To characterize attack type recurrence in relapsing myelin oligodendrocyte glycoprotein antibody disease (MOGAD)

Core MOGAD attack types include optic neuritis (ON), transverse myelitis (TM), and brain involvement. While population-level attack type frequencies vary with age, it remains unclear whether patients show tropism for recurrence of their initial attack type.

A retrospective cohort study was conducted of MOGAD patients meeting 2023 diagnostic criteria seen at Mass General Brigham and Mayo Clinic. Generalized additive models (GAMs) were fitted to first attacks to quantify age-expected attack type probabilities, and conditional GAMs were fitted separately for each initial attack type subgroup using subsequent attacks; deviations between the two were computed to isolate the effect of initial attack type. Permutation tests (1,000 iterations) compared observed attack type recurrence rates within four age-at-onset subgroups (0-12, 12-18, 18-40, 40+ years) against a null distribution drawn from a corresponding age-stratified pool of first attacks to assess whether age at onset affects tropism. Bonferroni correction was applied for multiple comparisons.

Of 493 patients meeting criteria, 359 had relapsing disease, contributing 1,245 attacks (median follow-up: 7.1 years; interquartile range: 3.4-12.3). Relapses involving new regions occurred in 145 patients (40.4%), who had a younger median age at onset (27 vs. 34 years, p<0.001) and more likely initially presented with TM or brain (p<0.001). Patients demonstrated tropism for recurrence of their initial attack type beyond age-matched reference probabilities. Tropism strengthened with older age at onset: ON tropism was significant in all age groups (risk ratios: 1.14-1.29), TM tropism emerged after 40 (RR: 1.96 [1.27-4.06]), and brain tropism emerged after 18 (RRs: 2.91-3.31).

Attack type recurrence is shaped by both age and initial attack locations. Tropism strengthens with age, persisting even after accounting for age-dependent shifts in prevalence. These findings can inform prognostication, clinical trial design, and studies of pathophysiology.