A Single, Early Magnetic Resonance Imaging Study in the Diagnosis of Multiple Sclerosis

Background  A diagnosis of multiple sclerosis in patients who present for the first time with a clinically isolated syndrome (CIS) can be established with brain magnetic resonance imaging (MRI) if the MRI demonstrates demyelinating lesions with dissemination in space (DIS) and dissemination in time (DIT).

Objective  To investigate the diagnostic performance of a single MRI study obtained within the first 3 months after symptom onset in a cohort of patients with a CIS suggestive of multiple sclerosis at presentation.

Design  Multicenter inception cohort with a follow-up of at least 24 months.

Setting  Referral hospitals.

Patients  Patients with CIS onset between April 1, 1995, and September 30, 2004, who fulfilled the following criteria were included: (1) age of 14 to 50 years and (2) clinical follow-up for at least 24 months after CIS onset or until development of clinically definite multiple sclerosis (CDMS), if this occurred within 2 years.

Main Outcome Measure  All patients underwent 2 comparable brain MRI examinations, the first within 3 months (early) and the second between 3 and 12 months (delayed) after CIS onset. We defined DIS using several existing MRI criteria, and DIT was inferred when there were simultaneous gadolinium-enhancing and nonenhancing lesions on a single MRI.

Results  Two hundred fifty patients were included in the study. The comparison of the diagnostic performance of various MRI criteria for identifying early converters to CDMS showed similar sensitivity and specificity between early and delayed MRIs. In addition, the use of less stringent criteria for DIS yielded better sensitivity and similar specificity, particularly when assessed in the first weeks after CIS onset.

Conclusion  A single brain MRI study that demonstrates DIS and shows both gadolinium-enhancing and nonenhancing lesions that suggest DIT is highly specific for predicting the early development of CDMS, even when the MRI is performed within the first 3 months after the onset of a CIS.

According to the McDonald criteria, a diagnosis of multiple sclerosis (MS) in patients who present for the first time with a clinically isolated syndrome (CIS) can be established with brain magnetic resonance imaging (MRI) if the MRI demonstrates demyelinating lesions with dissemination in space (DIS) and dissemination in time (DIT).^1,2 To demonstrate DIS, brain MRIs must fulfill the Barkhof-Tintoré criteria,^3,4 whereas DIT requires the demonstration of new T2 lesions compared with a reference MRI obtained at least 1 month after symptom onset or when an asymptomatic gadolinium-enhancing lesion is present on an MRI performed at least 3 months after onset.²

In 2006, Swanton et al⁵ proposed new MRI criteria for MS in which DIS is achieved by the detection of 1 or more subclinical T2 lesions in at least 2 of 4 locations considered characteristic for MS in the McDonald criteria (juxtacortical, periventricular, infratentorial, and spinal cord), and DIT requires a new T2 lesion on a delayed image, regardless of the timing of the early image. Compared with the McDonald criteria, these new MRI criteria are similarly (highly) specific for clinically definite MS (CDMS), more sensitive,⁶and based on the findings of T2-weighted images alone, without the need for gadolinium enhancement. However, because the development of a new T2 lesion is mandatory for demonstrating DIT, at least 2 MRIs are always needed to establish the MRI-based diagnosis of MS, although a minimum interval between symptom onset and the MRI study is not required.⁷

In patients with a CIS, gadolinium-enhancing lesions are often seen in association with a larger number of nonenhancing lesions, and this combination likely reflects a set of lesions in different stages of evolution. Hence, the presence of simultaneously enhancing and nonenhancing lesions on a single MRI obtained in a patient with a CIS could be a criterion to indicate a DIT process, regardless of the point when the MRI was obtained after CIS onset. If a single MRI acquired soon after the CIS event could be used to establish not only the required criteria for DIS but also those for DIT, there should be a favorable effect on health care cost and patient management: a prompt diagnosis would facilitate informed, supportive counseling for patients that includes discussion of the prognosis and potential treatment options. The purpose of this study was to investigate the diagnostic performance of a single MRI study obtained within the first 3 months after symptom onset in a large cohort of patients with a CIS.

Within a European, multicenter, collaborative research network dedicated to MRI research in MS (MAGnetic Resonance Imaging in MS [MAGNIMS]; http://www.magnims.eu/), MRI and clinical data were retrospectively collected from 5 European centers in which these data had been prospectively obtained from patients with a CIS following local standards (in all patients, the CIS onset was between April 1, 1995, and September 30, 2004). Each center was asked to identify patients who met the following inclusion criteria: (1) CIS suggestive of central nervous system demyelination involving the optic nerve, brainstem, spinal cord, or other area not attributable to other diseases; (2) age of 14 to 50 years; (3) clinical assessment within 3 months of CIS onset; (4) clinical follow-up of at least 24 months after CIS onset or until development of CDMS according to the Poser criteria,⁸ if this occurred within 2 years; (5) availability of 2 comparable brain MRI examinations, the first obtained within 3 months and the second between 3 and 12 months after CIS onset; and (6) no use of disease-modifying treatment before the last MRI study was performed.

Each center provided the following clinical information: age at CIS onset, sex, area and date of CIS (and second event if applicable), and dates of MRIs and last follow-up visit. The diagnosis of CDMS was established by an experienced neurologist at each center and was based on clinical information alone, according to the Poser criteria⁸ (ie, symptomatic and objective evidence of a second neurologic episode attributable to demyelination of >24 hours' duration >4 weeks after the initial attack). Alternative diagnoses were ruled out by appropriate tests according to the local protocols.

Ethics approval for the CIS studies at each center was obtained from the local institutional review boards. Because the present study was based on retrospective analysis of existing data, no additional approval was required.

The MRI studies were performed on 1.5-T (97%) or 1.0-T (3%) scanners and consisted of axial, fast spin-echo, proton density and T2-weighted, and postcontrast T1-weighted spin-echo (scan delay, 5-15 minutes) brain images. Section thickness varied from 3 mm (no gap) in 78.0% of patients to 5 mm (gap of 0%-10%) in the remaining 22.0% of patients, with a total of 24 to 44 sections that encompassed the entire brain. The acquisition scheme and contrast dose were kept constant in each patient for the follow-up MRIs, and internal landmarks were used to ensure that section positions were identical to those of the baseline MRI.

Spinal cord MRIs (performed in approximately half of the patients included in the study) were not used for the analysis because they were obtained at 1 time only and could not contribute to the study objective (ie, the comparison of diagnostic performance between 2 MRIs performed at different times after CIS onset). Areas of increased signal intensity present on T2-weighted images and proton density images were marked by consensus by 2 readers (E.H. and J.S.-G.). Both readers had been trained by experienced neuroradiologists (A.R. and K.M.) to identify and classify these lesions. The number and area of T2 and gadolinium-enhancing lesions were assessed.

Brain lesions were classified as periventricular, subcortical, juxtacortical, or infratentorial. Gadolinium-enhancing lesions identified on T1-weighted images were confirmed on proton density and T2-weighted images. Brain images were reported as normal when they showed no subclinical lesions consistent with demyelination (ie, brainstem lesions were not considered in the analysis in patients who presented with a brainstem syndrome).

Three sets of MRI-based criteria were used to demonstrate DIS: Barkhof-Tintoré,^3,4 Swanton,⁵ and Frohman⁹ (Table 1). The requirement of at least 1 asymptomatic gadolinium-enhancing lesion, in conjunction with 1 or more nonenhancing lesions, was used to demonstrate DIT. The various criteria used to demonstrate DIS were selected according to differences in their diagnostic performance in previous studies,^4,6,9 which indicated that the Barkhof-Tintoré criteria are more stringent than the Swanton and Frohman criteria. Early and delayed MRIs were classified as positive if they met the MRI criteria of DIS and DIT.

All data were entered into a database by a single person (E.H.) and checked for accuracy and consistency by confirmation of the validity of all entries. The value of imaging findings in early and delayed MRIs with regard to conversion to CDMS was expressed as sensitivity, specificity, positive predictive value, and negative predictive value. A true-positive result was defined as positive MRI findings and conversion to CDMS, a false-positive result as positive MRI findings in the absence of conversion to CDMS, a false-negative result as negative MRI findings but conversion to CDMS, and a true-negative result as negative MRI findings and no conversion to CDMS. The likelihood ratio for a positive test result was defined as sensitivity/(1 − specificity), and the likelihood ratio for a negative test result as (1 − sensitivity)/specificity. Statistical analyses were performed with commercially available software programs (SPSS, version 13.0 for Windows; SPSS Inc, Chicago, Illinois; and CIA software, version 2.1.1; Trevor Bryant, PhD, University of Southampton, Southampton, England).

A total of 250 patients were included in the study. The demographic characteristics of the cohort are given in Table 2 Early MRIs were obtained after a median of 32 days (range, 3-89 days) and delayed MRIs after a median of 141 days (range, 92-365 days) after CIS onset. The median interval between early and delayed MRIs was 98 days (range, 39-360 days). Early MRIs were obtained within the first 30 days after CIS onset in 119 patients (47.6%), between 31 and 60 days in 73 patients (29.2%), and between 61 and 90 days in 58 patients (23.2%). All of the clinical presentation types were similarly represented in each of these 3 groups.

Brain MRI showed asymptomatic focal lesions suggestive of demyelination in 193 patients (77.2%) on early studies and 196 (78.4%) on delayed studies. Two years later, 64 patients (25.6%) had developed CDMS; however, 22 of the 64 converters (34%) did so before the delayed MRI was performed (early converters).

Asymptomatic gadolinium-enhancing lesions were identified in 72 patients (28.8%) at early MRI and 62 (24.8%) at delayed MRI. None of the positive brain images at either the early or delayed study showed simultaneous contrast uptake in all the T2 lesions; that is, there were always nonenhancing lesions in addition to enhancing lesions on these images, fulfilling our proposed criterion for DIT.

Fulfillment of the various MRI criteria for DIS in early and delayed MRIs is indicated in Table 3. As would be expected, the number of patients who met the criteria for DIS increased when the less restrictive Frohman or Swanton criteria were applied. These criteria predicted early conversion to CDMS with good sensitivity, although specificity was rather low.

Fulfillment of criteria for DIS and DIT in early and delayed images is indicated in Table 4. Similar high specificity was demonstrated for all 3 sets of MRI criteria, but higher sensitivity was achieved with the less stringent criteria for DIS (Frohman and Swanton criteria). The requirement of DIT (at least 1 asymptomatic gadolinium-enhancing lesion) in addition to DIS decreases sensitivity but increases specificity.

The comparison between the diagnostic properties of a single, early MRI and a single, delayed MRI for identifying converters to CDMS showed similar results, with early MRI yielding slightly higher sensitivity. This difference disappeared when early converters (before delayed MRI was performed) were excluded from the early MRI analysis (Table 4).

We also assessed the diagnostic performance of the 3 criteria according to the interval between CIS onset and early MRI (0-30, 31-60, or 61-90 days after onset). Better sensitivity and specificity were obtained in patients in whom the early MRI was performed within the first 30 days after CIS onset (Table 5).

The present study shows that a single brain MRI study demonstrating criteria for DIS and at least 1 asymptomatic gadolinium-enhancing lesion may suffice to predict early conversion to CDMS, even when the MRI is performed during the first 3 months after the onset of symptoms. Although this MRI is highly specific for predicting early converters, the sensitivity is rather low; hence, many patients who will ultimately develop MS are not diagnosed with the early MRI. This low sensitivity mainly reflects the selectivity of the MRI diagnostic criteria for DIT (at least 1 asymptomatic gadolinium-enhancing lesion in addition to nonenhancing lesions).

When accompanied by gadolinium-enhancing lesions, the various criteria used to demonstrate DIS all showed similar high specificity, although sensitivity was slightly higher when the less stringent Swanton or Frohman criteria were applied. However, the fact that the Swanton criteria are based on the topographic characteristics of the lesions rather than the number of lesions, as occurs with the Frohman criteria, suggests that use of the Swanton criteria would result in fewer false-positive results in patients who present with the relatively frequent incidental or vascular focal white matter lesions. These white matter lesions commonly involve the subcortical hemispheric white matter but less often the juxtacortical white matter and rarely the spinal cord.^10,11

In clinical practice, MRI of the brain and sometimes of the spinal cord is often obtained within the first days after the onset of symptoms in patients with a CIS, particularly in brainstem and spinal cord syndromes, to rule out a nondemyelinating cause for the clinical symptoms and to confirm or exclude the presence of inflammatory or demyelinating central nervous system lesions, a feature that substantially influences the probability of developing CDMS in the long term.^12-16 However, according to the 2001 and 2005 McDonald criteria, in which a minimum of 3 months between symptom onset and an MRI study is required to demonstrate DIT, this early, single MRI cannot be used for establishing the diagnosis of MS.^1,2 The time interval of 3 months was selected to exclude acute disseminated encephalomyelitis (ADEM), typically a monophasic disease, but in which new lesions are sometimes seen on delayed MRI during the recovery period,¹⁷ yielding a mixed pattern of enhancing and nonenhancing lesions at the same time. Therefore, gadolinium-enhancing lesions unrelated to the initial event seen on an MRI obtained within the first few months after a CIS could conceivably be related to a monophasic disorder, such as ADEM, and not be a reflection of DIT. Nevertheless, the similar high specificity obtained in this study compared with previous studies^6,14 testing the diagnostic value of the McDonald and Swanton criteria indicates that overdiagnosis of MS because of inclusion of patients with a monophasic demyelinating disease such as ADEM is unlikely. Moreover, ADEM is rather unusual in the adult population, and there are several clinical, biological, and MRI features that can be used to differentiate patients with ADEM from those with MS.^18,19

The results of the present study support the concept that the simultaneous presence of nonenhancing and enhancing brain lesions on a single MRI, even if obtained within the first 3 months after CIS onset, suggests that they are in different stages of evolution and, therefore, in DIT. This concept was first proposed by Heun et al²⁰ in 1988 and supported by Tas et al,²¹ who reported that the presence of T2 lesions fulfilling the criteria outlined by Paty et al²² or Fazekas et al¹⁰ for DIS and at least 1 gadolinium-enhancing lesion on a single MRI within the first weeks after CIS onset provided high specificity and acceptable sensitivity for predicting early conversion to CDMS, with results comparable to those of the present study in a smaller series.²¹

In addition, we found that the sensitivity and specificity of a single, early MRI are slightly better when it is obtained within the first 30 days after symptom onset. This finding concurs with the idea that, although most enhancing lesions occur in the absence of clinical symptoms,²³ the likelihood of detecting an enhancing lesion is much higher when, or soon after, symptoms appear.²⁴

Some limitations of the present study should be considered. First, the diagnostic value of this early MRI was investigated in a CIS population older than 14 years, and only 5 (2.0%) were younger than 18 years; hence, the performance in a younger population remains to be tested. Second, the clinical data in this study were collected retrospectively from the patients' medical records; hence, we were not able to obtain sufficiently precise information on corticosteroid use, which is known to affect the intensity and duration of gadolinium enhancement of acute plaques.²⁵Finally, as commonly occurs in this type of study, the criteria were applied to cohorts recruited in third-level referral centers with particular interest and expertise in clinical and MRI assessment of patients with CIS suggestive of MS. Other diseases such as ADEM were likely to be excluded, and MRIs were obtained with highly standardized techniques that favor analysis and comparison and interpreted by 2 properly trained readers prepared to recognize the full range of brain abnormalities that suggest the diagnosis of MS. Therefore, we cannot be sure of the reproducibility of our results when the criteria are applied in general neurology or neuroradiology practices.

Our findings suggest that a single brain MRI study may suffice to identify a subset of patients with a CIS who are at high early risk of developing CDMS, even when the MRI is performed within the first 3 months after the onset of symptoms. In the simultaneous presence of gadolinium-enhancing and nonenhancing lesions that suggest DIT, better diagnostic performance was achieved by using the Swanton or Frohman criteria for DIS and by obtaining the MRI within the first 30 days after symptom onset. These data should be taken into account for future potential modifications of the MRI diagnostic criteria for MS aimed toward optimizing the examination in terms of ease, simplicity, and cost.

Correspondence: Alex Rovira, MD, Magnetic Resonance Unit, Department of Radiology, Hospital Vall d’Hebron, Passeig Vall d’Hebron 119-129, 08035 Barcelona, Spain (alex.rovira@idi-cat.org).

Accepted for Publication: November 20, 2008.

Author Contributions:Study concept and design: Rovira, Tintoré, Filippi, Frederiksen, Miller, and Montalban. Acquisition of data: Rovira, Swanton, Tintoré, Huerga, Barkhof, Filippi, Frederiksen, Langkilde, Polman, Rovaris, Sastre-Garriga, Miller, and Montalban. Analysis and interpretation of data: Rovira, Tintoré, Barkhof, Filippi, Langkilde, Miszkiel, Polman, Miller, and Montalban. Drafting of the manuscript: Rovira, Tintoré, Huerga, Miller, and Montalban. Critical revision of the manuscript for important intellectual content: Swanton, Tintoré, Barkhof, Filippi, Frederiksen, Langkilde, Miszkiel, Polman, Rovaris, Sastre-Garriga, Miller, and Montalban. Statistical analysis: Sastre-Garriga. Obtained funding: Barkhof, Polman, and Miller. Administrative, technical, and material support: Rovira, Swanton, Huerga, and Frederiksen. Study supervision: Rovira, Tintoré, Barkhof, Filippi, Miszkiel, Polman, Miller, and Montalban.

Financial Disclosure: None reported.