Longitudinally Extensive Transverse Myelitis (2024)

Importance Aquaporin 4 antibody (AQP4-Ab)–negative patients with longitudinally extensive transverse myelitis (LETM) behave differently from those with AQP4-Ab. Aquaporin 4 antibody–negative neuromyelitis optica (NMO) is rare when good assays are used.

Objective To assess if AQP4-Ab–negative patients with LETM share similar disease characteristics with AQP4-Ab–positive patients or whether they have distinct features and alternative diagnoses.

Design We collated clinical and paraclinical data on patients with LETM identified through the Oxford NMO clinical database. Aquaporin 4 antibodies were tested using 2 sensitive assays. We describe the features of patients with LETM, compare findings between patients with and without AQP4-Ab, and describe alternative diagnoses in AQP4-Ab–negative patients.

Setting Single specialist UK center for NMO.

Participants Seventy-six adult patients with LETM.

Main Outcomes and Measures Comparison of clinical and paraclinical data.

Results Fifty-eight percent of patients were AQP4-Ab positive. Alternative diagnoses could usually be identified in AQP4-Ab–negative patients, including those fulfilling NMO diagnostic criteria. Only 6.5% of patients had “true” seronegative NMO and 6.5% had idiopathic LETM. There were some important differences between AQP4-Ab–positive and –negative cases, including older onset age, higher proportion of females, lower incidence of simultaneous optic neuritis, lower frequency of conus involvement, and higher prevalence of coexisting autoimmune disorders in AQP4-Ab–positive cases. Attack severity and degree of recovery were similar in the 2 groups.

Conclusions and Relevance Patients with LETM without AQP4-Ab include a number of different diagnostic categories and it is not surprising therefore that they show important differences compared with AQP4-Ab–positive patients, even when considering only those fulfilling current NMO diagnostic criteria. Thus, we suggest that diagnoses such as myelin-oligodendrocyte glycoprotein antibody disease, multiple sclerosis, acute disseminated encephalomyelitis, and postinfectious disorders should be exclusions in the NMO diagnostic criteria and AQP4-Ab–positive and antibody–negative NMO/NMO spectrum disorder cohorts should be analyzed separately.

Until the discovery of aquaporin 4 antibodies (AQP4-Ab) in 2005,¹ neuromyelitis optica (NMO) was defined clinically as the coexistence of inflammatory myelitis and optic neuritis without symptomatic disease outside of these regions.² The high specificity of AQP4-Ab^3-5 has enabled broadening of the clinical phenotype of NMO; it has become clear that many patients have limited forms of disease such as monophasic or recurrent longitudinally extensive transverse myelitis (LETM) or less commonly bilateral or recurrent optic neuritis or brain disease, together encompassed by the term NMO spectrum disorders (NMOSD).⁶ We recently found that 47% of AQP4-Ab–positive patients had limited disease, most commonly monophasic or recurrent LETM.⁷ The NMO diagnostic criteria⁸ do not currently include such patients but most specialists agree that they should be treated the same as those with the full NMO phenotype, with early and aggressive immunosuppression. The diagnostic criteria need updating to reflect the expanding spectrum of NMO with AQP4-Ab. On the other hand, it is less clear how to classify and treat patients with NMO/NMOSD without AQP4-Ab. This is important for the correct definition of patients in clinical studies, which will influence outcome predictions and patient management.

Longitudinally extensive transverse myelitis is a hallmark of NMO/NMOSD. Oxford covers the southern half of the nationally funded UK NMO service and herein we describe the characteristics and diagnoses of patients referred to the clinical inflammatory team with an LETM. We set out to assess the final diagnoses of these patients after comprehensive investigations and follow-up, describe the “spectrum” of diseases causing LETM that are not due to AQP4-Ab disease, and identify useful features in the acute setting that can help guide diagnosis.

The study was approved by the regional ethics committee and patients provided written consent. The Oxford NMO clinical database was searched for patients who were referred to the neuroinflammatory specialist service with a history of LETM (defined as T2 hyperintensity on spinal magnetic resonance imaging [MRI] extending over ≥3 vertebral segments) or seen as inpatients with an LETM at the John Radcliffe Hospital, Oxford, England, between 2008 and May 2012. All adult patients with a history of LETM, including those with previous attacks of central nervous system inflammation, were included and only patients younger than 16 years were excluded. For patients with multiple LETM attacks, only features of the index LETM were analyzed.

Demographic and clinical data are prospectively collected in this database and information was collated retrospectively from this and case notes. The European Database for Multiple Sclerosis (EDMUS)⁹ scoring system was used to estimate disability during and after the attack. The MRI features were reviewed by a neuroradiologist with an interest in inflammatory disorders (G.Q. or W.K.). All patients were tested for AQP4-Ab at least twice by visualization of binding to transiently transfected AQP4-expressing human embryonic kidney cells and quantitative measurement of antibody binding by flow cytometry, as described previously.⁵ There was complete agreement in the positivity and negativity of all patients. Myelin-oligodendrocyte glycoprotein antibodies (MOG-Ab) were tested in all patients using a cell-based assay.¹⁰

Because this was a retrospective study, no routine panel of investigations was performed for every patient but clinical diagnoses were based on imaging (brain and spinal cord MRI), cerebrospinal fluid examination in most, and infection, autoimmune, paraneoplastic, and metabolic screening.

Statistical analysis was performed using GraphPad Prism version 4. The Mann-Whitney U test was used to compare continuous variables and the Fisher exact test was used to compare frequencies. Statistical significance was set at P < .05.

Seventy-six patients were included. Forty-four of these (58%) tested positive for AQP4-Ab, and 32 (42%) were AQP4-Ab negative by both assays. The median follow-up time since LETM was significantly longer in the AQP4-Ab–positive group (61.35 months [range, 2.3-260.2 months]) than the AQP4-Ab–negative group (25.04 months [range, 1.9-169.4 months]; P = .01).

There were differences between AQP4-Ab–positive and –negative cases (Table 1). Older age at LETM, a greater proportion of females, and higher frequency of coexisting autoimmune disorders were evident in the AQP4-Ab–positive cases. There was a trend toward more patients having had previous optic neuritis in the AQP4-Ab–positive group. By contrast, more patients experienced simultaneous or rapidly sequential optic neuritis with their LETM as part of a “classic” Devic syndrome (as originally described by Devic¹¹) in the AQP4-Ab–negative group. There were no differences in the proportions of patients with a preceding infection or preceding pain between the groups, although there was a trend toward preceding vomiting being more common in the AQP4-Ab–positive patients. The first neurological symptom was similar in both groups (sensory then motor disturbance). However, initial urinary retention occurred more frequently in AQP4-Ab–negative patients; all subsequently developed limb weakness and sensory disturbance.

The LETM attacks tended to be associated with severe disability in both groups (Table 1). The severity as measured by the EDMUS score was similar at nadir and maximum recovery (which included only patients with follow-up duration >6 months) in the AQP4-Ab–positive and –negative groups, and thus, the median improvement (ie, median change in EDMUS score) was similar (2 [range, 0-7] vs 3 [range, 0-8]). The proportions of patients left unable to walk 100 m unaided (EDMUS score ≥6) and wheelchair dependent (EDMUS score ≥8) were also similar (36% and 35% and 5% and 6% in the AQP4-Ab–positive and –negative groups, respectively). A breakdown of nadir and recovery EDMUS scores in the different diagnostic groups is shown in the eTable in Supplement.

Short-term treatment was comparable in the 2 groups. Thirty-four AQP4-Ab–positive patients (77%) and 25 AQP4-Ab–negative patients (78%) were treated with methylprednisolone and 11 AQP4-Ab–positive patients (25%) and 6 AQP4-Ab–negative patients (19%) additionally underwent plasma exchange. More AQP4-Ab–positive patients started long-term immunosuppression following the LETM attack than AQP4-Ab–negative patients (41% vs 13%; P = .005).

Patients with positive AQP4-Ab were significantly more likely to experience another episode of central nervous system inflammation after their LETM (74% vs 31%; P = .001), but this may be partly related to the longer follow-up period. Allowing for this, by performing survival analysis, showed a trend toward greater probability of relapsing over time in the AQP4-Ab–positive group (P = .07) (Figure 1), although the likelihood of early relapse (in the first 18 months) appeared similar. Most relapsing patients had a further LETM but 6 AQP4-Ab–positive patients and 1 AQP4-Ab–negative patient went on to have an optic neuritis. One AQP4-Ab–positive patient died of respiratory failure during the first LETM attack and a further 4 patients (all AQP4-Ab positive) have subsequently died.

Sixty-two spinal MRI scans from the acute attack were available for review (Table 1). The mean length of the longest lesions did not differ between AQP4-Ab–positive and AQP4-Ab–negative patients and there was no correlation between lesion length and nadir or maximum recovery EDMUS scores (data not shown). However, of possible practical significance, involvement of the conus was significantly more common in AQP4-Ab–negative patients (P < .01). The MRI brain scans from the acute attack were available from 61 patients and there were no differences in the proportions of patients with normal scans, nonspecific white matter lesions, or multiple sclerosis (MS)–like lesions between the 2 groups (Table 1). Two AQP4-Ab–negative patients with MS-like lesions fulfilled McDonald criteria for dissemination in space but there was no dissemination in time and the distribution of lesions was more typical for acute disseminated encephalomyelitis (ADEM), with involvement of both white and gray matter.

Cerebrospinal fluid results from the acute attack were available from 44 patients (Table 1). Cerebrospinal fluid oligoclonal bands were rare and only present in the AQP4-Ab–positive patients and there was a trend toward more pleocytosis in the negative group.

Twenty-two patients had visual evoked potentials performed during the acute attack. Three AQP4-Ab–positive patients and 2 AQP4-Ab–negative patients had delayed P100 latencies in asymptomatic eyes and the proportion of asymptomatic eyes with prolonged latencies was not significantly different between the 2 groups.

Aquaporin 4 antibodies are highly specific for NMO and its spectrum disorders and so patients testing positive for this antibody were assumed to have AQP4-Ab–mediated NMO/NMOSD. Most of the AQP4-Ab–negative patients had an alternative identifiable cause (Figure 2).

Six patients tested positive for MOG-Ab. Four of these patients have been described in detail elsewhere.¹⁰ These patients differed from AQP4-Ab–positive patients in that they had less of a female bias, were younger, often had a classic Devic phenotype at presentation, and had excellent recovery from the acute attack as described previously.¹⁰

Other diagnoses in AQP4-Ab–negative patients with LETM included central nervous system vasculitis (n = 1), a leptomeningeal syndrome associated with connective tissue disease (n = 1), infection (n = 4; 1 tuberculosis, 1 mycoplasma, 1 Epstein-Barr virus, 1 West Nile virus), paraneoplastic disease due to mesothelioma (n = 1), and spinal cord infarction (n = 1). Additionally, 5 patients had monophasic multifocal inflammation of the central nervous system consistent with ADEM, and 3 patients subsequently experienced a relapsing-remitting illness that was compatible with MS. Infectious etiologies were identified through high clinical suspicion, eg, relevant foreign exposure, together with positive serology results. Diagnoses of ADEM and MS were made on clinical grounds with supporting paraclinical criteria. All patients were seen by at least 2 neurologists with an interest in central nervous system inflammation. Patients diagnosed with ADEM had a preceding illness and/or supportive symptoms such as lethargy, irritability, headache, fever, and nausea and have remained relapse-free without treatment during the follow-up period. Four of the 5 had stored sera available from the acute attack and all were negative for MOG-Ab. Patients diagnosed with MS had relatively short LETMs (3 or 4 vertebral bodies in length), had MRI brain findings suggestive of, though not diagnostic of, MS at presentation and/or at follow-up, and experienced further relapses typical of MS over the follow-up period that included myelitis with short cord lesions or brain/brainstem attacks. One had a strong family history of MS and has entered a secondary progressive course. Testing of cerebrospinal fluid was performed later in the disease course in 1 patient and was positive for oligoclonal bands.

In 10 patients, despite extensive investigations that included a full autoimmune, infection, and metabolic screen, no cause for the LETM could be identified. Five of these patients had also experienced at least 1 optic neuritis attack and fulfilled the Wingerchuk revised diagnostic criteria for NMO⁸ and were felt to have seronegative NMO and 5 had disease limited to the spinal cord that was termed idiopathic LETM.

The patients with seronegative NMO had a similar female preponderance (4:1) and similar mean (SD) age at first LETM (38.15 [17.11] years) compared with AQP4-Ab–positive patients with NMO, although they composed a higher proportion of white patients (100% vs 60%; P = .01). There were no obvious differences in imaging or cerebrospinal fluid characteristics compared with AQP4-Ab–positive patients with NMO and no differences between nadir (median, 7.25) and maximum recovery (median, 3) EDMUS scores.

The 5 patients with idiopathic LETM were patients with monophasic (n = 2, follow-up duration 6 and 95 months; no immunosuppression received) and relapsing (n = 3, follow-up duration 15, 31, and 71 months) LETM. These patients included a significantly higher proportion of males compared with AQP4-Ab–positive patients (80% vs 14%; P = .004) but had a similar mean (SD) age at onset (45.61 [20.20] years); the mean (SD) lesion length on MRI was comparable with that seen in the group overall (7.2 [3.03] vertebral segments). The 2 patients with monophasic LETM were Afro-Caribbean whereas the relapsing patients were white (n = 2) and Asian (n = 1).

Thirty-five of 76 patients (46%) in the cohort fulfilled the 2006 Wingerchuk NMO diagnostic criteria⁸ and would be included in most clinical studies of NMO. However, only 20 of these were AQP4-Ab positive; 15 were AQP4-Ab negative. A comparison of these 2 groups is shown in Table 2, with a shorter follow-up time, younger age, and greater likelihood of a “classic” Devic presentation (ie, simultaneous optic neuritis and myelitis) in the AQP4-Ab–negative group. However, the AQP4-Ab–negative group comprised a mixed group of diagnoses, including 4 MOG-Ab–positive patients, 3 with ADEM, and 3 with MS, with only one-third actually felt to have seronegative NMO (eFigure in Supplement).

It has become clear since the discovery of AQP4-Ab that the spectrum of NMO encompasses patients with disease isolated to the spinal cord. Studies looking at differences in demographic and clinical features between AQP4-Ab–positive and –negative patients with NMO have largely compared patients with only the full NMO phenotype.^12-17 These studies have yielded conflicting data, which may in part be because most used the original tissue immunofluorescence assay¹⁸ to define cohorts.^12-14,17 This assay misses 17% to 25% of AQP4-Ab–positive cases compared with cell-based assays,^4,5,19 and therefore, it is likely that these studies included patients with false-negative results, which may have diluted any differences. In the current study, we used sensitive cell-based and flow cytometry AQP4-Ab assays to define AQP4-Ab–positive and –negative patients and sought to determine whether there are features of LETM that can help differentiate patients with AQP4-Ab from those without AQP4-Ab in the acute setting, when the final diagnosis is often not apparent.

We found that age at LETM and sex were important discriminators between AQP4-Ab–positive and AQP4-Ab–negative patients. The AQP4-Ab–positive patients were significantly older and showed a strong female preponderance. Women older than 45 years at LETM had an 88% chance of being AQP4-Ab positive whereas males younger than 35 years at LETM had an 82% chance of being AQP4-Ab negative. Other features that differed between AQP4-Ab–positive and –negative patients included a higher frequency of coexisting autoimmune disorders in AQP4-Ab–positive patients, as has been described previously.^16,20 This is not surprising, because although LETM in association with non–organ-specific autoimmune disorders is well recognized, it is associated with AQP4-Ab,²¹ suggesting that the co-association reflects a predisposition to autoantibody-mediated diseases generally. We found that attack severity and outcomes were not related to AQP4-Ab status, and although similar acute treatments were used in the 2 groups, we cannot rule out the possibility that lack of differences in disability outcomes may have been due to initiation of long-term immunosuppression in the AQP4-Ab–positive group. The follow-up time in AQP4-Ab–negative patients was shorter than in AQP4-Ab–positive patients, making it difficult to draw definitive conclusions about risk of relapse over time in the 2 groups, but there was a trend toward higher risk of relapsing disease in AQP4-Ab–positive patients, which has been shown previously.¹²

One of the hallmarks of NMO with AQP4-Ab is the longitudinally extensive nature of spinal cord lesions. However, in agreement with previous studies,^14,17,20,22 we found that the length of cord lesions in AQP4-Ab–negative patients with LETM is no different from those with AQP4-Ab. However, we have shown that conus involvement occurs more commonly in AQP4-Ab–negative LETM and this may account for the fact that urinary retention as the presenting neurological feature was only seen in these patients. A recent study²² found that central gray matter lesions were more frequent in patients with AQP4-Ab. Thus, there may be a number of radiological clues to differentiate AQP4-Ab–positive and –negative LETM that warrant validation in larger studies.

It is rare to see a “classic” monophasic Devic syndrome (as originally described by Devic¹¹) with simultaneous optic neuritis and LETM as the onset attack in AQP4-Ab–positive NMO/NMOSD.^7,20 Higher rates have been reported when seronegative patients are included in cohorts^15,23-25 and differences between patients with monophasic NMO and those with recurrent attacks were alluded to even before the discovery of AQP4-Ab.^2,26 In the current study, we found a monophasic Devic syndrome to be significantly more common in AQP4-Ab–negative patients with NMO, which is in concordance with other studies.^15,20 Thus, we feel that AQP4-Ab–negative patients with a monophasic Devic syndrome are different from patients with relapsing NMO and AQP4-Ab and should not be included in future clinical NMO studies. We have found MOG-Ab in some of these patients, and although the pathogenicity of MOG-Ab has not yet been proven, in our experience, MOG-Ab–positive patients have a characteristic phenotype¹⁰ that differs from true seronegative NMO and we believe these patients should be classed separately. However, further studies are required to characterize the role of MOG-Ab.

In addition to demonstrating useful differences between AQP4-Ab–positive and –negative patients, we also found that many AQP4-Ab–negative patients with LETM and NMO had alternative diagnoses. Only 57% of patients fulfilling the current clinical NMO diagnostic criteria were AQP4-Ab positive but many AQP4-Ab–negative patients had alternative diagnoses. Thus, it is vital that patients presenting with LETM are screened for other causes before being labeled as having NMO. These include infectious and inflammatory etiologies, particularly MOG-Ab disease. Vascular causes such as spinal cord infarction and dural fistulae should also be borne in mind. We have recently encountered 2 patients with the latter diagnosis who were initially treated as having inflammatory LETM. The NMO diagnostic criteria are currently being updated to reflect the significant advances in our understanding of NMO/NMOSD in recent years.

The number of patients with “true” seronegative NMO (relapsing NMO where other causes have been excluded) in this cohort was small and we believe that when good assays are used to define AQP4-Ab status, seronegative NMO is rare. The “true” AQP4-Ab–seronegative patients had similar features to AQP4-Ab–positive patients and they may have very low-titer AQP4-Ab, less than the detection limit of our assay. Additionally, some patients had received corticosteroids prior to AQP4-Ab testing and we cannot exclude the possibility that this may have produced a false-negative result, although in our experience it is rare for AQP4-Ab status to become negative following treatment. Despite extensive investigations, the cause of LETM could not be determined in a minority of patients. Idiopathic LETM may be a distinct entity; a recent study found that the HLA-DRB1*13 genotype was overrepresented in such patients.²⁷ The proportion of AQP4-Ab–positive patients in our cohort was much higher than that reported in this study. The discrepancies are likely related to differences in referral patterns and sensitivities of the assays used.

Our study has several limitations. First, some of the study data were collected retrospectively, which means that some data were incomplete. Second, all patients had been referred to our specialist neuroinflammatory team and thus there may be an element of referral bias and overrepresentation of inflammatory causes. Third, we acknowledge that differentiating ADEM and MS from seronegative NMO cannot be made with absolute certainty in the absence of pathological confirmation. Fourth, we did not include patients with NMOSD and bilateral or recurrent optic neuritis because they are commonly looked after by the Neuro-ophthalmology Service. Comparison of patients with NMOSD and recurrent optic neuritis with or without AQP4-Ab is warranted. Finally, we have not adjusted the P values for multiple comparisons because of the small sample sizes for some of the comparisons, and clearly, larger studies are required to confirm the trends found.

In conclusion, AQP4-Ab–negative patients with LETM, even if fulfilling the Wingerchuk diagnostic criteria, have a high chance of having alternative diagnoses. Thus, patients testing negative for AQP4-Ab using sensitive assays should not be labeled as having NMO/NMOSD without first pursuing an alternative cause. In this clinically and immunologically well-defined cohort, we have found features that may distinguish between AQP4-Ab–positive and –negative patients with LETM in the acute setting. The AQP4-Ab–negative patients were younger than those with AQP4-Ab, more likely to present with a classic Devic phenotype, and more likely to have conus involvement and early urinary retention. However, the difficulty in distinguishing some of these conditions poses a challenge and the planned revisions of the NMO diagnostic criteria will help clarify things.

Corresponding Author: Jacqueline Palace, DM, Nuffield Department of Clinical Neurosciences, Level 3 West Wing, Oxford University Hospitals NHS Trust, University of Oxford, Headley Way, Oxford OX3 9DU, England (jacqueline.palace@ndcn.ox.ac.uk).

Accepted for Publication: June 14, 2013.

Published Online: September 2, 2013. doi:10.1001/jamaneurol.2013.3890.

Author Contributions: Dr Palace had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Kitley, Leite, Palace.

Acquisition of data: All authors.

Analysis and interpretation of data: All authors.

Drafting of the manuscript: All authors.

Critical revision of the manuscript for important intellectual content: All authors.

Conflict of Interest Disclosures: Dr Kitley has received support for scientific meetings from Biogen Idec, Novartis, and Teva and is supported by the NHS National Specialised Commissioning Group for Neuromyelitis Optica. Dr Leite is involved in AQP4-Ab and MOG-Ab testing, is supported by the NHS National Specialised Commissioning Group for Neuromyelitis Optica and the NIHR Oxford Biomedical Research Centre, and has received speaking honoraria from Biogen Idec and travel grants from Novartis. Dr George has received support for scientific meetings from Biogen Idec. Dr Waters is a named inventor on patents for antibody assays and will receive royalties and has received a speaker honorarium from Biogen Idec Japan. Dr Woodhall is involved in AQP4-Ab and MOG-Ab testing and is supported by the NHS National Specialised Commissioning Group for Neuromyelitis Optica. Prof Vincent and the Nuffield Department of Clinical Neurosciences hold patents and receive royalties and payments for antibody tests. Dr Palace has received unrestricted grants and support for scientific meetings and scientific advisory honoraria from Merck Serono, Teva, Biogen Idec, Bayer Schering, and Novartis, has held MS Society grants, is a clinical lead for the UK Department of Health RSS, and is supported by the NHS National Specialised Commissioning Group for Neuromyelitis Optica. No other disclosures were reported.