Biomarkers and Bioinformatics Poster Presentation

P0118 - Neurofilament light chain concentration predicts risk of relapse in participants with relapsing multiple sclerosis in phase 3 ozanimod trials (ID 1211)

  • S. Harris
  • S. Harris
  • G. Comi
  • B. Cree
  • D. Arnold
  • L. Steinman
  • J. Sheffield
  • H. Southworth
  • L. Kappos
  • J. Cohen
Presentation Number
Presentation Topic
Biomarkers and Bioinformatics



In patients with multiple sclerosis (MS), neurofilament light chain concentration (NfL-c) is increased in blood and cerebrospinal fluid (CSF). Plasma and CSF NfL-c correlate and may serve as a biomarker for neurologic damage and disease activity in relapsing MS (RMS). Ozanimod, a sphingosine 1-phosphate receptor 1 and 5 modulator, reduced annualized relapse rate (ARR) and plasma NfL-c (pNfL-c) vs interferon β-1a (IFN) in phase 3 RMS trials.


To expand upon previously reported data showing that baseline (BL) pNfL-c predicts risk of on-treatment relapse, we analyzed the relationship between changes in pNfL-c and on-treatment risk of relapse with ozanimod vs IFN.


In this exploratory, post hoc analysis, pNfL-c was measured at BL and after 12 and 24 months of treatment with oral ozanimod 0.46 or 0.92 mg/d or intramuscular IFN β-1a 30 µg/wk in the phase 3 SUNBEAM (NCT02294058; ≥12 mo) and RADIANCE (NCT02047734; 24 mo) trials. The effect of treatment on pNfL-c was assessed, as were the relationships between BL pNfL-c and relapse rate and pNfL-c change from BL and ARR. NfL data are reported as median percentage change from BL. Poisson generalized linear models were used to fit the number of relapses as a function of BL pNfL-c and treatment group with an offset for duration. We calculated a predictive model of expected ARR based on median percentage change from BL in pNfL-c.


At end of treatment, median pNfL-c was significantly reduced from BL by 20‒23% (P<0.01) and 23‒27% (P≤0.0001) with ozanimod 0.46 and 0.92 mg, respectively, and by 13‒15% with IFN. Higher BL p-NfL-c was associated with an increase in number of relapses (P<0.0001). Over a 12-month period, participants treated with either dose of ozanimod had significantly fewer relapses than those treated with IFN (P<0.05); the greatest effect was observed with ozanimod 0.92 mg. Further analyses reveal that treatment groups associated with a greater median reduction from BL pNfL-c are also associated with lower ARR. Predictive modeling estimated that a 25% reduction in pNfL-c, similar to that observed with ozanimod 0.92 mg, predicts an ARR (standard error [SE]) of 0.18‒0.23 (0.4); a 13% reduction, which was similar to that observed with IFN, predicts an ARR (SE) of 0.29‒0.37 (0.04).


Our findings further support pNfL-c as a biomarker for RMS disease activity. BL pNfL-c is related to relapse rate, as are changes in pNfL-c during treatment. Ozanimod causes dose-dependent reductions in pNfL-c and ARR compared with IFN.