The <scp>CSF</scp> Profile Linked to Cortical Damage Predicts Multiple Sclerosis Activity

Abstract

Objective Intrathecal inflammation correlates with the grey matter damage since the early stages of multiple sclerosis (MS), but whether the cerebrospinal fluid (CSF) profile can help to identify patients at risk of disease activity is still unclear. Methods We evaluated the association between CSF levels of 18 cytokines, previously found to be associated to grey matter damage, and the disease activity, among 99 patients with relapsing‐remitting MS, who underwent blinded clinical and 3 T magnetic resonance imaging (MRI) evaluations for 4 years. Groups with evidence of disease activity (EDA) or no evidence of disease activity (NEDA; occurrence of relapses, new white matter lesions, and Expanded Disability Status Scale [EDSS] change) were identified. Cortical lesions and the annualized cortical thinning were also evaluated. Results Forty‐one patients experienced EDA and, compared to the NEDA group, had at diagnosis higher CSF levels of CXCL13, CXCL12, IFNγ, TNF, sCD163, LIGHT, and APRIL ( p &lt; 0.001). In the multivariate analysis, CXCL13 (hazard ratio [HR] = 1.35; p = 0.0002), LIGHT (HR = 1.22; p = 0.005) and APRIL (HR = 1.78; p = 0.0001) were the CSF molecules more strongly associated with the risk of EDA. The model, including CSF variables, predicted more accurately the occurrence of disease activity than the model with only clinical/MRI parameters (C‐index at 4 years = 71% vs 44%). Finally, higher CSF levels of CXCL13 (β = 4.7*10 −4 ; p &lt; 0.001), TNF (β = 3.1*10 −3 ; p = 0.004), LIGHT (β = 2.6*10 −4 ; p = 0.003), sCD163 (β = 4.3*10 −3 ; p = 0.009), and TWEAK (β = 3.4*10 −3 ; p = 0.024) were associated with more severe cortical thinning. Interpretation A specific CSF profile, mainly characterized by elevated levels of B‐cell related cytokines, distinguishes patients at high risk of disease activity and severe cortical damage. The CSF analysis may allow stratifications of patients at diagnosis for optimizing therapeutic approaches. ANN NEUROL 2020;88:562–573