Short segment myelitis as a first manifestation of neuromyelitis optica spectrum disorders

Introduction

Longitudinally extensive transverse myelitis (LETM) is generally described as a continuous spinal cord lesion spanning the length of three or more vertebral segments. LETM is one of the most characteristic features of neuromyelitis optica spectrum disorder (NMOSD), which is a rare, disabling autoimmune disease affecting the central nervous system (CNS) and associated with serum aquaporin-4 immunoglobulin G antibodies (AQP4-IgG).^{1 –3}

Although it is important to exclude other causes of myelopathy,^4,5 LETM has been a part of the diagnostic criteria for this disease. ⁶ Approximately 34%–45% of patients with NMOSD present with myelitis as an initial manifestation of disease^{7 –11} and 82%–91% of patients experience LETM during their disease course.^{8 –10}

By contrast, it has been shown that patients presenting with LETM are at lower risk of developing multiple sclerosis (MS).^12,13 Accordingly, LETM is useful for distinguishing between NMOSD and MS in patients presenting with acute myelitis. ¹⁴ Nevertheless, NMOSD patients may also present with spinal cord lesions spanning less than three vertebral segments (short transverse myelitis, STM) on magnetic resonance imaging (MRI), which poses a diagnostic challenge.

In this study, we evaluated the frequency of STM among patients who presented with myelitis as their first manifestation of NMOSD. We also analysed the clinical and radiographic characteristics of STM in NMOSD patients.

Patients and methods

We retrospectively identified patients who presented with myelitis as a first manifestation of NMOSD between June 2005 and March 2015 from three referral centres in Korea. The diagnosis of NMOSD was made based on the 2015 International Consensus Diagnostic Criteria for NMOSD. ¹⁵ We recruited patients who met the following inclusion criteria: (1) seropositivity for AQP4-IgG, (2) an initial presentation with myelitis without a prior history of neurological symptoms or signs and (3) a spinal cord MRI performed within 1 month of myelitis onset. We excluded patients with MS, as defined by the 2010 McDonald Criteria. ¹⁶

We also excluded patients with AQP4-IgG-negative NMOSD. The serostatus of AQP4-IgG was confirmed by two different methods, in-house enzyme-linked immunosorbent assay (ELISA) ¹⁷ and cell-based assay (CBA) with a commercial slide kit (Euroimmun, Luebeck, Germany), ¹⁸ particularly in patients with STM.

All initial MRIs were performed using a 1.5T or 3T scanner at various centres, and thus the imaging protocols could not be fully unified. Medical records were used to collect demographic, clinical (sex, age at onset and nadir disability) and laboratory (cerebrospinal fluid (CSF) and autoantibodies for screening co-existing autoimmunity) parameters at the time of the first attack. We classified the nadir disability at the time of the initial myelitis episode as either low disability (Expanded Disability Status Scale (EDSS) score < 6) or high disability (EDSS score ⩾ 6). The lengths of the spinal cord lesions were initially determined independently by two readers (S.-Y.H., J.-W.H.). All cases were then re-evaluated independently by an experienced neuroradiologist (S.-H.L.).

In sagittal sections, the length of each spinal cord lesion was expressed as the number of vertebral segments over which each lesion extended. STM was defined as a lesion extending fewer than three vertebral segments. ¹⁹ If a patient was determined to have STM, additional data were collected: axial T2-weighted images; gadolinium-enhanced T1-weighted spinal cord scans; brain MRI at first attack, if available; and clinical course and treatment history during follow-up. On axial T2-weighted imaging, the STM lesions were delineated in terms of location, size and appearance. If more than one axial image of the same lesion was available, we analysed the axial plane with the largest lesion.

The data were analysed using Graph-Pad PRISMS (San Diego, CA, USA). For non-normally distributed variables, differences between LETM and STM lesions were assessed using Mann–Whitney U tests. Unpaired t-tests were used when comparing two groups. A two-tailed Fisher’s exact test was used to compare proportions. This study was approved by the institutional review board at each centre.

Results

A total of 76 patients with myelitis as one of the presenting signs of NMOSD satisfied our inclusion criteria and were included in this study. All cases were Korean. Of these, a total of 69 LETM lesions were identified in 65 patients (85.5%, median length of seven segments, range: 3–16) and the remaining 11 patients (14.5%) collectively had 15 STM lesions (median length of two segments, range: 1–2.5). Concurrently, two STM lesions are shown in four patients (Figure 1). Table 1 shows the comparison of demographic, clinical and laboratory features between STM and LETM patients with AQP4-IgG. The mean age of NMOSD onset in patients with LETM was 38 ± 11.8 years (range, 17–66 years), and 41 ± 9.8 years (range, 22–58 years) in those with STM (p = 0.284). The female-to-male ratio was 4:1 for the LETM cohort and 1.2:1 for the STM group (p = 0.078). No significant differences were observed in current age, age at myelitis onset, the sex ratios or serological signs of co-existing autoimmunity. However, the STM group had fewer patients with high disability scores (EDSS ⩾ 6) than the LETM group (9% vs 41%, respectively, p = 0.047) at nadir of first attack. Two of 47 patients (4.2%) with LETM were positive for oligoclonal bands (OCBs), whereas OCBs were negative in all patients with STM.

OPEN IN VIEWER

Figure 1.

Two discrete STM lesions on sagittal T2-weighted MRI in a patient with NMOSD.

OPEN IN VIEWER

Table 1.

Comparison of demographics of LETM and STM in patients with positive AQP4 antibody.

Demographic	STM (N = 11)	LETM (N = 65)	p value
Median length, median (range)	2 (1–2.5)	7 (3–16)	<0.001
Mean age of onset, median ± SD (range), years	41 ± 9.8 (22–58)	38 ± 11.8 (17–66)	0.284
Female-to-male ratio	4:01	1.2:1	0.078
Interval symptom onset to MRI, median ± SD (range), days	9 ± 8.37 (3–26)	10 ± 9.3 (0–30)	0.569
High disability score (EDSS ⩾ 6) at nadir of first attack	9%	41%	0.047
Serological signs of co-existing autoimmunity, no. (%)	2 of 11 (18%)	21 of 55 (38%)	0.304
CSF oligoclonal bands, no. (%)	0 of 10 (0%)	2 of 47 (4.2%)	1

LETM: longitudinally extensive transverse myelitis; STM: short transverse myelitis; SD: standard deviation; MRI: magnetic resonance imaging; EDSS: Expanded Disability Status Scale; CSF: cerebrospinal fluid.

Expectedly, the median length of cord lesion was significantly different between groups although the median time from symptom onset to MRI was not different: 9 days (range, 3–26 days) in STM group and 10 days (range, 0–30 days) in LETM group, respectively.

On sagittal imaging, five lesions spanned 2.5 segments in length, six were two vertebral segments long, three were 1.5 segments long and one lesion was confined to a single segment.

Nine lesions were located within the cervical cord, while the remaining six lesions were in the thoracic cord.

On axial imaging, all STM lesions involved the central grey matter. A total of 12 STM lesions were predominantly located in the central cord area (Figure 2(a)–(l)). Three STM lesions from two different patients were not as centrally focused, but they involved the central cord area and crossed the grey commissure opposite the spinal cord (Figure 3). One STM lesion had a bright and spotty appearance. ²⁰

OPEN IN VIEWER

Figure 2.

(a–l) Common STM lesions in nine patients with NMOSD. T2-weighted image axial spinal cord MRI depicting centrally located grey matter lesions with extension across more than half of the cross-sectional cord area were found in NMOSD patients presenting with STM.

OPEN IN VIEWER

Figure 3.

(a and b) Less common STM lesions in two patients with NMOSD. The lesions spanned less than half of the cross-sectional cord area with involvement of the grey commissure (arrow). The central canal was also involved on axial view, and spindle-shaped lesions with blurred margins were noted on the sagittal T2-weighted image, neither of which is characteristic of MS lesions.

A total of 11 lesions extended across more than half of the cross-sectional area of the cord, whereas 4 lesions covered less than half of the cross-sectional area.

Gadolinium-enhanced images were available in 13 STM lesions, of which 9 showed enhancement (5 in diffuse and 4 in nodular patterns) and 4 STM lesions had no enhancement. Two lesions with focal enhancement were observed in the white matter area, despite being classified as central grey matter lesions by axial T2-weighted imaging.

Of the eight patients who had a brain MRI at the time of the initial myelitis attack, six did not have any brain lesions and two had brain lesions characteristic of NMOSD, including diencephalic and tegmental localization.

Three of 11 patients had recurrent STM during their NMOSD disease courses. Early detection of AQP4-IgG allowed six patients to receive immunosuppressive therapy with azathioprine, mycophenolate mofetil or rituximab, five of whom were relapse-free for a median of 26 months. Only one of the six patients with an early diagnosis relapsed with STM within 3 months, but no further relapse occurred with continuous immunosuppressive therapy.

The diagnosis of NMOSD was delayed in five patients, all of whom experienced relapses over the course of their disease. Serological measurement of AQP4-IgG was obtained in these patients at a mean of 49 months (range, 12–91 months) after the first attack. Three of the five patients with a delayed NMOSD diagnosis were initially misdiagnosed with MS. These patients were treated with interferon (IFN)-β and went on to develop recurrent myelitis or optic neuritis.

Discussion

In the present cohort study, a significant proportion of NMOSD patients (14.5%) presented with STM as their first manifestation. Two studies provided the relevant data linking the presence of STM to NMOSD. One study reported that 7.3% of Caucasian patients with NMOSD did not have MRI findings of LETM at the time of the first myelitis episode. ⁷ Another study reported that STM was observed in 14% of seropositive NMOSD patients at the initial myelitis attack, not necessarily initial presentation of NMOSD. ¹⁹ These patients might experience preceding optic neuritis, area postrema or brainstem syndrome, and the recognition of antecedent NMOSD-characteristic features might provide relevant diagnostic information. By contrast, our study highlights a population of NMOSD patients presenting with STM as a first manifestation of NMOSD, when the diagnostic uncertainty is the highest and suggests that initial manifestation of STM does not exclude the possibility of NMOSD diagnosis. Nevertheless, our analysis suggests NMOSD-associated STM lesions can be distinguishable from the spinal cord lesions of MS in several ways. First, in no instance was the STM lesions found to be predominantly localized to the peripheral white matter in cross-sectional views. All STM lesions occupied the central grey matter on axial images, and 72% of STM lesions took up more than half of the cross-sectional area. Although three STM lesions extended across less than half of the spinal cord on axial scan, they also involved the grey commissure as well as the central canal of the spinal cord (Figure 3). This finding is not specific for NMOSD and central canal lesion can also be seen in other diseases, such as neurosarcoidosis. ²¹ Nevertheless, it may help to differentiate NMOSD spinal cord lesions from those in MS, as it is rarely seen in MS.

Second, the overall length of STM lesions is still longer in NMOSD than that of usual MS spinal cord lesions, which tend to exhibit peripheral involvement with 90% of lesions spanning fewer than two vertebral segments, and usually do not affect the entire cross-sectional area.^22,23 In our cohort, however, only four lesions (4.7%) in three patients were less than two vertebral segments in length and none of the lesions involved less than one segment, which implies that small spinal cord lesions expanding less than two vertebral segments are rare in NMOSD. Regarding accompanying brain lesions, 18.2% of patients with STM (2/11) had asymptomatic brain lesions involving the corticospinal tract and periependymal lesions surrounding the third ventricles, all of which are consistent with characteristic brain lesions of NMOSD.^24,25 Furthermore, none of the brain MRIs in our patients demonstrated dissemination in space, thus there was not sufficient evidence to satisfy the 2010 McDonald criteria. ¹⁶

In summary, patients with STM lesions concentrated in the central grey matter on axial images with either a normal brain MRI or a characteristic brain lesion of NMOSD and insufficient evidence to satisfy the diagnostic criteria for MS should be tested for the AQP4 antibody to confirm the diagnosis of NMOSD.

Clinically, 54.4% (6/11) of the patients with STM relapsed after a median interval of 18 months, while the remaining five patients continued to be relapse-free after early (i.e. <3 months) initiation of immunosuppressant treatments. Among the STM patients, 45.5% (5/11) experienced a delayed diagnosis or were misdiagnosed as MS; however, at least two core clinical characteristics of NMOSD occurred later in all of these patients during follow-up. The median EDSS score at the end of follow-up was 0 (range, 0–1.0) in patients with an early diagnosis of NMOSD, whereas patients with a delayed NMOSD diagnosis scored a mean of 3.0 (range, 0–7.0). Notably, all three patients initially diagnosed with MS and treated with IFN-β, which may worsen the disease,^26,27 experienced relapses and worsening of disability. These findings illustrate the importance of early, accurate diagnosis and appropriate treatment and how delayed diagnosis by misconception or non-appreciation of potential clinical features negatively impact on patient’s prognosis.

Several limitations apply to our study, which include retrospective design and absence of direct comparison with AQP4-IgG seronegative STM including other CNS demyelinating diseases. In addition, each patient had a different interval between symptom onset and the spinal cord MRI, which were performed at different centres. This may have impacted the measurement of the spinal cord lesions to some degree. Nevertheless, to minimize this limitation, we only assessed the MRI studies that were performed less than 1 month from the appearance of symptoms.

Our study illustrated the importance of recognizing STM in patients with NMOSD, particularly in Asia and regions closer to the equator where the likelihood of an inflammatory demyelinating episode being NMOSD is high, given the diagnostic and treatment implications.^28,29

Transverse MRI imaging of the spinal cord at the onset of symptoms can be helpful in the differentiation between NMOSD and MS. Serological testing for the AQP4 antibody and brain MRI findings are essential to making an accurate and timely diagnosis in patients presenting with STM.