Cerebrospinal fluid biomarkers as predictors of multiple sclerosis severity.

Background: Prognostic biomarkers at multiple sclerosis (MS) onset to predict disease severity may help guide initial therapy selection for people with MS. Over 20 disease-modifying treatments (DMTs) of varying levels of risk and efficacy now exist. The ability to predict MS severity would help to identify those patients at higher risk where a highly effective, but potentially risky, therapy would be optimal. The goal of this project was to determine if cerebrospinal fluid (CSF) soluble markers obtained near time of diagnosis can predict disease severity in people with relapsing remitting MS (RRMS).

Methods: We identified 42 RRMS subjects with 4 or more years of clinical follow-up at our center, 8 subjects with other inflammatory neurological diseases (OIND), and 4 subjects with non-inflammatory neurological diseases (NIND) who had donated CSF samples collected for disease diagnosis. This study evaluated soluble CSF biomarkers chosen to reflect neuroinflammation (chemokine ligand 13 - CXCL13), microglia activity (soluble triggering receptor expressed on myeloid cells 2 - sTREM2), demyelination (myelin basic protein -MBP), axon injury and loss (neurofilament light, heavy, and intermediate chains - NFL, NFH, internexin-alpha - INT-α) and neuronal loss (parvalbumin - PVALB) to determine whether any of these CSF factors might predict future MS disease severity. The main outcome measure was MS Severity Score (MSSS), which takes into account disability accumulation (expanded disability status scale - EDSS) and duration of disease. EDSS at last clinical visit was a secondary outcome measure. Univariate and multivariable regression models were used for analysis. Spearman correlations were performed to evaluate correlation between laboratory and clinical variables.

Results: Forty-two RRMS patients with mean 9.4 years follow-up since lumbar puncture (LP) contributed data. Higher NFH, NFL, and sTREM2 each predicted worse MSSS using both univariate and multivariable regression models. Older age at the time of LP predicted worse MSSS both in the univariate and multivariable models. NFL correlated with NFH, and both were positively correlated with sTREM2 and CXCL13. In the combined OIND and NIND comparator group, NFH correlated with both NFL and CXCL13.

Conclusion: These data support that CSF sTREM2, NFH, and NFL are predictors of MSSS, a measure of MS disease aggressiveness. This study adds to a growing literature implicating microglial activity and axonal injury in MS progression, starting from early stages of the disease.