Prognostic value of serum/plasma neurofilament light chain for COVID-19-associated mortality

Abstract

Objective: Given the continued spread of coronavirus 2, the early predictors of coronavirus disease 19 (COVID-19) associated mortality might improve patients' outcomes. Increased levels of circulating neurofilament light chain (NfL), a biomarker of neuronal injury, have been observed in severe COVID-19 patients. We investigated whether NfL provides non-redundant clinical value to previously identified predictors of COVID-19 mortality.

Methods: We measured serum or plasma NfL concentrations in a blinded fashion in 3 cohorts totaling 338 COVID-19 patients.

Results: In cohort 1, we found significantly elevated NfL levels only in critically ill COVID-19 patients. Longitudinal cohort 2 data showed that NfL is elevated late in the course of the disease, following the two other prognostic markers of COVID-19: decrease in absolute lymphocyte count (ALC) and increase in lactate dehydrogenase (LDH). Significant correlations between ALC and LDH abnormalities and subsequent rise of NfL implicate that the multi-organ failure is the most likely cause of neuronal injury in severe COVID-19 patients. The addition of NfL to age and gender in cohort 1 significantly improved the accuracy of mortality prediction and these improvements were validated in cohorts 2 and 3.

Interpretation: A substantial increase in serum/plasma NfL reproducibly enhanced COVID-19 mortality prediction. Combined with other prognostic markers, such as ALC and LDH that are routinely measured in ICU patients, NfL measurements might be useful to identify the patients at a high risk of COVID-19-associated mortality, who might still benefit from escalated care.

Figure 1

Patient selection, objectives, and experiment outlines of 3 independent cohorts. Cohort 1 aims to analyze NfL cross‐sectionally across disease diagnosis and severity categories. In cohort 2, the objective was to analyze NfL levels in critically ill COVID‐19 patients, longitudinally at three different timepoints (T1, T2, and T3: collected averagely at 5‐ to 10‐day interval, within 30 days of hospitalization). Observed additional prognostic value of NfL with traditional demographic factors (age and gender) from cohorts 1 and 2, was independently validated in cohort 3. NfL, neurofilament light chain; COVID‐19, coronavirus disease 19. This figure was generated using BioRender.com software. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 2

In cohort 1, serum (A) NfL, (B) ALC, (C) CRP, and (D) LDH levels were compared across HC versus COVID‐19 disease severity and multiple sclerosis disease activity subgroups using Kruskal–Wallis ANOVA; **p < 0.005 and ****p < 0.0001. The dotted line on each plot indicates the median of HC. NfL, neurofilament light chain; ALC, absolute lymphocyte count; CRP, C‐reactive protein; LDH, lactate dehydrogenase; COVID‐19, coronavirus disease 19; HC, healthy controls. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 3

In cohort 2, (A) plasma NfL levels at 3 different time points (T1, T2, and T3: collected on average at 5‐to‐10‐day intervals, within 30 days of hospitalization) in critically ill COVID‐19 patients were compared (survived vs. died) using Kruskal–Wallis ANOVA; ****p < 0.0001. The dotted line indicates the median of the HC. (B) Longitudinal plasma NfL levels in critical COVID‐19 patients who died, plotted with respect to the number of days before death. Each line represents data from an individual patient. The dotted line represents upper limit in HC (i.e., mean + 3 × SD = 20 pg/mL). (C) Correlations between systemic biomarkers' measurements at earlier time points (T1 and 2) and NfL measurements at later time points (T2 and T3) were assessed using linear regression analysis. R ² and p‐value are represented on respective correlation plots. The dotted line indicates a 95% confidence interval. NfL, neurofilament light chain; COVID‐19, coronavirus disease 19; HC, healthy controls. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 4

In cohort 2, longitudinal plasma NfL (blue), ALC (orange), CRP (yellow), and LDH (green) levels in critically ill COVID‐19 patients those who died, plotted with respect to the number of days before death. Each plot represents individual patient data. The respective color dotted lines represent upper (for NfL, CRP, and LDH) or lower (for ALC) limit for HC for the respective biomarker (NfL: 20 pg/mL, ALC: 100 × 10/μL, CRP: 5 mg/L and LDH: 280 U/L). NfL, neurofilament light chain; ALC, absolute lymphocyte count; CRP, C‐reactive protein; LDH, lactate dehydrogenase; COVID‐19, coronavirus disease 19; HC, healthy controls. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 5

Comparisons of 4 predictive models of COVID‐19 associated mortality: continuous NfL measurement, Age plus Gender, Age plus Gender plus dichotomized NfL, and Age plus Gender plus ALC plus LDH plus CRP in 3 independent cohorts; (A) cohort 1, (B) cohort 2, and (C) cohort 3. The dotted line on each plot represents the optimal cut‐off for respective model predictor. The numerical values beside respective subgroups (survived or died) on each plot represents the percentage of correctly classified patients. In validation cohort 2, one subject from Died subgroup was repeated from training cohort 1, so that subject was excluded during validation. In validation cohort 3, all three laboratory (ALC, LDH, and CRP) values were available only for 62 subjects (survived = 38 and died = 24), so only these subjects were included in validation (cohort 3: column 4). NfL, neurofilament light chain; ALC, absolute lymphocyte count; CRP, C‐reactive protein; LDH, lactate dehydrogenase; COVID‐19, coronavirus disease 19. [Colour figure can be viewed at wileyonlinelibrary.com]