73–90% of neuromyelitis optica spectrum disorder (NMOSD) patients diagnosed according to the 2015 International panel on NMO diagnosis have aquaporin-4 antibodies (AQP4-IgG) [1, 2]. It is presumed that at least a proportion of the remaining 10–27% of patients, classified as seronegative NMOSD have another disease specific antibody. Antibodies to myelin oligodendrocyte glycoprotein (MOG-IgG) have been increasingly reported in a variety of CNS neuroinflammatory conditions including patients with phenotypes typical for NMOSD [3]. We aimed to determine the prevalence of MOG-IgG in AQP4-IgG-negative NMOSD.


The Walton Centre Neurosciences NHS Trust in Liverpool, United Kingdom, is a tertiary neurology hospital that hosts one of the two national multidisciplinary specialist clinics for patients with NMOSD and non-MS demyelinating disorders as part of the UK NMOSD service. We systematically reviewed all patients seen in this clinic over the last 4 years (after the availability of MOG-IgG testing), including clinical information, MRI, and antibody tests. Both AQP4-IgG and MOG-IgG were tested using a validated live cell-based assay with high specificity (John Radcliffe Hospital, Oxford, UK) [4, 5]. This study was approved by Research Ethics Service, NRES Committee London—Hampstead, Ref. no. 15/LO/1433.


261 unique patients with non-MS/atypical CNS inflammatory conditions attended the clinic and were assessed for NMOSD. All patients were tested for AQP4-IgG. 132 cases satisfied the 2015 NMOSD diagnostic criteria. Of these, 96 (73%) were AQP4-IgG positive and 36 (27%) AQP4-IgG negative. These 36 patients, were tested for MOG-IgG and 15/36 (42%) tested positive. This would account for 11% (15/132) of the total cohort of NMOSD patients (Fig. 1; Table 1). All MOG-IgG-negative patients were Caucasians with a median age of onset of 18 years (8–44 years) and median disease duration of 4.7 years (2–16 years). The predominant clinical phenotype of the demyelinating event was ON (60%), TM (21%), brain (12%), and brainstem (4%).

Fig. 1
figure 1

Classification of non-MS/atypical demyelination based on 2015 NMOSD criteria, AQP4-IgG, and MOG-IgG testing. NMOSD neuromyelitis optica spectrum disorder, AQP4 IgG Antibody to aquaporin 4, MOG-IgG antibody to myelin oligodendrocyte glycoprotein, OSD optico-spinal demyelination with normal brain MRI

Table 1 Demographic, clinical, and radiological characteristics of the 15 NMOSD patients with MOG-IgG

While we tested all AQP4-IgG-negative patients for MOG-IgG (n = 36), only a proportion (33%) of AQP4-IgG-positive patients (n = 32) were tested (as double positives are exceptionally rare) (Fig. 1). None were definitely positive. However, one patient was ‘low positive/possibly negative. This patient with one episode of long myelitis also had antinuclear antibodies (1/80 titre with homogenous pattern (nuclear antigens all negative) and was ‘low positive’ for anti-glycine antibodies too. The significance of the MOG-IgG in the context of these additional antibodies is uncertain and may reflect a heightened humoral autoimmune response rather than truly pathogenic dual positivity. This patient has not been included in the MOG cohort in this paper.

We also tested the majority of patients with a demyelinating syndrome referred to the service who did not fulfill the NMOSD criteria (125/129, 97%). Twenty-five (20%) were positive for MOG-IgG. Details of these cases will be the subject of an upcoming separate research paper and are not discussed further here.

We also assessed how many of the MOG-IgG patients with NMOSD phenotype had a relapsing course. Thirteen patients (86%) had a relapsing course. However, a relapsing course was the reason for referral to the clinic in the first place (n = 13/13). The median duration of illness for the relapsing patients was 4.7 years (2–16 years). The median inter-attack interval was 1 year (0.16–17) and median EDSS in the relapsing MOG group at last follow-up was 3 (0–9, Table 1). All relapsing patient are on immunosuppressants (Table 1).

We also assessed the proportion of patients with optic neuritis and long myelitis who fulfill Wingerchuk 2006 criteria [6] that are MOG-IgG positive, as this is a clinical question often posed. Of the whole cohort of 261 patients, 75 patients had long myelitis and optic neuritis. Of these 49 were AQP4-IgG positive (66%) and 10 were MOG-IgG positive (13%, or 38% of AQP4-IgG-negative patients) and 16 remained seronegative (21%). Serial testing where done in 14/15 patients (13 relapsing); MOG-IgG was detected in all. Treatment with steroid or immunosuppression does not seem to have an effect on MOG-IgG serostatus in this cohort of predominantly relapsing patients (Table 2).

Table 2 MOG-IgG testing in relation to disease course and immunosuppressive treatment. NA: not available


In a cohort of well-characterised NMOSD patients (n = 132), 73% were AQP4-IgG and 11% were MOG-IgG seropositive and 16% remained seronegative. MOG-IgG disease accounts for 42% of the AQP4 IgG-negative seronegative cohort. MOG-IgG was present in 38% of patients with long myelitis and optic neuritis who do not have AQP4 IgG.

86% (13/15) of our patients who satisfy criteria for NMOSD who are MOG-IgG-positive patients have relapsing disease, similar to a recent study [7] who reported that 80% of their MOG-IgG-positive cohort (n = 50) followed a relapsing course. However, a relapsing course was the reason for referral to the clinic in the first place (n = 13/13) making this a biased sample. Long-term follow-ups of a cohort of MOG-IgG-positive patients after the very first event is required to obtain the true risk of relapse.

Importantly, 20% of patients with non-MS/atypical demyelination who do not satisfy criteria for NMOSD tested positive for MOG-IgG (Fig. 1). Double positive cases (both AQP4-IgG and MOG-IgG) are rare [8,9,10] with none of the tested patients were definite positives. Since we have tested only 52% (68/132) of the total NMOSD cohort for MOG-IgG, this requires further clarification in future studies.

In conclusion, our study provides the best possible answers at the current time on several questions on the frequency of MOG-IgG patients: NMOSD who are AQP4-IgG negative and MOG-IgG positive (42%), NMO (as per Wingerchuk 2006) with optic neuritis and long myelitis who are AQP4-IgG negative but MOG-IgG (13%). We also found that MOG-IgG is found in 20% of non-NMOSD/non-MS demyelination. It is also estimated that at least 11% of all NMOSD (as per 2015 criteria) is MOG-IgG positive.

Our study has important practical implications. First, the definite diagnosis of MOG-IgG-associated disease offers patients and physicians a better diagnostic label than seronegative NMOSD. Second, as nearly one in every two of seronegative NMOSD, and 1/5 of atypical non-MS demyelination is MOG-Ig positive, testing for these cohorts will be of high yield and worthwhile, compared to testing every demyelination (which in most Caucasian predominant populations is likely to be MS) with attendant costs and risk of false-positive results. Third, it is likely that the long-term disease course and therefore treatment strategies of AQP4-IgG and MOG-IgG is different. If this is the case, MOG-IgG status, should be part of inclusion/exclusion criteria or a variable for stratification in clinical trials. The latter issue may have importance for currently recruiting trials that include seronegative NMOSD.