Methods

Patients

A group of 660 cases with spinal cord MRI were selected from a larger cohort of 983 patients with demyelinating disease in the PDDD: the diagnoses were CMS 574, NMO 27, isolated myelopathy 45, optic neuritis 6 and atypical demyelinating disease 8. After reviewing the spinal MRI studies 26 patients with LEM were identified. The following clinical parameters were analysed: age, gender, onset features, disease duration, relapse number and relapse rate, and last Extended Disability Status Scale (EDSS) score. EDSS scores were divided into three ranges: 0–3, mild disability; 3.5–5.5, moderate disability; 6.0–10, severe disability.

MRI

All 26 patients had cerebral and spinal MRI on one or more occasions on a 1.5 or 3.0 T scanner. T1 with and without gadolinium enhancement, T2, fluid attenuated inversion recovery (FLAIR) and short tau inversion recovery (STIR) sequences were performed in most cases. Spinal MRI was performed within 4 weeks of the onset of myelopathic symptoms. Lesions on spinal MRI were analysed for anatomical location (sagittal and axial) in the cord, length and the presence of cord swelling and gadolinium enhancement.

Aquaporin-4 antibody

The AQP4-IgG assay was performed blinded by Dr Takuya Matsushita at Kyushu University by immunofluorescence assay using green fluorescent protein (GFP)-AQP4 fusion protein transfected HEK-293T cells as described previously.8 Sera were taken during a clinical relapse in five patients and during remission in 15 patients.

Statistics

Statistical analyses of age-at-onset and at examination were performed using an analysis of the Student t test. Comparisons of relapse number, disease duration, relapse rate, EDSS score and length of MRI spinal cord lesions were performed using the Mann–Whitney U test between groups. Differences in gender, OCB positivity and MRI features between groups were tested by Fisher exact test. All statistical analyses were performed using the SPSS V16.0 for Mac. p Values <0.05 were considered statistically significant.

This study was approved by the Sir Charles Gairdner Hospital Human Research Ethics Committee, and informed consent was obtained from all study participants.

Results

Demographic and clinical features

As shown in table 1, there were 15 females and 11 males and all were Caucasian. Thirteen patients were classified as CMS, seven as NMO, and six as isolated LEM (one monophasic and five recurrent). All CMS patients fulfilled the 2001 McDonald criteria for definite MS and had a relapsing-remitting course.2 Cases classified as NMO all fulfilled the criteria of Wingerchuk et al 1999, and all but one had recurrent episodes.3 All isolated LEM cases had normal VEPs and an extensive workup, to exclude other specific causes, was negative except for one patient with an elevated anti-nuclear antibody titre (1:320).

The average onset age in the whole group was 37.1 years (18–63 years) and was significantly lower in the CMS group (30.6; p<0.05). Average disease duration in the whole group was 11.3 years (range 0–32 years): 14 patients >10 years, 9>15 years and 5>20 years. Six of the 26 patients were followed up from disease onset to death after mean periods of 1–21 years (mean 6.8 years).

The severity of disability was variable but was greater in the NMO group: mean last EDSS was 8.0 in the NMO group and 5.1 in the CMS group (p<0.05). Six of the seven NMO patients had a final EDSS score ≥6.0. Eight patients were classified in the mild disability group, three in the moderate disability group and 15 in the severe disability group. Two patients had a benign course with an EDSS score <3.0 more than 10 years after the onset of the LEM.

LEM was the initial presentation in 13 of the 26 patients: of these one had monophasic disease, while the remaining 12 relapsing cases were classified as CMS (4), NMO (3) and recurrent LEM (5). In the other 13 cases (nine CMS, four NMO) LEM occurred after intervals of 1–27 years following the initial presentation.

MRI findings

The extent and location of spinal cord MRI lesions are summarised in table 1. The mean lesion length was longer in the NMO group than in CMS group: 7.1 (range 3–12 segments) versus 4.5 (range 3–7 segments) and was longest in the recurrent LEM cases: 10.2 (range 4–18 segments) (p<0.05). Extremely long lesions spanning more than 10 vertebral segments were seen in two NMO and two recurrent LEM cases (figure 1). Lesions in CMS were mostly located in the lower cervical cord (C4–C7), but only rarely in the upper thoracic cord (T1–T4), which was significantly different from the NMO and recurrent LEM cases (p<0.05) in which thoracic lesions were more common. On axial MRI LEM lesions occupied the centre of the cord or the whole cord in all six isolated LEM cases, 5/7 NMO cases and 9/13 CMS cases. Cord swelling and enhancement were more common in the NMO and recurrent LEM groups.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Sixty-five-year-old male patient, who presented with hiccough and rapidly developed complete paraplegia, sphincter dysfunction and respiratory failure. He had three episodes of recurrent myelopathy and died of respiratory failure. Spinal MRI demonstrated a T2 hyperintense lesion extending from the medulla to approximately T9 (A) with gadolinium enhancement on the T1 image (B). Brain MRI was normal except for the contiguous medullary lesion. Antibody against aquaporin-4 (AQP4-IgG) was negative.

All except one of the CMS patients had typical cerebral lesions fulfilling the Barkhof criteria.2 Of the NMO patients, one had a medullary lesion, and three with normal MRIs at onset later developed cerebral lesions, one a tumefactive lesion and the others typical MS lesions. Of the six isolated LEM patients, one developed tumefactive lesions in the left occipital and temporal lobes 2 years after onset, a brain biopsy confirmed the diagnosis of demyelinating disease, and another developed a medullary lesion.

Aquaporin-4 antibody

AQP4-IgG was tested in 20 patients (four patients were deceased, and in two patients samples could not be obtained): 10 CMS, four NMO and six isolated LEM. Only one (relapsing LEM) was seropositive (titre 1:16).

Discussion

As in previous studies, LEM was not confined to cases of NMO or isolated myelopathy but also occurred in CMS with a frequency of 2.3%. This is similar to the figure of 3% reported by Tartaglino et al in a study of 68 MS cases in the USA.7 These figures contrast markedly with a frequency up to 30% in Asian MS cases.6 This may reflect the higher incidence of optic-spinal forms of MS in Asia.9 The overlapping of clinical findings in the present study lends support to the concept of a single disorder with a heterogeneous spectrum of manifestations.6 9–14

Our findings indicate that LEM may be the initial presentation of NMO or CMS, or may develop later in patients with clinically isolated syndromes. Based on our findings, when LEM is the first manifestation, a number of MRI features may be at least partially helpful in predicting whether the clinical course will be that of CMS or NMO. First, longer lesions are more predictive of NMO, particularly when located in the thoracic cord and associated with cord swelling and contrast enhancement, whereas shorter lesions in the cervical cord are more predictive of CMS. Unlike Matsuoka et al8 we did not find the axial location and extent of the cord lesion helpful in distinguishing cases of NMO and CMS. Second, the presence of cerebral lesions at the initial presentation is suggestive of CMS, and it is noteworthy that none of our patients with LEM onset and normal brain MRI went on to develop CMS over the period of follow-up of 3 to 11 years. In contrast with the typical brain lesions in cases of CMS, the cerebral lesions in NMO and isolated LEM may be unusual, such as tumefactive and medullary lesions.

Contrary to expectations, the prognosis of patients with LEM was found to be variable. Although in NMO cases LEM was usually associated with a severe disability and poor outcome, mild or moderate disability was observed in 42% of patients in the group as a whole, and 8% of cases had a benign course even after intervals of over 10 years. A benign prognosis has also been reported in children with LEM.15 The present findings therefore indicate that LEM is not always associated with a poor outcome, particularly in patients with CMS.

It has been suggested that monophasic or recurrent isolated LEM, as well as recurrent isolated optic neuritis, should be included in a broader NMO spectrum based on the presence of AQP4-IgG,16 and the term “NMO spectrum” has frequently been used. In our cohort, AQP4-IgG testing was not as helpful as expected, and Marignier et al found similarly that in their cohort AQP4-IgG positivity did not correlate with any clinical or radiological features in their NMO cases.10 Only one of 20 of our cases tested had AQP4-IgG (0 of 4 with NMO), but these results are entirely consistent with other reports from Canada, India and Singapore using equally sensitive assays.17–19 The low seropositivity of AQP4-IgG in our study may be due to immunogenetic differences, as the 20 patients tested for AQP4-IgG were exclusively Caucasian. In contrast, 26% of LEM in a Japanese cohort were seropositive assayed in the same laboratory.8 Moreover, the Mayo Clinic cohort was about 50% non-Caucasian with variable amounts of African and American Indian origin patients.16 Although six of the 26 LEM patients were not tested for AQP4-IgG, this did not affect our overall results, which are of significance for further investigation of AQP4-IgG in LEM in other Caucasian cohorts, as it is evident that there remain numerous unexplained discrepancies in the published AQP4-IgG data.14

In summary, LEM did occur in CMS, as well as NMO, could be the initial presentation of either condition and could also occur in isolation. Patients with LEM had highly heterogeneous clinical characteristics and a low incidence of AQP4-IgG seropositivity in this population.