Multicentre appraisal of amyotrophic lateral sclerosis biofluid biomarkers shows primacy of blood neurofilament light chain

Abstract

The routine clinical integration of individualized objective markers of disease activity in those diagnosed with the neurodegenerative disorder amyotrophic lateral sclerosis is a key requirement for therapeutic development. A large, multicentre, clinic-based, longitudinal cohort was used to systematically appraise the leading candidate biofluid biomarkers in the stratification and potential therapeutic assessment of those with amyotrophic lateral sclerosis. Incident patients diagnosed with amyotrophic lateral sclerosis (n = 258), other neurological diseases (n = 80) and healthy control participants (n = 101), were recruited and followed at intervals of 3-6 months for up to 30 months. Cerebrospinal fluid neurofilament light chain and chitotriosidase 1 and blood neurofilament light chain, creatine kinase, ferritin, complement C3 and C4 and C-reactive protein were measured. Blood neurofilament light chain, creatine kinase, serum ferritin, C3 and cerebrospinal fluid neurofilament light chain and chitotriosidase 1 were all significantly elevated in amyotrophic lateral sclerosis patients. First-visit plasma neurofilament light chain level was additionally strongly associated with survival (hazard ratio for one standard deviation increase in log₁₀ plasma neurofilament light chain 2.99, 95% confidence interval 1.65-5.41, P = 0.016) and rate of disability progression, independent of other prognostic factors. A small increase in level was noted within the first 12 months after reported symptom onset (slope 0.031 log₁₀ units per month, 95% confidence interval 0.012-0.049, P = 0.006). Modelling the inclusion of plasma neurofilament light chain as a therapeutic trial outcome measure demonstrated that a significant reduction in sample size and earlier detection of disease-slowing is possible, compared with using the revised Amyotrophic Lateral Sclerosis Functional Rating Scale. This study provides strong evidence that blood neurofilament light chain levels outperform conventional measures of disease activity at the group level. The application of blood neurofilament light chain has the potential to radically reduce the duration and cost of therapeutic trials. It might also offer a first step towards the goal of more personalized objective disease activity monitoring for those living with amyotrophic lateral sclerosis.

Keywords: amyotrophic lateral sclerosis; biomarker; motor neuron disease; neurofilament; trial.

Graphical Abstract

Figure 1

Cross-sectional biomarker levels. (A) Cross-sectional levels of CSF NFL and CHIT1 (pg/ml). CSF levels of NFL and CHIT1 were elevated in the first-visit CSF samples of ALS patients compared with healthy (NFL P < 0.0001, CHIT1 P < 0.0001) and disease controls (NFL P < 0.0001, CHIT1 P < 0.0001). (C) Cross-sectional levels of plasma NFL (pg/ml), serum ferritin (μg/L), plasma CK (IU/l) and serum C3 (g/l) were elevated in the first-visit CSF samples of ALS patients compared with healthy (for NFL P < 0.0001, ferritin P < 0.0001, CK P < 0.0001, C3) and disease controls (NFL P < 0.0001, ferritin P = 0.003, CK P < 0.001, C3). Serum C4 (g/l) (B) and plasma CRP (mg/l) (D) levels were similar between ALS compared with healthy controls (CRP P = 0.612, C3 P = 0.064, C4 P = 0.861) and disease controls (CRP P = 0.883, C3 P = 0.209, C4 P = 0.899). All data are displayed as median ± IQR; P-values given for ANCOVA of log-transformed analyte levels, controlling for age, sex and study site. ALS, amyotrophic lateral sclerosis; DC, disease control; HC, healthy control; NFL, neurofilament light chain; CHIT1, chitotriosidase 1; CK, creatine kinase; CRP, C-reactive protein; C3, complement C3 protein; C4, complement C4 protein. *P < 0.05; ***P < 0.001, NS, not significant.

Figure 2

Survival analysis. Kaplan–Meier survival estimates for patients with ALS stratified by tertile of (A) plasma NFL and (B) baseline PR (participants with data for plasma NFL, n = 237) with log-rank test. P-value indicated for NFL or PR tertiles. NFL, neurofilament light chain; PR, progression rate; ALS, amyotrophic lateral sclerosis.

Figure 3

Longitudinal analysis. Longitudinal analysis of NFL levels in plasma (A and B) and CSF (C and D) of patients with ALS. Linear mixed-effects models were fitted to data from baseline visit (A and C) and from symptom onset (B and D). Separate models of plasma neurofilament from symptom onset were constructed for samples taken within 12 months of symptom onset and those 12–48 months from symptom onset. Longitudinal data for other analytes are given in Supplementary Fig. 4. NFL, neurofilament light chain; ALS, amyotrophic lateral sclerosis.

Figure 4

Power simulations for randomised placebo-controlled trial. Illustrating a treatment effect corresponding to a 40% reduction in baseline PR, comparing difference in ALSFRS-R change (red) or plasma NFL (green) between groups using mixed models for repeated measures, demonstrating improved statistical power (1 − β) for a given sample size when using plasma NFL as an outcome measure. Points indicate simulated power at given sample size for 1000 iterations with α = 0.05. ALSFRS-R, revised ALS Functional Rating Scale score; NFL, neurofilament light chain; n, number of participants per group.