← Back
NeuroTrials.ai
Neurology Clinical Trial Database

MOG DISCONTINUATION

Optimal strategies for treatment discontinuation in MOG antibody-associated disease

Year of Publication: 2026

Authors: Yeh WZ, Francis A, Cooper S, ..., Palace J

Journal: Brain

Citation: Yeh WZ, et al. Optimal strategies for treatment discontinuation in MOG antibody-associated disease. Brain. 2026. doi:10.1093/brain/awag006

Link: https://doi.org/10.1093/brain/awag006


Clinical Question

What is the optimal duration of maintenance immunomodulatory treatment in MOGAD before discontinuation, and what factors predict post-discontinuation relapse?

Bottom Line

In MOGAD, prolonged maintenance IT beyond 3 months significantly reduces post-discontinuation relapse risk. Treatment durations of 10-18 months (after a single/monophasic attack) and 20-30 months (in relapsing disease) balance relapse reduction against IT-related risks; relapsing disease course at discontinuation nearly doubles relapse risk (HR 1.95).

Major Points

  • Largest cohort study to date of MOGAD IT discontinuation: 190 patients, 236 discontinued IT intervals from Oxford NMO Highly Specialised Service (2010-2025).
  • 39% of discontinuations were followed by relapse at median 5.4 months.
  • Relapsing disease course at time of IT discontinuation independently doubled relapse risk (adjusted HR 1.95, 95% CI 1.25-3.06, P=0.003).
  • Longer IT duration prior to discontinuation was significantly protective; effect plateaued at 10-15 months (monophasic) and ~20 months (relapsing).
  • Estimated optimal treatment duration: 10-18 months for monophasic patients, 20-30 months for relapsing patients.
  • 5-year relapse-free survival for monophasic patients: 87.7% (15-mo IT) vs 47.9% (3-mo IT).
  • 5-year relapse-free survival for relapsing patients: 62.1% (24-mo IT) vs 24.3% (3-mo IT).
  • Negative pre-discontinuation MOG IgG1 was associated with lower relapse hazard in sensitivity analysis (HR 0.38, 95% CI 0.15-0.99).
  • Timing of onset attack acute treatment (early/intermediate/late) did NOT affect subsequent relapse risk after discontinuation.
  • Cohort was predominantly steroid-treated (84.7% used corticosteroids alone).

Design

Study Type: Retrospective single-service cohort study with prospective data collection

Randomization:

Blinding: None (observational)

Allocation: Not applicable (observational cohort)

Enrollment Period: January 2010 - May 2025

Follow-up Duration: Median disease duration 4.9 years (IQR 2.5-8.2) at last follow-up; minimum 12 months follow-up required

Centers: 1

Countries: United Kingdom

Sample Size: 190

Analyzed: 190

Analysis: Univariable and multivariable Cox regression with robust standard errors clustered by patient ID. Pre-discontinuation IT duration modelled with restricted cubic splines (3 knots) referenced to 3 months. Multiple imputation (100 datasets) for missing data. Two-tailed P<0.05 significant. R v4.4.1 (tidyverse, survival, mice, rms).


Inclusion Criteria

  • Fulfillment of 2023 MOGAD diagnostic criteria
  • Seen at Oxford Neuromyelitis Optica Highly Specialised Service between January 2010 and May 2025
  • Consent to participate in the Demyelinating Research Tissue Bank
  • ≥12 months of follow-up
  • Commenced and discontinued eligible maintenance IT (oral corticosteroids, azathioprine, mycophenolate, methotrexate, IVIG, or monoclonal antibody therapies)

Exclusion Criteria

  • Did not meet 2023 MOGAD criteria
  • <12 months follow-up
  • Never commenced or never discontinued eligible maintenance IT

Baseline Characteristics

Total patients: 190

Total discontinued IT intervals: 236

Patients contributing 1 interval: 153/190 (80.5%)

Patients contributing up to 3 intervals: 37/190 (19.5%)

Median age at onset (years): 31.5 (IQR 16.2-41.9)

Female: 115 (60.5%)

Median disease duration at last follow-up (years): 4.9 (IQR 2.5-8.2)

Relapsing course at last follow-up: 86 (45.3%)

Discontinuations when monophasic: 150/236 (63.6%)

Discontinuations when relapsing: 86/236 (36.4%)

Median relapses before discontinuation (relapsing group): 1 (IQR 1-2)

Relapse-free on treatment at discontinuation: 229/236 (97%)

Corticosteroids alone (all intervals): 200/236 (84.7%)

Corticosteroids alone (monophasic intervals): 139/150 (92.7%)

Corticosteroids alone (relapsing intervals): 61/86 (70.9%)

Non-steroid IT (alone or combined): 36/236 (15.2%)

Planned discontinuations: 210/236 (89%)

Discontinuations due to adverse events: 7 (3.0%)


Arms

FieldMonophasic course at IT discontinuationRelapsing course at IT discontinuation
N15086
InterventionDiscontinuation of maintenance IT (mostly oral corticosteroids, 92.7%) after a single attack and before any relapseDiscontinuation of maintenance IT (corticosteroids in 70.9%; non-steroid IT such as azathioprine or mycophenolate in the remainder) after ≥1 relapse
DurationVariable IT duration; optimal estimated 10-18 monthsVariable IT duration; optimal estimated 20-30 months

Outcomes

OutcomeTypeControlInterventionHR / OR / RRP-value
Time-to-relapse (new clinical attack >30 days after prior attack) following IT discontinuationPrimaryMonophasic at discontinuation: 27.8% relapsedRelapsing at discontinuation: 50% relapsed1.950.003
Overall post-discontinuation relapse rateSecondary92/236 (39.0%) discontinuations followed by relapse at median 5.4 months (IQR 1.4-20.1)
Univariable HR for relapsing vs monophasic course at discontinuationSecondary2.02<0.001
Univariable HR for negative MOG IgG1 pre-discontinuationSecondary0.430.045
Multivariable HR for negative MOG IgG1 (main analysis)Secondary0.530.14
Sensitivity analysis (MOG IgG1 available within 12 months of discontinuation, n=171 intervals): univariable HR for negative MOG IgG1Secondary0.340.024
Sensitivity analysis multivariable HR for negative MOG IgG1Secondary0.380.49
Monophasic subgroup: estimated 2-year relapse-free survival with 15-mo ITSecondary92.8% (95% CI 86.0-100%)
Monophasic subgroup: estimated 5-year relapse-free survival with 15-mo ITSecondary87.7% (95% CI 77.0-100%)
Monophasic subgroup: estimated 2-year relapse-free survival with 3-mo ITSecondary65.5% (95% CI 50.1-85.6%)
Monophasic subgroup: estimated 5-year relapse-free survival with 3-mo ITSecondary47.9% (95% CI 29.9-76.5%)
Relapsing subgroup: estimated 2-year relapse-free survival with 24-mo ITSecondary70.8% (95% CI 50.3-99.6%)
Relapsing subgroup: estimated 5-year relapse-free survival with 24-mo ITSecondary62.1% (95% CI 38.6-99.9%)
Relapsing subgroup: estimated 2-year relapse-free survival with 3-mo ITSecondary35.9% (95% CI 19.3-66.5%)
Relapsing subgroup: estimated 5-year relapse-free survival with 3-mo ITSecondary24.3% (95% CI 10.3-57.1%)
Pre-discontinuation MOG IgG1 positive/low-positive prevalenceSecondaryRelapsers 89.8% vs non-relapsers 71.5% (P=0.005)
Timing of onset attack acute treatment (early/intermediate/late)SecondaryNo significant effect on subsequent relapse risk
Number of relapses before discontinuation (relapsing subgroup)SecondaryNot associated with post-discontinuation relapse risk
Discontinuations due to adverse eventsSafety7/236 (3.0%)

Subgroup Analysis

Prespecified analyses in monophasic-at-discontinuation (n=150), relapsing-at-discontinuation (n=86), and an incident cohort (148 patients, 168 discontinuations; 44/148 [29.7%] relapsing at last follow-up). Incident-cohort analysis confirmed a minimum ~10-month IT duration was significantly protective vs 3 months in monophasic patients.


Criticisms

  • Single-center, retrospective observational design susceptible to selection and referral bias.
  • Cohort predominantly treated with corticosteroids (84.7%); limited data on steroid-sparing agents and monoclonal antibodies.
  • Missing pre-discontinuation MOG IgG1 status in 16.9% of intervals required multiple imputation.
  • Optimal duration estimates depend on modeling assumptions (restricted cubic splines with 3 knots) and reference to 3-month cutpoint.
  • Some patients contributed multiple discontinuation intervals (up to 3), addressed by clustered robust standard errors but still a potential source of within-patient correlation.
  • IT duration and time from last attack to discontinuation were highly correlated (r=0.87), limiting separation of their effects.
  • Wide confidence intervals for long-duration survival estimates (e.g., 5-year monophasic 15-mo IT CI up to 100%) reflect diminishing precision.
  • Discrepancy in sensitivity multivariable P-value (0.49 reported) versus HR (0.38, 95% CI 0.15-0.99) suggests possible typographical error in source.

Based on: MOG DISCONTINUATION (Brain, 2026)

Authors: Yeh WZ, Francis A, Cooper S, ..., Palace J

Citation: Yeh WZ, et al. Optimal strategies for treatment discontinuation in MOG antibody-associated disease. Brain. 2026. doi:10.1093/brain/awag006

Content summarized and formatted by NeuroTrials.ai.