Cerebrospinal fluid cytokines after autologous haematopoietic stem cell transplantation and intrathecal rituximab treatment for multiple sclerosis

Abstract

Multiple sclerosis has been established as an inflammatory disease of the central nervous system. Many aspects of the pathophysiology are still unknown and it is presently unclear how different treatments affect the immunopathology of multiple sclerosis. In this study, we explored cytokines discriminating between individuals with multiple sclerosis and healthy controls and then how these cytokines were affected by treatment intervention with autologous haematopoietic stem cell transplantation or intrathecal rituximab. CSF from individuals with multiple sclerosis and healthy controls were analysed with a proximity extension assay to simultaneously determine the level of 92 cytokines and other inflammation-related proteins. In total, CSF from 158 multiple sclerosis patients and 53 healthy controls were analysed. Sixty-four patients with relapsing-remitting multiple sclerosis and 27 with progressive multiple sclerosis took part in a cross-sectional study and underwent lumbar puncture on a single occasion. Forty-five patients with relapsing-remitting multiple sclerosis were treated with autologous haematopoietic stem cell transplantation and underwent lumbar puncture at baseline and then at follow-up visits made at 1-, 2- and 5 years. Twenty-two patients with progressive multiple sclerosis were treated with intrathecal rituximab and followed with lumbar punctures at baseline and then at follow-up visits made at 3-, 6- and 12 months. Of the 92 studied cytokines, 16 were found to be altered in multiple sclerosis and 11 were decreased after treatment with autologous haematopoietic stem cell transplantation. None of the studied cytokines was affected by treatment with intrathecal rituximab for progressive multiple sclerosis. Some proteins were highly associated with each other. Therefore, a cluster analysis was made and then the highest-ranked protein from the four highest-ranked clusters was used for the subsequent analyses. CCL3, IL-12B, CXCL10 and IL-8 discriminated between multiple sclerosis patients and controls, but only IL-12B differed between patients with relapsing-remitting and progressive multiple sclerosis. The CSF concentrations of CCL3, IL-12B and CXCL10 were decreased after autologous haematopoietic stem cell transplantation, whereas IL-8 appeared to be unaffected by this intervention. High concentrations of IL-8 were associated with worse outcome in both treatment groups. Overall, the results suggest a profound effect of autologous haematopoietic stem cell transplantation on the inflammatory milieu of the CSF in multiple sclerosis.

Keywords: CSF; autologous haematopoietic stem cell transplantation; cytokines; multiple sclerosis; rituximab.

Graphical abstract

Figure 1

Overview of protein analysis and selection.

Figure 2

Correlation matrix of the 16 proteins differentiating between MS patients and healthy controls. Data from the exploration and replication cohort were combined and then the correlations between 16 proteins differentiating between MS patients and healthy controls were computed using Pearson correlation. Hierarchical clustering was performed using average linkage and the correlation distance $\sqrt{1-r}$, where r is the correlation coefficient. Three clusters of highly correlating (r > 0.75, P < 0.001) proteins could be identified. One cluster contained IL-12B, sCD5 and TNFRSF9, another CXCL10, CXCL11 and MCP-2, and the third cluster IL-8 and CXCL1. CCL3, with the lowest P-value in both the discovery and replication cohort, did not cluster with any other protein and was considered its own cluster.

Figure 3

CSF concentrations of CCL3, IL-12B, IL-8 and CXL10 in healthy controls, patients with relapsing-remitting MS and progressive MS. Using linear models adjusted for sex and age, CCL3, IL-12B, IL-8 and CXCL10 were identified as the most important proteins differentiating between MS patients and healthy controls. Pairwise comparisons between groups using the estimated marginal means from the linear regression model showed that IL-12B was the only protein that was consistently higher in patients with RRMS than patients with PMS. Each datapoint represents a patient’s protein NPX value. NS = not significant, *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 4

CSF concentrations of CCL3, IL-12B, IL-8 and CXL10 in healthy controls, untreated patients and relapsing-remitting MS patients treated with first- and second-line treatment. Treatment effects on baseline CCL3, IL-12B, IL-8 and CXCL10 were estimated using linear regression models adjusted for age and sex. Pairwise comparisons between treatments were estimated using the marginal means from the linear regression model. In general, patients with first-line treatment had higher concentrations than untreated patients and patients treated with second-line treatment. Each datapoint represents a patient’s protein NPX value. NS = not significant, *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 5

CSF concentrations of CCL3, IL-12B, IL-8 and CXL10 in healthy controls and relapsing-remitting MS patients treated with AHSCT. Linear mixed effects models adjusted for sex and age were used to estimate difference over time for the proteins CCL3, IL-12B, IL-8 and CXCL10. Pairwise comparisons between time points were estimated using the marginal means from the linear mixed linear models. After treatment with AHSCT, the levels of IL-12B decreased rapidly to approximately the level of healthy controls. CCL3 and CXCL10 also decreased, but more gradually and did not reach the levels of healthy controls. The levels of IL-8 were unaffected by AHSCT. The association with EDA was estimated within each time point using the marginal means from the model. Patients with evidence of EDA during follow-up had higher levels of CCL3, IL-8 and CXCL10. Each datapoint represents a patient’s protein NPX value at the given time point. NS = not significant, *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 6

CSF concentrations of CCL3, IL-12B, IL-8 and CXL10 in healthy controls and progressive MS patients treated with ITRTX. Linear mixed effects models adjusted for sex and age were used to estimate the difference over time for the proteins CCL3, IL-12B, IL-8 and CXCL10. The levels of these cytokines were unaffected by treatment with ITRTX. The association with disability progression was estimated within each time point using the marginal means from the model. Patients with confirmed disability progression had higher levels of IL-12B and IL-8 at baseline. Each datapoint represents a patient’s protein NPX value at the given time point. NS = not significant, *P < 0.05, **P < 0.01.