Normative Values for Serum Neurofilament Light Chain in US Adults

Abstract

Background and Purpose

Neurofilament light chain (NfL) levels serve as a marker of neuroaxonal injury and can be measured in both cerebrospinal fluid and serum. Although serum NfL (sNfL) levels have been shown to increase with the progression of various neurological conditions, normative values for healthy individuals have not yet been established. This study was undertaken to determine age-specific normative values for sNfL and evaluate the associations between sNfL and sociodemographic characteristics.

Methods

A retrospective analysis was conducted using population-based data collected by the National Health and Nutrition Examination Survey between 2013 and 2014. The sera of 2071 adult participants were collected. General linear models were used to examine the associations between sNfL levels and sample characteristics.

Results

The data analysis revealed a significant positive association between age and sNfL levels (p<0.001). Sex was also associated with sNfL levels (p=0.04) after controlling for age. The mean sNfL levels for males and females were 17.99 pg/mL (95% confidence interval [CI]=15.43–20.17) and 15.78 pg/mL (95% CI=13.00–18.55) respectively, after controlling for age.

Conclusions

These results suggest that sNfL levels increase with age and are affected by sex. The findings of this study provide a useful baseline for comparing sNfL levels in clinical practice and future research.

INTRODUCTION

Neurofilaments are structural proteins that form the neural cytoskeleton and are comprised of light, medium, and heavy chains.¹,² Neurofilament light chain (NfL) levels have recently been found to have potential utility due to their high solubility facilitating release into both cerebrospinal fluid and blood.¹ Serum NfL (sNfL) is subject to tight homeostatic regulation with negligible variation.³ Elevated NfL levels have shown promise as a surrogate marker for neuroaxonal injury associated with various neurological conditions such as multiple sclerosis (MS).⁴,⁵,⁶

Verde et al.⁷ recently found that sNfL levels were higher in patients with COVID-19, suggesting that subclinical axonal damage can occur even in the absence of clinically significant neurological manifestations. Importantly, sNfL levels have not been convincingly demonstrated to be related to cognitive performance or the neuropsychiatric symptoms of fatigue, depression, or anxiety.⁷,⁸ While sNfL levels have been found to increase with disease progression and age, age-specific reference ranges have not yet been established in a large population-based study.²,⁹

Previous reports have focused on small samples of individuals with neurological disease such as MS.¹⁰,¹¹,¹²,¹³ It is essential to understand how sNfL levels evolve with age in order to correctly interpret how those levels are associated with axonal damage resulting from injury or disease. The aim of this study was therefore to provide normative values for sNfL by age and sex in a representative sample of noninstitutionalized US adults.

METHODS

A retrospective analysis was conducted using cross-sectional, stratified, multistage probability sample data collected by the National Health and Nutrition Examination Survey between 2013 and 2014. The design and weighting methodology of the survey have been previously described.¹⁴,¹⁵ Participants completed a home-based interview followed by serum collection and a physical examination performed by trained medical personnel at a mobile examination center.

The NfL levels in sera were measured in 2071 participants using the fully automated Atellica Solution Immunoassay & Clinical Chemistry Analyzer system. Details of the laboratory methodology have been published elsewhere.¹⁶ Briefly, samples were first incubated with acridinium-ester (AE)-labeled antibodies as they bound to the NfL antigen. Paramagnetic particles coated with capture antibody were then added to form AE-labeled antibody–antigen complexes. Chemiluminescence was initiated and the consequent light emission was measured after unbound AE-labeled antibodies were removed. The lower limit of quantification (LLOQ) was 3.9 pg/mL for the assay. The value of LLOQ divided by the √2 was imputed for analytic results below the LLOQ (n=36). No samples exceeding the upper detection limit were detected.

Self-reported participant characteristics included age, sex, race/ethnicity, education level, and income. Body mass index (BMI) was calculated using the height and weight measured during the physical examination. Nicotine exposure was assessed using serum cotinine concentrations. Testosterone and estradiol levels in the collected serum were also assessed.

Weighted estimates were determined using subsample-specific weights. Prevalence estimates are reported with their 95% Wald confidence intervals (CIs). Variances following complex sampling procedures were estimated using Taylor series linearization. General linear models (GLMs) were constructed to assess the potential relationships between sNfL and demographic characteristics. All statistical tests were performed with p<0.05 set as the significance level. Analyses were conducted using SPSS Complex Samples software (SPSS version 25, IBM Corp., Armonk, NY, USA).

This study was deemed by the IRB to be exempt from the need to obtain informed consents due to the use of deidentified and publicly available data. The source NHANES 2013-2014 survey was approved by the NCHS IRB (NCHS IRB/ERB protocol number: #201117).

RESULTS

The sample consisted of 2,071 participants representing 212 million US residents aged 20–75 years. The mean age of the sample was 45 years (95% CI=44–46 years). There was no significant difference in the proportions of males (n=990, 48.7%, 95% CI=46.8%–50.6%) and females (n=1,081, 51.3%, 95% CI=49.4%–53.2%). The overall mean sNfL concentration was 16.76 pg/mL (95% CI=14.45–19.07 pg/mL).

GLM results indicated a significant association between age and sNfL levels (p<0.001, r²=0.06). Fig. 1 shows the observed sNfL levels by age with a quadratic best-fit line. Independent of age, sex was also associated with sNfL level (p=0.04). The mean sNfL levels independent of age for males and females were 17.99 pg/mL (95% CI=15.43–20.17 pg/mL) and 15.78 pg/mL (95% CI=13.00–18.55 pg/mL), respectively. Table 1 lists sNfL levels for selected percentiles by age range and sex. Subsequent analyses revealed no relationship between sNfL and total testosterone levels (p=0.24). However, a negative relationship was found between sNfL and estradiol levels (p<0.02).

Fig. 1. Mean serum neurofilament light chain by age.

Table 1. Serum neurofilament light chain levels (in pg/mL) by sex and age.

Sex	Age (years)	No. of participants	Percentile
			10th	25th	50th	75th	90th
Male
	20–29	167	4.7	6.4	8.8	13.7	18.4
	30–39	183	5.2	7.3	9.8	14.2	24.9
	40–49	197	6.1	8.1	11.4	16.0	24.4
	50–59	188	8.9	10.8	14.6	23.9	38.9
	60–69	185	11.7	13.5	18.6	28.6	41.1
	≥70	70	12.0	16.8	24.7	33.1	37.4
Female
	20–29	179	4.3	5.2	7	9.6	14.6
	30–39	187	4.9	6.6	9.1	12.5	22.9
	40–49	223	5.8	7.7	10.5	14.5	26.8
	50–59	205	7.3	10.2	13.5	17.7	24.3
	60–69	196	11	13.5	17.5	24.9	40.4
	≥70	91	14	17.3	22.5	32.3	45.4

No associations were observed between sNfL levels and race/ethnicity, family income, education level, BMI, or nicotine use (all p>0.05). Despite previous reports showing an association between BMI and sNfL levels, no such significance was noted in these data even after controlling for both age and sex (p=0.30).

DISCUSSION

The goal of this report was to describe normative values by age group and thus provide context when assessing the sNfL level of an individual. The observed levels were relatively stable in individuals younger than 35 years, but began to increase nonlinearly as age progressed. While age is the most significant factor to consider when examining sNfL levels, sex should not be discounted.

The negative relationship between estradiol and sNfL suggests a potential mechanism for mitigation of axonal damage or degradation. Examination of the 90th percentile sNfL levels suggested that differences between males and females diminish with age, particularly after 40 years. Further research is needed to determine whether the smaller differences are attributable to biological factors such as hormonal changes, or sociobehavioral factors.

The main strength of this study was its analysis of a representative sample of the US population, which helped to reduce selection bias. The sample was weighted to represent more than 212 million noninstitutionalized adults. Although the nonlinear increase in sNfL levels associated with age confirmed the findings of previous studies that used small samples from various European regions, it is worth noting that the previously identified levels were uniformly lower than these data.¹⁷

One important limitation of this study was the inability to screen out existing neuroaxonal-injury-associated disease (e.g. MS), potentially confounding disease processes (e.g. diabetes), and acute injury. Another limitation of this study was that the blood samples of each participant were only collected at a single time point. Therefore it was not possible to analyze sNfL fluctuations due to changes in the disease state. Finally, the assay used to measure sNfL levels relied on chemiluminescence rather than a single-molecule array. Previous research has indicated that the latter is higher sensitivity in detecting NfL in patients with MS compared with electrochemiluminescence-based assays.¹⁸

In conclusion, the present data suggest that sNfL levels increase nonlinearly with age and are affected by sex. The findings of this study provide general baseline values for comparison when assessing potential axonal damage. Finally, these findings highlight the importance of considering potential confounders that may impact the proposed biomarkers.

Availability of Data and Material

The datasets generated or analyzed during the study are available from the corresponding author on reasonable request.