Serial gadolinium-enhanced MRI of the brain and spinal cord in early relapsing-remitting multiple sclerosis

Serial MRI studies of the brain have greatly enhanced our understanding of the natural history and pathophysiology of MS. Studies using either serial T_2-weighted ^[1-3] or T_1-weighted images after the administration of gadolinium-diethylenetriaminepenta-acetic acid (Gd-DTPA) ^[4-10] have demonstrated that the disease is often active even in the absence of clinical change; in patients with relapsing-remitting and secondary progressive MS, disease activity is found on MRI up to 10 times as frequently as clinical relapse. ^[1,8-11] Brain MRI has gained acceptance as a useful marker of disease activity in the context of clinical trials. ^[12-16] However, much of the symptomatology of MS implicates the spinal cord, a site not studied by MRI in comparable detail. Although previous studies have shown that it is possible to detect MS lesions within the spinal cord using MRI, ^[17-22] especially using multiarray receiver coils and fast spin echo (FSE) pulse sequences, ^[23] there are few published serial data. Weibe et al ^[24] studied 29 patients over 6 months, but only scanned them three times a month or during a clinical relapse. Capra et al ^[9] imaged 10 patients every 2 weeks for 3 months.

We carried out monthly Gd-DTPA-enhanced brain and spinal cord MRI scans over 1 year in 10 patients with relapsing-remitting MS to establish how frequently new spinal cord lesions appear and how often they are symptomatic; the frequency and duration of Gd-DTPA enhancement; and the relationship between spinal cord disease and brain disease. From this natural history data, we hope to establish the role (if any) of serial spinal cord imaging in treatment trials. We previously reported a parallel study of progressive MS ^[25].

Methods.

We studied 10 women, aged 22 to 45 years, with clinically definite or laboratory-supported definite MS, according to Poser et al.'s criteria. ^[26] Ethical approval was obtained for the study and all patients gave their written informed consent to participate.

At the outset of the study, a detailed history was obtained and a full neurologic examination was carried out. Disability was rated according to Kurtzke's functional system and expanded disability status scale (EDSS). ^[27] MRI and clinical examination were then performed monthly over 1 year. At each visit any new symptoms or signs were elicited and the Kurtzke scores were reassessed by a single observer (J.T.), who was blinded to the MRI results. A relapse was defined as the occurrence of a symptom of neurologic dysfunction lasting more than 24 hours. ^[26] Remission was defined as an improvement in signs or symptoms, or both, lasting for at least 1 month.

All imaging was performed on a 1.5-tesla Signa system (IGE Medical Systems, Milwaukee, WI). Brain MRI was carried out using a standard quadrature head coil. Pilot scans were obtained in axial, coronal, and sagittal planes to ensure adequate repositioning. Axial dual echo (proton density and T_2-weighted) fast spin echo (FSE_3500/18,90, echo train length of 8), and, 5 minutes after the administration of Gd-DTPA (Magnevist 0.1 mmol/kg, Schering), T_1-weighted spin echo (SE_575/19) images were obtained For all cranial images, contiguous interleaved 4-mm slices, one excitation, and a 192 times 256 matrix with 24-cm field of view (FOV) were used. Spinal imaging using a spinal multiarray receiver coil, was carried out immediately after brain MRI Sagittal T_1-weighted spin echo images (SE_500/19, 4-mm contiguous interleaved slices, 256 times 512 matrix, 24 times 48-cm rectangular FOV, two excitations) were first obtained, approximately 10 to 15 minutes after Gd-DTPA injection. Then, two sets of sagittal FSE images with different degrees of T₂ weighting were obtained, the parameters used were FSE_2500/51 and FSE_2500/102, echo train length of 16, 3-mm contiguous interleaved slices, 512 times 512 matrix, 48 times 48-cm FOV, two excitations. On the first and last visits only, axial proton density-weighted gradient echo images (GE_300/15, flip angle of 15 degrees, 5-mm slice thickness, 256 times 256 matrix, 20-cm FOV, four excitations) at four vertebral levels (C5, T2, T7, T11) were obtained for measurement of spinal cord atrophy Total examination time was approximately 1 hour.

At the end of the study one neuroradiologist (I.M.) reviewed all the brain images sequentially and another (B.K.) reviewed the spinal cord studies. Both were blinded to the clinical history. Lesions were counted on the initial T_2-weighted FSE images and a weighted lesion load was calculated by scoring lesions according to size: Table 3 Enhancement of lesions was recorded on all the T_1-weighted images; on the follow-up images it was noted whether enhancement was new or persisting from a preceding imaging study. The T_2-weighted images were scrutinized for any lesions that had appeared without showing enhancement on the T_1-weighted images, as well as for any lesions that had changed in size or disappeared. For the purpose of analysis, an active lesion was defined as one showing gadolinium enhancement on a T_1-weighted image or one that had newly appeared or enlarged on a T_2-weighted image in the absence of gadolinium enhancement. A set of images was considered active if it contained active lesions.

Measurements of spinal cord area from the gradient echo images were made manually, as previously described ^[23] on a Sun Workstation using image analysis soft-ware (D L Plummer, Department of Medical Physics, University College, London) Each set of images was measured twice by a single observer (J.T.) who could not tell to which patient they belonged nor whether they were from the beginning or end of the study.

Statistical comparisons were carried out using the chisquare test (chi (²)) with Yates' correction or, for small numbers (expected values less than 5), Fisher's exact test. Comparisons of medians were performed using the Mann-Whitney U test. Spearman's rank correlation was used to assess correlation between nonparamentric data.

Results.

Follow-up MRI.

One hundred eighteen follow-up examinations were carried out. Two patients showed no change in either brain or spinal cord MRI over the 1-year period. In the remaining eight patients, there were 144 new contrast-enhancing brain lesions (81 small, 28 medium, and 9 large) and nine new nonenhancing (5 small, 3 medium, and 1 large). Eleven new enhancing spinal cord lesions (7 small, 2 medium, and 2 large) were found (8 cervical and 3 thoracic), including two within a preexisting T₂ lesion, as well as seven new nonenhancing lesions (5 small and 2 medium), all within the thoracic spinal cord. None of the spinal cord lesions showed contrast enhancement on more than one monthly scan, whereas 23 brain lesions (9 small, 11 medium, and 2 large) showed enhancement on two consecutive examinations and four lesions (2 medium and 2 large) on three examinations. Although lesions that had enhanced were often smaller on subsequent T_2-weighted scans, no brain lesions completely disappeared. Four spinal cord lesions (1 small enhancing, 2 small, and 1 medium nonenhancing) became invisible on follow-up. No lesions became enlarged in the absence of contrast enhancement.

Overall, activity (new or persisting gadolinium enhancement or new or enlarging nonenhancing lesions) was more common on brain than spinal cord imaging: 57 of 128 of brain studies were active, compared with 15 of 128 spinal studies (chi (²) equals 32.48, p less than .001), contrast enhancement was found on 56 of 128 brain scans and 10 of 128 spinal cord scans. There was a clear relationship between the presence of activity in the brain and the spinal cord Figure 1. On the 113 occasions where there was no spinal cord activity, there was no brain activity on 69. Conversely, of the 15 occasions upon which there was spinal cord activity, brain activity was found on all but two scans. The difference was highly significant (chi (²) equals 10.36, p less than .002), and remained significant even when the data from the two patients who showed no activity were excluded (chi² equals 5.57, p less than.02).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Concordance of brain and spinal cord activity in a 27-year-old woman with a recent sensory relapse with deafferentation of the left hand. (A) T_2-weighted fast spin echo (FSE_2500/102) of the spinal cord. There is a large lesion between C2 and C4. (B) Gadolinium-diethylenetriaminepenta-acetic acid (Gd-DTPA)-enhanced T_1-weighted spin echo (SE_500/19) of the spinal cord. The lesion in the upper cervical cord shows patchy enhancement. (C) Gd-DTPA-enhanced T_1-weighted spin echo (SE_575/19) of the brain. There are several enhancing lesions.

Discussion.

In this study we attempted to characterize the dynamics of relapsing-remitting MS in the spinal cord using monthly Gd-DTPA-enhanced MRI. Of a total of 485 lesions detected on the baseline images, 42 (9%) were within the spinal cord. Overall, we saw 186 active lesions during the study period, 19 of which (10%) were within the spinal cord. Thus, if lesion count on brain MRI alone is used as a marker of disease activity in the context of a treatment trial in relapsing-remitting MS, approximately 10% of disease activity will be overlooked. Given the significant time penalty involved in spinal imaging, this is probably acceptable, especially since there seems to be no obvious pathophysiologic mechanism by which a treatment would have a different effect in the brain than in spinal cord. Furthermore, because there was spinal cord activity alone on only two of the 59 occasions on which MRI indicated any activity, the percentage of imaging studies revealing activity would fall very little (from 46 to 45%) if spinal MRI were omitted.

As found in many other studies, ^[1-11] there were many more active lesions in the brain than clinical relapses: Overall, there were 167 active brain lesions and only 11 relapses, a ratio of 15 to 1. Furthermore, only one active brain lesion was probably symptomatic. Conversely, at least six (31%) of the active spinal cord lesions produced symptoms, which is not surprising given the tight packing of critical ascending and descending white matter tracts within the spinal cord. Wiebe et al., ^[24] who studied 29 patients with relapsing-remitting and progressive MS with three MRI studies at three-month intervals, and additional imaging during clinical relapses, made similar observations. Of the 25 changes they detected on spinal cord imaging, 10 (40%) were symptomatic. Compared to the current study, however, they found relatively little asymptomatic brain disease. This is to be expected, because small lesions, which had ceased enhancing within 3 months, may have been overlooked on the unenhanced images. ^[29] In the study by Capra et al., ^[9] two of three contrast-enhancing spinal cord lesions (detected on biweekly imaging for 3 months in 10 patients) were symptomatic, compared with only one of 93 enhancing brain lesions.

New spinal cord lesions were less likely to enhance (61%), particularly in the thoracic region (30%), than new brain lesions (94%) and, unlike brain lesions, never demonstrated enhancement on more than one study. This is likely a result of the greater technical difficulty in producing high-quality, artifact-free T_1-weighted images of the spinal cord and the small size of most of the spinal cord lesions, rather than to biological differences in the permeability of the blood-brain and blood-spinal cord barriers in new lesions.

Brain MRI activity was much higher around the time of clinical relapse. This confirms the findings of other authors. ^[30-32] Of greater significance was the finding that brain activity was more common in the presence of spinal cord activity. The fact that new lesions tend to develop simultaneously at widely spaced anatomic sites can be used to argue strongly in favor of a generalized, systemic stimulus, rather than a local trigger to lesion formation, providing a further rationale for systemic treatment strategies to combat MS.

No patient developed further significant, fixed disability over the study period, apart from one patient probably just entering the progressive phase. There was no discernible progressive atrophy of the spinal cord over the study period, as measured from axial gradient echo images. This is compatible with a lack of significant axonal loss ^[23] and contrasts with the findings in progressive (both primary and secondary) MS where there was a significant decrease in the spinal cord area over time, in association with an increase in locomotor disability, despite few new developing lesions. ^[25] This lends credence to the notion that, following an initial relapsing-remitting phase in which there are recurrent episodes of acute perivascular inflammation associated with blood-brain (and blood-spinal cord) barrier breakdown and demyelination but relatively little axonal loss, there can follow a phase of increasing axonal loss (with resultant atrophy) in the absence of acute inflammatory episodes. Further longitudinal studies incorporating newer MR techniques that are sensitive to axonal loss or tissue disruption, such as proton spectroscopy ^[33,34] and magnetization transfer imaging, ^[35,36] are needed to clarify this important issue. Furthermore, future advances in imaging, for example, with improved design of multiarray coils, higher matrices, and implementation of three-dimensional FSE allowing for the acquisition of thin (1 mm) axial slices, are likely to improve the MRI resolution of the spinal cord. Further serial studies using such techniques will be needed to evaluate their sensitivity to pathologic changes and the correlation of disease activity with clinical events.