gms | German Medical Science

Background: Neurofilament light (NfL) is assumed to be released from dying neurons. Detection of NfL in serum is a quantitative measure of neuro-axonal injury. Tau, a microtubule-associated protein in neurons, is complementary to NfL because of its strong association with the process of overall brain degeneration and structural brain volume loss. So far, serum NfL and tau have mainly been investigated for their prognostic value in patient populations; studies in the general population are missing. We investigated if serum NfL and tau could serve as prognostic markers for overall mortality in elderly individuals without pre-defined neurological disease.

Methods: In 1997, 385 inhabitants of Augsburg who were aged 65 years and older were included in the Memory and Morbidity in Augsburg Elderly (MEMO) study. They participated in a face-to-face medical interview and a magnetic resonance imaging (MRI) scan of the brain to assess brain volume and white matter lesions (WML). They also provided a non-fasting blood sample. Vital status of all MEMO participants and time of exit was documented prospectively until December 31, 2015. Biomarkers in the serum (NfL and tau) were measured using highly sensitive single molecule array assays.

Associations between the biomarkers and all-cause mortality were investigated using Cox models including subgroup analyses by sex. All regression analyses were adjusted for age, sex, years of education, and morbidity (any history of myocardial infarction, heart failure, vascular surgery, pacemaker, cancer, diabetes, joint replacement or trauma with loss of function based on medical records). Concordance (c) statistics based on the predicted 5-year survival probabilities (estimated with 10-fold cross-validation) were calculated for the base model (age, sex, years of education, morbidity) and its extensions (by successively adding biomarkers, atrophy score, and total volume of subcortical WML). Jackknife estimation was used to assess the c statistic improvement between nested models.

Results: The two biomarkers were slightly correlated (Pearson's ρ = 0.12). NfL (HR 1.26 per standard deviation (10.46 pg/mL), 95% CI [1.12, 1.41]) and tau (HR 1.20 per standard deviation (0.45 pg/mL), 95% CI [1.06, 1.35]) were independently associated with all-cause mortality. Effect estimates were higher among men than among women. When adding biomarkers to the base model (sociodemographic characteristics and morbidity, c statistic 0.68 [0.62, 0.75]), the c statistic did not improve (NfL 0.70 [0.63, 0.76], p=0.092; tau 0.69 [0.62, 0.76], p=0.61; both biomarkers 0.70 [0.63, 0.77], p=0.23). The biomarkers did also not improve the c statistic when added to a model based on sociodemographic characteristics, morbidity, and MRI measurements.

Conclusion: In population-based settings, NfL and tau can be used to identify individuals at higher risk of mortality relative to a reference group. In terms of absolute risk, no biomarker increased the discriminative ability of a model based on sociodemographic characteristics, morbidity, and MRI measurements. If, however, MRI measurements or a full (sub-)clinical morbidity profile are not available, serum tau and NfL can be used instead for prognostic models and seem to harbour important health information in the general population.

The authors declare that they have no competing interests.

The authors declare that a positive ethics committee vote has been obtained.