Associations of cumulative exposure and longitudinal trajectories of neurofilament light chain with 1-year adverse events in amyotrophic lateral sclerosis: A prospective cohort study.
TL;DR
Early repeated NFL measurements summarized as cuNFL and longitudinal trajectory groups were associated with short-term adverse events in this cohort, and support further external validation of longitudinal NFL-based risk stratification.
Abstract
Background
Baseline neurofilament light chain (NFL) predicts amyotrophic lateral sclerosis (ALS) outcomes, but it does not summarize early repeated biomarker information. We evaluated whether cumulative NFL exposure (cuNFL), calculated over an early monitoring window, and longitudinal NFL trajectory groups were associated with subsequent adverse events in a prospective cohort.
Methods
This prospective study enrolled patients with ALS from 32 provinces in China between January 2024 and January 2025. Plasma NFL was measured at baseline, month 1, and month 2. We used K-means clustering to describe longitudinal NFL patterns and multivariable survival models to evaluate prognostic associations. Inverse-probability weighting and causal-forest analyses were used as exploratory, confounding-adjusted analyses of high versus low cuNFL.
Results
K-means clustering identified three longitudinal NFL groups: Low-Stable (n = 183), Moderate (n = 225), and High-Progressive (n = 100). The High-Progressive group had the shortest event-free survival (P < 0.001). Each 1-SD increment in cuNFL was associated with higher odds of adverse events (adjusted OR 1.99; 95% CI 1.52-2.60; P < 0.001). Using the manuscript-defined 127-event outcome, apparent discrimination was AUC 0.730 for cuNFL and 0.717 for baseline NFL. The paired DeLong comparison was not statistically significant (P = 0.084). Exploratory machine-learning analyses suggested heterogeneity in the association of high cuNFL with outcome by age and respiratory function.
Conclusion
Early repeated NFL measurements summarized as cuNFL and longitudinal trajectory groups were associated with short-term adverse events in this cohort. These findings support further external validation of longitudinal NFL-based risk stratification. They do not establish a causal biological effect or a basis for routine repeated measurement.