Validation of the international MOGAD panel proposed criteria: a single-centre US study

Abstract

Background:

Myelin Oligodendrocyte Glycoprotein (MOG) antibody-associated disease (MOGAD) is a demyelinating disorder of the central nervous system. We aimed to evaluate the diagnostic performance of recently proposed MOGAD diagnostic criteria in a real-world patient cohort at a tertiary referral center.

Methods:

We identified all patients who were evaluated at Johns Hopkins and were MOG-IgG seropositive by cell-based assay. We retrospectively applied the proposed MOGAD diagnostic criteria.

Results:

Among the 122 patients included in this study, 109 fulfilled the diagnostic criteria. Of 64 patients with clear positive MOG-IgG titer, 63 patients also satisfied the supporting clinical or MRI features. Of 58 patients with low positive or unknown MOG-IgG titer, 46 met criteria by fulfillment of the supporting features. The medical records were independently reviewed by two investigators with expertise in demyelinating disease, and patients were assigned empirical clinical diagnoses, with agreement with the application of the MOGAD diagnostic criteria in the majority of cases (90%).

Conclusions:

Our findings support the diagnostic utility of the proposed MOGAD diagnostic criteria. Patients with MOGAD met the supporting clinical or MRI features almost universally, which suggests that the criteria can be used to accurately differentiate MOGAD from mimics with low-titer MOG-IgG seropositivity.

Introduction

Myelin Oligodendrocyte Glycoprotein (MOG) antibody-associated disease (MOGAD) is a distinct demyelinating disorder of the central nervous system (CNS).[1] Clinical manifestations include optic neuritis (ON), transverse myelitis (TM), acute disseminated encephalomyelitis, brainstem or cerebellar attacks and cerebral cortical encephalitis. MOG-IgG is detected by cell-based assays (CBA) using full-length MOG-IgG in its native conformational state.[2] While MOGAD shares features with multiple sclerosis (MS) and aquaporin-4 (AQP4)-IgG seropositive neuromyelitis optica spectrum disorder (NMOSD), the underlying pathophysiology and approach to treatment differ, hence accurate and timely diagnosis is critical.[3]

Criteria for the diagnosis of MOGAD were recently proposed by the International MOGAD Panel.[4] We sought to evaluate the diagnostic performance of the proposed MOGAD diagnostic criteria in a real-world cohort of patients evaluated for MOGAD at our tertiary referral center.

Methods

We identified all patients who were evaluated at Johns Hopkins and were MOG-IgG seropositive by CBA. We retrospectively reviewed their medical records and extracted pertinent data, including demographics, clinical characteristics, and magnetic resonance imaging (MRI) findings. Patients with insufficient records were excluded. MOG-IgG was tested in the serum by CBA for all patients per routine clinical practice. For patients with MOG-IgG tested on multiple occasions, the highest available titer was used. The classification of MOG-IgG titer to high vs low was based on proposed cut-offs depending on assay type (≥1:100 considered high titer for the assays used in our cohort).[4]

The proposed 2023 MOGAD diagnostic criteria were retrospectively applied to our cohort to evaluate their diagnostic performance. For the purposes of this work, we evaluated the performance of combined criterion A (presence of core demyelinating event) and criterion B (MOG-IgG serology and supporting clinical or MRI features) alone, as well as in combination with criterion C (exclusion of other diagnoses), as this step requires clinical interpretation and expertise in neuroimmunological conditions.

The medical records were also independently reviewed by two investigators with expertise in demyelinating diseases and patients were assigned empirical clinical diagnoses based on their clinical presentations, para-clinical findings and clinical course (96% concordance). Disagreements were resolved by consensus discussion.

Statistical analyses were performed using Stata version 16 (StataCorp, College Station, TX). The study was approved by the Johns Hopkins Institutional Review Board and patient consent was waived for this retrospective chart review.

Results

We identified 1,890 unique patients with MOG-IgG CBA testing results available between July 2019 and September 2023, of whom 130 (6.9%) were MOG-IgG seropositive on at least one occasion. Eight were excluded due to insufficient records and the remaining 122 were included in the analysis (Figure 1). Demographics for this cohort were as follows: 54% female, mean±SD age at initial attack 33.7±17.9 years (Supplemental Table). There were 27 pediatric patients (22%) at initial attack. MOG-IgG was tested with live CBA in 88 patients (72%) and fixed CBA in 34 patients (28%). There were 64 patients with clear positive MOG-IgG titer (57/64; 89% tested via live CBA), 52 patients with low positive titer (25/52; 48% tested via live CBA) and 6 patients with unknown titer. There were 63 patients (52%) with a relapsing disease course. Median (IQR) time from attack to MOG-IgG testing was 34 (5–281) days. Median time from symptom onset to MRI of the part of the neuraxis corresponding to the clinical attack was 12 days for TM, 8 days for cerebral/brainstem attacks and 7 days for ON. Median (IQR) duration of follow-up since first attack was 2.3 (0.3, 5.1) years.

Figure 1.

Flowchart of study cohort and application of MOGAD criteria. JH: Johns Hopkins; MOG: myelin oligodendrocyte glycoprotein; CBA: cell-based assay; MOGAD: MOG antibody-associated disease; MS: multiple sclerosis; CIS: clinically isolated syndrome. a 57/64 tested via live CBA; b 25/52 tested via live CBA; c 6/6 tested via live CBA; d 1/3 tested via live CBA; e 3/4 tested via live CBA; f 57/63 tested via live CBA; g 25/46 tested via live CBA; h 53/59 tested via live CBA; i 22/42 tested via live CBA. CBA, cell-based assay; CIS, clinically isolated syndrome; JH, Johns Hopkins; MOG, myelin oligodendrocyte glycoprotein; MOGAD, MOG antibody-associated disease; MS, multiple sclerosis.

Among the 122 patients included in this study, 109 fulfilled the diagnostic criteria either at the time of initial presentation (n=96) or during subsequent attacks (n=13). For all 13 patients who did not meet criteria during their first attack, the attack was remote with inadequate clinical information and/or MRI data (median (IQR) time to subsequent attack that led to MOGAD diagnosis was 10.7 (0.9–15.8) years). Of 64 patients with clear positive MOG-IgG titer, one patient did not fulfill MOGAD diagnostic criteria as they had not experienced a core demyelinating event, while 63 patients had experienced a core demyelinating event and fulfilled criteria. All 63 patients also satisfied the supporting clinical or MRI features. In accordance with the proposed criteria, patients with positive MOG-IgG but unknown titer were handled similarly to low-titer, with requirement of supporting features to meet criteria. Of 58 patients with low positive or unknown MOG-IgG titer, 46 met criteria by fulfillment of the supporting features and 12 did not meet criteria (although 5 had missing or insufficient investigations; details below) (Figure 1).

The empirical clinical diagnosis agreed with the application of the MOGAD diagnostic criteria in 90% of cases (110/122): 101 patients met diagnostic criteria and were also assigned an empirical clinical diagnosis of MOGAD (“true positive”), and nine patients did not meet diagnostic criteria and were also assigned alternative clinical diagnoses (“true negative”; 1 migraines with non-specific white matter changes, 1 functional vision loss, 2 MS, 1 retinal vasculitis, 1 spondylotic myelopathy, 1 epidural lipomatosis, 2 unclear diagnosis) (Supplemental Figure).

There were four patients that were assigned a clinical diagnosis of MOGAD but did not meet diagnostic criteria (“false negative”). All four patients had low titer MOG-IgG seropositivity and insufficient investigations: one had clinically diagnosed unilateral ON and TM and did not meet supporting criteria given negative MRI but MRI was delayed and obtained after pulse steroids, and three otherwise met criteria but AQP4-IgG was not tested or not available.

There were eight patients who met criteria but were either assigned an alternative clinical diagnosis or an alternative clinical diagnosis could not be excluded (“false positive”). The most common alternative diagnosis was MS. Five patients were given an empiric clinical diagnosis of MS or MS could not be ruled out (Table 1; Supplemental Figure). All five patients had CNS demyelinating lesions with appearance typical for MS, positive oligoclonal bands (OCB), met 2017 McDonald criteria for MS diagnosis[5], and two accrued additional asymptomatic lesions over time. For three additional patients with low-titer MOG-IgG seropositivity, an alternative diagnosis could not be excluded (Table 1; Supplemental Figure): one patient was diagnosed with MOGAD vs neurosyphilis, one patient was diagnosed with MOGAD vs clinically isolated syndrome and one patient had unilateral ON consistent with MOGAD but also simultaneous cauda equina post-gadolinium enhancement with marked albumino-cytological dissociation which are features that are uncommon in MOGAD, hence an alternative clinical diagnosis could not be excluded, although such features have been described in MOGAD.[6]

Table 1.

Alternative clinical diagnoses for patients who met MOGAD diagnostic criterion A & B

Patients with clear positive MOG-IgG titer (n=4)

Patients with low positive or unknown MOG-IgG titer (n=4)

• MS (31 y.o. with brainstem attack with ill-defined pontine lesion, additional CNS demyelinating lesions with appearance typical for MS, positive OCB) • MOGAD vs MS (45 y.o. with unilateral longitudinally extensive ON with perineuritis and short segment TM, CNS demyelinating lesions with appearance typical for MS, positive OCB) • MOGAD vs MS (32 y.o. with sequential ON with optic disc edema; CNS demyelinating lesions with appearance typical for MS, positive OCB, accrued additional asymptomatic lesions over time) • MOGAD vs MS (35 y.o. with brainstem attack with ill-defined pontine lesion that became less prominent on subsequent imaging, additional ill-defined periventricular and juxtacortical CNS demyelinating lesions, positive OCB)

• MS (46 y.o. with simultaneous bilateral ON, CNS demyelinating lesions with appearance typical for MS, positive OCB, accrued additional asymptomatic lesions over time) • MOGAD vs neurosyphilis (52 y.o. with bilateral ON with perineuritis with a slowly progressive clinical course, limited response to corticosteroids, positive syphilis testing) • MOGAD vs clinically isolated syndrome (30 y.o. with unilateral short-segment ON, no OCB, no additional CNS lesions, only met the supporting clinical feature of optic disc edema). • Unclear diagnosis (29 y.o. with unilateral ON consistent with MOGAD but also simultaneous cauda equina enhancement with marked albumino-cytological dissociation which are features that are uncommon in MOGAD)

MOG: myelin oligodendrocyte glycoprotein; MOGAD: MOG antibody-associated disease; MS: multiple sclerosis; CNS: central nervous system; OCB: oligoclonal bands; ON: optic neuritis; TM: transverse myelitis

Cerebrospinal fluid (CSF) analysis was available for 99 patients, 84 of which were given a diagnosis of MOGAD. CSF-specific OCB were absent in 76/84 (90%) patients who were given a diagnosis of MOGAD vs 8/15 (53%) patients who were given an alternative diagnosis. All patients who were given a clinical diagnosis of MS had CSF-specific OCB.

Overall, for patients with sufficient investigations, the sensitivity of the proposed MOGAD diagnostic criteria was 100%. The specificity of combined criterion A and B was 53%, while the specificity of the diagnostic criteria as proposed (including criterion C which requires clinical interpretation to exclude alternative diagnoses) was 100%.

Discussion

Our findings support the diagnostic utility of the proposed MOGAD diagnostic criteria. In our cohort, the criteria had excellent sensitivity and captured almost all patients with an empirical clinical MOGAD diagnosis. Additional supporting clinical or MRI features are needed to establish a diagnosis of MOGAD for patients with low positive MOG-IgG titer, since low titers are frequently discordant between assays and less reproducible.[2] In our study, patients with high MOG-IgG titers and historical MOGAD diagnosis met the supporting features almost universally, which suggests that the selection of supporting criteria is appropriate and they can be used to differentiate MOGAD from mimics with low-titer MOG-IgG seropositivity. These findings are similar to studies reporting the application of the MOGAD criteria in cohorts from Korea and the UK.[7,8] Pursuing a complete investigation (e.g. dedicated orbital fat-saturated images of the optic nerves and fundoscopy in ON cases) and ruling out AQP4-NMOSD is crucial in cases with possible MOGAD, however, in real-world settings, investigations may be incomplete, leading to lower “real-world” sensitivity of the proposed diagnostic criteria. Clinical judgement is necessary in patients with delayed imaging and/or delayed MOG-IgG serologic testing, since MOG-IgG titers may decrease or become negative over time, and MRI lesions are often transient in MOGAD, hence the diagnosis might be missed in patients with delayed investigations.[8,9] Conversely, lesion evolution over time can be helpful diagnostically, since complete lesion resolution is frequent in MOGAD but rare in AQP4-NMOSD and MS.[10]

The criteria rely on clinical interpretation to exclude other diagnoses including MS. In our cohort, this was a crucial step, since the specificity of combined criterion A and B was 53%, but increased to 100% when clinical judgement was used to exclude alternative diagnoses. A prior smaller study also showed only modest specificity (55.5%) of the criteria, but criterion C was not taken into account.[11] In our cohort, MS represented the majority of the alternative diagnoses. However, differentiating MS from MOGAD may be challenging in some cases, as there may be overlapping clinical and imaging features (such as optic disc edema in ON or ill-defined brainstem lesions) and MOG-IgG seropositivity has been reported in 0.3–2% of MS patients.[12,13] Of note, the 2017 McDonald criteria also recommend that alternative diagnoses should be excluded prior to MS diagnosis.[5,14] A multicenter European study reported findings from a cohort of 28 patients who met diagnostic criteria for MS and were MOG-IgG seropositive but 21 (75%) also had at least one supporting feature compatible with MOGAD diagnosis.[15] Both in their cohort and our cohort, patients with discordant diagnoses had positive CSF-specific OCB. This highlights the importance of clinical interpretation and expertise to establish accurate diagnoses in equivocal cases, especially since it is unclear at this time if MOGAD-MS overlap syndromes represent a distinct clinical entity. This may also present challenges for arriving at accurate diagnoses in settings lacking this clinical expertise.

Our study has several limitations. This was a retrospective study from a single tertiary referral center. Our study population was limited to patients with MOG-IgG seropositivity which is a sine qua non for the diagnosis of MOGAD. Naturally, this approach enriches the cohort for those with a likely diagnosis of MOGAD, however rates of MOG-IgG seropositivity were rather low amongst those tested (6.9%), increasing the likelihood of false-positive results. As such, our performance estimates of the specificity of the proposed diagnostic criteria relate to their application to a MOG-IgG seropositive population when MOG-IgG testing is performed broadly in a population suspected of underlying demyelinating disease. Furthermore, we studied a real-world cohort with variable intervals between clinical attacks and imaging/serologic MOG-IgG testing. MOG-IgG was tested with variable assays, including both fixed and live CBA assays. This is an important limitation since fixed CBA may have lower sensitivity and specificity than live CBA, which may result in misdiagnosis.[16,17] Additional studies are needed to establish the impact of discordant assay results on the application of the diagnostic criteria.[15] Moreover, given the lower proportion of pediatric cases in our cohort (22%), generalizability of our results may be limited in pediatric MOGAD populations. Nevertheless, we believe that our work provides a robust and systematic evaluation of the diagnostic performance of the proposed MOGAD diagnostic criteria in a real-world cohort.

Overall, the proposed MOGAD diagnostic criteria are a key step towards accurate identification of patients with MOGAD and were aligned with empirical clinical MOGAD diagnosis in our cohort.