Article Text

Download PDFPDF

Overlapping CNS inflammatory diseases: differentiating features of NMO and MS
  1. Maciej Juryńczyk,
  2. Matthew Craner,
  3. Jacqueline Palace
  1. Department of Neurology, Oxford University Hospitals National Health Service Trust, Oxford, UK
  1. Correspondence to Jacqueline Palace, Department of Neurology, Oxford University Hospitals NHS Trust, West Wing Level 3, John Radcliffe Hospital, Oxford, OX39DU, UK; jacqueline.palace{at}


Neuromyelitis optica (NMO) has long been considered as a variant of multiple sclerosis (MS) rather than a distinct disease. This concept changed with the discovery of serum antibodies (Ab) against aquaporin-4 (AQP4), which unequivocally differentiate NMO from MS. Patients who test positive for AQP4-Abs and present with optic neuritis (ON) and transverse myelitis (TM) are diagnosed with NMO and those who show an incomplete phenotype with isolated ON or longitudinally extensive TM (LETM) or less commonly brain/brainstem disease are referred to as NMO spectrum disorders (NMOSD). However, many patients, who have overlapping features of both NMO and MS, test negative for AQP4-Abs and may be difficult to definitively diagnose. This raises important practical issues, since NMO and MS respond differently to immunomodulatory treatment and have different prognoses. Here we review distinct features of AQP4-positive NMO and MS, which might then be useful in the diagnosis of antibody-negative overlap syndromes. We identify discriminators, which are related to demographic data (non-white origin, very late onset), clinical features (limited recovery from ON, bilateral ON, intractable nausea, progressive course of disability), laboratory results (cerebrospinal fluid (CSF) pleocytosis with eosinophils and/or neutrophils, oligoclonal bands, glial fibrillary acidic protein in the CSF) and imaging (LETM, LETM with T1 hypointensity, periependymal brainstem lesions, perivenous white matter lesions, Dawson's fingers, curved or S-shaped U-fibre juxtacortical lesions). We review the value of these discriminators and discuss the compelling need for new diagnostic markers in these two autoimmune demyelinating diseases of the central nervous system.


Statistics from


Neuromyelitis optica (NMO) and multiple sclerosis (MS) are both idiopathic, autoimmune and demyelinating diseases of the central nervous system (CNS) typically following a relapsing course. Cardinal NMO manifestations, namely transverse myelitis (TM) and optic neuritis (ON), are not specific for NMO and are also common in MS, which explains why NMO has been considered by many as a variant of MS and not as a separate disease.1This concept changed when antibodies targeting the water channel protein AQP4 were identified as a highly specific biomarker and a trigger of autoimmunity in NMO.2–5 The discovery of AQP4-Abs proved that NMO has a distinct immunopathogenesis being a primary autoantibody-mediated astrocytopathy and is thus separate from MS. It also broadened the spectrum of NMO disorders (NMOSD) to include limited forms of NMO, such as isolated recurrent ON, longitudinally extensive TM (LETM) and antibody-positive brain/brainstem syndromes.1

AQP4-Abs considerably facilitate distinction between NMO and MS, but there are still many patients with ON and/or TM who test negative and are at the interface of NMO and MS. The differential diagnosis is difficult due to an overlap of clinical and paraclinical features. Some of these patients fulfil the criteria for seronegative NMO, but have monophasic disease and thus assumed to have distinct pathomechanisms from either relapsing NMO or MS. It is not uncommon that antibody-negative relapsing NMO is misdiagnosed as MS due to its relative rarity in the western world and because of relatively high prevalence of brain lesions. In these cases the diagnosis of NMO may only be made after worsening with MS disease-modifying treatments.6 There is therefore a compelling need for algorithms and new biomarkers to further differentiate between NMO and MS. In this paper we review distinct demographic, clinical, laboratory and radiological features of NMO and MS to identify valuable discriminators in the differential diagnosis of both diseases. We refer to all patients with NMO and NMOSD as NMO and because of the heterogeneity of diseases within the antibody-negative groups we focus on the features of those with AQP4-Abs.7 ,8

Epidemiology and demography

The differences in demographic characteristics of NMO and MS might be potentially useful in the process of differential diagnosis. Whereas MS has a marked geographical distribution, NMO appears to be a worldwide disease. In high MS prevalence areas, like Western Europe and North America, the ratio of MS to NMO is around 50–100, whereas in low prevalence areas it is much lower.1 For example opticospinal MS in Japan accounts for 15–40% of MS cases and over half of these are AQP4-positive.2 ,9–11 While MS predominates in the white population, NMO has a mixed race distribution. In UK and US studies 20–50% of patients with NMO were of non-white origin.12 ,13

Another distinguishing feature of MS and NMO is the age at onset of symptoms. Patients with NMO are on average 10 years older than patients with MS on disease presentation with a mean onset at around 40 for NMO and 30 for MS.1 ,12 ,14 While late-onset MS (ie, starting after the age of 50) is not exceptional, a very late onset MS (>70) is considered a rarity.15 ,16 This is in contrast to NMO, where 12% of AQP4-positive tests have come from participants >6517 and in the UK cohort >20% Caucasian patients were >60 at onset.12 In our practice we encountered several AQP4-positive patients with NMO who presented in the eighth or ninth decade of life (unpublished observations). Thus, a very late onset might be considered as a discriminator in the differential diagnosis between NMO and MS.

Although both diseases have a female predominance, this is much greater in AQP4-positive NMO with a female to male ratio between 8 and 9 in most studies.12 ,18 This disparity does not however hold true of seronegative NMO, which is equally distributed among sexes.18–20

MS predisposition is known to be influenced by genetic factors and co-occurrence of the disease in family members is not uncommon.21 NMO has also been reported to aggregate within families and several loci have been identified as potentially relevant22 ,23 Large multicenter genome-wide associated studies are needed to provide further insights into the genetic background of NMO.

Clinical phenotypes

Importantly, disability in NMO is relapse associated due to the poor recovery from relapses and the lack of a progressive phase.24 This is in contrast to MS, where relapses are milder and recovery is generally good and thus the predominant disability accumulates during progressive disease.1

Optic neuritis

Isolated ON is a feature of both NMO and MS, but may also occur in other immune-mediated disorders, such as single isolated ON or recurrent inflammatory ON.25 ,26 ON in NMO is usually severe and has limited recovery.26 In one of the studies the mean follow-up visual acuity was at 6/18.26 It has been estimated that at 5 years from onset 41% of AQP4-positive patients would be blind in one or both eyes and 9% of patients in both eyes.20 This is in line with a report on 106 AQP4-positive patients from UK and Japan, which described permanent bilateral visual disability (visual acuity of <6/36 in the best eye) in 18% of patients having median disease duration of 75 months and nearly half of those with ON at disease onset were under 30 years and 42% of Afro-Caribbean decent.12 In contrast, ON in MS usually has a good prognosis. In the ON Treatment Study 77% of patients with MS had a visual acuity of at least 6/6 at the 15-year follow-up from acute unilateral ON.27 Only 2/294 patients had a visual acuity of less than 6/12. The pattern of visual field abnormalities following ON might also be diagnostically relevant. This holds particularly true of altitudinal deficits, which were encountered in NMO (>10% patients), but not in MS.28 ,29 This type of visual field defect is typical of ischaemic ON and might suggest involvement of vascular mechanisms in the pathogenesis of NMO-associated ON.28 ,29

Isolated simultaneous bilateral optic neuritis (BON) is another syndrome typically associated with NMO and rarely encountered in MS. According to NMO studies BON is present at onset in around 6–8% AQP4 NMO cases and occurs at any time in up to 20%.18 ,20 In a study focused on the frequency of ON in MS only two patients out of 472 presented with BON.30

The imaging appearances of ON might help in the diagnosis. In paediatric patients long-segment, sometimes bilateral, optic nerve enhancement with extension into optic chiasm was typical of NMO.31 In adults a trend toward involvement of posterior parts of the optic nerve in patients with NMO as compared to patients with MS was also demonstrated.32 ,33 OCT studies showed that ON in NMO causes more severe neuronal damage and greater thinning of the retinal nerve fibre layer (RNFL) than ON in MS.34–36 Interestingly RNFL thinning in MS showed clear temporal preponderance, while in NMO it was more widespread.34 ,37

Tranverse myelitis

NMO is characterised by a severe TM manifesting with paraplegia, sphincter disturbances and a sensory level. The symptoms are typically accompanied by extensive lesions spanning at least three segments of the spinal cord. Such radiological presentation is termed LETM and is a hallmark of NMO and thus incorporated into the diagnostic criteria.1 ,38 In recent studies LETM was found in around 90% of AQP4-positive patients with NMO.12 ,19 LETM lesions in patients with NMO are generally located centrally and involve the white and the grey matter of the spinal cord (figure 1).39 MS-associated TM usually affects a small segment of spinal cord and results in milder symptoms.40 ,41 In the MRI this corresponds to peripherally located spinal cord lesions, which are typically shorter than one segment.41 Two to three per cent of patients with MS might develop LETM.42–44 In a recent study LETM lesions in patients with MS were shorter than those of patients with NMO, less oedematous and less likely to show contrast enhancement.43 LETM in MS very rarely exhibits T1-weighted hypointensity in the central part of the cord, which is typical of NMO.41 ,45

Figure 1

Neuromyelitis optica (NMO)/multiple sclerosis (MS) discriminators associated with the spinal cord involvement on MRI. Longitudinally extensive transverse myelitis from C1 to T7 (arrow) with high signal on T2-weighted images (A) and central hypointensity (arrow) on T1-weighted imaging (B) in a patient with aquaporin-4-positive NMO. The lesion is holocord and involves both the grey and the white matter of the spinal cord (C, arrow). Short spinal cord lesion at C3 (D) located peripherally (E) in a patient with MS.

Advances in non-conventional MRI of the spinal cord might also provide new diagnostic tools. Diffusion tensor imaging (DTI) revealed that NMO causes more profound spinal cord pathology than MS as evidenced by an increase in radial diffusivity within lesioned tracts.46 Interestingly, non-lesioned spinal tracts were frequently abnormal in patients with MS, while in NMO they were not different from controls.46 MR spectroscopy of the cervical cord demonstrated lower myo-inositol values within NMO lesions in comparison to MS lesions, which may be related to the astrocytic pathology.47

In line with more severe spinal cord pathology in NMO than MS, patients with NMO have worse motor outcomes and require ambulatory help at earlier stages of the disease. After median disease duration of 75 months 34% of AQP4-positive patients in the UK and Japan study developed permanent motor disability (EDSS ≥6.0), >60% in those with disease onset >50 years, and 23% became wheelchair user.12 It is worth noting that walking difficulties in patients with NMO might partially result from severe proprioceptive deficits. It has been shown using DTI that posterior columns of the spinal cord are more affected in NMO compared with MS.48

A Devic-type presentation

The classical NMO phenotype described by Devic is characterised by a simultaneous or sequential occurrence of ON and TM at disease onset.49 Such presentation is atypical of MS but also much less common in AQP4-positive than in seronegative NMO. In recently published reports Devic phenotype was observed at disease onset in 24–32% of AQP4-negative patients and only 4–6% of AQP4-positive patients.12 ,19 ,20 However, in around 20% of AQP4-positive patients this type of attack occurs at a later time in the disease course.50

Brain involvement

Brainstem symptoms such as intractable nausea and vomiting with or without hiccup are well described in NMO, often at onset, due to isolated brainstem lesions or from rostral extension of cervical myelitis.51 ,52 Less frequently alteration of appetite and symptoms such as inappropriate antidiuretic hormone secretion indicative of hypothalamic involvement are reported.53 This goes along with the regions of high AQP4 expression, that is, around the third and fourth ventricle (area postrema, hypothalamus) and the aqueduct of Sylvius (figure 2).33 ,54 Although hypothalamic lesions occur in up to 13% of patients with MS this clinical picture is rare in MS55 and should indicate the need to test for AQP4 antibodies. Of note the brainstem involvement is often reversible clinically and on imaging in contrast to NMO ON and TM relapses.

Figure 2

Neuromyelitis optica (NMO)/multiple sclerosis (MS) discriminators associated with brain involvement on MRI. Bilateral hypothalamic (arrow) lesions (A) and a lesion at the floor of fourth ventricle (B) in a patient with aquaporin-4-positive NMO. Dawson's finger-type lesions (C) and cortical lesions in a patient with MS (D). An example of an S-shaped U-fibre lesion in a patient with MS (E). Section E of this figures was taken from supplementary data from reference.57 Permission to use this image has been granted by Wolters Kluwer Health.

Brain lesions are reported in 60% of patients with NMO56 and vary from non-specific changes, NMO typical lesions (reported in 7–9% of patients) to MS-like appearances.33 ,57 It has been shown that the Barkhof MS criteria for dissemination in space are fulfilled by 10–42% of patients with NMO.56–59 Thus the need for better discriminating criteria on brain MRI led to a study which suggested that the presence of at least 1 lesion adjacent to the body of the lateral ventricle and in the inferior temporal lobe; or the presence of a subcortical U-fibre lesion; or a Dawson's finger-type lesion could distinguish patients with MS from those with NMOSD more accurately with 92% sensitivity and 96% specificity.57

Alongside conventional brain MRI, non-conventional MRI techniques might be also of use in the diagnostic process. A 7T MRI study revealed that lesions centred by a small vein are typical for MS (92% of all MS plaques), but merely absent in NMO.60 Cortical lesions on double inversion recovery imaging also differentiated MS from NMO as they were found exclusively in patients with MS.61

Autoantibody associations with NMO and MS

There is a greater association with antibody-mediated autoimmunity in NMO than in MS with 20–30% of NMO patients having other autoimmune disorders and 40–50% having autoantibodies.38 ,62–64 Among more than 20 autoimmune diseases, which were described in patients with NMO, Sjögren's syndrome, systemic lupus erythematosus, thyroid autoimmune diseases and myasthenia gravis (MG) seem to be the most common.63–65 From the differential diagnosis point of view, MG is particularly interesting, as it has not been epidemiologically linked with MS, while it is not uncommon in patients with NMO.64 ,66 ,67 Two to five per cent of patients with NMO have symptomatic MG, while 11% have been shown to have serum antibodies against acetylcholine receptor (AChR).64 ,68 AChR-Abs might therefore serve as a red flag in the differential diagnosis of NMO and MS.

CSF findings

The distinct pathogenesis of NMO and MS is reflected in the CSF. In patients with NMO CSF usually shows ≥50, while in MS <50 WCC/mm3.62 ,69 CSF of patients with NMO might contain neutrophils (typically ≥5/mm3) and eosinophils, which are absent in the CSF of patients with MS characterised by the predominance of lymphocytes and macrophages.62 In a study focusing on the CSF in patients with NMO neutrophils were found in 44% and eosinophils in 10% of examined samples.70 Oligoclonal bands (OCB) are found in up to 95% of patients with relapsing–remitting MS, but only in 10–25% of patients with NMO.1 ,19 ,70–72 Moreover, the presence of OCB in the CSF might be transitional in NMO and restricted only to relapses.70 In the face of a relative scarcity of biomarkers differentiating NMO from MS, OCB remain one of the most helpful parameters.

The discovery of AQP4 as a target of immune responses in NMO led to astrocyte-specific biomarker measurements, including glial fibrillary acidic protein (GFAP). In an important study GFAP levels in the CSF of NMO patients during acute relapse were much higher than in patients with MS (mean 2 477 vs 0.8 ng/mL) allowing differentiation of NMO and MS with 91% sensitivity and 97% specificity.73 Proinflammatory cytokines might also serve as potential biomarkers. One of the candidates is interleukin-6 (IL-6), which was consistently linked with NMO pathomechanisms.74 In several studies IL-6 was increased in the CSF of patients with NMO, but not in MS.75 ,76

The diagnostic relevance to treatment

Since NMO and MS are pathologically distinct, treatment responses differ and might provide important diagnostic clues. This particularly concerns severe NMO relapses which occur in patients misdiagnosed with MS treated with MS-modifying drugs. Exacerbations of NMO have been reported after treatment with interferon β, natalizumab and fingolimod.77–82 Importantly, NMO relapses triggered by MS drugs often occurred shortly after the initiation of treatment, had catastrophic course and were associated with the development of extensive brain lesions.

Conclusions and future prospects

It is now well established that patients with ON and/or TM who test positive for AQP4-Abs should be diagnosed as NMO/NMOSD. These patients would be treated with long-term immunosuppressive therapy such as corticosteroids with or without add-on therapy with the aim of preventing relapses. The remainder of patients are usually categorised as MS, seronegative NMO or monophasic conditions such as ADEM or acute inflammatory TM, unless they present with symptoms of connective tissue diseases, neurosarcoidosis or other diseases, which may mimic NMO. The differential diagnosis between NMO and MS is crucial, since both chronic diseases have distinct immunopathogenesis and differrential immunomodulatory treatment response.6 ,81 However, the diagnostic process in such patients is challenging since the clinical and radiological manifestations of NMO and MS often overlap.

In this review we focused on distinct features of AQP4-positive NMO and MS which then might be useful in recognising antibody-negative patients with a disease that relates to NMO (table 1). We identified discriminators, which are related to demographic data (non-white origin, very late onset, male sex), clinical features (limited recovery from ON, BON, intractable nausea, progressive course of disability), laboratory results (CSF pleocytosis with eosinophils and/or neutrophils, OCB, CSF GFAP, serum AChR-Abs) and imaging (LETM, LETM with T1 hypointensity, area postrema and/or periependymal brainstem lesions, perivenous white matter lesions, Dawson's fingers, U-fibre juxtacortical lesions). Some of these NMO/MS differentiators are specific and thus useful when present (very late onset, BON, Devic phenotype, intractable nausea, periependymal brainstem lesions in NMO, cortical lesions in MS) or sensitive and thus useful when absent (OCB in MS), or at best both specific and sensitive (AQP4 antibodies, LETM, CSF GFAP in NMO, U-fibre or Dawson's finger-type lesion in MS). While the diagnostic value of some discriminators is well established (AQP4 antibodies, LETM, OCB), the relevance of others requires confirmation in independent cohorts (brain lesion distribution criteria, GFAP in the CSF at relapse).57 ,73 It is worth noting that the usefulness of discriminators varies with the relative prevalence of NMO and MS in a given population. It poses diagnostic difficulties in western countries, where MS is approximately 100 times more common and some NMO specific features might occur in a similar absolute number of patients with MS.

Table 1

Clinical and paraclinical features, which are exceptional in either NMO or MS, and thus valuable in the differential diagnosis of both diseases

Despite the utility of current discriminators, there is a compelling need for novel biomarkers in CNS demyelinating diseases. In response to this need a considerable effort has been made to discover new autoantibodies in MS and NMO. Recently, two independent groups detected antibodies against myelin oligodendrocyte glycoprotein in a subset of patients with NMOSD.7 ,8 ,50 New biomarkers might also be discovered with the use of novel imaging techniques (box 1). Non-conventional MRI is of special interest, since it might hopefully provide new tools for identifying the pathogenic process without performing CNS tissue biopsies, which carry many risks and are often inconclusive.

Box 1

Potential discriminators associated with novel imaging techniques

Pattern of retinal nerve fibre layer thinning in OCT

Pattern of optic nerve involvement on the MRI

Diffusivity of spinal cord tracts on Diffusion tensor imaging

Metabolic pattern of tissue damage on MRI spectroscopy of lesions

Location and morphology of brain lesions on ultra-high field (7T) MRI

Non-conventional imaging techniques holding promise for recognising the details of disease process in overlapping Neuromyelitis optica/multiple sclerosis syndromes.



  • Contributors MJ drafted the manuscript. MC and JP revised it. All authors contributed to the conception of the work.

  • Competing interests MJ received research fellowship from the Polish Ministry of Science and Higher Education programme Mobliność Plus (1070/MOB/2013/0).

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement No additional unpublished data are available for this study.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.