Combined Cerebrospinal Fluid Neurofilament Light Chain Protein and Chitinase-3 Like-1 Levels in Defining Disease Course and Prognosis in Multiple Sclerosis

Abstract

Background: Neurofilament light chain protein (NFL) and chitinase3-like1 (CHI3L1) have gained importance recently as prognostic biomarkers in multiple sclerosis (MS). Objectives: We aimed to investigate NFL and CHI3L1 cerebrospinal fluid (CSF) profiles in multiple sclerosis and the informative and prognostic potential of the individual and combined measures. Methods: CSF NFL and CHI3L1 levels were measured in a cross-sectional cohort of 157 MS patients [99 relapsing-remitting (RRMS), 35 secondary progressive (SPMS), and 23 primary progressive (PPMS)]. Clinical relapse and/or gadolinium-enhanced lesions (GEL) in MRI within 90 days from CSF collection by lumbar puncture (LP) were registered and considered as indicators of disease activity. Longitudinal treatment and disability data were evaluated during medical visits with a median follow-up of 50 months. Results: CSF levels of NFL and CHI3L1 were higher in MS patients compared to non-MS controls. In RRMS and SPMS patients, increased NFL levels were associated with clinical relapse, and gadolinium-enhanced lesions in MRI (p < 0.001), while high CHI3L1 levels were characteristic of progressive disease (p = 0.01). In RRMS patients, CSF NFL, and CHI3L1 levels correlated with each other (r = 0.58), and with IgM-oligoclonal bands (p = 0.02 and p = 0.004, respectively). In addition, CSF CHI3L1 concentration was a predictor for 1-point EDSS worsening {HR = 2.99 [95% CI (1.27, 7.07)]} and progression during follow-up {HR = 18 [95% CI (2.31, 141.3)]}. The pattern of combined measure of biomarkers was useful to discriminate MS phenotypes and to anticipate clinical progression: RRMS more frequently presented high NFL combined with low CHI3L1 levels, compared to SPMS (HR 0.41 [0.18-0.82]), and PPMS (HR 0.46 [0.19-0.87]), while elevation of both biomarkers preceded diagnosis of clinical progression in RRMS patients (log rank = 0.02). Conclusions: Individual measures of CSF NFL and CHI3L1 are biomarkers of disease activity and progression, respectively. The pattern of combined measure discriminates MS phenotypes. It also predicts the subset of RRMS patients that will progress clinically allowing early intervention.

Keywords: CHI3L1; NFL; YKL-40; gadolinium-enhancing lesions; progressive multiple sclerosis.

Figure 1

Flow chart of study cohort.

Figure 2

CSF levels of NFL and CHI3L1 in MS in association to phenotype, inflammatory activity, and other influencing variables. (A) Patients with disease activity had higher CSF NFL levels compared to patients without documented disease activity (defined as clinical attack and/or GEL in the MRI within 90 days of CSF collection) (Student T-test p < 0.001). CHI3L1 levels were not different with regards to disease activity (Student T-test p = 0.705). (B) NFL levels were not different between MS forms (ANOVA, p = 0.6) while CHI3L1 were increased in progressive MS compared to RRMS (ANOVA p = 0.01; Bonferroni post-hoc RRMS vs. PPMS p = 0.009). (C) Both NFL and CHI3L1 were higher in the presence of OCMB in the RRMS cohort (Student T-test p = 0.02 and p = 0.004, respectively). (D) NFL and CHI3L1 correlated between each other (Pearson's r = 0.58; p < 0.001) in the RRMS cohort. NFL levels were lower as time from clinical relapse passed (Pearson's r = 0.29; p = 0.04). Sample size for each condition is expressed between brackets in the diagram. Student T-test and correlations were performed with log-transformed NFL and CHI3L1. NFL, neurofilament light chain; CHI3L1, chitinase-3-like-1; RRMS, relapsing-remitting multiple sclerosis; SPMS, secondary progressive multiple sclerosis; PPMS, primary progressive multiple sclerosis. GEL, gadolinium-enhanced lesions; CSF, cerebrospinal fluid; MRI, magnetic resonance imaging.

Figure 3

Probability of conversion to SPMS according to biomarker profile. Patients were cathegorized according to NFL and CHI3L1 median values (calculated in the RRMS cohort without disease activity) into four groups: Group 1: both NFL and CHI3L1 below median value; Group 2: high NFL; Group 3: High CHI3L1; Group 4: NFL and CHI3L1 above median values. Patients included in group 4 had more probability to be diagnosed of SPMS during follow-up than patients pertaining to the other groups (log rank p = 0.02). SPMS: secondary progressive multiple sclerosis.

Figure 4

Density maps representing paired NFL and CHI3L1 values in CSF of non-MS controls and MS patients. The black lines in each diagram represent the median log values of CSF NFL and CHI3L1 in the whole cohort. Note that each subgroup has the peak density of patients in a different position in reference with the median log values. (A) In non-MS control group, almost all patients fell in the left-inferior quadrant. (B) In PPMS subgroup, CHI3L1 levels were over the median in almost all patients, and two groups were distinguished with regards to NFL levels. (C) In the RRMS group, black triangles represented patients that at the end of the observation period evolved to progressive disease. (D) SPMS patients constitute an intermediate density map with three regions of distribution. NFL, neurofilament light chain; CHI3L1, chitinase-3-like-1; RRMS, relapsing-remitting multiple sclerosis; SPMS, secondary progressive multiple sclerosis; PPMS, primary progressive multiple sclerosis; Switch, diagnosis of SPMS during follow-up.

Figure 5

Marginal effect plots. This plot is the graphical representation of the multinomial regression model performed and shows the probability to be comprised in a MS clinical form, including the probability of conversion to SPMS, with regards to CSF NFL, and CHI3L1 levels. Thus, this diagram shows the interaction between both biomarkers. (A) When levels of NFL are low, low CHI3L1 levels imply higher probability to be a non-MS control (pink) or a RRMS patient (red). High CHI3L1 levels in these patients are associated with the diagnosis of progressive MS (green and brownish green). Probability of switching to SPMS is very low. (B) With average levels of NFL, the probability of pertaining to a non-MS control decreases, and CHI3L1 levels discriminate better between RRMS (low CHI3L1), and PPMS (high CHI3L1). (C) With high NFL levels, the probability of being comprised in the RRMS group is the highest. High NFL and CHI3L1 are more probable in patients diagnosed of SPMS during follow-up, and in a subgroup of PPMS patients, likely PPMS patients with active disease. NFL, neurofilament light chain; CHI3L1, chitinase-3-like-1; RRMS, relapsing-remitting multiple sclerosis; SPMS, secondary progressive multiple sclerosis; PPMS, primary progressive multiple sclerosis; Switch, diagnosis of SPMS during follow-up.