Objective:

To investigate whether longitudinal neurofilament light chain (NfL) trajectory is associated with survival in ALS across multiple cohorts and complementary analytical frameworks.

Background:

Elevated baseline NfL predicts shorter survival in ALS, but the prognostic value of longitudinal NfL trajectory requires validation to support its use as a surrogate endpoint. The absence of NfL measurements at the time of death and through survival makes its use difficult and requires robust analytical strategies.

Design/Methods:

We analyzed longitudinal NfL and survival data from three independent ALS cohorts: the German APST ALS Data Repository, ANSWER ALS, and the HEALEY ALS Platform Trial. The primary analysis fitted a Weibull accelerated failure time model with time-varying NfL change from baseline of natural log-transformed NfL values. The secondary analysis employed joint longitudinal-survival modeling. Supportive Cox proportional hazards models with empirical Bayes (EB) slope estimates were performed across all cohorts. Sensitivity analyses used inverse probability weighting and landmark analyses.

Results:

In the primary time-varying Weibull AFT models, increasing NfL was significantly associated with shorter survival (APST: N=4,750 observations, 470 deaths, β = −0.382, SE=0.063, p < 0.0001, γ=1.54; ANSWER ALS: N=1,118 observations, 157 deaths, β = −0.639, SE=0.180, p = 0.0004, γ=1.78). Each unit increase in Δln(NfL) was associated with 32–47% shorter survival time. Secondary joint modeling confirmed the association (HR=1.10–1.12 per unit Δln(NfL), both p≤0.001). Supportive Cox EB proportional hazards analyses confirmed consistent prognostic associations across all three cohorts (all p<0.05). NfL reduction was associated with a 6–11% instantaneous mortality hazard reduction per 10% decline. Sensitivity analyses demonstrated concordant associations.

Conclusions: