Serum neurofilament light chain predicts long term clinical outcomes in multiple sclerosis

Abstract

Serum neurofilament light chain (NfL) is emerging as an important biomarker in multiple sclerosis (MS). Our objective was to evaluate the prognostic value of serum NfL levels obtained close to the time of MS onset with long-term clinical outcomes. In this prospective cohort study, we identified patients with serum collected within 5 years of first MS symptom onset (baseline) with more than 15 years of routine clinical follow-up. Levels of serum NfL were quantified in patients and matched controls using digital immunoassay (SiMoA HD-1 Analyzer, Quanterix). Sixty-seven patients had a median follow-up of 18.9 years (range 15.0-27.0). The median serum NfL level in patient baseline samples was 10.1 pg/mL, 38.5% higher than median levels in 37 controls (7.26 pg/mL, p = 0.004). Baseline NfL level was most helpful as a sensitive predictive marker to rule out progression; patients with levels less 7.62 pg/mL were 4.3 times less likely to develop an EDSS score of ≥ 4 (p = 0.001) and 7.1 times less likely to develop progressive MS (p = 0.054). Patients with the highest NfL levels (3rd-tertile, > 13.2 pg/mL) progressed most rapidly with an EDSS annual rate of 0.16 (p = 0.004), remaining significant after adjustment for sex, age, and disease-modifying treatment (p = 0.022). This study demonstrates that baseline sNfL is associated with long term clinical disease progression. sNfL may be a sensitive marker of subsequent poor clinical outcomes.

Figure 1

Patient flow.

Figure 2

Raw serum NfL levels and baseline patient characteristics. Serum NfL levels were higher in patients who had had a documented relapse within 90 days of sampling compared to those who had not (p = 0.033, A) and controls (p = 0.0007). Patients who had not had a relapse within 90 days still had higher NfL levels than controls (p = 0.033). We did not find differences in NfL levels between different disease subtypes determined at baseline, although NfL levels were higher in both PPMS and RRMS patients compared to controls (p = 0.010 and 0.015 respectively). Y axis scale is Log(2) to most accurately represent the distribution of the data. Lines and bars depict median and interquartile range. Key: PPMS primary progressive MS, SPMS secondary progressive MS, CIS clinically isolated syndrome.

Figure 3

Raw serum NfL levels and subsequent disease progression. Serum NfL levels were higher in patients who reached EDSS ≥ 4 compared to those who had an EDSS < 4 (p = 0.0094, A) and controls (p = 0.0001), although was no significant difference comparing EDSS < 4 to controls. Similarly, baseline sNfL levels were higher in MS patients compared to controls, both in patients who developed progressive disease (p = 0.0005, B) and patients with relapsing disease (p = 0.041). Y axis scale is Log(2) to most accurately represent the distribution of the data. Lines and bars depict median and interquartile range.

Figure 4

Baseline NfL tertile and rate of EDSS progression. Average rate of EDSS progression (median and Interquartile range) over time in subgroups of MS patients based on serum NfL tertiles at baseline (Kruskal–Wallis p = 0.020, df 2).

Figure 5

Baseline NfL tertile and EDSS trajectories. Trajectory of EDSS over long-term follow-up in subgroups of MS patients based on serum NfL tertiles at baseline (overall repeated measures ANOVA p = 0.031, df 2 F value 3.43).

Figure 6

NfL tertile and likelihood of clinical progression. Kaplan–Meier survival curves to compare hazard of progression to EDSS ≥ 4 (A) (Cox regression overall p = 0.012 df 2 Chi-square 8.9) and conversion to SPMS (B) (Cox regression overall p = 0.0113 df 2 Chi-square 4.4) after long-term follow-up in subgroups of MS patients based on serum NfL tertiles at baseline.