Plasma neurofilament light chain in the presenilin 1 E280A autosomal dominant Alzheimer's disease kindred: a cross-sectional and longitudinal cohort study

Abstract

Background: Neurofilament light chain (NfL) is a promising biomarker of active axonal injury and neuronal degeneration. We aimed to characterise cross-sectional and longitudinal plasma NfL measurements and determine the age at which NfL concentrations begin to differentiate between carriers of the presenilin 1 (PSEN1) E280A (Glu280Ala) mutation and age-matched non-carriers from the Colombian autosomal dominant Alzheimer's disease kindred.

Methods: In this cross-sectional and longitudinal cohort study, members of the familial Alzheimer's disease Colombian kindred aged 8-75 years with no other neurological or health conditions were recruited from the Alzheimer's Prevention Initiative Registry at the University of Antioquia (Medellín, Colombia) between Aug 1, 1995, and Dec 15, 2018. We used a single molecule array immunoassay and log-transformed data to examine the relationship between plasma NfL concentrations and age, and establish the earliest age at which NfL concentrations begin to diverge between mutation carriers and non-carriers.

Findings: We enrolled a cohort of 1070 PSEN1 E280A mutation carriers and 1074 non-carriers with baseline assessments; of these participants, longitudinal measures (with a mean follow-up of 6 years) were available for 242 mutation carriers and 262 non-carriers. Plasma NfL measurements increased with age in both groups (p<0·0001), and began to differentiate carriers from non-carriers when aged 22 years (22 years before the estimated median age at mild cognitive impairment onset of 44 years), although the ability of plasma NfL to discriminate between carriers and non-carriers only reached high sensitivity close to the age of clinical onset.

Interpretation: Our findings further support the promise of plasma NfL as a biomarker of active neurodegeneration in the detection and tracking of Alzheimer's disease and the evaluation of disease-modifying therapies.

Funding: National Institute on Aging, National Institute of Neurological Disorders and Stroke, Banner Alzheimer's Foundation, COLCIENCIAS, the Torsten Söderberg Foundation, the Swedish Research Council, the Swedish Alzheimer Foundation, the Swedish Brain Foundation, and the Swedish state under the ALF-agreement.

FIGURE 1. Cross-sectional plasma NfL levels start to divert between mutation carriers and non-carriers 22 years before estimated clinical onset (44 years)

A. Log-transformed cross-sectional plasma NfL values of non-carriers (blue triangles, n = 1074) and mutation carriers (red circles, n = 1070) as a function of age. B. Differences in log-transformed plasma NfL (i.e., the space between the carrier and non-carrier mean values at any given age) between carriers and non-carriers as a function of age. Non-carrier levels are set at zero. The curves and credible intervals are drawn from the actual distributions of model fits derived by the Hamiltonian Markov chain Monte Carlo analyses. Baseline plasma NfL increases in PSEN1 E280A mutation carriers began to differ from non-carriers at age 22, 22 years before the carriers’ estimated mean age of 44 at MCI onset. For A and B the shaded areas represent the 99% credible intervals around the model estimates. Numbers below 1A indicate number of participants breaking into separate age ranges matching x-axis label. Pg/ml= Picograms Per Milliliter

FIGURE 2. Rate of change in Plasma NfL levels as a function of age in mutation carriers and non-carriers: Log-transformed longitudinal data

A. Longitudinal change rates of plasma NfL as a function of age. B. Change rate differences between carriers & non-carriers as a function of age (i.e., the space between the carrier and non-carrier mean values at any given age). Non-carrier rates are set at zero. The rate of NfL increases in PSEN1 E280A mutation carriers began to differ from non-carriers at age 22, 22 years before the carriers’ estimated median age of 44 at MCI onset. Log-transformed data were modeled using Linear Mixed Effects Models (LMEMs), a restricted cubic spline, & Hamiltonian Markov chain Monte Carlo (MCMC) analyses. The shaded areas represent 99% credibility intervals. The underlying modeling procedures used in the generation of these representations of the longitudinal changes rates of plasma NfL are very similar to those used to generate the representations in Figure 1 (cross-sectional plasma NfL between carriers and non-carriers) but the dependent variable (i.e. the rate of change in NfL) of the models was different between the analyses. Pg/ml= Picograms Per Milliliter

FIGURE 3. Correlations between baseline and longitudinal plasma NfL and subsequent clinical and cognitive decline for unimpaired and impaired carriers

Baseline plasma NfL levels & annual change rates are associated with annual clinical and cognitive decline rates in impaired & unimpaired mutation carriers. A. Correlations between baseline plasma NfL and annual decline in MMSE scores. B. Correlations between annual increase in plasma NfL and annual decline in MMSE scores. C. Correlations between baseline plasma NfL and annual decline in CERAD word delayed recall scores. D. Correlations between annual increase in plasma NfL and annual decline in CERAD word delayed recall scores. Impaired carriers are shown in red circles and unimpaired carriers are shown in blue triangles. Pg/ml= Picograms Per Milliliter