Predictive power of serum Neurofilament light chain (sNFL) for SARSv2 associated morbidity and mortality

National Institute Of Allergy And Infectious Diseases

Investigators

Abstract

Using validated, high-sensitivity assay (Single Molecule Array; SIMOA) we measured serum or plasma NFL in total of 3 cohorts of subjects. All measurements were performed on coded samples and the diagnostic code was broken only after NFL levels were measures and quality controlled. Cohort #1 investigated question whether NFL is elevated in COVID19 subjects and whether its levels correlate with COVID19 severity. As negative controls we used healthy individuals (n=28) and individuals with acute respiratory infection but negative SARS-Cov2 serology (n=10). As positive controls we used subjects with neuroinflammatory disease multiple sclerosis (MS) with active inflammation visible on brain MRI (i.e., active; n=35) or without neuro-imaging visible inflammation (i.e., inactive; n=35). We analyzed NFL levels in 30 SARS-Cov2+ individuals hospitalized with COVID19: 10 with moderate disease severity, 10 in critical condition but who eventually survive and 10 in critical condition who died. Upon unblinding diagnostic codes, we observed significantly elevated NFL levels (i.e., beyond healthy donor range) only in COVID19-subjects in critical condition. COVID19 patients with moderate disease severity had NFL levels comparable to subjects with inactive MS, even though the comparison with healthy individuals did not reach statistical significance in this small cohort. ON the other hand, subjects with COVID19 in critical condition and especially subjects who eventually succumbed to the disease had NFL levels elevated way beyond active MS patients. As these results showed that NFL has prognostic value in COVID19 morbidity and may even differentiate patients in critical condition who will survive from those who may eventually die, we asked whether serum/plasma NFL levels provide prognostic value beyond traditional prognostic markers such as age, gender and comorbidities. We observed that adding NFL to demographic prognostic factors enhanced accuracy of the combined mortality classifier from the area under receiver operator characteristics curve (AUC of ROC) from 0.8 to 0.85. We used Cohort#2 to validate the added value of NFL to predict COVID19-associated morbidity, but also to explore, in longitudinal samples, the dynamics of CNS injury in relationship to other systemic biomarkers of COVID19 severity and organ damage: Lactate dehydrogenase (LDH), lymphopenia (Absolute lymphocyte count), C reactive protein (CRP) and d-dimer. We analyzed, again blindly, 60 plasma NFL samples, 3 each derived from 20 COVID19 patients in critical conditions with average of 10 days interval between each sample. Upon unblinding the data, we observed that predictive model that included NFL was able to differentiate subjects who eventually died from those who survives at second (p=0.0048) and third time points (p = 7e-4). However, the models that included only demographic prognostic markers could not predict COVID19 mortality in the same cohort with statistical significance (p=0.06 for both time-points). We also observed that all other systemic biomarkers of COVID19 severity exhibited large day-to-day fluctuations and generally rose several days/weeks before increase in NFL. In contrast, NFL levels only increased or remained stable between different time-points, consistent with known dynamic of this biomarker in acute brain injury: it increases within 24-72h and remains elevated for at least 3 months. This dynamic make NFL uniquely valuable as biomarker of acute CNS injury, as single blood value reliably reflects the amount of axonal injury in preceding 3 months, while all other biomarkers studied here must be measured daily to identify rapid rise and equally rapid decline in abnormal values. We conclude that serum/plasma NFL significantly enhances prognostic value of demographic biomarkers in COVID19 patients. Because this conclusion was still based on small cohort of critically ill patients (n=20), to provide unequivocal evidence of clinical value of single NFL measurement in COVID19 patients, we used Cohort#3 of 288 critically ill COVID19 patients. Upon unblinding, we observed that NFL values differentiated patients who eventually survived (n=113) from those who dies (n=175) with p<0.001. We conclude that serum/plasma NFL provides strong, non-redundant prognostic value of COVID19-ssociated morbidity and mortality.