Multisequence MRI in clinically isolated syndromes and the early development of MS

Methods.

Results.

We recruited 60 patients (mean age, 30 years; range, 16 to 49 years), 34 of whom were women and 26 of whom were men. EDSS score at baseline, as a result of their isolated syndrome, ranged from 0 to 8. The presenting symptom was optic neuritis in 38 patients (63%; acute unilateral in 36 patients and bilateral consecutive in 2 patients), a brainstem syndrome in 16 patients (27%), and a spinal cord syndrome in 6 patients (10%). Thirty of the patients are included in a previous preliminary report of baseline MRI findings alone.25

Discussion.

Our cohort contained a large proportion of patients with optic neuritis due to the high ascertainment rate achieved through the Physicians Clinic at Moorfields Eye Hospital. Earlier studies have suggested that in the general population the first site of the lesion is in the spinal cord in 50%, optic nerves in 25%, and brainstem in 15% of patients.26 However, because these conditions appear to be due to the same pathologic process, and the difference between the clinical presentations is determined by the location of a new lesion, the proportion of each type of condition should not affect the overall results.

We found 73% of patients with clinically isolated syndromes to have asymptomatic, disseminated, high-signal lesions on 3-mm-thick FSE brain images. This compares with 62% in the previous study performed at 0.5 T using CSE with 10-mm-thick images in our unit,3 and probably reflects the greater sensitivity of the later MRI technique. The studies of FSE brain lesions by region (see table 3) showed that lesions in the corpus callosum had the greatest positive predictive factor (57%) and a high specificity (84%) for the development of MS, confirming this as a characteristic location for plaques of demyelination. Discrete cerebral lesions also had a high specificity (70%) for MS, whereas subcortical/cortical and periventricular lesions did not. It is possible that nondemyelinating pathology is more common in the latter regions, although it may also emerge that with longer follow-up (as more patients develop clinical MS), the specificity of white matter lesions in all regions will increase. As others have described previously,27 a greater number of lesions on the baseline T2 image increases the specificity for early development of clinical MS, but at the expense of decreasing sensitivity.

fFLAIR is a heavily T2-weighted inversion-recovery technique that nullifies the CSF signal, enabling subtle periventricular or subcortical lesions to become more prominent, and increases the detection rate of MS lesions in some areas of the brain.14-16 In fact, in our cohort we found fFLAIR to be slightly less sensitive than FSE, both above and below the tentorium, but overall detected abnormalities in the same percentage of patients. The lower sensitivity of fFLAIR in the posterior cranial fossa is consistent with previous studies in established MS14-16; however, these same studies showed fFLAIR to be the more sensitive sequence in the supratentorial region. There could be several explanations for our different finding: First, we used thinner slices than previous studies (3 mm versus 5 mm), which may improve detection of smaller lesions more so on FSE than fFLAIR. Second, with the generally small lesion load in patients with clinically isolated syndromes, detection of subtle high-signal foci on FSE is likely to be more accurate. And third, the analysis of FSE and fFLAIR images side by side (unlike the independent analyses in previous studies) sometimes highlighted lesions that were subtle on FSE but more clearly seen on fFLAIR. However, the reverse was also true. The addition of fFLAIR did not reveal any lesions in patients judged to have a normal FSE study, and all patients who developed clinically definite or probable MS during follow-up had an abnormal FSE study at baseline. In light of these results, we suggest there is no need to perform fFLAIR in addition to thin-slice FSE in the investigation of patients with clinically isolated syndromes.

In the spinal cord, the combination of high-field (1.5-T), phased-array coils and FSE has improved markedly the detection of demyelinating lesions.17 Asymptomatic high-signal lesions were detected on FSE images of the spinal cord in 42% of patients. It has been revealed that some patients with MS can have multifocal MRI abnormalities compatible with demyelination in the spinal cord and still have a normal result on brain MRI.28 This was not seen in any of the patients in this study with clinically isolated syndromes. An important question is whether the combination of brain and spinal cord MRI findings will be able to assign a more definite prognosis for the future development of clinically definite MS. Our preliminary experience suggests that it may. After 1 year, 10 of 21 patients (48%) with abnormal brain and spinal cord MRI had developed MS compared with only 3 of 17 patients (18%) with abnormal brain images alone (see table 3).

The administration of Gd-DTPA produces enhancement on T1-weighted images if there is disruption of the blood–brain barrier, as is often the case in early inflammatory lesions.18,19 This enhancement continues for approximately 1 month,20 and so the combination of enhancing with older, nonenhancing lesions suggests that dissemination in time may also have occurred. For practical reasons it was not always possible to scan the patients within the first month of their symptoms. Fifty percent of the patients were scanned within 4 weeks of the onset of their symptoms, and 90% were scanned within 8 weeks. This delay may have resulted in an underdetection of enhancing lesions in the group. Nevertheless, a combination of Gd enhancing and nonenhancing brain lesions was seen in 50% of patients with abnormal brain FSE images at baseline. Of these patients, 11 of 22 (50%) were scanned in the first 4 weeks, and 86% within 8 weeks. Our results are in keeping with earlier reports of contrast enhancement in 46 to 59% of patients with abnormal imaging in comparable groups of patients.27,29,30 The presence of Gd enhancement at baseline had a high sensitivity (77%) and specificity (81%) for development of MS after 1 year. This finding supports that of Barkhof et al.27 that the presence of Gd enhancement in patients presenting with an isolated syndrome is one of the most predictive features of the early development of MS.

A subgroup of T2-weighted lesions can be seen as hypointense areas on T1-weighted MRI. There is evidence that these represent lesions with more severe tissue destruction, including axonal loss.21 They occur more frequently in the secondary progressive phase in MS, and one study demonstrated that the rate of accumulation correlated more strongly with progression of disability than did changes on T2-weighted imaging.22 Low-signal lesions were commonly seen on T1-weighted images at baseline in our cohort, and their presence was associated with a relatively high frequency of new lesions on FSE at the 3-month follow-up (16 of 23; 70%) and development of MS after 1 year (10 of 23; 43%). Long-term follow-up is needed to determine whether the presence of low-signal lesions early on is predictive of later outcome, especially the risk of developing secondary progressive MS and increasing disability.

The early follow-up imaging demonstrated new FSE lesions in 44% of patients, but new contrast enhancement was seen in only 55% of these patients. This ratio of activity contrasts with monthly studies in patients with established relapsing–remitting or secondary progressive MS: In such cohorts, new contrast-enhancing lesions are identified twice as often as new or enlarging lesions on T2-weighted images.31 The higher yield of new lesions on FSE images in our study is due to the fact that these lesions represent activity occurring over the whole 3-month period, whereas contrast-enhancing lesions represent activity likely to have occurred only within the last month. Also, new lesions on T2-weighted images are easier to detect because of the much smaller background lesion load in this group of patients compared with those with established MS.

One important, although admittedly much rarer condition, that can give MRI appearances identical to those seen in MS is the monophasic disorder ADEM. Two follow-up studies, with intervals ranging from 2 months to 10 years, of small ADEM cohorts using serial MRI have reported that partial resolution of lesions, without new lesion formation, is the rule.32,33 It seems appropriate, therefore, to seek MRI evidence of dissemination in time before suggesting a diagnosis of MS based on MRI findings in patients presenting with a clinically isolated syndrome. On FSE images, just less than half of our patients exhibited pathologic dissemination in time as well as in space within a 3-month period. Of these, more than half had developed clinically definite or probable MS at 12 months, and the sensitivity and specificity of this MRI profile for developing MS was especially high at 92% and 81% respectively. These data suggest that it may be of value to develop a new diagnostic category of MRI-supported MS, which could be defined on the basis of serial MRI after a suitable interval, in patients with a clinically isolated syndrome. Such a category might become relevant in selecting patients for trials of new drugs that aim to prevent the development of clinical MS.

Longer clinical follow-up of this cohort is necessary to define fully the prognostic value of the MRI sequences described in this study. Larger numbers of patients also need to be studied and other novel techniques should be applied, such as MRS, magnetization transfer imaging, diffusion tensor imaging, and quantitation of brain and spinal cord volumes. The predictive power both individually and in combination of MRI sequences should be evaluated to determine whether it is possible to identify, at the earliest stage, patients likely to experience future disability and in whom early therapeutic intervention will be most appropriate.