European Radiology

, Volume 19, Issue 8, pp 2066–2074

MR spectroscopy (MRS) and magnetisation transfer imaging (MTI), lesion load and clinical scores in early relapsing remitting multiple sclerosis: a combined cross-sectional and longitudinal study


  • J. Bellmann-Strobl
    • Cecilie Vogt Clinic for NeurologyCharité—University Medicine Berlin and Max Delbrueck Center for Molecular Medicine
  • H. Stiepani
    • Department of NeuroradiologyCharité—University Medicine Berlin
  • J. Wuerfel
    • Cecilie Vogt Clinic for NeurologyCharité—University Medicine Berlin and Max Delbrueck Center for Molecular Medicine
    • Institute of Neuroradiology, Campus LuebeckUniversity Schleswig-Holstein
  • G. Bohner
    • Department of NeuroradiologyCharité—University Medicine Berlin
  • F. Paul
    • Cecilie Vogt Clinic for NeurologyCharité—University Medicine Berlin and Max Delbrueck Center for Molecular Medicine
  • C. Warmuth
    • Department of RadiologyCharité—University Medicine Berlin
  • O. Aktas
    • Cecilie Vogt Clinic for NeurologyCharité—University Medicine Berlin and Max Delbrueck Center for Molecular Medicine
  • K. P. Wandinger
    • Department of NeurologyCharité—University Medicine Berlin
  • F. Zipp
    • Cecilie Vogt Clinic for NeurologyCharité—University Medicine Berlin and Max Delbrueck Center for Molecular Medicine
    • Department of NeuroradiologyCharité—University Medicine Berlin

DOI: 10.1007/s00330-009-1364-z

Cite this article as:
Bellmann-Strobl, J., Stiepani, H., Wuerfel, J. et al. Eur Radiol (2009) 19: 2066. doi:10.1007/s00330-009-1364-z


The purpose of this study was to correlate magnetic resonance imaging (MRI)-based lesion load assessment with clinical disability in early relapsing remitting multiple sclerosis (RRMS). Seventeen untreated patients (ten women, seven men; mean age 33.0 ± 7.9 years) with the initial diagnosis of RRMS were included for cross-sectional as well as longitudinal (24 months) clinical and MRI-based assessment in comparison with age-matched healthy controls. Conventional MR sequences, MR spectroscopy (MRS) and magnetisation transfer imaging (MTI) were performed at 1.5 T. Lesion number and volume, MRS and MTI measurements for lesions and normal appearing white matter (NAWM) were correlated to clinical scores [Expanded Disability Status Scale (EDSS), Multiple Sclerosis Functional Composite (MSFC)] for monitoring disease course after treatment initiation (interferon β-1a). MTI and MRS detected changes [magnetisation transfer ratio (MTR), N-acetylaspartate (NAA)/creatine ratio] in NAWM over time. EDSS and lesional MTR increases correlated throughout the disease course. Average MTR of NAWM raised during the study (p < 0.05) and correlated to the MSFC score (r = 0.476, p < 0.001). At study termination, NAA/creatine ratio of NAWM correlated to the MSFC score (p < 0.05). MTI and MRS were useful for initial disease assessment in NAWM. MTI and MRS correlated with clinical scores, indicating potential for monitoring the disease course and gaining new insights into treatment-related effects.


Multiple sclerosisMR spectroscopyMagnetisation transfer imagingWhite matter


Multiple sclerosis (MS) is the most common demyelinating disease worldwide and the most common neurological disorder within the northern hemisphere in young adults, showing a prevalence of 0.1% [1]. Different MS disease courses as well as varying therapeutic regimens require early and subtle assessment of acute and chronic parenchymal inflammation.

Magnetic resonance imaging (MRI) has proven to be the imaging technique of choice in establishing the diagnosis as well as for disease monitoring. MRI plays an important role in MS diagnostics, as underlined by the so-called revised McDonald criteria of the International Panel on the Diagnosis of Multiple Sclerosis [2].

Advanced MRI techniques such as MR spectroscopy (MRS) and magnetisation transfer imaging (MTI) have shown the unprecedented ability to depict and quantify changes in lesions and normal appearing white matter (NAWM) over time [35]. MRS provides metabolic information and indicates neuronal loss by decrease of marker molecule concentrations such as N-acetylaspartate (NAA) [6], whereas MTI detects alterations in macromolecular components, e.g. of myelin [7], which cannot be depicted by conventional MRI.

This study aimed at evaluating the suitability of MRS and MTI in monitoring neuroinflammatory parenchymal brain damage in correlation with conventional MRI as well as different clinical disability scores at the time of initial diagnosis and throughout the disease course after initiating interferon β (IFN β) therapy in patients with early relapsing remitting multiple sclerosis (RRMS). Only a few studies have been carried out with mostly small sample sizes and lack of placebo control using either MRS or MTI for evaluating IFN-β treatment effects. Whether or not MRS is a suitable technique for therapy monitoring could not be clarified by these studies [811]. In a single small trial on RRMS patients treated with interferon β–1b (IFN β–1b), magnetisation transfer ratio (MTR) was not altered during a period of 6 months [12]. However, lesional MTR was reported to evolve at a faster rate during IFN therapy and may be a surrogate measure of resolving demyelination [13]. To our knowledge, this is the first longitudinal study combining MRS and MTI to investigate RRMS patients treated with IFN β-1a).

Materials and methods


Seventeen patients (ten women, seven men; mean age 33.0 ± 7.9 years, range 17–46 years) were included in our prospective study according to the following criteria:
  • Diagnosis of MS according to the McDonald criteria [14] at the time of study inclusion

  • No corticoidsteroid and/or immunmodulatory treatment throughout the preceding 6 months

  • No clinical signs of an acute inflammatory relapse

The patients in the study group were assessed clinically and by MRI at the time of initial MS diagnosis (i.e. the time of study inclusion) as well as at monthly intervals for 1 year and a single follow-up at 24 months.

The control group consisted of 17 healthy subjects (ten women, seven men; mean age 30.5 ± 7.0 years, range 20–45 years) with no clinical signs nor a history of neurological disorders. None of the participants had been treated with corticosteroids and/or other immunomodulatory drugs.

The study was approved by the local ethics committee. All subjects gave their written informed consent before inclusion in the study.


All MRI studies were performed using a 1.5-Tesla system (Magnetom Vision plus, Siemens Medical Solutions, Erlangen, Germany).

The patients head was positioned in a vacuum pillow to avoid head malrotation, and MRI measurements were subsequently performed as shown in Table 1.
Table 1

MR sequence parameters


TR (ms)

TE (ms)

Flip angle


FoV (mm)

Number of slices

Slice thickness (mm)

Distance factor (mm)






256 × 192










256 × 192










512 × 192










256 × 192









8 × 8





Lesion count and volumes were quantified based on T2-weighted double-echo sequences. These sequences were also applied for the segmentation of normal brain matter, NAWM and cerebrospinal fluid (CSF). The T1-weighted spin-echo sequence was used for determining the volume of so-called “black holes", corresponding to T1-hypointense, post-inflammatory lesions. As based on the available literature, a significant correlation between clinical scores and lesion enhancement could not be expected and, due to the tight time-line of re-scans, the decision was made not to apply contrast-enhanced measurements.


1H-MRS was performed by using chemical-shift imaging (CSI). Voxel positioning included the callosal body parallel to a line drawn through the splenium and genu, also including parts of the periventricular white matter (WM) (Fig. 1). The volume of interest (VOI) had a length of 80 mm and a thickness of 20 mm. Local shimming was performed to a maximum of 10 Hz full width at half maximum of the water peak. Water suppression was adjusted in order to assess the peaks at a maximum amplification of 114.85 dB in the phase image. Subsequently data acquisition was performed and phase and baseline correction as well as peak fitting were achieved by using the MR system’s software module LUISE (Siemens Medical Solutions, Erlangen, Germany). Altogether, 64 voxels of 10 × 10 × 20 mm size were acquired in each patient’s measurement. The spectroscopy map acquired by multivoxel imaging was visually assessed for signal quality and for obvious MS lesions in the T2-weighted scout image. Voxels indicative of MS lesions were excluded from post-processing. The following metabolite ratios were calculated: NAA to creatine (Cr) (NAA/Cr) and choline (Cho) to creatine (Cho/Cr) ratio. A spectral map was generated to show single spectra in relationship to their anatomical background and to allow for removal of those voxels of either insufficient signal intensity or lesion contamination. Altogether, 12,096 voxels were acquired and evaluated.
Fig. 1

Scout view for positioning the area of interest in multivoxel 1H-MRS (chemical shift imaging). a Axial T2-weighted MR image; the internal square (light blue) shows positioning for multivoxel spectroscopy. The smaller square (red) indicates the voxel actually evaluated. b Sagittal scout view, indicating VOI placement for multivoxel spectroscopy, lateral view


MTI measurements allowed for determining the average transfer ratio of lesions and WM. An off-resonance radiofrequency pulse with 1.5 kHz distance to the Larmor frequency of hydrogen, 16.4 ms duration, 850° flip angle and 768 Hz bandwidth was chosen. Measurements were programmed in a way that a gradient-echo sequence without off-resonance pulses was immediately followed by the one using off-resonance pulses.

Image post-processing

MS lesions were identified based on double-echo T2- and proton-density-weighted images, (Magic View, Siemens Medical Solutions, Erlangen, Germany) and quantified using an established software tool (AMIRA 3.01, Konrad Zuse Zentrum, Berlin, Germany). Lesions as well as NAWM were segmented using a threshold-based algorithm including a spatial connectivity paradigm and prior manual selection of regions of interest (ROIs) (Fig. 2). Subsequent to lesion masking, NAWM was segmented. Small areas that were not included in the first segmentation step could be included in a second automated segmentation procedure with limited maximum volume, in order to protect non-NAWM areas such as the insular region from inclusion. For MTR calculation purposes, MTI and double-echo studies were spatially co-registered.
Fig. 2

ac Segmentation of lesions and whole brain NAWM for MTI measure in various plains. Red areas indicate already demarcated MS lesions and are excluded form WM segmentation. d Surface-shaded three-dimensional view of WM volume, segmented on the basis of the algorithm shown in ac

Subsequently, the following parameters were determined in an automated manner: lesion count and volumes of T2-weighted studies, average lesion MTR and average MTR of whole brain WM. Altogether, 189 lesions were analysed throughout the study period. Average time consumption per session post-processing amounted to 120 min.

Clinical assessment

Each patient was assessed clinically on the MRI study day by an experienced neurologist; this included the Expanded Disability Status Scale (EDSS) and the Multiple Sclerosis Functional Composite (MSFC). EDSS quantifies disability in MS with respect to eight functional systems by allowing the neurologist to assign a Functional System Score (FSS) in each of these. EDSS steps 0 to 3.5 refer to patients with MS who are fully ambulatory, EDSS steps 4.0 to 9.5 are defined by the impairment to movement.

The EDSS test was supplemented by the Multiple Sclerosis Functional Composite (MSFC) test as this is able to monitor subtle changes in the disease course as well as to reflect cognitive impairment induced by MS [15, 16]. MSFC is a multidimensional clinical outcome measure that includes quantitative tests of leg function/ambulation [Timed 25-Foot Walk (TWT)], arm function [9-Hole Peg Test (9-HPT)], and cognitive function [Paced Auditory Serial Addition Test (PASAT)]. From these three tests, so-called z-scores are derived representing the differences between the patients performance compared with the cohort investigated.

Therapeutic regimen

Ten out of 17 patients completed the study period of 24 months. They were treated with either IFN β-1a 22 μg s.c. (n = 7), IFN β-1a 30 μg i.m. (n = 1), glatiramer acetate (n = 1) or remained untreated (n = 1).

After a baseline scan, therapy start was initiated by applying IFN β-1a 22 μg s.c. three times per week, by injecting IFN β-1a 30 µg once per week or by daily injections of glatiramer acetate.


Statistical evaluation as well as diagram generation was achieved by using SPSS 11.5 (SPSS, Chicago, Ill., USA). In order to assess cross-sectional data differences at initial diagnosis the Mann-Whitney U-test as well as the Spearman’s rank correlation coefficient (rs) were used. For longitudinal data evaluation, the non-parametric Friedman test and the Wilcoxon signed-rank test for repeated measurements on a single sample were used as well as the Spearman’s rank correlation coefficient.



Seventeen patients with RRMS were included into the study, ten of which could be followed up to 24 months (longitudinal study) subsequently. Seven patients terminated their study participation prematurely for personal reasons, such as relocation etc.

One out of ten patients in the longitudinal study part was excluded from data assessment because of disease conversion from relapsing-remitting to a secondary progressive course at 15 months.

Another three of the remaining nine patients in the 2-year follow-up had to be excluded from data analysis because of differences in medication; one of these three patients remained untreated and was also followed also over a period of 24 months.

This added up to a group of six patients treated with IFN β-1a 22 μg and monitored in monthly intervals for 1 year with a follow-up after 24 months.

Control group

MRI studies of all control subjects remained without abnormal findings at any time-point throughout the study period. For whole brain NAWM in the control group, a mean MTR of 41.3% was determined.

Cross-sectional study


Conventional MRI

All patients showed WM lesions on T2-weighted images with an average of 14.8 ± 11.8 lesions, 3.0 ± 3.9 ml in T2-weighted lesion load and 0.8 ± 1.3 ml in T1-weighted hypointense lesion volumes.

At the time of initial diagnosis, a broad range with respect to lesion count and volumes in T2- and T1-weighted studies was noted.

T2- and T1-weighted lesion volumes were correlated (rs = 0.871, p < 0.001). Lesion load and neurological scores (EDSS, MSFC) did not cohere significantly, whereas the z-score of 9-HPT, a MFSC subtest, did correlate with the T1-weighted lesion volumes (rs = -0.636, p < 0.05). Lesion count and volumes did not correlate with MRS nor MTI measurements.


Compared with healthy controls (NAA/Cr 2.07; Cho/Cr ratio 0.91), the NAA/Cr ratio in the periventricular WM and callosal body of patients was decreased to 1.89 ± 0.15 (p < 0.001) and the average Cho/Cr ratio was increased to 1.05 ± 0.09 (p < 0.001).

The average MTR of NAWM amounted to 38.51 ± 1.05%, average lesion MTR to 34.45 ± 3.09%. These results differed significantly from those of healthy controls (MTR 41.30 ± 0.74%; p < 0.001).

Neurological testing

A statistical trend was observed towards a correlation between MSFC and MRS (NAA/Cr ratio) (rs = 0.439, p < 0.058). A correlation was noted between 9-HPT, a subtest of MSFC, and MRS (Cho/Cr ratio) (rs = 0.675, p < 0.01).

EDSS and MSFC scores did not correlate with MTI study results nor with conventional MRI data.

Longitudinal measurements

All 17 control subjects completed the follow-up period of 24 months. In order not to obscure treatment effects by differences in medication or disease conversion, only RRMS patients with the standard medication of IFN β-1a 22 μg (n = 6) were included in the longitudinal data analysis. An overview of the longitudinal MRI study results is given in Table 2. The individual data profile of the untreated patient is shown in Fig. 3; at study termination a decrease of MTR as well as NAA/Cr ratio became apparent.
Fig. 3

Individual data profile of the untreated patient over a course of 24 months. Note the decline of whole-brain NAWM and lesion MTR as well as NAA/Cr ratio over time

Table 2

Mean NAA/Cr ratio and MTR values (%) in NAWM at baseline, 12 and 24 months after treatment initialisation (*p < 0.05)



Month 12

Month 24



1.93 ± 0.06

2.04 ± 0.06

2.06 ± 0.05*


2.17 ± 0.07

2.12 ± 0.06

2.15 ± 0.05

MTR (%) (whole-brain NAWM)


39.33 ± 0.67

40.76 ± 0.94

40.97 ± 1.09*


41.57 ± 0.80

41.58 ± 0.61

41.89 ± 0.71


Conventional MRI

Neither lesion counts nor lesion volumes changed significantly throughout the study period.


A predominantly continuous increase in WM NAA/Cr ratio throughout the entire follow up period was observable, although during the first 3 months a slight decrease in NAA/Cr ratio had been noted in three patients. At the time of study termination increases in NAA/Cr ratio (Fig. 4a) were encountered in all patients (p < 0.05). The Cho/Cr ratio decreased on average without gaining statistical significance.
Fig. 4

Longitudinal study results of NAA/Cr (a, b) as well as MTR measurements in NAWM (c, d) of RRMS patients (a, d) treated with IFN-β-1a 22 μg s.c. (n = 6) and healthy controls (b, d) (n = 17). At the time of study termination patients showed increases of NAA/Cr ratio (a) and MTR in NAWM (c) (p < 0.05), whereas no changes were delineated for healthy controls

Throughout the study period a continuous MTR increase in NAWM (Fig. 4c) was delineated (p < 0.05), whereas there were no corresponding changes seen in the control group (Fig. 4d).

Neurological testing

EDSS values decreased (p < 0.05) and MSFC figures increased (p < 0.05) during the follow-up period of 24 months in the patient group.

EDSS correlated with MTR in NAWM (p < 0.001) as well as within lesions (p < 0.05).

MSFC correlated with MTR of NAWM (p < 0.001) as well as with the T2-weighted lesion load (p < 0.01). At 24 months MSFC correlated with NAA/Cr ratio (p < 0.05), as well as did 9-HPT at 2, 12 and 24 months (p < 0.05). EDSS showed a relation to the Chol/Cr ratio at 24 months (p < 0.05). EDSS and MSFC scores did not correlate with conventional MRI data at any time-point.


Advanced MRI techniques, including MTI and proton MR spectroscopy have been shown to provide information about structural and biochemical alterations within and outside MS WM lesions [1618].

Whereas the MTR reflects important aspects of macromolecular changes during de- and remyelination in MS [19], MRS has proven useful in assessing neuronal damage, i.e. axonal loss, as NAA is exclusively localised in the neuronal compartment [20].

Recent surveys underline the potential of MTI and MRS in elucidating widespread tissue damage in NAWM at early MS stages, monitoring various aspects of inflammation in MS, characterising new, stable and resolved lesions, predicting the individual clinical course and monitoring treatment-related effects on myelin repair and neuroprotection [18, 2123].

Also, correlations between MRS/MTI and clinical disability scores in MS patients have been reported [2426].

These results strongly support the use of MTI and MRS in clinical MS studies.

In our investigation, MR spectroscopy revealed significant changes in patients’ NAA/Cr as well as in Cho/Cr ratios of the periventricular WM and callosal body in comparison to healthy controls, most probably reflecting different pathophysiological mechanisms such as axonal damage (NAA/Cr ratio) [27], and inflammatory cell membrane turnover (Cho/Cr) [6]. These results did not correlate with conventional lesion counts and volumes assessed on T1-/T2-weighted MR studies, indicating that advanced and conventional MR techniques, although both effective in differentiating patients and controls in our study, deliver different qualities of information in MS [28]. Dissociation between MRS findings indicative of axonal loss and T2 lesion volumes at early stages of the disease have also been encountered in other studies [29].

Diffusely increased Cho levels in NAWM of RRMS patients [30] as well as decreased NAA concentrations in patients with clinically isolated syndromes (CIS) suggestive of MS have been described [31], presumably representing inflammatory changes and significant axonal damage early in the disease course. MR spectroscopy, i.e. NAA levels, might be a prognostic marker for CIS patients with a higher risk of conversion into definite MS, as signs of early axonal damage at baseline was more prominent in this group of patients [26].

Neither lesion counts nor lesion volumes changed significantly throughout the study period, whereas MRS depicted a predominantly continuous and significant increase in WM NAA/Cr ratio. This is in line with two MRS studies of IFN β treatment in RRMS [9, 11]; however, similar studies by Pascual-Lozano et al. [8], and Sarchielli et al. [10] could not confirm these results. There is growing evidence that in MS inflammation is not only limited to lesions but also a diffuse, chronic low-level process affecting the entire brain. Therefore, the rise of NAA/Cr is discussed to be not only a sign of axonal repair but also the expression of resolving axonal dysfunction due to the anti-inflammatory effects of IFN β [9].

These findings are supported by clinical studies which implicated that IFN β may prevent and potentially reverse axonal injury [32].

Correlations between MRS findings and clinical scores have been reported in both cross-sectional and longitudinal studies [33]. In the cross-sectional part of our study, a statistical trend was observed towards a correlation between NAA/Cr ratio and MSFC (rs = 0.439, p < 0.058), in accordance with the findings reported by Adalsteinsson et al. [34], as well as well as a significant correlation between Cho/Cr ratio and 9-HPT, a subtest of MSFC (p < 0.01). Also in the course of 24 months MSFC correlated with NAA/Cr (p < 0.05). As there were no correlations between changes in conventional MRI parameters (T1 and T2 lesion count and volume) and disability, MRS seems to be a potential surrogate marker of disease progression.

Thus, advanced MRI techniques may be helpful in therapeutic decision making, e.g. with respect to early initiation of treatment with IFN β-1a as well as IFN β-1b, both of which were shown to be beneficial in counteracting disability development [3537].

MTI proved suitable for detecting macromolecular tissue damage in the WM of MS patients without lesional evidence in conventional MR studies [7]. Chen et al. [38] were able to validate the MTR changes in an MS lesion indicative of remyelination and demyelination by histopathological studies.

MTR changes in the NAWM of patients that correlated with clinical decline assessed by EDSS, MSFC and additional neuropsychological tests have recently been reported by other authors [25, 39, 40]. In our study, such correlations were only delineated during follow-up, not in the cross-sectional study part. The early stage of MS as well as a heterogeneous lesion distribution within our cross-sectional patient cohort might at least partially account for these differences. In addition, clinical alterations with respect to attention, information processing speed, memory, inhibition and conceptualisation which occur during early stages of RRMS [39] may not be addressed sensitively enough by the standard MS disability scores applied in this study.

The MTI findings of a continuous and significant MTR increase in NAWM support the assumption of brain parenchymal recovery. MTI provided the only imaging parameter (MTR) that correlated with the improvements seen in both clinical scores. Other studies report no change of MTR during treatment with IFN β [12]. However, this discrepancy could be due to the fact that their patients suffered from more severe disease state with high frequency of enhancing lesions. In addition, in our data the NAA/Cr ratio as a second independently derived parameter also showed an increase, implicating a restoration of axonal integrity.

Yet, in studies of untreated MS patients, the MTR of NAWM showed a slowly progressive decrease that started at disease onset—as we observed in the untreated patient of our investigation—and accelerated rapidly in focal areas just prior to lesion appearance on conventional MRI, probably reflecting different disease courses [41].

Limitations of our study are the relatively small number of patients fulfilling the 2-year longitudinal study. Additionally, grey matter (GM) changes have not been addressed in our study, although there is growing evidence for the concept that MS comprises a neurodegenerative component, with different patterns of atrophy evolution in GM and WM tissue compartments [42]. MT MRI assessment in GM has been shown to detect subtle brain tissue changes that are associated with mild clinical impairment as well as short-term disability accumulation in patients with RRMS [43, 44]. In addition, our data do not permit a statistically meaningful comparison between treated and untreated patients (only one patient remained untreated); yet, it may be speculated that the comparable MRS and MTR figures of the untreated patient and the healthy control subjects for the longest part of our study period suggest interindividual differences in susceptibility to acute inflammatory episodes in MS patients.


Our results suggest that advanced MR imaging techniques (MTI, MRS) perform better than conventional MRI in terms of detecting early parenchymal damage as well as reflecting the patients’ clinical status in relapsing remitting multiple sclerosis. Further studies with larger cohorts of patients are necessary to clarify the significance of these methods in monitoring therapeutic efficacy.

Copyright information

© European Society of Radiology 2009