GFAP and NfL Fall During C5 Inhibition in NMOSD

Key Takeaways

• Adults with AQP4-Ab+ NMOSD in the PREVENT and CHAMPION-NMOSD biomarker cohorts had significantly higher pretreatment GFAP and NfL than matched external healthy donors.
• High GFAP was significantly associated with older age at presentation, more recent relapse, and greater ambulation impairment in CHAMPION-NMOSD, but corresponding adjusted associations were not significant in PREVENT.
• GFAP fell significantly early within both C5 inhibitor treatment groups, whereas NfL declines reached significance only at selected ravulizumab visits.
• Among sampled PREVENT relapses, GFAP rose less in eculizumab extension cases than in placebo primary-period cases, a descriptive comparison across different treatment periods.
• Baseline GFAP and NfL were not significantly associated with time to first adjudicated relapse among PREVENT placebo recipients.

Relapses in aquaporin-4 antibody–positive (AQP4-Ab+) neuromyelitis optica spectrum disorder (NMOSD) can leave lasting neurologic disability. Serum glial fibrillary acidic protein (GFAP) reflects astrocytic injury, whereas neurofilament light chain (NfL) reflects axonal injury.

Investigators examined serum from adults with AQP4-Ab+ NMOSD, a prior relapse, and an Expanded Disability Status Scale (EDSS) score of ≤7 in the PREVENT and CHAMPION-NMOSD biomarker analysis. PREVENT serum was assessed retrospectively for 41 of 143 participants assigned eculizumab or placebo, subject to remaining samples and consent; CHAMPION-NMOSD serum was measured prospectively for 55 of 58 ravulizumab recipients. Samples were taken before and during treatment, with relapse samples collected 24–48 hours after symptom onset. Demographically matched external healthy donors provided reference sera and a donor-based threshold for classifying biomarker levels as high. Quanterix single-plex single molecule array (Simoa) assays measured both markers. Adjusted models examined clinical correlates in CHAMPION-NMOSD and separately in PREVENT; investigators also assessed within-arm changes and placebo-arm time to relapse.

Pretreatment GFAP and NfL were significantly higher than in external donors in each cohort. GFAP met the donor-based high classification in 29% of each cohort, while high NfL was less frequent. In adjusted analyses, high GFAP tracked older age at initial presentation, shorter time since last relapse, and greater impairment on the Hauser Ambulation Index (HAI) in CHAMPION-NMOSD; corresponding PREVENT associations were not significant. High NfL tracked older age at first dose in both cohorts and higher HAI scores significantly only in CHAMPION-NMOSD. Among PREVENT eculizumab recipients, median GFAP fell from 120.0 to 86.95 pg/mL by week 4 versus pretreatment baseline (p = 0.0068).

Among CHAMPION-NMOSD ravulizumab recipients, median GFAP fell from 129.0 to 107.0 pg/mL at week 6 versus pretreatment baseline (p = 0.0325), with another significant within-group decline at a later visit. NfL fell significantly at selected ravulizumab visits, while neither PREVENT arm had a significant NfL change and placebo recipients had no significant GFAP change. In available PREVENT relapse samples, mean GFAP rose 8.72-fold in placebo primary-period cases versus 2.67-fold in eculizumab open-label extension cases; these relapse subsets were not randomized against each other. NfL did not differ from baseline at relapse, and baseline GFAP and NfL were not significantly associated with time to first adjudicated relapse in the placebo arm.

PREVENT biomarker availability depended on remaining serum and consent. Relapse-linked samples were scarce, placebo and eculizumab relapse samples came from different treatment periods, and primary-period eculizumab relapses lacked biomarker samples. CHAMPION-NMOSD had no adjudicated on-trial relapses for a corresponding assessment. External healthy donors were not randomized trial controls, and differences between the separate drug cohorts do not test comparative efficacy. Biomarker movement does not establish patient-level treatment monitoring or relapse prediction, and these findings do not provide new estimates of clinical relapse protection.

GFAP may provide a pharmacodynamic readout of C5 blockade in these trial-derived samples, according to the investigators. The prognostic and clinical utility of GFAP and NfL remains to be defined prospectively.

Clinician Questions

How were high serum GFAP and NfL defined in AQP4 antibody–positive NMOSD biomarker cohorts?

For PREVENT and CHAMPION-NMOSD participants with AQP4 antibody–positive NMOSD, high pretreatment serum GFAP or NfL meant at least 2 standard deviations above the mean in matched external healthy-donor sera. The cutoffs were GFAP ≥164.27 pg/mL and NfL ≥16.11 pg/mL. These were reference-based classifications, not established thresholds for initiating treatment.

Did cerebrospinal fluid NfL follow serum NfL during ravulizumab treatment for AQP4-positive NMOSD?

In an optional CHAMPION-NMOSD cerebrospinal fluid (CSF) subset of four people with AQP4 antibody–positive NMOSD receiving ravulizumab, CSF NfL fell alongside serum NfL in three. In the remaining participant, CSF NfL was relatively stable despite a slight serum decrease; baseline CSF NfL was unavailable, and CSF observation was shorter than serum observation. This limited exploratory observation did not validate serum NfL for patient-level monitoring.