Longitudinal measurement of serum neurofilament light in presymptomatic familial Alzheimer's disease

Abstract

Background: To investigate how serum neurofilament light (NfL) concentration changes through the course of disease in familial Alzheimer's disease (FAD) and to assess when NfL concentration first increases.

Methods: NfL was measured using an ultrasensitive immunoassay in 117 serum samples from 61 individuals from families with PSEN1 or APP mutations in a longitudinal study (mean ± SD = 1.9 ± 1.1 visits/patient; inter-visit interval = 1.8 ± 1.1 years). The relationship between NfL concentration and estimated years to/from symptom onset (EYO) was modelled using linear regression, including all time points and robust standard errors to allow for repeated measurements, adjusting for age at visit and sex. Also, for the 27 participants who became symptomatic (during or before the study), NfL concentration was also modelled against known actual years to/from onset (AYO).

Results: There were 15 non-carriers and 46 mutation carriers (21 symptomatic; 25 presymptomatic). NfL concentration was increased (p = 0.045) in mutation carriers compared with non-carriers 15 years prior to expected symptom onset, increasing progressively thereafter. There was a significant inter- and intra-individual variability in the longitudinal pattern of change. Modelling NfL for the 27 mutation carriers with known AYO also showed a progressive increase over time.

Conclusions: There is evidence that serum NfL is increased more than a decade before the onset of clinical symptoms in FAD and rises thereafter. While there is variability in change over time, both within and between individuals, and more work is needed to understand the sources of this variability, serum NfL remains a promising, accessible biomarker of early neurodegeneration in presymptomatic Alzheimer's disease.

Keywords: Search terms; [111] blood; [111] longitudinal; [111] neurofilament light; [26] Alzheimer’s disease; [91] autosomal dominant.

Fig. 1

Serum NfL against EYO. Mutation carriers are represented in red and non-carriers in black. a Measured values of serum NfL concentration against EYO. Those measurements that belong to the same individual are connected by a line. To ensure it is not possible to identify any of the individual asymptomatic participants (based on their EYO) and so determine their mutation status, five outlying participants have been removed and a jitter of up to ± 2 years has been applied to all remaining participants. Also, for the three individuals who donated more than three samples, only their first three measurements are shown. b Geometric mean NfL modelled against EYO, for someone of average age (40.9 years) and sex (half male/female). Dotted lines indicate 95% confidence intervals. EYO estimated years to/from symptom onset. The geometric mean is the exponential of a mean calculated on the log scale

Fig. 2

Serum NfL against AYO. a (In red) measured serum NfL concentration against AYO, for the 27 mutation carriers who have already developed symptoms (either before or during the study). Those measurements that belong to the same individual are connected by a line. No jitter has been applied. The black broken line represents the mean serum NfL concentration for non-carriers, with dotted lines for ± 1.96 SD, calculated on the log scale and back transformed. By showing ± 1.96 SD, we are indicating a reference range within which 95% of the observed values are expected to lie. b Geometric mean NfL modelled against AYO, for someone of average age (40.9 years) and sex (half male/female). The black broken line represents the geometric mean serum NfL concentration for non-carriers. Dotted lines indicate 95% confidence intervals (i.e. on repeated random sampling, 95% of CIs calculated in this way will include the true population geometric mean). AYO actual years to/from symptom onset. The geometric mean is the exponential of a mean calculated on the log scale