Positive Predictive Value of Myelin Oligodendrocyte Glycoprotein Autoantibody Testing

Key Points Question What is the positive predictive value of myelin oligodendrocyte glycoprotein (MOG)–IgG1 testing in a clinical setting? Findings Of 1260 consecutive patients tested for MOG-IgG1 at the Mayo Clinic over 2 years, 92 (7.3%) were positive, 26 (28%) of whom had their results independently designated as false positive by 2 neurologists. The positive predictive value was 72% and varied with autoantibody titer (≥1:1000, 100%; 1:100, 82%; 1:20-40, 51%) and clinical–magnetic resonance imaging phenotypes at testing (pretest probability: high, 85%; low, 12%). Meaning False-positive MOG-IgG1 results are encountered in clinical practice; caution is advised before assigning a MOG-IgG1–associated disorder diagnosis in patients with low-titer positive results and atypical phenotypes.

M yelin oligodendrocyte glycoprotein (MOG)-IgG1associated disorder (MOGAD) is a distinct central nervous system (CNS)-demyelinating disease characterized by attacks of optic neuritis, myelitis, brain or brainstem inflammation, or combinations thereof. 1 Seropositivity for MOG-IgG1 confirms the diagnosis with a compatible clinical and radiologic phenotype, 2 with important therapeutic and prognostic implications. 3,4 International comparative studies of different MOG-IgG1 assays have shown that live cell-based assays yield the highest specificity for MOGAD. 5,6 However, MOGAD is rare, and indiscriminate testing for MOG-IgG1 may lead to falsepositive results despite high specificity. 7 The positive predictive value (PPV), which provides the likelihood that a positive test result is truly positive for the disease of interest, is arguably of greater clinical utility. Studies examining the PPV of MOG-IgG1 testing in clinical practice are lacking yet crucial to better interpret test results. We studied the PPV of MOG-IgG1 testing in a large cohort from a tertiary referral center.

Methods
The Mayo Clinic institutional review board approved the study. All patients provided written informed consent. Patients without research authorization were excluded.

Study Population
We included consecutive patients who were seen at the Mayo Clinic between January 1, 2018, and December 31, 2019, and tested for MOG-IgG1 as part of routine clinical care. Details are in the eFigure in the Supplement.

Autoantibody Testing
Testing for MOG-IgG1 was performed with a live cell-based flow cytometry or fluorescence-activated cell-sorting assay with fulllength MOG in its conformational form. Serum samples were screened at 1:20 dilution, and if the IgG-binding index (IBI; a ratio of median fluorescence intensities of MOG-transfected vs MOG-nontransfected cells) was 2.5 or more, they were diluted at 1:20, 1:40, 1:100, and 10-fold thereafter to establish end-titer values (last dilution with an IBI ≥2.5; reference value, <1:20). 5

Pretest Probability
Medical records were initially reviewed by 2 investigators (E.S. and M.B.) blinded to MOG-IgG1 serostatus to determine demographic, clinical, magnetic resonance imaging (MRI), and cerebrospinal fluid characteristics of patients at the time of testing. The pretest probability for MOGAD was considered high with acute attacks (nadir ≤1 month) of (1) optic neuropathy, (2) myelopathy, (3) brain or brainstem demyelination, (4) unilateral cortical encephalitis, or (5) multifocal CNS demyelination. 2,8 Patients with typical multiple sclerosis (MS) lesions on brain MRI 9,10 or other phenotypes were designated as having low pretest probability.

True-Positive vs False-Positive Assessment
The medical records and MRIs of individuals positive for MOG-IgG1 at last follow-up were independently reviewed by 2 neu-rologists (E.S. and E.P.F.), and true-positive results were defined per current international recommendations on diagnosis 2 or more recently recognized unilateral cortical encephalitis syndrome. 8 Alternative diagnoses or phenotypes inconsistent with MOGAD were designated as false-positive results. Diagnoses of MS were based on the revised McDonald criteria, 11 including typical MS lesions on MRI. 9 Consensus was reached for cases in which disagreement existed.

Statistics
Continuous and categorical variables were compared using Wilcoxon rank sum and Fisher exact tests, respectively. A P value less than .05 was considered statistically significant. Correlations were assessed by Spearman ρ. The PPV (truepositive results divided by total positive results) and specificity (true-negative results divided by true-negative results plus false-positive results) were reported; 95% CIs were calculated using the score method (SAS version 9.4 [SAS Institute]). Graphs were built with R version 3.6.2 (R Foundation for Statistical Computing).

Key Points
Question What is the positive predictive value of myelin oligodendrocyte glycoprotein (MOG)-IgG1 testing in a clinical setting?
Meaning False-positive MOG-IgG1 results are encountered in clinical practice; caution is advised before assigning a MOG-IgG1-associated disorder diagnosis in patients with low-titer positive results and atypical phenotypes.

True-Positive and False-Positive Results
The 2 independent raters agreed on 91 of 92 cases (99%) for true and false positivity. The single discordant case, with steroid-responsive subacute progressive encephalitis and brain biopsy results showing demyelination, remyelination, and perivascular lymphocytic inflammation, was designated as having a true-positive result after consensus. Twenty-six patients (28%) had false-positive results, including 11 with typical MS, 3 with infarction (1 in the optic nerve, 1 in the brainstem, and 1 in the spinal cord); 2 with B 12 deficiency, 2 with biopsyproven neoplasia (1 with a histiocytic neoplasia and 1 with glioma), 1 with genetically confirmed adrenomyeloneuropathy, 1 with an isolated caudate lesion (which was noninflammatory on biopsy), 1 with postradiation myelopathy, 1 with varicella-zostervirus-associated myelitis, 1 with probable neurosarcoidosis, 1 with idiopathic progressive cerebellar degeneration, 1 with idiopathic progressive myelopathy, and 1 with a nonneurologic syndrome. The median antibody titer was higher with true-positive results (1:100 [range, 1:20-1:

Discussion
This study confirms MOG-IgG1 as a highly specific biomarker of MOGAD in clinical practice. The increasing PPV with higher titers and pretest probability, and excellent agreement of independent raters on MOGAD phenotypes supports it being a distinct disease. 2 However, more than one-quarter of positive results might be false in a high-throughput setting, and given the frequency of MOG-IgG1 testing requests (~20 000 samples/year in our laboratory), clinicians should be aware. Indiscriminate MOG-IgG1 testing is not recommended, and caution is advised when interpreting low-titer positivity with atypical phenotypes. While the absolute requirement of MOG-IgG1 positivity for MOGAD diagnosis hinders sensitivity calculation, identification of alternative diagnoses or incompatible clinical-MRI phenotypes allows recognition of false-positive results for speci-ficity and PPV calculation. Our specificity was similar to those of prior studies (97.8% vs 99.6%-100% 5,6 ) that used experimental populations, but PPV in clinical practice was lower (72% vs 95.5%-100% 5,6 ) because it was affected by disease prevalence and ordering practices. 2 Increasing the MOG-IgG1 IBI screening cutoff value would increase the PPV but exclude many true MOGAD cases, and stratifying results by low (1:20-1:40) and high (≥1:100) titers might be preferred.
Multiple sclerosis was overrepresented among patients with false-positive results, which should dissuade clinicians from uniform ordering of MOG-IgG1 testing in patients with typical MS. 9 Prior studies 12-14 have shown a lower frequency of MOG-IgG1 positivity via live cell-based assays among patients with MS (0%-2% vs 2.5% in our study), possibly from differences in assay cutoff values, inclusion criteria, or referral bias. In clinical practice, MOG-IgG1 positivity requires careful evaluation for typical MOGAD clinical-MRI characteristics and red flags that argue against MOGAD (eg, a progressive course). 2 Our findings have major implications for future updates of international consensus diagnostic criteria in MOGAD. More stringent clinical, radiologic, and laboratory requirements accompanying MOG-IgG1 positivity would help prevent misdiagnosis, inappropriate treatment, and enrollment of individuals with false-positive results in clinical trials that could hinder their success.

Limitations
Our study has limitations. We did not evaluate cerebrospinal fluid, in which isolated MOG-IgG1 positivity may occur, 15 but serum testing is generally recommended. 2 The PPV observed in this study depends on the titer cutoff value used for MOG-IgG1 positivity and selection of different cutoff values would yield different PPVs. The PPV is also strongly associated with the population tested and ordering practices, and in other populations, in which MOGAD is less represented, the PPV could be lower. Similar to diagnostic criteria for other CNS-demyelinating diseases, the definition of true MOGAD was supported by exclusion of alternative diagnoses, and future identification of new CNS-demyelinating syndromes might reduce the PPV over time. Positivity for MOG-IgG1 was not confirmed in a second laboratory, but our assay is comparable with those of other centers, 5,6 and repeated testing in clinical practice may not be available or practical, still requiring careful assessment for MOGAD phenotypes.

Conclusions
Assessment of MOG-IgG1 by live cell-based assay is highly specific for MOGAD diagnosis but has a potential risk of false-positive results when tested indiscriminately in clinical practice. Future multicenter efforts should focus on assay improvements that could reduce false-positive results.