Serum Protein Signature May Predict Relapse in Neuromyelitis Optica Spectrum Disorder Months in Advance, Study Finds

A proteomic analysis of longitudinally collected serum samples from patients enrolled in the CIRCLES cohort has identified a 10-protein signature capable of stratifying relapse risk in neuromyelitis optica spectrum disorders (NMOSD), with findings published in Neurology: Neuroimmunology & Neuroinflammation.¹ If validated prospectively, the signature could provide clinicians with a meaningful window to intervene before neurological damage accrues.

NMOSD is a rare autoimmune CNS disorder in which disability accumulates with each relapse. Virtually all patients with anti-aquaporin-4 antibody (AQP4)-seropositive disease experience relapses, and even with effective immunosuppressive therapy they occur unpredictably, carrying the risk of permanent vision loss, paralysis, and cognitive decline. By the time a relapse is confirmed through biomarkers like GFAP or neurofilament light chain, the window for meaningful prevention has already closed. The CIRCLES study team set out to find something earlier.

Using high-resolution liquid chromatography/mass spectrometry, investigators characterized 305 longitudinally collected serum samples from 126 CIRCLES-enrolled patients with documented relapses. The CIRCLES cohort, the largest NMOSD biorepository in North America, enrolled 599 patients across 15 US sites between 2013 and 2020. Led by several neurologists, including Michael R. Yeaman, PhD, MSc, median time from sample collection to subsequent relapse in the analytic cohort was 243 days, a deliberate feature: investigators wanted signals appearing months before clinical deterioration, not days.

Six modeling frameworks were applied to the proteomic data, spanning logistic regression, naive Cox proportional hazards, and time-dependent Cox models, each run in single-protein and multiprotein configurations across 100 randomly sampled patient subsets. The same 10 proteins emerged consistently across at least four of the six approaches: coagulation factor XI (F11), surfactant protein B (SFTPB), C1RL, filamin A, cholesteryl ester transfer protein, cathelicidin antimicrobial peptide, C4A, transferrin receptor, immunoglobulin kappa constant, and serum amyloid A2 protein.

Factor XI and surfactant protein B were the two most robust candidates, remaining statistically significant after false discovery rate correction at q = 0.05, a threshold the other eight proteins did not individually reach. Both showed increasing serum abundance in the period preceding relapse. Each unit increase in F11 z-score corresponded to a 23% to 34% increase in relapse hazard depending on the model; each unit increase in SFTPB corresponded to a 22% to 37% increase. Complement proteins C4A and transferrin receptor moved in the opposite direction, with higher abundance associated with 19% to 29% decreases in relapse hazard.

READ MORE: Monoclonal Antibodies Show Greater Relapse Reduction in AQP4-IgG Seropositive vs Seronegative NMOSD, Meta-Analysis Finds

When the full 10-protein signature was applied to stratify patients into high and low relapse risk groups, the separation was highly significant by log-rank test (P = 9.9 × 10⁻¹⁰), a substantial improvement over using only the top two proteins (P = 1.5 × 10⁻⁵). The biology underlying several of these proteins fits plausibly with known NMOSD pathophysiology. The complement system is central to AQP4-antibody-mediated astrocyte injury, and the directional associations observed for C4A and C1RL align with that established mechanism. Factor XI's coagulation role has been increasingly linked to autoimmune neurological conditions in prior literature.

The study's authors were careful to temper the findings. A logistic regression model trained to predict relapse within 120 days showed a precision of 0.14 and a recall of 0.05, underscoring that the signature's current utility lies more in longer-range risk stratification than short-range prediction. The cohort was predominantly female (85.7%) and anti-AQP4-seropositive, and while the signature significantly stratified relapse risk in Black and African American patients as well as White patients, it did not achieve significance in smaller subgroups, likely reflecting power limitations. The absence of healthy controls or comparative neurological disease cohorts also leaves the signature's specificity unestablished.

Four FDA-approved therapies now exist for NMOSD relapse prevention, including eculizumab, inebilizumab, satralizumab, and ravulizumab.² If a validated relapse-prediction tool existed, clinicians could intensify monitoring or adjust therapy before a relapse rather than after. The investigators propose that the 243-day median forecasted lead time would, if validated, be sufficient for precisely that kind of proactive clinical decision-making.

REFERENCES
1. Bach E, Breschi A, Wang Y, et al. Prognostic biomarker signature of relapse in neuromyelitis optica spectrum disorders in the CIRCLES observational study. Neurol Neuroimmunol Neuroinflamm. 2026;13(4):e200611. doi:10.1212/NXI.0000000000200611. https://doi.org/10.1212/NXI.0000000000200611
2. Levy M, Fujihara K, Palace J. New therapies for neuromyelitis optica spectrum disorder. Lancet Neurol. 2021;20(1):60-67. doi:10.1016/S1474-4422(20)30392-6. https://doi.org/10.1016/S1474-4422(20)30392-6