Background

Over the past few years, the role of immunoglobulin G serum antibodies to myelin oligodendrocyte glycoprotein (MOG-IgG) in patients with inflammatory CNS demyelination has been revisited. While antibodies to MOG were originally thought to be involved in multiple sclerosis (MS), based on results from enzyme-linked immunosorbent assays employing linearized or denatured MOG peptides as antigen, more recent studies using new-generation cell-based assays have demonstrated a robust association of antibodies to full-length, conformationally intact human MOG protein with (mostly recurrent) optic neuritis (ON), myelitis and brainstem encephalitis, as well as with acute disseminated encephalomyelitis (ADEM)-like presentations, rather than with classic MS [1,2,3,4,5,6,7,8,9,10,11].

Based on evidence from (a) immunological studies suggesting a direct pathogenic impact of MOG-IgG, (b) neuropathological studies demonstrating discrete histopathological features, (c) serological studies reporting a lack of aquaporin-4 (AQP4)-IgG in almost all MOG-IgG-positive patients, and (d) cohort studies suggesting differences in clinical and paraclinical presentation, treatment response and prognosis, MOG-IgG is now considered to denote a disease entity in its own right, distinct from classic MS and from AQP4-IgG-positive neuromyelitis optica spectrum disorders (NMOSD), which is now often referred to as MOG-IgG-associated encephalomyelitis (MOG-EM) [11,12,13].

Importantly, however, MOG-EM and MS show a relevant phenotypic, i.e., clinical as well as radiological, overlap [3, 14]: like MS, MOG-EM follows a relapsing course in most cases [3, 6], at least in adults, and 33 and 15% of adult patients with MOG-EM meet McDonald’s and Barkhof’s criteria for MS, respectively, at least once over the course of disease [3, 14]. Accordingly, many patients with MOG-EM were falsely classified as having MS in the past [3, 4]. However, such misclassification has potential therapeutic implications: (a) similar to what has been observed in AQP4-IgG-positive NMOSD, some drugs approved for MS might be ineffective or even harmful in MOG-EM owing to differences in immunopathogenesis [3, 4, 15,16,17]; (b) MOG-EM is associated with a high risk of flare-ups after cessation of steroid treatment for acute attacks and may thus require close monitoring and careful steroid tapering [3, 18,19,20,21,22]; and (c) patients positive for MOG-IgG might be particularly responsive to antibody-depleting treatments for acute attacks such as plasma exchange or immunoadsorption [3, 4, 9, 14, 23, 24], to B cell-targeted long-term therapies such as rituximab, to treatment with intravenous immunoglobulins (IVIG) (especially in children [25]), and to immunosuppressive treatments [3, 6, 14, 25, 26]. Therefore, increasing numbers of patients with suspected or established MS are currently being screened for MOG-IgG.

However, screening of large unselected populations for rare biomarkers generally decreases the positive predictive value of diagnostic tests by increasing the rate of false-positive results [27, 28]. Even if assays with high specificity (≥99%) are used, true-positive (TP) results can easily be outnumbered by false-positive (FP) results if the prevalence of a marker is low and the number of samples tested is high. This also applies to MOG-IgG testing. Based on a hypothetical prevalence of 1% genuinely MOG-IgG-positive cases among all patients currently diagnosed with MS, testing of 100,000 patients with an almost flawless, 99% specific and 100% sensitive assay would result in an unacceptable ratio of 990 FP results to 1000 TP results. Therefore, unselected screening of all patients with suspected or established MS for MOG-IgG should be discouraged and more specific criteria for MOG-IgG testing are urgently needed.

In this paper, we propose for the first time indications for MOG-IgG testing based on expert consensus. In addition, we give a list of conditions considered atypical for MOG-EM (“red flags”) that should prompt physicians to challenge the validity of a positive MOG-IgG test result. Finally, we provide recommendations regarding assay methodology, specimen sampling, and data interpretation.

Methods

PubMed was searched for articles published between February 2007 and February 2017 using the following search term: (“myelin oligodendrocyte glycoprotein” OR MOG) AND (antibody OR antibodies OR IgG). All articles identified by this means were reviewed by a core group of physicians (S.J., B.W., F.P., K.R.) for clinical and paraclinical findings that have been frequently reported in association with MOG-IgG seropositivity in patients with CNS demyelination and which, therefore, may justify MOG-IgG testing, as well as for potential “red flags”, i.e., conditions that are typically found in inflammatory disorders of the CNS but have been reported to be absent or very rare in MOG-IgG-positive patients and thus may indicate diagnoses other than MOG-EM. Based on core group consensus, a first set of recommendations was formulated and then circulated to a broader panel of experts in the field from Australia, Denmark, France, Germany, Italy, Japan, South Korea, Spain, UK, and the USA for discussion and refinement. Panel members were invited by the core group based on eminence and previous contributions to the field. Based on several rounds of core group-led peer-to-peer discussions of the individual recommendations with all individual members of the panel, a final set of evidence- as well as eminence-based recommendations was drawn up to which all members gave their approval. All recommendations given here should be considered as expert consensus.

Recommendations on MOG-IgG testing

In Table 1, we propose indications for MOG-IgG testing based on clinical and paraclinical findings that are typical of MOG-EM and/or atypical for MS and were considered by the panel members to be associated with pre-test odds high enough to justify MOG-IgG testing or that demand MOG-IgG testing because of potentially significant therapeutic consequences of a positive test result according to expert consensus. These recommendations apply to all patients with suspected CNS demyelination of putative autoimmune etiology and an either monophasic or relapsing disease course. Given the very low pre-test probability [29], we recommend against general MOG-IgG testing in patients with a progressive disease course. In Table 2, we give a number of case vignettes of patients considered to be at high risk of MOG-EM to illustrate the broad spectrum of symptoms associated with that syndrome and the practical feasibility and relevance of the proposed criteria. In Table 3, we give a number of recommendations regarding assay selection, specimen sampling, and data interpretation. Finally, Table 4 lists conditions (“red flags”) that we believe are atypical for MOG-EM and should thus prompt physicians to challenge a positive MOG-IgG test result and seek a better explanation for the patients’ clinical and paraclinical findings.

Table 1 Recommended indications for MOG-IgG testing in patients presenting with acute CNS demyelination of putative autoimmune etiology
Full size table
Table 2 Case vignettes of patients at risk of MOG-IgG seropositivity (examples)
Full size table
Table 3 Recommendations on methodology, test parameters, specimen sampling and data interpretation
Full size table
Table 4 “Red flags”: conditions that should prompt physicians to challenge a positive test result (consider re-testing the patient, ideally using an alternative, i.e., methodologically different cell-based assay; in case of doubt, consider seeking expert advice from a specialized center)
Full size table

In practice, many patients diagnosed with AQP4-IgG-negative NMOSD according to the IPND 2015 criteria [30] will meet also the criteria for MOG-IgG testing given in Table 1 and should thus be tested. However, MOG-IgG testing should not be restricted to patients with AQP4-IgG-negative NMOSD. While this approach seems to offer simplicity, it would be inappropriate for several reasons: (1) The IPND criteria for AQP4-IgG-negative NMOSD demand dissemination in space, which would prevent testing of many patients with syndromes compatible with MOG-EM (e.g., patients with isolated longitudinally extensive transverse myelitis [LETM], isolated bilateral ON, or isolated brainstem encephalitis). (2) They include magnetic resonance imaging (MRI) criteria that are based on lesion distribution patterns observed in AQP4-IgG-positive NMOSD, some of which reflect areas of high AQP4 expression; however, AQP4 is not the target antigen in MOG-EM. Accordingly, lesion distribution may differ between NMOSD and MOG-EM and some MOG-EM patients do not satisfy these criteria (e.g., patients with recurrent bilateral non-longitudinal ON without chiasm involvement plus non-NMOSD-typical brain lesions, those with severe and recurrent non-longitudinally extensive myelitis, and those with ADEM-like presentation with severe brain and brainstem involvement but no area postrema lesion). (3) Such a recommendation would imply testing of all patients for AQP4-IgG before they were tested for MOG-IgG, which might unnecessarily delay diagnosis and treatment. (4) The criteria for AQP4-IgG-negative NMOSD require exclusion of other diagnoses; this would constitute a logical repugnancy, since a negative test for MOG-IgG would be a prerequisite for MOG-IgG testing. (5) Finally, but less importantly, using NMOSD criteria for diagnosing MOG-EM would, in addition to resulting in a substantial loss in sensitivity and specificity, also be confusing to non-experts, given that AQP4-IgG-positive NMOSD and MOG-EM are distinct diseases with different target antigens (AQP4 vs. MOG), pathophysiology (astrocytopathy vs. primary demyelination), and clinical spectra.

Alternatively, should we restrict MOG-IgG testing to patients with AQP4-IgG-negative NMO according to Wingerchuk’s 2006 criteria [31]? This would again result in a substantial loss of patients at high risk of MOG-EM, since those criteria require a history of both ON and myelitis and would thus be inappropriate. Of note, MOG-IgG testing in patients with seronegative NMO according to the 2006 criteria is already covered by our recommendation to test all patients with LETM for MOG-IgG (see Table 1), since the 2006 criteria strictly require a history of LETM in patients negative for AQP4-IgG.

Instead, we propose to base the indication for MOG-IgG testing in patients with suspected CNS demyelination on the presence of specific clinical and paraclinical findings that are considered typical for MOG-EM and/or atypical for conventional MS (see Table 1).

During the consensus-finding process, concerns were raised regarding inclusion of the following treatment-related indications for MOG-IgG testing in Table 1:

  1. a.

    Particularly good response to antibody-depleting therapies (plasma exchange [PEX], immunoadsorption [IA])

  2. b.

    Particularly good response to B cell-depleting therapies (rituximab, ocrelizumab, ofatumumab) but relapse immediately after re-occurrence of B cells

It was argued by some members of the panel that good responses to PEX, IA, or B cell depletion have also been observed in conventional MS. However, consensus was achieved that if present in addition to any of the indications listed in Table 1, good response to antibody or B cell-depleting treatments or IVIG further increases the pre-test likelihood of MOG-EM and thus supports the decision to test for MOG-IgG.

Taking into account that MOG-IgG serum concentrations depend on disease activity (with higher concentrations during acute attacks) and treatment status (with lower concentrations while on immunosuppression) as well as on assay sensitivity, we recommend re-testing patients during acute attacks or during treatment-free intervals and/or in a second cell-based assay if MOG-IgG was negative at first examination but MOG-EM is still suspected based on the list of indications given in Table 1 [3].

Only sparse data are available on the usefulness of regular monitoring of antibody titers in individual patients known to be positive for MOG-IgG. Median MOG-IgG titers have indeed been shown to be significantly higher during relapse than during remission [2], making regular MOG-IgG testing a potentially promising method for predicting attacks and monitoring treatment efficacy. However, there are several limitations: While titers > 1:2560 were found only during acute attacks in a recent study using a live cell-based assay [2], some patients still had relatively low titers during acute attacks and others had relatively high titers during remission, suggesting that additional factors such as blood–CSF barrier damage, T cell activation, antibody affinity, or complement-activating activity may be involved, with no general cut-off value for relapse induction [2]. In addition, treatment effects could play a role. Finally, intervals sufficient to detect imminent attacks in time have not yet been defined. Based on experience from studies on AQP4-IgG-positive NMOSD, in which serum antibody levels rise only very shortly before an attack [32], very close testing intervals may be required, which would make monitoring both expensive and challenging from a practical point of view. Accordingly, no general recommendation for regular monitoring of MOG-IgG titers for relapse prediction or treatment monitoring can currently be made.

Of note, some patients have been reported in whom MOG-IgG disappeared over time [2, 33,34,35]. Interestingly, many of these patients had monophasic disease. By contrast, MOG-IgG was detectable at the last follow-up in all patients (n = 18) with a relapsing disease course and available follow-up samples (mean interval 33 months since first testing; maximum follow-up period 10 years) in a recent study [2]. Disappearance of MOG-IgG after the initial attack might thus have prognostic implications, and re-testing of MOG-IgG-positive patients 6–12 months after the first attack might therefore be useful. However, there are some limitations: Most of the reported monophasic patients were children or juveniles, and most had ADEM. Moreover, no long-term data were provided for most cases. This is important, since titers may fall below cut-off temporarily following treatment with steroids, plasma exchange, or immunosuppressants (or even spontaneously) and rise again at a later disease stage; accordingly, (transient) seroconversion has also been observed in a few patients with relapsing disease [2, 14, 33]. It would therefore be challenging to base long-term treatment decisions solely on whether MOG-IgG disappears or not after a first attack. If long-term treatment with immunosuppressants or oral steroids is abandoned by reason of conversion to seronegativity, close monitoring of the patient’s MOG-IgG serostatus is highly recommended to confirm seronegativity in the long-term course. Before making a diagnosis of “monophasic” MOG-EM and thus a decision against long-term treatment, one should also take into account that the interval between first and second attack in relapsing MOG-EM varies considerably among patients, with the second clinical attack occurring only after an interval of several years in some cases [3].

Diagnostic criteria for MOG-EM

There is an unmet need for diagnostic criteria for MOG-EM. However, no specific clinical or radiological findings (except for the general requirement of a demyelinating CNS lesion) have yet been identified that are present in all MOG-IgG-positive patients and which would thus represent a diagnostic sine qua non. A lack of Dawson’s finger lesions and ovoid/round lesions on brain MRI have been proposed to be typical for MOG-EM, but this awaits confirmation in independent and larger cohorts [36, 37]. We propose that for the time being MOG-EM should be diagnosed in all patients who meet all of the following criteria:

  1. 1.

    Monophasic or relapsing acute ON, myelitis, brainstem encephalitis, or encephalitis, or any combination of these syndromes

  2. 2.

    MRI or electrophysiological (visual evoked potentials in patients with isolated ON) findings compatible with CNS demyelination

  3. 3.

    Seropositivity for MOG-IgG as detected by means of a cell-based assay employing full-length human MOG as target antigen

In patients with conditions considered “red flags” as defined in Table 4 and in whom MOG-IgG has not yet been confirmed in a second (and third if necessary), methodologically different cell-based assay, a diagnosis of “possible MOG-EM” should be made, especially in the context of clinical studies and treatment trials.

Limitations and caveats

It is a limitation that all recommendations given here are necessarily based on expert consensus, owing to a lack of systematic and prospective studies. Moreover, as a general caveat, it should be stressed that before a diagnosis of MOG-EM is made, all available information, including clinical, radiological, electrophysiological, and laboratory data, need to be taken into account, and differential diagnoses, some of which are listed in Table 4, need to be excluded. Most of the information given in a previous consensus paper on differential diagnosis in MS [38] is also pertinent to MOG-EM. Finally, while the criteria proposed here can certainly help in identifying pediatric patients at high risk of being positive for MOG-IgG, they are primarily intended for use in adults and adolescents. Indications for MOG-IgG testing in children do not need to be as rigorous as in adults, since MOG-IgG is thought to be much more common in children with acquired demyelinating disease (up to 70% depending on age) than in their adult counterparts (≤ 1% in Western countries; probably ≤ 5% in Japan and other Asian countries because of lower MS prevalence). In consequence, the risk of an unfavorable ratio of FP to TP results outlined above is lower in children. While ADEM is the predominant clinical association in young children, in older children with MOG antibodies there is a shift towards presentation with ON, myelitis, and/or brainstem symptoms [11].

Conclusion

Here, we give for the first time indications for MOG-IgG testing and propose preliminary criteria for the diagnosis of MOG-EM. While we believe that our recommendations are highly timely considering the large numbers of patients currently being tested, we are well aware that they reflect current knowledge in an evolving field and may need to be adjusted when new clinical and paraclinical data emerge and novel and optimized assays become available.