Investigating the 2023 MOGAD Criteria in Children and Adults With MOG-Antibody Positivity Within and Outside Attacks

Abstract

Background and objectives: The 2023 criteria for myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD) perform well in adults but have not been assessed in children. Methods: This prospective observational nationwide study includes children and adults with demyelinating syndromes or encephalitis, whose serum or CSF was found MOG-immunoglobulin G (IgG) positive at Institut d'Investigacions Biomèdiques August Pi i Sunyer-Hospital Clínic of Barcelona (Spain). Exclusion criteria were lack of clinical information and follow-up <1 year, and serum unavailable for antibody testing. The primary outcome was to assess the accuracy of the 2023 MOGAD criteria, using as gold standard the most plausible diagnosis after a follow-up >1 year. MOGAD criteria were retrospectively applied assessing core syndromes, supportive clinical-radiological features, and MOG-IgG titers. Patients tested ≤3 months of a disease attack (acute phase) or afterward (remission) were considered separately. The positive predictive value (PPV) of the criteria (true-positive [patients classified as MOGAD and MOGAD diagnosis last follow-up] divided by total positive [all patients classified as MOGAD]), and its 95% CI, was calculated with the Wilson procedure. Results: A total of 257 patients (133 children) were included in the study (median age 15 years [interquartile range 6-38], 54% female). Among 202 patients assessed during a disease attack, 158 (78%) had high MOG-IgG serum titers, 36 (18%) low titers, and 8 (4%) antibodies only in CSF. No differences were identified between patients with high and low titers, but those with low titers were more likely to have an alternative diagnosis at last follow-up (2/36 [6%] vs 0/158, p = 0.012). Supportive features were present in 230 of 257 (89%) patients, regardless of age, MOG-IgG titers, and core syndromes except for optic neuritis in adults whose assessment with orbital MRI was not systematic. Overall, 240 of 257 (94%) patients were well classified by the MOGAD criteria (e.g., 236 eventually having MOGAD and 4 alternative diagnoses), and 17 were wrongly classified (e.g., 11 eventually having MOGAD and 6 alternative diagnoses). Although the criteria classified better during disease attacks than during remissions (187 [96%] vs 49 [89%] serum MOG-IgG-positive patients were well-classified, p = 0.038), the PPV was high in both settings (99% [95% CI 97-100] vs 98% [95% CI 89-100]).