To assess the strength of reported clinical prognosticators for developing relapsing myelin-oligodendrocyte-glycoprotein antibody associated disease (R-MOGAD) in 101 patients with MOGAD (86% with onset in adulthood) from three UK Specialist centres using a prospectively planned analysis design with validation analysis.

It is currently difficult to accurately predict who, after a first clinical attack of MOGAD, will develop R-MOGAD. Several clinical features have been reported as possibly predictive, but none are validated in clinical practice.

A multivariable binary logistic regression model using variables identified from a scoping literature review was fitted in an observational retrospective clinical dataset of patients with MOGAD (Nottingham MS & Neuroinflammation Centre - NUH), with validation analysis in two independent datasets (Walton NMOSD Specialist Centre-WSC; Imperial College London-ICL). Secondary analysis investigated time to first relapse using Cox proportional hazards on the combined cohort (n=101).

In the NUH dataset (n=33), a persistently positive MOG-IgG status, age at onset, optic neuritis at onset, and sex were not significant predictors of developing R-MOGAD. Only treatment with steroids≥10mg≥three months after the inaugural relapse was a significant negative predictor of developing R-MOGAD(p=0.006) with sensitivity 83% (95%CI:59-96) and specificity 73%(95%CI:45-92). To assess generalisability this predictor was tested in the two independent datasets (WSC: n=39; ICL: n=29) giving sensitivity and specificity estimates for not developing R-MOGAD of 63%(95%CI:38-84) and 85%(95%CI:62-97) (OR=9.7, 95%CI:2.1-45.4) in WSC, and 73%(95%CI:39-94) and 50%(95%CI:26-74) (OR=2.7, 95%CI:0.5-13.4) in ICL respectively . For the combined cohort(n=101), not receiving prednisolone ≥10mg≥3m had an OR=6.2 (95%CI:2.6-14.8; p<0.0001) of developing R-MOGAD, with hazard of relapsing β 0.47 (95%CI:0.24-0.91; p=0.024).