MRI criteria for multiple sclerosis

Methods.

This was a retrospective study of a cohort of children followed in a multidisciplinary pediatric MS clinic at The Hospital for Sick Children, Toronto, Canada. All children have fulfilled the criteria for CDMS as defined by Poser et al.,9 i.e., all had more than one clinical demyelinating event, involving more than one area of the CNS (proven on examination), separated in time by more than 1 month. Clinical records were reviewed to identify the dates and clinical features of the patients’ first attack and subsequent MS-defining attack. MRI studies performed at the time of each attack were graded according to the criteria of Paty et al.,5 Fazekas et al.,6 and McDonald et al.8 (table 1).

Results.

Twenty-four children with CDMS were identified. Four were excluded because applicable MRI studies were unavailable for review. Of the remaining 20 children, 11 were female. The mean age at diagnosis of CDMS was 12 years (range 6 to 18 years). The median interval between the first attack and the MS-defining attack was 10 months (range 2 to 32 months). At the time of the first attack, clinical features included one or more of the following: hemiparesis (10 patients), ataxia (8 patients), optic neuritis (6 patients), brainstem syndromes (6 patients), encephalopathy (3 patients), headache (3 patients), spinal cord syndromes (2 patients), and hemisensory symptoms (1 patient). At the time of the MS-defining attack, clinical features included one or more of the following: hemiparesis (7 patients), ataxia (4 patients), optic neuritis (4 patients), hemisensory symptoms (4 patients), brainstem syndromes (3 patients), spinal cord syndromes (3 patients), and encephalopathy (1 patient).

Of the 20 children with MRI scans available for review, 15 underwent MRI at both attacks, 2 underwent MRI only at the first attack, and 3 underwent MRI only at the MS-defining attack. Therefore, in total, 35 MRI scans were reviewed, most of which were performed at our institution. All scans included a T2-weighted sequence in one or more planes, 16 studies included a fluid-attenuated inversion recovery sequence, and 20 studies were performed with gadolinium. MRI of the spine was performed in eight children who had symptoms referable to the spinal cord, and four of these studies were performed with gadolinium. In every patient, MRI demonstrated one or more lesions in the brain or spinal cord at the time of the first attack, MS-defining attack, or both. In the remaining patients, MRI lesions involved the following anatomic sites: periventricular white matter (in 28 MRI scans), juxtacortical white matter (in 25), corpus callosum (in 21), brainstem (in 18), deep white matter (in 14), deep gray matter (in 13), cerebellum (in 12), spinal cord (in 8), and cortex (in 6). Four MRI studies demonstrated tumefactive lesions, defined by the presence of surrounding edema and mass effect. No child demonstrated normal imaging at both time points.

The number of children meeting each of the established MRI criteria for MS is given in table 2. At the time of their first attack, children were most likely to meet the Paty et al. criteria, followed by the Fazekas et al. criteria and finally the McDonald MRI criteria. At the time of their MS-defining attack, children were most likely to meet the Fazekas et al. criteria, followed by the Paty et al. criteria and finally the McDonald MRI criteria.

Discussion.

This study represents the first attempt to evaluate existing MRI criteria for MS in a pediatric MS population. Our findings indicate that a substantial number of children with CDMS do not meet established MRI criteria for the disease. In our cohort, the McDonald DIS criteria were fulfilled by only 53% of children at the time of their first attack, and 67% of children at the time of their MS-defining attack. The Fazekas et al. and Paty et al. criteria were fulfilled by a higher percentage of children because these criteria are less stringent than the McDonald DIS criteria.

In order to identify why some children in this cohort failed to meet the McDonald DIS criteria, we further analyzed each of the McDonald DIS subcriteria individually (table 3). It appeared that no single subcriterion stood out as being particularly sensitive or insensitive in pediatric MS patients. Of the eight children who failed to meet the McDonald DIS criteria at the time of their first attack, three children demonstrated only infratentorial lesions, one demonstrated only a juxtacortical lesion, one demonstrated only optic nerve enhancement, one demonstrated only a deep gray matter lesion, and one child had both a juxtacortical and three periventricular lesions. None of these eight children demonstrated nine T2-hyperintense lesions or a gadolinium-enhancing lesion. Although not all MRI scans were performed with gadolinium, we found only one case in which administration of gadolinium and identification of an enhancing lesion would have caused the child to meet the McDonald DIS criteria.

Recent studies have evaluated the validity of the McDonald MRI criteria in cohorts of patients with clinical isolated syndromes.3,4⇓ Although it is tempting to compare our study to these reports, our results are not directly comparable to this literature because our study included only patients with CDMS. Furthermore, because this was not a prospective study in which MRI scans were obtained at predetermined time intervals, we could not assess the validity of the McDonald criteria for dissemination in time.

Our findings indicate that existing MRI criteria for the diagnosis of adult-onset MS, particularly the McDonald DIS criteria, may not apply as well to the pediatric MS population. Pediatric MS patients appear to have fewer white matter lesions at the time of their MS diagnosis than do newly diagnosed adults with MS. We hypothesize that this is because children with MS are closer to the true biologic onset of the disease, and therefore have had less time to accrue numerous white matter lesions. Furthermore, MS in childhood begins before myelinogenesis is complete, which may influence lesion appearance, size, and distribution. The influence of myelin maturation on lesion distribution within the CNS, and the impact of demyelination on immature white matter, will be the subject of a planned future study.

The accurate and timely diagnosis of MS in children will be supported by the development of MRI criteria that are validated in the pediatric MS population. Future prospective MRI studies, with appropriate control groups, are required to develop these pediatric MS criteria. In the interim, we recommend caution when applying established MRI criteria, particularly the McDonald DIS criteria, to children with suspected MS.