Predictive Value of Serum Neurofilament Light Chain Levels in Anti-NMDA Receptor Encephalitis

Determinants of disease activity and prognosis are limited in anti-NMDA receptor (NMDAR) encephalitis. Neurofilament light chains (NfL) are markers of axonal damage and have been identified as valuable biomarkers for neurodegenerative and other neuroinflammatory disorders. We aimed to investigate serum NfL levels in patients with anti-NMDAR encephalitis as a biomarker for disease severity and outcome.

In this retrospective study, NfL values were measured in all available pretreatment serum and paired CSF samples of the nationwide anti-NMDAR encephalitis cohort. The values were analyzed in duplicate using single-molecule array and compared with measurements in healthy references. Follow-up sera were tested to analyze longitudinal responsiveness, if at least available from 2 time points after diagnosis. Serum NfL levels were compared with data on disease activity (seizures, MRI, and CSF findings), severity (modified Rankin Scale [mRS] score, admission days, and intensive care unit admission), and outcome (mRS score and relapses), using regression analysis.

We have included 71 patients (75% female; mean age 31.4 years, range 0-85 years) of whom pretreatment serum samples were analyzed. Paired CSF samples were available of 33 patients, follow-up serum samples of 20 patients. Serum NfL levels at diagnosis were higher in patients (mean 19.5 pg/mL, 95% CI 13.7-27.7) than in references (mean 6.4 pg/mL, 95% CI 5.8-7.2, p < 0.0001). We observed a good correlation between serum and CSF NfL values (R = 0.84, p < 0.0001). Serum NfL levels and age correlated in patients (Pearson R = 0.57, p < 0.0001) and references (R = 0.62, p < 0.0001). Increased NfL values were detected in patients post-herpes simplex virus 1 encephalitis (mean 248.8 vs 14.1 pg/mL, p < 0.0001) and in patients with brain MRI lesions (mean 27.3 vs 11.1 pg/mL, p = 0.019). NfL levels did relate to the long-term follow-up (mRS score at 12 months; β_NfL = 0.55, p = 0.013), although largely explained by the effect of age on NfL levels and prognosis. In serial samples, NfL values did roughly follow clinical disease activity, albeit with delay.

Increased serum NfL levels reflect neuroaxonal damage in anti-NMDAR encephalitis. No relationship was identified with disease severity, whereas the association with outcome was confounded by age. The implied role of sampling timing on NfL levels also limits the applicability of NfL as a prognostic marker.