Publications & Posters

Neurofilament Light Chain Serum Levels Correlate with 10-year MRI Outcomes in Multiple Sclerosis


Chitnis T, Gonzalez C, Healy BC, Saxena S, Rosso M, Barro C, Michalak Z, Paul A, Kivisakk P, Diaz-Cruz C, Sattarnezhad N, Pierre IV, Glanz BI, Tomic D, Kropshofer H, Häring D, Leppert D, Kappos L, Bakshi R, Weiner HL and Kuhle J

Annals of Clinical and Translational Neurology. 2018;0.




To assess the value of annual serum neurofilament light (NfL) measures in predicting 10‐year clinical and MRI outcomes in multiple sclerosis (MS).


We identified patients in our center’s Comprehensive Longitudinal Investigations in MS at Brigham and Women’s Hospital (CLIMB) study enrolled within 5 years of disease onset, and with annual blood samples up to 10 years (n = 122). Serum NfL was measured using a single molecule array (SIMOA) assay. An automated pipeline quantified brain T2 hyperintense lesion volume (T2LV) and brain parenchymal fraction (BPF) from year 10 high‐resolution 3T MRI scans. Correlations between averaged annual NfL and 10‐year clinical/MRI outcomes were assessed using Spearman’s correlation, univariate, and multivariate linear regression models.


Averaged annual NfL values were negatively associated with year 10 BPF, which included averaged year 1–5 NfL values (unadjusted P < 0.01; adjusted analysis P < 0.01), and averaged values through year 10. Linear regression analyses of averaged annual NfL values showed multiple associations with T2LV, specifically averaged year 1–5 NfL (unadjusted P < 0.01; adjusted analysis P < 0.01). Approximately 15–20% of the BPF variance and T2LV could be predicted from early averaged annual NfL levels. Also, averaged annual NfL levels with fatigue score worsening between years 1 and 10 showed statistically significant associations. However, averaged NfL measurements were not associated with year 10 EDSS, SDMT or T25FW in this cohort.


Serum NfL measured during the first few years after the clinical onset of MS contributed to the prediction of 10‐year MRI brain lesion load and atrophy.


Multiple sclerosis (MS) is a demyelinating and degenerative disease with a heterogeneous disease course.1 Patients experience periodic relapses and varying ranges of disability accrual over their lifetime.

Neurofilament light chain (NfL) is a major component of the neuronal cytoskeleton and is important for axonal growth, stability, and intracellular transport.23 NfL are released upon axonal or neuronal damage or degeneration, and can be found as a consequence, in the CSF and blood. Prior studies have shown that NfL concentrations in cerebrospinal fluid (CSF) are associated with the occurrence of MRI lesions, relapses, neurological disability, and treatment status in MS.47 Additional studies have demonstrated predictive value of CSF neurofilament light or heavy chain levels with clinical outcomes,810 and MRI measures.11More recently, single molecule array (SIMOA) based assays, which offer improved sensitivity for detection of molecules, have been used to measure NfL in serum samples. SIMOA‐based assays of serum NfL have demonstrated high correlation with CSF values71213 and potentially provide a more accessible means to monitor MS patients. Serum NfL measurements by SIMOA correlate with disease state as well as short‐term outcomes in MS,1218 however, the associations of serum NfL levels in predicting longer term outcomes, have not been explored.

In this study, we assessed serum NfL levels collected annually for 10 years in a cohort of MS patients with first sample within the first 5 years of disease onset. We assessed correlation with clinical, cognitive, and MRI outcomes at 10 years.



The MS subjects included in this study were patients enrolled in the Comprehensive Longitudinal Investigation of MS at the Brigham and Women’s Hospital (CLIMB, This study has enrolled over 2100 patients since 2000, and patients are followed longitudinally with biannual standardized clinical exams, annualized MRI scans, and stored blood samples. Subjects in this analysis met additional specific inclusion criteria: (1) enrolled in the quality of life (QOL) subgroup of the CLIMB study; (2) met the diagnostic criteria of MS by the 2010 McDonald criteria at last visit 20; (3) first blood drawn within 5 years of first symptom onset; (4) at least 8/10 annual blood draws from first collection to year 10; (4) provided consent for sample sharing. EDSS and T2FW are collected in all CLIMB subjects annually. Subjects in the QOL subgroup of CLIMB annually completed several patient reported outcomes (PROs) and this analysis included a fatigue measurement (modified fatigue impact scale, MFIS) 21 and cognition (symbol digit modalities test, SDMT).22

Standard Protocol Approvals, Registrations, And Patient Consents

Institutional Review Board approval was granted by the Partners Human Research Committee, and participants provided written informed consent for participation.

NFL Measurements

Serum samples were collected at annual CLIMB visits and were stored at −80°C following standardized procedures. The NfL serum samples were shipped on dry ice from Boston to Basel in a temperature controlled container and were measured by SIMOA assay as previously described.13 Inter‐assay coefficients of variation (CV) for three native serum samples were 10.8%, 8.3%, and 5.7% for control samples with mean concentrations of 9.2 pg/mL, 24.4 pg/mL, and 101.4 pg/mL, respectively. The mean intra‐assay CV of duplicated determinations for concentration was 5.1%. Repeat measurements were performed for few samples with intra‐assay CV above 20%. 4 samples showed an NfL value below 1.3 pg/mL (i.e., the lower limit of quantification), these were extrapolated from the standard curve and 12 values were measured as zero.

Untransformed NfL levels were used in all analyses. Several subjects were missing NfL measurements at some timepoints, and these subjects were removed from analyses related to that specific timepoint. In some analyses, NfL values were averaged across multiple time points (e.g., averaged yearly 1–2 NfL was calculated by the sum of the year 1 and year 2 NfL, then were divided by (2). If subjects were missing one or more of the values for the interval, the average was calculated using the available measurements. In additional analyses not presented in this paper, log‐transformed NfL levels were also analyzed, and we converted all 0’s to 1 prior to log transformation.

Clinical Outcomes

The primary clinical outcome for our analyses was disability measured by the Expanded Disability Status Scale (EDSS) at year 10. Secondary outcomes at year 10 were SDMT, MFIS, and Timed 25‐Foot Walk (T25FW). The SDMT tested executive function and processing speed which was a sensitive early marker of longitudinal cognitive changes in MS. The MFIS was a commonly used measure of fatigue for MS patients, and has three subscale scores (physical, mental, and psychosocial) as well as a total fatigue score. For the T25FW, there were 16 (1.62%) individuals who had high values for T25FW or were unable to complete the walk. For these patients, a score of 25 was assigned to limit the impact of those extreme observations on the analysis. The SDMT, MFIS, and T25FW measurements closest to the 10‐year sample were used for analysis. Also, a calculation in the difference between year 10 and year 1 SDMT, T25FW, and MFIS were performed.

MRI Acquisition And Processing

Brain MRI acquisition protocol was performed on a 3T unit (Siemens Skyra) which used a 20‐channel head coil, comprised of 3 sagittal sequences, and covered the whole head with 1 mm3 isotropic voxel sizes. This included a 3D T1‐weighted gradient echo (TE/TR = 2.96/2300 msec, TI = 900 msec, flip angle = 9 deg), 3D T2 spin echo (TE/TR = 300/2500 msec, echo train length = 160), and 3D T2‐FLAIR (TE/TR = 389/5000 msec, TI = 1800 msec, echo train length = 248). The sequences were optimized in contrast for depicting brain‐cerebrospinal fluid (CSF) interfaces and white matter lesions. The main steps of the fully automated quantitative analysis pipeline were outlined in Meier et al.23 Key steps were co‐registration of the three MR sequences, anatomical parcellation with heuristic misclassification correction, and an expectation‐maximization algorithm. The output provided brain T2 hyperintense lesion volume (T2LV) and brain parenchymal fraction (BPF), a surrogate of whole brain atrophy. This pipeline showed high accuracy and reliability.23 Intraclass correlation coefficients of 0.95, 0.91, and 0.86 were obtained for T2LV, CSF, and BPF accuracy. A scan‐rescan reliability experiment showed coefficients of variation (COVs) of 8%, 2%, and 0.4% for T2LV, CSF volume, and BPF. For this study, BPF values were multiplied by 100 to yield interpretable estimates for our analyses, and T2LV was log transformed due to skewness.

Statistical Analysis

MS patients: To assess the potential long‐term association between NfL and clinical/MRI outcomes, correlations between each NfL sample year and the 10‐year clinical/MRI outcomes were assessed using Spearman’s correlation and linear regression. Beyond the individual NfL measurements, the association between averaged yearly NfL values from specific intervals, the 10‐year clinical/MRI outcomes, and the year 1 and year 10 differences were also assessed using linear regression models. In addition, multiple linear regression models adjusted for sex, age, and disease duration at baseline. Additionally, in order to quantify the additional variance explained by adding NfL levels to the multiple regression model, we reported the R‐squared from reduced (the absence of the averaged yearly NfL) and full models. To further investigate the relationship between NfL and clinical disability, logistic regression was used to compare the relationship of NfL values with year 10 EDSS measurement (±1.5 years). We additionally performed all analyses using log NfL values and the results were generally similar compared to the untransformed NfL values presented in this paper. Also, given we completed 21 comparisons for each outcome, the Bonferroni corrected alpha level was 0.0024. P‐values for all analyses will be compared to 0.05 as well as 0.0024 to account for multiple comparisons. All analyses were performed using the Statistical Analysis System (SAS) 9.4 (Cary, NC).


Patients And NFL Characteristics

The baseline demographic and clinical characteristics of our MS patient cohort are shown in Table 1. 66% of patients were treated with a DMT at year 1 NfL measurement and the proportion of treated patients increased in year 2 to 85%. The arithmetic mean of NfL values per year show the highest levels at years 1 and 5 (Fig. 1A), and a spaghetti plot of individual MS patient trajectories showed variability (Fig. 1B).