Abstract
Objective: To evaluate the contribution made by cervical cord damage, assessed using a fast short-tau inversion recovery (fast-STIR) sequence and magnetization transfer ratio (MTR) histogram analysis to the clinical manifestations of MS.
Background: Previous studies have failed to show significant correlations between the number and extent of T2 spinal cord lesions and the clinical status of patients with MS. Fast-STIR is more sensitive than T2-weighted imaging for detecting cervical cord MS lesions. MTR histogram analysis provides estimates of the overall disease burden in the cervical cord with higher pathologic specificity to the more destructive aspects of MS than T2-weighted scans.
Methods: We obtained fast-STIR and magnetization transfer (MT) scans from 96 patients with MS (52 with relapsing-remitting [RRMS], 33 with secondary progressive [SPMS], and 11 with primary progressive [PPMS] MS) and 21 control subjects. Dual-echo scans of the brain were also obtained and lesion load measured.
Results: Eighty-one of the patients with MS had an abnormal cervical cord scan. Patients with SPMS had more cervical cord lesions and more images with visible cervical cord damage than did patients with RRMS or PPMS (p = 0.04). The entire cohort of patients with MS had lower average MTR of the cervical cord (p = 0.006) than control subjects. Compared to control subjects, patients with RRMS had similar cervical cord MTR histogram-derived measures, whereas those with PPMS had lower average MTR (p = 0.01) and peak height (p = 0.02). Patients with SPMS had lower histogram peak height than did those with RRMS (p = 0.03). The peak position and height of the cervical cord MTR histogram were independent predictors of the probability of having locomotor disability. We found no correlation between brain T2 lesion load and any of the cervical cord MTR histogram metrics.
Conclusions: This study shows that the amount and severity of MS pathology in the cervical cord are greater in the progressive forms of the disease. An accurate assessment of cervical cord damage in MS gives information that can be used in part to explain the clinical manifestations of the disease.
The spinal cord frequently is involved in MS, with postmortem examination showing cord lesions in 86% of randomly selected patients with MS1 and MRI cord abnormalities seen in 47% to 90% of the patients studied.2-7 Although our ability to detect lesions in the spinal cord of patients with MS improved with the introduction of new MRI technology, particularly after the introduction of phased-array coils, previous studies3,5 unexpectedly failed to show significant correlations between the number and extent of spinal cord lesions and the clinical status of patients with MS.
There are at least three possible reasons for this discrepancy between clinical and MRI findings. First, some lesions may go undetected when imaging the cord with conventional T2-weighted sequences. Fast short-tau inversion recovery (fast-STIR) sequences give higher lesion contrast in the cervical cord2 and have been shown to detect on average 66% more cord abnormalities in MS than T2-weighted scans.7 Second, T2-weighted imaging lacks specificity to the heterogeneous pathologic substrates of MS lesions.8 Abnormalities seen on T2-weighted images reflect an increased water content, which can accompany edema, demyelination, remyelination, reactive gliosis, and axonal loss. Clearly, these different pathologic processes are likely to be associated with different neurologic deficits. Third, conventional T2-weighted imaging gives no information about the changes that are known to occur in the so-called normal-appearing white matter (NAWM) in patients with MS. A recent histologic study9 showed that such changes also include axonal loss and might, therefore, contribute to the development of disability. MTI is a promising technique that may help to overcome the two latter limitations.
MTI is based on the interactions between protons in a relatively free environment and those in which motion is restricted. In the nervous tissue, these two states correspond to the protons in tissue water and in the macromolecules of myelin and cell membranes. Off-resonance irradiation is applied, which saturates the magnetization of the less-mobile protons, but this is transferred to the mobile protons, thus reducing the signal intensity from the observable magnetization. The degree of signal loss depends on the density of the macromolecules in a given tissue. Thus, low magnetization transfer (MT) ratio (MTR) indicates a reduced capacity of the macromolecular protons to exchange magnetization with the surrounding water protons, reflecting damage to myelin or to the axonal membrane. In patients with MS, average brain lesion MTR10 correlates better with physical disability than does the volume of abnormalities on conventional T2-weighted MRI. In addition, it has been shown that in the brain, estimates of the amount and severity of microscopic and macroscopic disease burden can be obtained using MTR histograms,11 which may provide a more global picture of disease burden in MS. Brain MTR histogram-derived measures can easily distinguish patients with MS from control subjects11-14 and are correlated with the clinical manifestations of MS.12-14
The cervical area is where cord MS lesions are most commonly located,3,4,8,15 and MS-related changes in the cervical cord are likely to contribute significantly to the patients’ clinical status. Despite this, only one recent preliminary study has assessed MTR changes in the cervical cord from patients with MS; this showed that the mean MTR measured from a region in the cervical cord of 12 patients with MS was significantly lower than that from 12 age- and sex-matched healthy volunteers.16 The current study was performed in a large cohort of patients with MS to evaluate in the cervical cord 1) the presence and extent of macroscopic lesions using a fast-STIR sequence and 2) the overall disease burden (i.e., microscopic and macroscopic) using MTR histogram analysis. The ultimate aim of the study was to use MR metrics to assess the contribution that cervical cord damage makes to the clinical manifestations of MS.
Patients and methods.
Patients.
Patients included had had clinically definite MS17 for at least 2 years. They were classified as relapsing–remitting MS (RRMS; clearly defined disease relapses with either full recovery or sequelae but without disease progression during periods between the relapses), SPMS (initial relapsing-remitting course followed by progression with or without occasional relapses, minor remissions, or plateaux), or primary progressive MS (PPMS; disease progression from onset with occasional plateaux and temporary minor improvements) according to the criteria of Lublin and Reingold.18 None of the patients had undergone immunosuppressive or immunomodulating treatments for at least 12 months before entry to the study. In addition, they had neither relapses nor steroid treatment during the preceding 3 months. When MRI was performed, patients were assessed neurologically by a single physician, unaware of the MRI results, and disability was measured using the expanded disability status scale (EDSS).19 Sex- and age-matched control subjects with no history of neurologic diseases and with a normal neurologic examination were also studied. Local Ethical Committee approval and written informed consent from all the patients and control subjects were obtained before study initiation.
MRI.
MRI scans were obtained from all the patients and control subjects using a 1.5-T system (Vision; Siemens, Erlangen, Germany). With a tailored cervical spine-phased array coil for signal reception, the following pulse sequences were used:
-
1. Fast-STIR (repetition time [TR] = 2,288 msec, echo time [TE] = 60 msec, inversion time [TI] = 110 msec, echo train length = 11, field of view [FOV] = 280 × 280 mm, matrix size = 264 × 512, number of signal averages = 4). These acquisition parameters were chosen to match those suggested as optimal by previous studies.2,7
-
2. Two-dimensional gradient-echo (GE) (TR = 640 msec, TE = 10 msec, flip angle = 20°, FOV = 250 × 250 mm, matrix size = 192 × 256, number of signal averages = 2) both with and without a saturation pulse. The saturation pulse was an off-resonance radio frequency pulse centered 1.5 kHz below the water frequency with a Gaussian envelope of 7.68-msec duration and a flip angle of 500°.
For the fast-STIR, eight contiguous, interleaved sagittal slices were obtained with 3-mm-slice thickness and an interslice gap of 0.3 mm. For the GE scans, 20 contiguous, interleaved, axial slices with a thickness of 5 mm were obtained. From the two GE images, with and without the saturation pulse, MTR images were derived pixel-by-pixel according to the following equation: MTR = (M0-MS)/M0 × 100%, in which M0 is the signal intensity for a given pixel without the saturation pulse and MS is the signal intensity for the same pixel when the saturation pulse is applied.
In all the patients, we also obtained 24 contiguous, interleaved axial images of the brain using a dual-echo fast spin-echo sequence (TR = 3,300 msec, TE = 16/98 msec, echo train length = 5, slice thickness = 5 mm, FOV = 192 × 256 mm, matrix size = 188 × 250, number of acquisitions = 1) to estimate the brain T2 lesion burden.
Image review and quantification.
Lesions in the cervical cord were identified by agreement by two experienced observers, without knowing to whom the scans belonged, on the hard copies of the fast-STIR scans from all patients and control subjects. The extent of each lesion was measured by counting the number of slices where it was seen. A score for the total cervical cord damage was found for each patient by summing the extent of all the lesions seen for that patient.
Cervical cord MTR histograms were obtained for each of the patients with MS and control subjects as follows. First, the two GE images (i.e., with and without the MT saturation pulse) were coregistered. Registration was performed using an automated technique based on pixel similarity measures.20 Next, an MTR image was calculated according to the formula above from the registered GE images. Then, the entire cervical cord was segmented from the MTR images by a single observer, without knowing to whom the scans belonged, using a segmentation technique based on local thresholding.21 Figure 1 illustrates this process on a sample image, where it can be seen that the GE sequence generally is robust in the face of cord and CSF pulsatile movement with little motion artifact. Any slight patient movement between the two GE images is effectively corrected by the registration process. Finally, MTR histograms (with bins 1% in width) were created. We excluded from the analysis all the pixels with MTR values <10% to eliminate CSF. To correct for the between-patient differences in cord volume, each histogram was normalized by dividing it by the total number of pixels included. For each histogram, the following measures were derived: the relative peak height (i.e., proportion of pixels at the most common MTR value) and the peak position (i.e., the most common MTR). To take into account cervical cord atrophy, we also measured the absolute peak height (i.e., the peak height of the non-normalized MTR histograms). All histogram-derived measures were from the whole of the cervical cord tissue, thus including both MS lesions and normal-appearing cervical cord tissue. To calculate intraobserver coefficients of variation (COVs) for cervical cord MTR-derived measures, the cord MTR histograms from 10 randomly selected patients were created on a second occasion (separated from the first by an interval of at least 1 month) by the same observer, who was masked as to the results of the previous analysis using the same method described above.
Figure 1. Axial gradient-echo images of the cervical cord without (top left) and with (top right) the magnetization transfer pulse applied. The bottom right panel shows the corresponding magnetization transfer ratio (MTR) map obtained from the two previous images, and the bottom left is the segmented cord tissue that entered the MTR histogram analysis.
Hard copies of the proton density brain images from all patients and control subjects were also reviewed in random order by agreement between the same two observers, without knowing to whom the scans belonged, to identify hyperintense lesions. The T2-weighted images were always used to increase confidence in lesion identification. Lesion volume measurements were then performed by a single observer, again without knowing to whom the scans belonged. A semiautomated segmentation technique based on local thresholding was used for lesion segmentation21 using the marked hard copies as a reference. The intraobserver COV of this technique for measuring MRI abnormalities in MS is <5%.21
Statistical analysis.
The two-tailed Student’s t-test for unpaired data was used to compare MTR histogram-derived measures between control subjects and the entire cohort of patients with MS and between MS patients with and without cervical cord lesions. The Kruskal–Wallis test was used to compare the numbers of cervical cord lesions and fast-STIR slices showing lesions between the different MS phenotypes. Bonferroni correction was used to adjust for multiple comparisons. A one-way analysis of variance model with a priori contrasts was used to compare the cervical cord MTR histogram parameters from patients with different clinical phenotypes; this model was corrected for the number of pixels introduced into the analysis to minimize the effect of cord atrophy. The following a priori contrasts were chosen: control subjects versus RRMS, control subjects versus PPMS, and RRMS versus SPMS. The number of the a priori contrasts was determined by the available degrees of freedom, and their nature was chosen on the basis of the clinical evolution of the disease (i.e., MS onset is either as a RRMS or as PPMS, and RRMS can then evolve to SPMS). Univariate correlations were performed using the Spearman rank correlation coefficient. A multivariate logistic model was used to evaluate the effect of cervical cord MTR histogram measures (including the number of segmented pixels used to obtain MTR histograms), the number and extent of lesions in the cervical cord, and the brain lesion load on the probability that patients belonged to the groups with or without locomotor disability (i.e., EDSS ≥ 4.0 versus EDSS < 4.0). The same logistic regression model was applied to the subgroup, including only patients with RRMS and SPMS. The intraobserver variability of the cervical cord MTR histogram-derived parameters was assessed using a COV, which is defined as the SD of a random variable divided by its mean value. The standard errors (SE) of the COVs were estimated using the bootstrap resampling technique.
Results.
Clinical data.
Ninety-six patients (53 women and 43 men) were included in the study. Their mean age was 37.7 years (SD = 10.3), median disease duration was 7 years (range = 2–34 years), and median EDSS score was 2.5 (range = 0.0–7.5). Fifty-two patients were classified as having RRMS (mean age [SD] = 34.2 [8.6] years, median disease duration [range] = 4 [2–19] years, median EDSS score [range] = 1.0 [0.0–4.0]) 33 with SPMS (mean age [SD] = 43.4 [10.1] years, median disease duration [range] = 10 [4–34] years, median EDSS score [range] = 6.0 [3.0–7.5]), and 11 with PPMS (mean age [SD] = 37.4 [11.9] years, median disease duration [range] = 7 [2–15] years, median EDSS score [range] = 4.5 [2.0–6.5]). Twenty-one healthy volunteers (12 women and 9 men; mean age [SD] = 36.9 [7.2] years) served as control subjects.
MRI results.
No cervical cord abnormalities were found on the fast-STIR scans from control subjects. Eighty-one (84.4%) patients with MS had an abnormal cervical cord scan (41 patients with RRMS [78.8%], 31 with SPMS [94.0%], and 9 with PPMS [81.8%]). In the entire cohort of patients with MS, the mean number of cervical cord lesions was 2.0 (SD = 1.4) per patient, and the mean number of cervical cord slices showing lesions was 3.3 (SD = 2.5) per patient. The mean numbers of cervical cord lesions per patient were 1.7 (SD = 1.3) for RRMS, 2.5 (SD = 1.4) for SPMS, and 1.8 (SD = 1.2) for PPMS. The mean numbers of cervical cord slices showing lesions per patient were 2.7 (SD = 2.0) for RRMS, 4.4 (SD = 2.8) for SPMS, and 3.2 (SD = 2.7) for PPMS. Patients with SPMS had more cervical cord lesions (p = 0.04) and damaged cervical cord slices (p = 0.04) than did patients with the other two phenotypes.
The mean intraobserver COV was 1.3% (SE = 0.3%) for average cervical cord MTR, 4.4% (SE = 1.2%) for peak height, and 1.2% (SE = 0.3%) for peak position. In figure 2, the average MTR histograms from control subjects and the entire cohort of patients with MS are shown. Average cervical cord MTR was 45.8% (SD = 1.4%) for control subjects and 44.3% (SD = 2.3%) for patients with MS (p = 0.006). Mean absolute peak height was 112.4 (SD = 29.8) for control subjects and 94.0 (SD = 22.9) for patients with MS (p = 0.01). Mean relative peak height and location were also lower in patients with MS than in control subjects, but these differences were not statistically significant. In the table, the measures derived from the normalized MTR histogram of control subjects and each MS phenotype studied are reported. The results of the a priori contrast analysis showed no significant differences between the cervical cord MTR histogram-derived measures from control subjects and patients with RRMS, whereas patients with PPMS had lower average MTR (p = 0.01) and relative peak height (p = 0.02) than did control subjects. Patients with SPMS had lower relative peak height than did those with RRMS (p = 0.03).
Figure 2. Magnetization transfer ratio histograms of the cervical cord from control subjects (black line) and the entire MS cohort (gray line).
Cervical cord magnetization transfer ratio (MTR) histogram metrics in the control subjects and the different patient groups
No significant differences were found in the MTR histogram-derived measures between patients with and without lesions visible on the fast-STIR scans (data not shown). There was no significant correlation between the number or extent of fast-STIR lesions and any of the cervical cord MTR histogram metrics. In the entire cohort of patients with MS, the median brain lesion volume was 9.7 mL (range, 0.2–69.5 mL). There was no significant correlation between the brain lesion volume and the number of fast-STIR lesions or any of the MTR-derived measures, whereas a moderate correlation (r = 0.22, p = 0.04) was found between brain lesion volume and the number of fast-STIR slices showing lesions.
The multivariate regression model showed that the peak position (odds ratio [OR] = 0.81, p = 0.02) and the peak height (OR = 0.95, p = 0.03) of the cervical cord MTR histogram, the number of segmented pixels (OR = 0.99, p = 0.04), the number of damaged cervical cord slices (OR = 1.30, p = 0.03), and the brain lesion volume (OR = 1.05, p = 0.02) predicted the likelihood of patients belonging to the group with locomotor disability. After the patients with PPMS were removed from the analysis, the probability of patients with RRMS and SPMS having locomotor disability was predicted by the peak position of the cervical cord MTR histogram, the number of damaged cervical cord slices, and the brain lesion volume.
Discussion.
MS commonly affects the spinal cord,1-7 and it is likely that such damage contributes to the clinical manifestations of the disease. However, previous studies3,5 have failed to show strong correlations between the cord lesion load, as measured on conventional T2-weighted scans, and clinical disability. Of the several possible factors that could lead to such poor correlations, we believe that there are three important ones. First, a substantial number of cord lesions may be missed when using conventional T2-weighted scans. Fast-STIR sequences improve the conspicuity of cord lesions2 and, as a consequence, enable more lesions to be seen.7 Second, it is likely that even the cord tissue not involved with macroscopic MS lesions might be damaged by microscopic abnormalities, as is the case in the brain.22-24 Measures of damage based on MTR histograms encompass both the macroscopic and microscopic aspects of MS pathology in the brain but have never been used to assess cord pathology. Third, MS abnormalities as measured on T2-weighted scans are pathologically aspecific8 and do not separately quantify demyelination and axonal loss, which are likely to be the most disabling aspects of MS. This is not the case for reduced MTR, which has been shown to correlate more strictly with severe axonal loss and demyelination in a postmortem study.25 Animal studies have also shown a correlation between low MTR values and histopathologic findings of myelin loss and axon destruction,26-28 while edematous lesions resulted in slightly increased MTR values.22 Dramatically reduced MTR values are also found in the “pure” demyelinating lesions of patients with progressive multifocal leukoencephalopathy29 or central pontine myelinolysis.30 Improving MS lesion detection in spinal cord MRI and giving better specificity to the heterogeneous aspects of MS pathology may lead to a better understanding of the clinical manifestations of the disease.
This study shows that the amount and severity of MS pathology within the cervical cord are relevant factors in the clinical manifestations of the disease. First, we found that the average MTR of the whole cervical cord tissue is significantly different from that in control subjects. This confirms a previous preliminary study in which MTR was measured in a relatively small region of interest in the cervical cord.16 Second, we saw more lesions, and with a greater extent, in patients with SPMS than in those with RRMS or PPMS. In addition, the cervical cord MTR histogram measures were lower in more disabled patients with the progressive forms of the disease. This is contrary to the lack of correlation between cord MTR measurements and disability found by Silver et al.,16 a discrepancy that might be explained by the limitations of the previous study.16 In the work by Silver et al.,16 only 12 patients were assessed, and a relatively small and variable amount of cervical cord tissue analyzed (between 114 and 121 mm2). Admittedly, we also limited our investigation to the cervical cord, since MTI of the entire cord would have been extremely time-consuming. However, MS-related damage3,4,6,15 is more frequent in the cervical cord than in any other cord region and, therefore, we believe that our findings illustrate the most important parts of MS cord pathology.
An EDSS score ≥ 4.0 indicates that a patient has a limited ability to walk, whereas lower scores are not related to disability but to neurologic impairment in one or more of the EDSS functional systems.19 Because cervical cord damage is likely to have an impact on locomotor capabilities, we chose to assess whether different MRI metrics were relevant factors in such disability. We showed that measures of cervical cord damage are useful in predicting the risk of patients with MS having locomotor disability and that such measures contain information that is independent from the extent of T2 lesions in the brain. This suggests a role for measures derived from cervical cord MRI as surrogate markers in monitoring treatment of the progressive forms of the disease.
Previous studies have shown that cervical cord atrophy, derived from measures of cord cross-sectional area at C2 or C5, frequently is seen in patients with MS and is correlated with disability.31,32 This is confirmed by the current study, since we found that the peak height of the non-normalized histograms was lower in patients with MS than in control subjects and that the number of segmented pixels used to create MTR histograms was an independent predictor of the presence of locomotor disability. When atrophy is present, it is more likely that pixels at the edge of the cord would include a contribution from CSF in the MTR histograms. This would inevitably result in a reduction in all the MTR histogram-derived measures. However, in the current study, we ran the a priori contrast analysis and the multivariate regression analyses, correcting for the number of pixels used to create MTR histograms. Although the number of segmented pixels in the MTR maps is not the standard approach for measuring spinal cord size, we believe that with such an approach, we minimized the effect of atrophy on our results. We also excluded from the analysis pixels with an MTR lower than 10%, thus removing those pixels most severely affected by partial-volume averaging with CSF.
Despite lesions being visible on the fast-STIR scans, patients with RRMS had MTR histogram-derived measures similar to those from control subjects. This suggests that the pathologic changes occurring in cervical cord lesions and NAWM are mild in patients with early, mildly disabling RRMS. In SPMS, we found a deterioration of the MTR histogram metrics and an increase in the number of lesions seen on fast-STIR slices and the number of slices showing damage. This suggests that the extent and severity of cervical cord damage are important in determining the evolution from RRMS to SPMS. However, because this is a cross-sectional study, caution must be exercised before drawing firm conclusions about the evolution of MS.
Patients with PPMS have cord MTR histogram-derived measures significantly lower than those from control subjects. This provides further evidence that cervical cord damage is important in the progressive accumulation of disability in MS. Interestingly, the cord MTR histogram-derived metrics from patients with PPMS were similar to those from patients with SPMS, even though less damage is apparent on the fast-STIR scans of patients with PPMS. This may result from two factors, which are not mutually exclusive: more severe damage in the lesions or more widespread damage within the NAWM. Unfortunately, global MTR histogram analysis is limited by its inability to show the site of the abnormalities and, therefore, it cannot tell us the relative contributions of these two factors. Nevertheless, it would be both interesting and challenging to obtain separate MTR measures for MS lesions and NAWM in the spinal cord. However, both MTI14 and MRS33 studies of the brain have already suggested that widespread NAWM damage is important in patients with PPMS, and diffuse cord abnormalities have been shown on their T2-weighted scans.6
We found no correlation between cord MTR histogram-derived measures and the extent of macroscopic abnormalities in the brain and cord, suggesting that cervical cord damage in MS is not merely a result of Wallerian degeneration secondary to brain damage. This is particularly apparent in patients with PPMS who have the lowest cord MTR histogram-derived metrics and yet very low brain lesion loads.34 A more strict relationship between brain and cord changes might have been obtained by a more pathologically specific assessment of tissue damage in the brain using T1-weighted35 or MTI10,12-14,24 sequences. However, the high correlation between brain T2 lesion load and hypointense T1 lesion load35 or MTR-derived metrics36 makes it unlikely that the strength of correlation we found would have changed dramatically.
The absence of a correlation between macroscopic lesion load and cervical cord MTR histogram metrics, together with similarity of cord MTR histograms in patients with and without cord lesions visible on fast-STIR scans, suggests that NAWM changes in the cervical cord may be as relevant as those in the brain37 in determining the clinical manifestations of MS.
Acknowledgments
Supported by a grant from TEVA Italy (M.P.S.) and a grant from the Neurology School of the University of Chieti (G.I.).
Acknowledgment
The authors thank C. Pereira for his assistance in performing MRI scans.
- Received April 16, 1999.
- Accepted August 9, 1999.
References
Letters: Rapid online correspondence
REQUIREMENTS
You must ensure that your Disclosures have been updated within the previous six months. Please go to our Submission Site to add or update your Disclosure information.
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
You May Also be Interested in
Alert Me
Recommended articles
A magnetization transfer histogram study of normal-appearing brain tissue in MS
MRI and magnetization transfer imaging changes in the brain and cervical cord of patients with Devic’s neuromyelitis optica
Comparison of MS clinical phenotypes using conventional and magnetization transfer MRI
Pathologic damage in MS assessed by diffusion-weighted and magnetization transfer MRI