Evaluating the utility of serum NfL, GFAP, UCHL1 and tTAU as estimates of CSF levels and diagnostic instrument in neuroinflammation and multiple sclerosis

Background: This study aimed to evaluate the utility of neurofilament light chain (NfL), glial fibrillary acidic protein (GFAP), ubiquitin C-terminal hydrolase L1 (UCHL1) and total tau (tTAU) serum concentrations as approximation for cerebrospinal fluid (CSF) concentrations of the respective biomarkers in the context of neuroinflammation and multiple sclerosis (MS).

Methods: NfL, GFAP, UCHL1 and tTAU concentrations in serum and CSF were measured in 183 patients (122 with neuroinflammatory disease and 61 neurological or somatoform disease controls) using the single molecule array HD-1 analyzer (Quanterix, Boston, MA). Spearman's rank correlations were computed between serum and CSF concentrations. In a second step, the effects of age, BMI, gadolinium-enhancing lesions in MRI, integrity of the blood-brain barrier (BBB) and presence of acute relapse were accounted for by computing partial correlations. The analyses were repeated for a subsample consisting of MS phenotype patients only (n = 118). EDSS, MS disease activity and acute relapse were considered as additional covariates. Receiver operating characteristic (ROC) analysis was performed for each serum/CSF biomarker concentration to assess how well the particular biomarker concentration differentiates MS patients from somatoform disease controls. Correlations between serum and CSF levels as well as area under the curve (AUC) values were compared for the different biomarkers using z-test statistics.

Results: Serum concentrations correlated positively with CSF levels for NfL (r = 0.705, p < 0.01) as well as for GFAP (r = 0.259, p < 0.01). Correlation coefficients were significantly higher for NfL than for GFAP (z = 5.492, p < 0.01). We found no significant serum-CSF correlations for UCHL1 or tTAU. After adjusting for covariates, the results remained unchanged. In the analysis focusing only on MS patients, the results were replicated. ROC analysis demonstrated similarly acceptable performance of serum and CSF NfL values in differentiating MS phenotype patients from somatoform disease controls. AUC values were significantly higher for serum and CSF NfL compared to other biomarkers.

Conclusion: NfL and GFAP but not UCHL1 or tTAU serum concentrations are associated with CSF levels of the respective biomarker. NfL exhibits more robust correlations between its serum and CSF concentrations as compared to GFAP independently from BBB integrity, clinical and radiological covariates. Both serum and CSF NfL values differentiate between MS and controls.