DS-NMOSD COHORT
Comparative Effectiveness of Disease-Modifying Treatments in Double Seronegative Neuromyelitis Optica Spectrum Disorder
Bottom Line
In a multicenter international cohort of 103 patients with DS-NMOSD, anti-CD20 therapy (mostly rituximab) was associated with a ~98% lower relapse incidence rate ratio than the pretreatment period and substantially outperformed nonspecific immunosuppressants and MS DMTs, supporting B-cell depletion as first-line therapy.
Major Points
- Largest reported cohort of double seronegative NMOSD (n=103) from 6 countries; median follow-up 6 years.
- Anti-CD20 agents reduced relapse IRR to 0.02 (95% CI 0.01-0.04) vs pretreatment; nonspecific immunosuppressants 0.09 (0.07-0.13); MS DMTs 0.13 (0.06-0.30).
- Estimated adjusted ARR: anti-CD20 0.17, NSIS 0.76, MS DMTs 1.07; Bonferroni-corrected anti-CD20 vs NSIS p=0.003 and vs MS DMTs p=0.002.
- Cox HR for relapse: anti-CD20 vs MS DMTs 0.06 (95% CI 0.02-0.24, p<0.001); log-rank for anti-CD20 vs NSIS p=0.0001.
- Within most-used drugs, relapse-free status: rituximab 84.6%, mycophenolate 74.1%, azathioprine 46% (chi-squared p=0.00012); azathioprine vs rituximab HR 4.95 (p=0.00016).
- Traditional MS DMTs (interferon, glatiramer, fumarates, alemtuzumab) did not appear as harmful in DS-NMOSD as in AQP4+ NMOSD, but still inferior to anti-CD20.
- Class IV evidence supporting anti-CD20 as preferred first-line relapse prevention in DS-NMOSD.
Design
Study Type: Retrospective multicenter international cohort study (Class IV evidence)
Randomization:
Blinding: Unblinded (retrospective chart review)
Enrollment Period: Data collection December 2023 - November 2024
Follow-up Duration: Median 6 years (IQR 4.1-10.9)
Centers: 10
Countries: United States, Brazil, United Kingdom, Thailand, Turkiye, China
Sample Size: 103
Analysis: Negative binomial mixed-effects regression with adjustment for covariates; Cox proportional hazards; estimated marginal means with Bonferroni correction; treatment exposure modeled as time-varying
Inclusion Criteria
- Adults >=18 years at the time of data collection
- Fulfilled IPND-2015 AQP4-IgG-seronegative NMOSD criteria (2 core clinical syndromes, at least one supporting neuroimaging feature, no alternative diagnosis)
- Negative serum AQP4-IgG via cell-based assay at least once with subsequent consistently negative testing
- Negative serum MOG-IgG via cell-based assay at least once with subsequent consistently negative testing
- At least 12 months of clinical follow-up
- Diagnosis of DS-NMOSD confirmed as leading diagnosis at last follow-up by senior neuroimmunologists
Exclusion Criteria
- Fewer than 12 months of follow-up
- Missing treatment data
- Combination therapy with 2 or more concurrent DMTs
- Documented medication adherence concerns
- Suspicion for an alternative diagnosis such as atypical multiple sclerosis or neurosarcoidosis
Arms
| Field | Anti-CD20 | Nonspecific Immunosuppressants (NSIS) | MS DMTs | Control |
|---|---|---|---|---|
| Intervention | Rituximab or ocrelizumab (B-cell depletion) | Azathioprine, mycophenolate, methotrexate, cyclophosphamide, or mitoxantrone | Interferon-beta, glatiramer, fumarates, or alemtuzumab | No chronic DMT (time from disease onset to first DMT or full follow-up for untreated patients) |
| Duration | Variable, time-varying exposure modeled per patient | Variable, time-varying exposure modeled per patient | Variable, time-varying exposure modeled per patient | Variable |
Outcomes
| Outcome | Type | Control | Intervention | HR / OR / RR | P-value |
|---|---|---|---|---|---|
| Incidence rate ratio (IRR) of relapses on each DMT class compared with the pretreatment period (negative binomial mixed-effects regression, adjusted for covariates) | Primary | <0.001 for all DMT groups vs pretreatment | |||
| Adjusted annualized relapse rate (ARR) - Anti-CD20 | Secondary | Value: 0.17 (95% CI 0.07-0.40) | |||
| Adjusted ARR - NSIS | Secondary | Value: 0.76 (95% CI 0.40-1.43) | |||
| Adjusted ARR - MS DMTs | Secondary | Value: 1.07 (95% CI 0.39-2.93) | |||
| Pairwise ARR anti-CD20 vs NSIS (Bonferroni-corrected) | Secondary | 0.003 | |||
| Pairwise ARR anti-CD20 vs MS DMTs (Bonferroni-corrected) | Secondary | 0.002 | |||
| Pairwise ARR NSIS vs MS DMTs | Secondary | Not significant | |||
| Cox HR for relapse - anti-CD20 vs MS DMTs | Secondary | 0.06 (95% CI 0.02-0.24) | <0.001 | ||
| Cox HR for relapse - NSIS vs MS DMTs | Secondary | 0.35 (95% CI 0.12-1.04) | 0.059 | ||
| Kaplan-Meier anti-CD20 vs NSIS log-rank | Secondary | 0.0001 | |||
| Relapse-free status during rituximab treatment periods | Secondary | Value: 84.6% (44/52) | |||
| Relapse-free status during mycophenolate treatment periods | Secondary | Value: 74.1% (20/27) | |||
| Relapse-free status during azathioprine treatment periods | Secondary | Value: 46% (23/50) | |||
| Azathioprine vs rituximab (univariate Cox) | Secondary | 4.95 (95% CI 2.16-11.38) | 0.00016 | ||
| Mycophenolate vs rituximab (univariate Cox) | Secondary | 2.03 (95% CI 0.71-5.79) | 0.18 | ||
| Mycophenolate vs azathioprine (univariate Cox) | Secondary | 0.41 (95% CI 0.18-0.94) | 0.03 | ||
| Early (first-line) vs late rituximab | Secondary | 1.28 (95% CI 0.3-5.4) | 0.73 | ||
| Pretreatment duration (each additional month) | Secondary | <0.001 | |||
| NSIS discontinuation - driven by relapses | Adverse | 46% of NSIS discontinuations | |||
| NSIS discontinuation - adverse events | Adverse | 29% of NSIS discontinuations | |||
| NSIS discontinuation - pregnancy planning | Adverse | 14% of NSIS discontinuations | |||
| Anti-CD20 discontinuation | Adverse | Rare; too few events for class-level rate comparison | |||
| Note | Adverse | Detailed adverse event tables (e.g., infusion reactions, infections) were not reported in this retrospective cohort; safety data were limited to discontinuation reasons recorded in charts | |||
Subgroup Analysis
Sex, race, ethnicity, age at onset, suspicion for MOGAD, and clinical onset syndrome (ON, TM, simultaneous ON+TM, area postrema, brainstem, or cerebral) were not significantly associated with relapse rate after adjustment. Longer interval from disease onset to first DMT was associated with a slightly lower IRR (0.98 per month, p<0.001), interpreted as confounding by indication (milder disease delays DMT).
Criticisms
- Retrospective observational design with potential residual confounding and confounding by indication despite multivariable adjustment.
- Class IV evidence; no randomization or blinding.
- Roughly 23% of patients tested with fixed (rather than live) cell-based assays, which have lower sensitivity for MOG-IgG and AQP4-IgG; cannot fully exclude misclassified MOGAD or AQP4+ cases.
- Median time from onset to antibody testing was long (33.6 months for AQP4-IgG and 47.1 months for MOG-IgG), and only 27% were tested before any DMT; immunotherapy can cause MOG-IgG sero-reversion.
- Neuroimaging was not required for attack definition and MRI was available in only ~60% of attacks; possible inclusion of pseudorelapses.
- Corticosteroid maintenance regimens excluded from exposure modeling due to inconsistent documentation; acute rescue therapy data also incompletely captured.
- Patients on combination DMTs excluded, possibly skewing the cohort toward milder phenotypes; selection bias from IPND-2015 diagnostic requirement of 2 core syndromes biases toward relapsing disease.
- Small number of patients on MS DMTs (n=7), eculizumab (n=1), tocilizumab (n=2), and IVIg (n=3) limits inference for these classes.
- Adverse event reporting was limited to documented discontinuation reasons rather than systematic safety capture.
Funding
Not specified in the available full text
Based on: DS-NMOSD COHORT (Neurology Neuroimmunology & Neuroinflammation, 2026)
Authors: Mahler JV, Vallejos GB, Mikami T, et al.
Citation: Neurol Neuroimmunol Neuroinflamm 2026;13(2):e200514
Content summarized and formatted by NeuroTrials.ai.