MEDEN-NMOSD NMA
Comparative Evaluation of Rituximab Versus Approved Therapies in Aquaporin-4-IgG-Positive Neuromyelitis Optica Spectrum Disorder: A Systematic Review and Network Meta-analysis
Bottom Line
This network meta-analysis of 8 trials suggests hazard-ratio point estimates numerically favoring eculizumab and ravulizumab over rituximab for time to first relapse, but all comparisons were statistically non-significant with extremely wide confidence intervals — the analysis is hypothesis-generating and does not establish superiority of any therapy over rituximab.
Major Points
- PRISMA-guided systematic review across PubMed, Scopus, CINAHL, EMBASE, Web of Science, Cochrane Library, and gray literature through 31 October 2024, updated 1 November 2025; frequentist network meta-analysis with random-effects model using the R netmeta package.
- From 6337 records screened, 8 trials were included: RIN-1 and Nikoo et al. (rituximab), SAkuraSky and SAkuraStar (satralizumab), PREVENT (eculizumab), N-MOmentum (inebilizumab), TANGO (tocilizumab), and CHAMPION-NMOSD (ravulizumab).
- Combination ± monotherapy network (5 trials): rituximab HR 5.00 (95% CI 0.25–101.01) vs ravulizumab ± IST, HR 1.17 (0.12–10.89) vs eculizumab ± IST, HR 0.29 (0.04–2.23) vs satralizumab ± IST.
- Monotherapy network (6 trials): rituximab HR 3.33 (0.13–83.16) vs ravulizumab, HR 1.59 (0.05–50.17) vs eculizumab, HR 0.31 (0.04–2.31) vs inebilizumab, HR 0.27 (0.03–2.21) vs satralizumab.
- P-score rankings placed ravulizumab and eculizumab highest but with high uncertainty; authors emphasize these should not guide clinical decisions.
- Substantial between-trial heterogeneity: variations in prior treatment, baseline ARR (1.9 in eculizumab/ravulizumab trials vs 1.4–1.5 in rituximab/satralizumab trials), relapse definitions and adjudication (investigator vs blinded committee), and ethnicity (RIN-1 was Japanese-only).
- Rituximab is likely to remain a cornerstone of NMOSD therapy globally because of its substantially lower cost and wider accessibility, particularly in resource-limited settings.
Design
Study Type: Systematic Review and Network Meta-analysis (frequentist, random-effects)
Randomization:
Blinding: Not applicable (meta-analysis of RCTs and open-label trials)
Enrollment Period: Databases searched through 31 October 2024; updated 1 November 2025
Follow-up Duration: Varied across included trials (e.g., 72 weeks in RIN-1; event-driven in PREVENT; ~50 weeks in CHAMPION-NMOSD)
Countries: International (trials from Japan, USA, Europe, and multinational cohorts)
Sample Size: 8
Analysis: Frequentist network meta-analysis using R netmeta v4.5.1, random-effects model; continuity correction (0.5) applied for zero-event arms; P-scores used for treatment ranking; PRISMA reporting; JBI critical appraisal for quality; OSF registration UW5MQ.
Inclusion Criteria
- Population: patients with AQP4-IgG-positive neuromyelitis optica spectrum disorder (NMOSD).
- Intervention: rituximab, eculizumab, inebilizumab, ravulizumab, satralizumab, or tocilizumab.
- Comparator: placebo or any other active intervention.
- Study type: randomized controlled trials or open-label trials.
- Outcomes reported that could be pooled (primarily time to first relapse).
- Peer-reviewed publications; abstracts from ACTRIMS, ECTRIMS, AAN, and EAN meetings (2021–2024) also screened.
Exclusion Criteria
- Observational studies (cohort, case-control, cross-sectional, case series, case reports).
- Studies not comparing a medication with placebo or another medication.
- Reviews, animal studies, hypothesis papers, in vitro studies, and books.
- Systematic reviews, meta-analyses, and other network meta-analyses.
- Non-peer-reviewed articles.
- Qualitative studies.
- Patients who were AQP4-IgG-negative were excluded from the NMA.
Arms
| Field | Rituximab (index therapy) | Ravulizumab ± IST | Eculizumab ± IST | Inebilizumab | Satralizumab ± IST | Tocilizumab (TANGO) |
|---|---|---|---|---|---|---|
| Intervention | Anti-CD20 monoclonal antibody; RIN-1 dosing per Tahara 2020 (n=19); Nikoo et al. dosing per Nikoo 2017. | Long-acting terminal complement C5 inhibitor (CHAMPION-NMOSD). | Terminal complement C5 inhibitor (PREVENT). | Anti-CD19 monoclonal antibody (N-MOmentum, monotherapy). | Anti-IL-6 receptor monoclonal antibody (SAkuraSky combination, SAkuraStar monotherapy). | Anti-IL-6 receptor monoclonal antibody vs azathioprine; excluded from primary network because comparator was azathioprine. |
| Duration | Trial-specific (RIN-1: 72 weeks) | Median ~50 weeks | Event-driven | 197 days randomized period | Trial-specific | Trial-specific |
Outcomes
| Outcome | Type | Control | Intervention | HR / OR / RR | P-value |
|---|---|---|---|---|---|
| Time to first relapse (hazard ratio, rituximab vs each comparator) — combination ± monotherapy network (5 trials: RIN-1, PREVENT, CHAMPION-NMOSD, SAkuraSky, SAkuraStar). | Primary | P-score Ranking (combination ± monotherapy): Ravulizumab ± IST 0.927 > Eculizumab ± IST 0.667 > Rituximab 0.618 > Satralizumab ± IST 0.287 · Interpretation: Point estimates favor ravulizumab and eculizumab over rituximab; wide CIs cross HR=1.0; no statistically significant difference. | |||
| Time to first relapse — monotherapy network (6 trials) | Secondary | ||||
| Rituximab vs Ravulizumab (monotherapy) | Secondary | 3.33 | NS | ||
| Rituximab vs Eculizumab (monotherapy) | Secondary | 1.59 | NS | ||
| Rituximab vs Inebilizumab (monotherapy) | Secondary | 0.31 | NS | ||
| Rituximab vs Satralizumab (monotherapy) | Secondary | 0.27 | NS | ||
| P-score ranking (monotherapy) | Secondary | ||||
| Zero-event arms | Secondary | ||||
| ARR, EDSS, steroid reduction, AQP4 titers, HACA, JCV, CD19/CD20 lymphocyte subsets | Secondary | ||||
| Note | Adverse | Safety comparisons were not performed. Authors state meaningful safety comparisons would require observational and real-world data, which were beyond the scope of this NMA. | |||
Subgroup Analysis
Two pre-specified networks: (1) combination ± monotherapy including all seropositive patients (RIN-1, PREVENT, CHAMPION-NMOSD, SAkuraSky, SAkuraStar); (2) monotherapy only (adds N-MOmentum). AQP4-IgG-negative patients were excluded from the NMA.
Criticisms
- Small rituximab evidence base: the only eligible rituximab trial for time-to-relapse (RIN-1) enrolled just 38 patients (19 per arm), limiting precision.
- Zero-event arms in RIN-1, PREVENT (monotherapy), and CHAMPION required a 0.5 continuity correction, which may bias HR estimates.
- Extremely wide confidence intervals (some crossing 100) make all pairwise comparisons statistically non-significant — point estimates and P-score rankings are hypothesis-generating only.
- Substantial between-trial heterogeneity in relapse definitions and adjudication (investigator vs blinded committee; imaging-supported in RIN-1 and N-MOmentum only) violates NMA transitivity assumptions.
- Baseline population differences: RIN-1 was Japanese-only (a population with better outcomes) and had no prior-relapse requirement, whereas PREVENT/CHAMPION required 1–3 relapses in the prior 12–24 months, with baseline ARR 1.9 vs 1.4–1.5.
- CHAMPION-NMOSD used the PREVENT placebo group as an external comparator and was open-label, whereas most included trials were double-blind and placebo-controlled.
- Time to first relapse ignores relapse severity, long-term ARR, cumulative disability, and quality of life — endpoints that may distinguish therapies clinically.
- Extension phases were excluded (no placebo arm), limiting long-term inference.
- No safety analysis, no direct head-to-head comparisons available, and only one trial per therapy (except satralizumab).
Funding
No funding or sponsorship was received for the publication of this article.
Based on: MEDEN-NMOSD NMA (Neurology and Therapy, 2026)
Authors: Barzegar M, Samadzadeh S, Audoin B, et al. (MEDEN study group)
Citation: Neurol Ther. 2026 Jul 16 (online). doi:10.1007/s40120-026-00989-x
Content summarized and formatted by NeuroTrials.ai.