijl PC. Fiber tracking: principles and strategies - a technical review. NMR Biomed (2002) 15:468–80.[CrossRef][ISI][Medline]
Narayana PA, Doyle TJ, Lai D, Wolinsky JS. Serial proton magnetic resonance spectroscopic imaging, contrast-enhanced magnetic resonance imaging, and quantitative lesion volumetry in multiple sclerosis. Ann Neurol (1998) 43:56–71.[CrossRef][ISI][Medline]
Parker GJ, Haroon HA, Wheeler-Kingshott CA. A framework for a streamline-based probabilistic index of connectivity (PICo) using a structural interpretation of MRI diffusion measurements. J Magn Reson Imaging (2003) 18:242–54.[CrossRef][ISI][Medline]
Pierpaoli C, Barnett A, Pajevic S, et al. Water diffusion changes in Wallerian degeneration and their dependence on white matter architecture. Neuroimage (2001) 13:1174–85.[ISI][Medline]
Powell HW, Parker GJ, Alexander DC, et al. Hemispheric asymmetries in language-related pathways: a combined functional MRI and tractography study. Neuroimage (2006) 32:388–99.[CrossRef][ISI][Medline]
Provencher SW. Estimation of metabolite concentrations from localized in vivo proton NMR spectra. Magn Reson Med (1993) 30:672–9.[ISI][Medline]
Sastre-Garriga J, Ingle GT, Rovaris M, et al. Long-term clinical outcome of primary progressive MS: predictive value of clinical and MRI data. Neurology (2005) 65:633–5.
Schmierer K, Wheeler-Kingshott CA, Boulby PA, et al. Diffusion tensor imaging of post-mortem multiple sclerosis brain. Neuroimage (2007) 35:467–77.[CrossRef][ISI][Medline]
Song SK, Sun SW, Ramsbottom MJ, Chang C, Russell J, Cross AH. Dysmyelination revealed through MRI as increased radial (but unchanged axial) diffusion of water. Neuroimage (2002) 17:1429–36.[CrossRef][ISI][Medline]
Song SK, Yoshino J, Le TQ, et al. Demyelination increases radial diffusivity in corpus callosum of mouse brain. Neuroimage (2005) 26:132–40.[CrossRef][ISI][Medline]
Suhy J, Rooney WD, Goodkin DE, et al. 1H MRSI comparison of white matter and lesions in primary progressive and relapsing-remitting MS. Mult Scler (2000) 6:148–55.
Tench CR, Morgan PS, Jaspan T, Auer DP, Constantinescu CS. Spinal cord imaging in multiple sclerosis. J Neuroimaging (2005) 15:94S–102S.[CrossRef][ISI][Medline]
Thottakara P, Lazar M, Johnson SC, Alexander AL. Application of Brodmann's area templates for ROI selection in white matter tractography studies. Neuroimage (2006) 29:868–78.[CrossRef][ISI][Medline]
Valsasina P, Rocca MA, Agosta F, et al. Mean diffusivity and fractional anisotropy histogram analysis of the cervical cord in MS patients. Neuroimage (2005) 26:822–8.[CrossRef][ISI][Medline]
van Walderveen MA, Barkhof F, Pouwels PJ, van Schijndel RA, Polman CH, Castelijns JA. Neuronal damage in T1-hypointense multiple sclerosis lesions demonstrated in vivo using proton magnetic resonance spectroscopy. Ann Neurol (1999) 46:79–87.[CrossRef][ISI][Medline]
Voss HU, Watts R, Ulug AM, Ballon D. Fiber tracking in the cervical spine and inferior brain regions with reversed gradient diffusion tensor imaging. Magn Reson Imaging (2006) 24:231–9.[ISI][Medline]
Wheeler-Kingshott CA, Hickman SJ, Parker GJ, et al. Investigating cervical spinal cord structure using axial diffusion tensor imaging. Neuroimage (2002) 16:93–102.[CrossRef][ISI][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| |||||||||||||||||||||||||||||||||||||||||||||||