RESUMO
BACKGROUND: Neurofilament light chain (NFL) is a biomarker for monitoring disease activity and treatment response in multiple sclerosis (MS). However, while most agree that NFL levels predict disease activity and worsening, the predictive value of NFL on future relapse risk remains uncertain. OBJECTIVE: The primary aim was to evaluate the predictive value of age-corrected serum NFL (sNFL) ratio on relapse risk in highly active relapsing-remitting MS patients (RRMS) treated with natalizumab. A secondary aim was to investigate the predictive value of sNFL ratios for MRI activity. METHODS: From January 1, 2006, to December 31, 2010, 355 patients initiated natalizumab treatment at the Danish Multiple Sclerosis Center. 305 patients were anti-natalizumab antibodies negative and had at least one blood sample available for sNFL analysis using single molecule array analysis at baseline, three, six, or 12 months. The patients were either treatment-naïve (n = 8), switching from interferon-ß or glatiramer acetate (n = 253), or switching from mitoxantrone (n = 44). An age-corrected ratio was calculated for sNFL. Time to first relapse was calculated from baseline and after re-baseline at 90 days. Data were collected from baseline until the two-year follow-up or end of treatment and included disease duration, expanded disability status scale, previous treatments, relapses 12 months prior to natalizumab initiation, smoking intensity, body mass index, and body weight. In addition, the patients underwent annual MRI of the brain. RESULTS: The sNFL ratio was increased in 173 of 287 samples (60.3 %) at baseline, in 119 of 246 samples (48.8 %) at month three, in 109 of 287 samples (38.0 %) at month six, and in 82 of 270 samples (30.4 %) at month 12. The sNFL ratio continuously declined over 12 months with significant decreases for every measuring timepoint: baseline vs. three months p = 3.0 × 10-6; three months vs. six months p = 3.2 × 10-5; six months vs. 12 months p = 0.002. Univariate Cox regression analysis showed that time to first relapse from 1) natalizumab initiation and from 2) re-baseline was associated with the number of relapses in the previous 12 months (hazard ratio 1.31 per relapse, 95 % CI = 1.2-1.5, p = 2.0 × 10-6; and 1.21 per relapse, 95 % CI = 1.1-1.4, p = 0.002, respectively). sNFL ratio at re-baseline was negatively associated with relapse risk (hazard ratio 0.82 per unit; 95 % CI = 0.7-1.0; p = 0.049). A multivariable Cox regression analysis of relapse risk from re-baseline showed that the number of relapses in the 12 months prior to natalizumab treatment (hazard ratio 1.29; 95 % CI = 1.1-1.5; p = 6.0 × 10-4) and smoking (hazard ratio 1.51 per 20 cigarettes per day; 95 % CI = 1.0-2.2; p = 0.030) were associated with increased risk of relapse; sNFL ratio was associated with a lower risk of relapse (hazard ratio = 0.736 per unit; 95 % CI = 0.6-0.9 p = 0.007). In univariate logistic regression analyses, the sNFL ratio at 12 months and values above the 75th and the 90th percentile predicted MRI activity in the following year (odds ratio [OR] = 2.0, 95 % CI = 1.2-3.6, p = 0.012; OR = 2.2, 95 % CI = 1.2-4.1, p = 0.014; and OR = 2.8, 95 % CI = 1.1-6.7, p = 0.026). CONCLUSION: In this highly active RRMS cohort, high sNFL ratios reflected previous relapse activity and decreased after initiation of treatment but were not associated with increased relapse risk in the following two years. Pre-treatment relapses and smoking on treatment were predictors of relapse risk after re-baselining at 90 days. MRI activity in year two was predicted by sNFL ratios at month 12.