Neurofilament light (NFL), a biomarker of axonal damage and disease activity in multiple sclerosis (MS), has been increasingly recognized for prognostic and therapeutic decisions.
To assess the sensitivity and specificity of cerebrospinal fluid (CSF) NFL for disease activity and to evaluate its prognostic value in predicting disease severity and conversion from relapsing-remitting (RRMS) to secondary progressive MS (SPMS).
Patients with a confirmed diagnosis of RRMS (n=769) who had determined NFL in CSF as part of the diagnostic work-up or from assessments as part from regular clinical visits between 2001 and 2018 were retrospectively identified in the Swedish MS registry. Measurements over time of disease activity (clinical relapses and lesion formation on MRI) and disability (EDSS) were retrieved. We assessed NFL levels in relation to concomitant clinical and MRI activity and as outcome for treatment response.
Patients with a concurrent clinical relapse had significantly higher NFL concentrations (median NFL no relapse 278 ng/L, relapsed 1122 ng/L, p<0.001) and a correlation with relapse severity was observed (p<0.001). Patients with gadolinium-enhancing lesions had higher median NFL levels (1414 ng/L) than those without (426 ng/L, p<0.001) and NFL levels correlated with the number of gadolinium enhancing lesions (p<0.001). CSF NFL showed sensitivity of 93.3% and specificity of 77.4% to disease activity (relapses and/or MRI activity). High NFL at diagnosis (n=414) was independently associated with worsening of disability and predicted progression to EDSSā„3 (n=128, p<0.001, HR 0.566 95% CI 0,413-0,711) and conversion to secondary progressive MS (n=39, p=0.0003, HR=0,380 95% CI 0,225-0,641). In a subset of patients (n=159), NFL was analysed at baseline and at follow-up (median time between LPs 23.7 months). Patients who switched from a first-line to a second-line therapy (n=65) exhibited significant reduction in NFL concentrations (p<0.001). However, patients who did not receive disease modifying therapy (DMT) (n=32), had first-line therapy (n=40), or switched to a third-line treatment (n=22) between baseline and follow-up did not show a significant decrease in their CSF NFL levels (p=0.975, p=0.658, and p=0.059 respectively).
Since 2001 CSF NFL has been part of the clinical assessment at Sahlgrenska University Hospital, Gothenburg. Thus, the results of this study are based on a real-life material and we can confirm the utility of NFL as a biomarker in MS. Our data underline NFL as a sensitive biomarker of disease activity, its usefulness for prediction of disability and clinical course and for monitoring DMT response. Serum/plasma NFL is most likely to show similar properties but probably at a lower precision.