← Back
NeuroTrials.ai
Neurology Clinical Trial Database

RIFUND-MS

Dimethyl Fumarate, But Not Rituximab, Reduces Serum GFAP Levels and PIRMA in Relapsing-Remitting MS

Year of Publication: 2026

Authors: Shawket F, Lycke J, Salzer J, ..., Svenningsson A

Journal: Annals of Clinical and Translational Neurology

Citation: Ann Clin Transl Neurol 2026;0:1-11

Link: https://doi.org/10.1002/acn3.70395


Clinical Question

Does rituximab or dimethyl fumarate better reduce astrocytic injury biomarkers and progression independent of relapse/MRI activity in RRMS?

Bottom Line

In RRMS, DMF but not RTX significantly reduced serum GFAP over 2 years, and RTX was associated with 3-fold higher risk of PIRMA despite superior MRI suppression. These findings challenge the assumption that MRI efficacy equals neuroprotection, and implicate GFAP as a clinically meaningful biomarker of DMT neuroprotective effects.

Major Points

  • DMF significantly reduced sGFAP by ~18% over 2 years while RTX produced no significant reduction (3.6%), despite RTX achieving far superior MRI lesion suppression
  • Both agents reduced sNfL comparably (~46–51%) with no significant difference between arms (p=0.06)
  • RTX-treated patients had 3.3-fold higher risk of PIRMA compared to DMF (HR 3.3, 95% CI 1.1–10, p=0.04)
  • PIRA (HR 2.6, 95% CI 0.9–7.3, p=0.06) and CDW (HR 2.6, 95% CI 0.9–7.3, p=0.07) trended toward DMF benefit but did not reach significance
  • Higher serum GFAP was an independent predictor of PIRMA (HR 2.60 per doubling, 95% CI 1.28–5.30, p=0.008)
  • DMF's Nrf2-activating mechanism may confer astrocyte-protective effects beyond its anti-inflammatory action
  • MRI efficacy alone does not predict neuroprotective biomarker profile or disability-independent progression risk

Design

Study Type: Biomarker sub-study of a rater-blinded, randomized, open-label trial (RIFUND-MS RCT)

Randomization: 1

Blinding: Rater-blinded; treating physicians and patients not blinded

Allocation: 1:1

Enrollment Period: July 2016 – April 2021

Follow-up Duration: 24 months

Centers: 17

Countries: Sweden

Sample Size: 200

Analyzed: 197

Analysis: Intention-to-treat; per-protocol and switch subgroup sensitivity analyses

Power Calculation: Not reported in this biomarker sub-study. sNfL was a predetermined secondary endpoint of the RIFUND-MS trial; sGFAP was analyzed post hoc.

Registration: NCT02746744


Inclusion Criteria

  • Diagnosis of RRMS or CIS
  • Eligible to start disease-modifying therapy (per RIFUND-MS main trial protocol; see Svenningsson 2022, Lancet Neurology)

Exclusion Criteria

  • Detailed exclusion criteria not reported in this biomarker sub-study; see main RIFUND-MS trial publication (Svenningsson 2022, Lancet Neurology)

Baseline Characteristics

CharacteristicRituximab (n=99)Dimethyl Fumarate (n=98)Biomarker baseline — Rituximab (n=83)Biomarker baseline — Dimethyl Fumarate (n=86)
Age (years), mean (SD)33.5 (7.7)33.4 (7.7)
Female, n (%)67 (68)64 (65)
MS duration (yrs from diagnosis), mean (SD)1.6 (3.2)1.8 (2.6)
EDSS, mean (SD)1.61.7
EDSS, median (IQR)1.5 (1–2.5)2 (1–2)
Treatment naïve ever, n (%)6 (6)4 (4)
≥1 relapse in last year, %7769
≥1 GD+ MRI lesion, n (%)41 (41)32 (33)
sNfL, geometric mean ng/L (95% CI)11.7 (10.3, 13.2)10.0 (8.7, 11.5)
sGFAP, geometric mean ng/L (95% CI)64.5 (57.0, 73.0)63.9 (57.9, 70.4)

Arms

FieldRituximabDimethyl Fumarate
N9998
InterventionRituximab 1000 mg IV at baseline, then 500 mg IV every 6 monthsDimethyl fumarate 240 mg twice daily orally
Duration24 months24 months

Outcomes

OutcomeTypeControlInterventionHR / OR / RRP-value
Reduction of serum NfL and serum GFAP in each treatment arm over 24 months (linear mixed model, log2-transformed concentrations). sNfL was a predetermined secondary endpoint of the RIFUND-MS trial; sGFAP was analyzed post hoc.PrimaryRTX: sGFAP −3.6% (NS); sNfL −50.7% (p<0.001)DMF: sGFAP −18.4% (p<0.001); sNfL −46.4% (p<0.001)Global treatment difference favors DMF for sGFAP; no significant difference for sNfLsGFAP DMF p<0.001, RTX p=0.81; global sGFAP treatment difference ITT p=0.02, PP p=0.004. sNfL both arms p<0.001; global sNfL treatment difference p=0.06
Serum NfL change over 24 months (ITT)SecondaryRTX: −50.7%DMF: −46.4%Both arms p<0.001; global between-arm p=0.06 (NS)
PIRMA (Progression Independent of Relapse and MRI Activity)SecondaryRTX: higher risk (reference arm DMF)DMF: reference0.04
PIRA (Progression Independent of Relapse Activity)SecondaryRTX: higher riskDMF: reference0.06 (NS)
Confirmed Disability Worsening (CDW)SecondaryRTX: higher riskDMF: reference0.07 (NS)
SDMT worseningSecondaryRTX vs DMFDMF: reference0.51 (NS)
sGFAP change (per doubling) as predictor of PIRMASecondary0.008
sGFAP change (per doubling) as predictor of PIRASecondary0.07
sGFAP change (per doubling) as predictor of CDWSecondary0.08
sNfL change (per doubling) as predictor of PIRMASecondary0.91

Subgroup Analysis

Per-protocol sensitivity analysis (n=96 RTX vs n=57 DMF) confirmed the sGFAP treatment difference (DMF −23.1% [−33.0, −11.7] p<0.001; RTX −3.8% [−13.9, 7.6] p=0.81 at month 24). Switch subgroup (n=41 DMF→RTX switchers vs PP RTX) showed no significant between-arm difference in sNfL or sGFAP; initial GFAP decline in switchers halted by 12 months and remained unchanged thereafter. PIRMA/PIRA/CDW hazard ratios were reported for the ITT treatment allocation only. No formal subgroup analyses by age, sex, or prior DMT were reported.


Criticisms

  • Open-label design (rater-blinded only) — potential for bias in subjective assessments
  • Small sample size (n=197) limits statistical power for disability progression endpoints
  • Short 24-month follow-up may be insufficient to detect meaningful disability differences in RRMS
  • No placebo arm — head-to-head comparison does not allow inference vs untreated patients
  • PIRMA definition incorporates MRI activity; its framing as 'independent' of MRI is conceptually contested
  • Wide confidence intervals for HRs reflect limited number of progression events (n=18)
  • Majority of patients (>90%) had prior DMT exposure, which may confound biomarker trajectories
  • sGFAP analyses were post hoc, not pre-specified — findings should be interpreted as hypothesis-generating
  • Safety outcomes and adverse event denominators are not reported in this biomarker sub-study

Funding

Swedish Research Council (grant 2016-00398), Swedish Brain Fund, and regional ALF funds (Uppsala, Gothenburg, Skåne, Stockholm, Västerbotten, Örebro, Östergötland, Västra Götaland)

Based on: RIFUND-MS (Annals of Clinical and Translational Neurology, 2026)

Authors: Shawket F, Lycke J, Salzer J, ..., Svenningsson A

Citation: Ann Clin Transl Neurol 2026;0:1-11

Content summarized and formatted by NeuroTrials.ai.