Multiple sclerosis (MS) has a highly variable disease course1 and knowledge of factors that predict subsequent disease course in individual patients with a first event suggestive of MS (also called patients with a clinically isolated syndrome: CIS) is limited.

In CIS patients, magnetic resonance imaging (MRI) characteristics have been described as a predictor of conversion to clinically definite MS (CDMS) and of the subsequent disease course. CIS patients with an abnormal cerebral MRI scan at presentation have a substantially higher long-term risk of conversion to CDMS than those with a normal cerebral MRI2. Diagnostic guidelines for MS include detailed MRI rules for the definition of dissemination in space of MS-specific pathology.34 In untreated5 and treated6 CIS patients fulfillment of these criteria is associated with a high risk of CDMS.

According to the recommendations of the International Panel (IP) on the Diagnosis of Multiple Sclerosis, disease dissemination in CIS patients can also be identified by clinical examination of symptoms and signs at the first clinical event.3 In contrast to a detailed algorithm on the use of MRI criteria, however, it was left unclear how clinical disease dissemination should be evaluated. Recently it was shown that the clinical assessment of disease dissemination can vary widely between physicians.7 To standardize these assessments, a clinical classification system was proposed.7 This system was centrally applied to patients of the BEtaferon^®/BEtaseron^® in Newly Emerging multiple sclerosis For Initial Treatment (BENEFIT) study, a study evaluating the impact of interferon beta-1b (IFNB-1b) in CIS patients. By analyzing baseline data from this study we have recently shown that patients with clinical dissemination in space (multifocal, indicating more than one clinical lesion) had more lesions on their MRI than monofocal patients (those exhibiting symptoms and signs from only one clinical lesion).8

In the present study we assessed the prognostic value of this clinical classification system7 for conversion to CDMS and the added value of potentially prognostic MRI parameters, by analyzing data obtained during the placebo-controlled treatment period of the BENEFIT study.

Four hundred and eighty-seven patients were randomized and 468 started treatment in the BENEFIT study. Four hundred and thirty-seven of these (93.6%) completed the placebo-controlled study. Two hundred and ninety-two patients received IFNB-1b and 176 received placebo. The main outcome data have previously been published9.

In a cross-sectional analysis of baseline data from CIS patients in the BENEFIT study we have recently demonstrated that clinical dissemination of the disease corresponds to more widespread subclinical CNS pathology as detected by cerebral MRI8. In the present study we addressed whether clinical disease dissemination in these patients also indicates an increased risk for subsequent disease activity, and whether the presence versus absence of clinical dissemination has an impact on the prognostic value of MRI parameters.

Patients with monofocal versus multifocal clinical presentation did not differ in terms of their risk for CDMS or with respect to the annualized number of NALs over the 2-year placebo-controlled period, as has been reported previously 11. However, we did find that MRI findings of subclinical disease dissemination or activity have a different prognostic value for development of CDMS in mono- versus multifocal CIS patients. The presence of at least nine T2 lesions or at least one Gd-enhancing lesion during screening was predictive for time to CDMS in monofocal patients though not in multifocal patients. Similar observations were made for the prognostic value of a new Gd-enhancing lesion on an MRI scan performed at month 3 or month 6. Thus, in monofocal, but not in multifocal patients the risk for CDMS depends on MRI findings. This differential impact of MRI findings in CIS patients with clinical monofocal versus clinical multifocal presentation was supported by a significant interaction between the impact of clinical mono-/multifocality and ≥ 9 T2 lesions at baseline and at least one Gd-enhancing lesion at month 3 on time to CDMS.

These findings strongly suggest that only in CIS patients with monofocal clinical presentation do MRI findings have prognostic value. We hypothesize that, whilst in monofocal CIS patients more pronounced subclinical disease dissemination might primarily reflect more active disease, similar findings in multifocal patients may be more indicative of prolonged subclinical disease evolution, and as such MRI adds less information in these patients.

To further elaborate on this hypothesis we expanded these comparisons to subgroups of multifocal patients: those by symptoms and those by signs, under the assumption that especially those patients multifocal by signs may have had an earlier event that was asymptomatic or forgotten, and therefore may have a longer and more benign form of the disease. We found similar baseline characteristics and only a nonsignificant difference in time to CDMS in these subgroups. The observation that multifocal placebo patients by symptoms tended to have more active MRI lesions during the study than multifocal placebo patients by signs further supports our hypothesis that the former may be considered more acute and at higher risk for future disease activity than a multifocal patient by signs in whom a longer subclinical disease history might be assumed. Differences with respect to patients showing at least one Gd-enhancing lesion (more in patients multifocal by symptoms) and patients showing at least one T1-hypointense lesion (more in patients multifocal by signs) as shown in Table 3, although not significant, are also supportive of our hypothesis.

We compared our results to those obtained in the Early Treatment of MS (ETOMS)12 and Controlled High-Risk Avonex^® Multiple Sclerosis Prevention Study (CHAMPS)13 studies, other interventional trials in CIS patients where comparable analyses were performed. In the ETOMS study, the presence of three or more MRI criteria as incorporated in the International Panel on the Diagnosis of Multiple Sclerosis guidelines3 was also predictive for CDMS only in patients who were classified as clinically unifocal6. Unlike our observation, multifocal patients in ETOMS had a higher risk for CDMS. This difference in the predictive value of "multifocality" in the BENEFIT and the ETOMS cohort may result from the different methods used to classify patients in the two studies. Evaluation of clinical dissemination in ETOMS was based on the local investigator's assessment, whilst multifocality in BENEFIT was based on a central assessment procedure according to a proposed classification system of all presenting clinical symptoms and signs. Thus, multifocality in BENEFIT was also assumed in patients who, in addition to a monosymptomatic presentation [e.g. optic neuritis], presented with additional clinical signs (e.g. pyramidal dysfunction as indicated by extensor plantar response) indicating an additional clinical lesion. Also, in a post hoc analysis of the CHAMPS study14 all patients were reclassified, taking into account the results of neurological examinations at baseline. In a multivariate analysis, classification by focality was not predictive of conversion to CDMS, which is in line with our results.

All patients in BENEFIT, ETOMS, and CHAMPS had a minimum number of asymptomatic T2 lesions; therefore it is unclear whether these results also can be applied to CIS patients with fewer or no lesions. Further limitations of our analyses should be considered. All subgroup analyses were performed post hoc and our results need confirmation, particularly the novel findings in the subgroups of multifocal CIS patients. However, we would like to emphasize the similarities between our findings and the ETOMS study6 in terms of the lack of impact of MRI findings in multifocal patients on the risk of CDMS.

To summarize, MRI lesions may generally be interpreted as indicators of past and future disease activity in patients with monofocal presentation, though not in multifocal patients, in whom their presence does not add to the risk as defined by the clinical evaluation only. Our findings show that a carefully performed neurological assessment of symptoms and signs in CIS patients is important to define the risk of conversion to CDMS and the potential added value of MRI investigations.

This study was sponsored by Bayer Schering Pharma AG, Berlin, Germany. Dr Nielsen has nothing to disclose. Dr Pohl has received personal compensation for activities with Schering AG as an employee. Consultancy for Schering, Aventis, UCB, Roche, Serono, Novartis. Dr Freedman has received personal compensation for activities with Bayer-Schering Pharmaceuticals, Merck-Serono, Pfizer Inc, Teva Neuroscience, Biogen Idec, Genentech, Inc. and BioMS as a consultant and advisory board member. Dr Edan has nothing to disclose. Prof Miller has received personal compensation for activities with Biogen Idec, GlaxoSmithKline, Inc., and Schering AG as a consultant. Prof Miller has received personal compensation in an editorial capacity for Journal of Neurology. Prof Miller has received research support from Biogen Idec., GlaxoSmithKline, Inc., and Schering AG. Dr Kappos has nothing to disclose. Dr Bauer has received personal compensation for activities with Schering AG as an employee. Dr Rupert Sandbrink received personal compensation from Schering AG as a salaried employee of this company. Prof Polman has received consulting fees from Biogen Idec, Schering AG, Teva, Serono, Novartis, GlaxoSmithKline, Novartis, and Teva. In the past year, dr. Uitdehaag has received compensation for consultancy from Novartis and Merck Serono. In the past year, dr. Uitdehaag received personal compensation from Ariez Medical Publishing for serving as a jounal editor. The institute for which dr Uitdehaag works received financieal support for research activities from Biogen Idec, Bayer Schering Pharma, GlaxoSmithKline, Novartis, Merck Serono and Teva.

JN drafted the manuscript, performed the statistical analysis and participated in the design of the study, CP participated in the design and coordination of the study, drafted the manuscript and performed the statistical analysis, CHP participated in the design and coordination of the study, and drafted the manuscript, FB commented on the manuscript and participated in the design and coordination of the study, MF commented on the manuscript and participated in the design and coordination of the study, GE commented on the manuscript and participated in the design and coordination of the study, DM commented on the manuscript and participated in the design and coordination of the study, LB commented on the manuscript and participated in the design and coordination of the study, RS commented on the manuscript and participated in the design and coordination of the study, LK commented on the manuscript and participated in the design and coordination of the study, BU drafted the manuscript, performed the statistical analysis and participated in the design of the study. All authors read and approved the final manuscript.