# Serum neurofilament light chain and glial fibrillary acidic protein predicting multiple sclerosis after clinically isolated syndrome

## TL;DR

Serum neurofilament light chain (NfL) is a better predictor of multiple sclerosis (MS) after a first neurological event than glial fibrillary acidic protein (GFAP), with HLA-DRB1*15:01 status improving prediction accuracy.

## Contribution

This study demonstrates that serum NfL, combined with HLA-DRB1*15:01 status, provides robust early risk stratification for MS after clinically isolated syndrome.

## Key findings

- High serum NfL levels independently predict earlier MS diagnosis after clinically isolated syndrome.
- HLA-DRB1*15:01 status improves predictive accuracy when combined with NfL.
- GFAP and anti-EBNA1 IgG add limited prognostic value for MS prediction.

## Abstract

Serum neurofilament light chain (NfL) and glial fibrillary acidic protein (GFAP) may synergistically enhance early risk stratification of multiple sclerosis (MS) diagnosis after clinically isolated syndromes (CIS). We investigated the prognostic value of combined NfL and GFAP for McDonald 2024 MS diagnosis after CIS and associations with key genetic and environmental risk factors.

CIS participants, within six months after symptom onset, were included in a prospective cohort. We measured baseline serum NfL and GFAP levels and calculated z-scores. We evaluated weighted genetic risk scores for MS susceptibility, HLA-DRB1*15:01 risk and measured Anti-Epstein Barr virus Nuclear Antigen-1 (anti-EBNA1) immunoglobulin G (IgG) antibodies. Associations with MS diagnosis were evaluated using Cox proportional hazards models and time-dependent receiver operating characteristic (ROC) analyses.

During follow-up, 162/221 CIS participants were diagnosed with McDonald 2024 MS. Separately, high NfL and GFAP associated with earlier MS diagnoses (hazard ratio (HR) 1.36, 95% confidence interval (CI) 1.12–1.66, p = 0.002, HR 1.12, 95% CI 1.02–1.42, p = 0.01, respectively). In combined models, only NfL remained independently predictive (HR 1.30, 95% CI 1.02–1.60, p = 0.01). Time-dependent ROC analyses showed similar results for NfL alone and combined with GFAP. HLA-DRB1*15:01-risk, but not GFAP or anti-EBNA1 IgG, improved predictive value.

Our study found that serum NfL outperformed GFAP in predicting early MS diagnoses after CIS. Baseline NfL, together with HLA-DRB1*15:01 status, provides robust early risk stratification for MS after CIS, whereas GFAP and anti-EBNA1 titres add limited prognostic value. Additional immunological and imaging markers are essential to further refine predictive models.

The online version contains supplementary material available at 10.1007/s00415-026-13667-7.