B-cell composition in the blood and cerebrospinal fluid of multiple sclerosis patients treated with dimethyl fumarate

BACKGROUND
B cells may contribute to the immunopathogenesis of multiple sclerosis (MS). Dimethyl fumarate (DMF) has recently been shown to reduce the frequency of memory B cells in blood, but it is not known whether the drug influences the cellular composition in the cerebrospinal fluid (CSF).


METHODS
A cross-sectional study examining the cellular composition in blood and cerebrospinal fluid (CSF) from 10 patients treated with DMF and 18 patients receiving other disease modifying drugs or no treatment.


RESULTS
Patients treated with DMF had reduced proportions of memory B cells in blood compared to other MS patients (p = 0.0007), and the reduction correlated with treatment duration (rs = -0.75, p = 0.021). In the CSF, the absolute number of mononuclear cells were significantly lower in DMF-treated patients compared to the other patients (p = 0.023), and there was a disproportionate decrease of plasmablasts (p = 0.031).


CONCLUSION
The results of this exploratory study support a B-cell mediated mechanism of action for DMF in both blood and CSF.

. Several potential mechanisms have been proposed, including antigen presentation, cytokine secretion, and antibody production 4 . In MS, the proportion of immunoglobulin class-switched CD27 + B cells is increased in the cerebrospinal fluid (CSF) compared to the peripheral blood 5 , and in particular the frequency of plasmablasts expressing CD38 and CD138 have been shown to be elevated 6,7 . Studies have demonstrated that immunoglobulin class-switched B cells in the CSF are related to B cells in the brain parenchyma 8 , cervical lymph nodes 9 , and peripheral blood 10,11 , indicating a dynamic exchange between these compartments.
Dimethyl fumarate (DMF) is an oral disease-modifying treatment (DMT) approved for treating relapsing-remitting MS (RRMS). The mechanism of action is not fully clarified 12 , but includes a reduction of peripheral blood CD19 + CD27 + memory B cells [13][14][15][16] . Upon intake, DMF is rapidly metabolized to monomethyl fumarate (MMF), which crosses the blood-brain barrier in animal models and in humans, however only at 15% of plasma concentration 17,18 .
The penetration of MMF into the CSF, and the exchange of B cells between the periphery, CSF and the central nervous system (CNS), suggest the possibility that DMF may exert a direct or indirect effect on intrathecal B cells. To address this, we examined blood and CSF Bcell subsets from MS patients treated with DMF and patients not receiving immunomodulatory treatment. We found that while DMF treatment is associated with a decrease of CD19 + CD27 + memory B cells in the periphery, it is associated with a significant reduction of CD19 + CD27 + CD38 + plasmablasts in CSF.

ACCEPTED MANUSCRIPT
A C C E P T E D M A N U S C R I P T 2 Materials and methods

Patients
MS patients (n = 28) were recruited at the Departments of Neurology at Akershus University Hospital and Oslo University Hospital; other data based on the same cohort have been previously published 19 . All patients met the 2010 McDonalds criteria for MS 20 ; one patient was classified as secondary progressive MS and the remaining as RRMS. All patients had intrathecal synthesis of immunoglobulin G (IgG) at the time of diagnosis, detected as oligoclonal IgG bands by isoelectric focusing. Lumbar puncture was performed either as a part or outside of diagnostic investigations, reflected by the variation in disease duration (Table 1). The study was approved by the Regional Ethical Committee South East (2009/23 S-04143a). All patients gave written informed consent before inclusion.

Cell subset analysis
For each individual included in this study, blood, serum and CSF samples were collected once, during the same consultation. Blood and CSF mononuclear cells were isolated as described previously, counted using a Neubauer improved hemocytometer, and immediately processed for flow cytometry, which was performed within two hours after sample collection 11,19 . For B-cell subset identification we used fluorochrome-conjugated antibodies as described 19 , including anti-human CD19, CD27, CD38, CD138, IgG, Ki-67, and HLA-DR, and additionally CD3 and CD14 for dump-channel purposes (BD Biosciences). Paired cell samples from CSF and blood were stained for surface antigens, and after fixation and permeabilization following the manufacturer's instructions (Fixation/Permeabilization Solution Kit, BD Biosciences), the cells were stained for Ki-67. The flow cytometry dataset was reanalyzed for the present study using FlowJo Version 10.2 (FlowJo, LLC). Of note, flow A C C E P T E D M A N U S C R I P T cytometry data from one DMF-treated and one untreated patient were excluded due to technical reasons.

Quantification of IgG and albumin
For each patient, blood and CSF were collected simultaneously. After centrifugation and removal of cells for immediate flow cytometry analysis, the serum and CSF supernatants were frozen at -80° C. The supernatants were thawed and analyzed as a single batch within 36 months after the collection and freezing of the first patient sample. The collection and freezing procedures followed the published consensus protocol for the standardization of cerebrospinal fluid collection and biobanking 21 . IgG and albumin in CSF were quantified by nephelometry (BN ProSpec Systems, Siemens). IgG and albumin in serum were determined by turbidimetry and colorimetry, respectively (Vitros 5.1 FS, Ortho Clinical Diagnostics).

Statistics
Statistical analyses were performed in JMP® pro 12.1 (StataCorp, LLC). Due to nonnormality of data, non-parametric 2-sample Wilcoxon exact tests were used to compare groups unless otherwise specified. The significance level was set at 5%, and the tests were two-sided. No correction for multiple testing was performed. Results are presented as median

Patients
Ten patients received treatment with DMF at inclusion, whereas 18 received other DMTs or no treatment (seven patients received glatiramer acetate, four received teriflunomide, and seven received no treatment). Clinical and demographic data at the time of CSF and blood ACCEPTED MANUSCRIPT Table 1. With the exception of treatment durations, which was shorter for the patients treated with DMF than other DMTs , there were no significant differences in clinical characteristics between DMF-treated patients and the other patients . All DMF-treated patients had normal total lymphocyte count before treatment start, with a median value of 1.7x10 9 /L [1.2-2.6], and two patients developed lymphopenia during DMF treatment (<0.9x10 9 /L).  (Fig S1 A). The CSF mononuclear cell count in DMF-treated patients tended to correlate inversely with the treatment duration, but this was not statistically significant (Spearman r s -0.6, p=0.07).

DMF reduces the frequency of memory B cells in blood and plasmablasts in the CSF
Previous studies have shown that DMF reduces the number of memory B cells in blood [13][14][15][16] , typically identified as CD19 + CD27 + cells. To identify these cells in blood and CSF, a gating strategy based on available data was devised ( Figure 1). Confirming previous studies, we found that DMF-treatment was associated with a relative reduction in CD19 + CD27 + CD38memory B cells in blood (12.8% [2.2-25.2] in DMF-treated vs 26.6% [6.9-46.6] in the other MS patients, p=0.0007, Figure 2A). Moreover, the reduction correlated strongly with treatment duration (r s = -0.75, p=0.021, Figure 2B). There were no significant differences between the two groups in the proportions of CD19 + CD27 + CD38 + plasmablasts in blood  Figure 2A). There were no differences between the two groups in the proportion of total CD19 + B cells in blood or CSF (Figure 2A). Including only DMF-treated and untreated patients in the analyses yielded similar results (Fig. S1 B).

DMF reduces HLA-DR expression among memory B cells in blood
To further characterize the B-cell subsets, we examined antigen-presenting potential (HLA-DR), IgG expression, proliferation status (Ki-67), and the expression of the differentiation marker CD138. In patients treated with DMF, we found that blood CD19 + CD27 + CD38 -

DMF treatment did not influence intrathecal IgG production
Since DMF-treated patients exhibited preferential reductions of CD19 + CD27 + CD38 + antibody-secreting cells in CSF, we compared the intrathecal IgG production among DMFtreated patients (n = 10) with that among untreated patients (n = 7). There were no differences in IgG or albumin levels or indices of these, also indicating a similar blood-brain barrier integrity (Table 1).

Discussion
This cross-sectional study demonstrates that while DMF-treatment causes a time-dependent reduction in CD19 + CD27 + CD38memory B cells in blood, it leads to a depletion of CD19 + CD27 + CD38 + plasmablasts in the CSF. Our findings in blood are in agreement with recent publications [13][14][15][16] , and the present work demonstrates how the peripheral effects of DMF translate into the CSF. These data point to a possible B-cell mediated mechanism of action for DMF, in addition to the previously discussed pluripotent immunomodulatory effects 12 .
The almost immediate effect of anti-CD20 treatment in MS is thought to be mediated by a reduction of B cells participating in antigen presentation and/or cytokine production 22,23 .
Accordingly, the intrathecal IgG production does not seem to be affected by the treatment in the short run 1,24 . Memory B cells express HLA class II molecules and are potent antigenpresenting cells 25 . Interestingly, we found that blood CD19 + CD27 + CD38memory B cells express lower levels of HLA-DR in DMF-treated than in untreated MS patients, suggesting that memory B cells in these patients are less efficient in presenting antigens. One could speculate that a reduced amount of memory B cells, with additionally reduced expression of HLA class II molecules, may result in less maturation and development of

A C C E P T E D M A N U S C R I P T
CD19 + CD27 + CD38 + antibody-secreting cells in the CSF. This possibility is supported by the dynamic exchange of B cells between the CSF and periphery [8][9][10][11] . However, a not mutually exclusive possibility is that the observed alterations in the composition of CSF B cells could be due to a local intrathecal effect of MMF.
Despite depletion of CD19 + CD27 + CD38 + plasmablasts in the CSF of DMF-treated patients, we did not observe a reduced intrathecal IgG production. This could be due to persistent secretion from end-differentiated plasma cells protected within tertiary lymphoid tissue in the meninges of MS patients 26 . In line with this thought, we found that all antibody-secreting cells in the CSF express high levels of Ki-67, compatible with recently derived proliferating plasmablast. Interestingly, a similar persistence of intrathecal IgG synthesis has been observed in patients receiving anti-CD20 therapies 27 . In accordance with the present findings, no difference in circulating IgG, IgA, or IgM in blood of DMF treated patients has been observed, indicating that end differentiated plasma cells could be protected from the effects of the drug also in the periphery in a short-term perspective 14 .
The strength of this study is the comparison of CSF B-cell subsets between two groups of MS patients similar in clinical and demographic characteristics. This was made possible by slight different treatment strategies between the two clinical departments contributing to the study.
Samples from both hospitals were transported directly to the same external laboratory and processed identically, and the findings are not influenced by differences in inflammatory activity. Although we cannot exclude that differences in treatment durations could have influenced the results, treatment durations were long enough to allow full immunological effect in the majority of patients receiving DMF and other DMTs.

A C C E P T E D M A N U S C R I P T
This study has several limitations. First, the number of patients is small. Given the risk of lumbar puncture headache, as well as rare but more serious risks such as infection, bleeding and herniation, we cannot perform lumbar puncture solely for the purpose of studying drug effects. While properly understanding the mechanism of action of DMF in MS patients is important, such studies must due to the above reasons be part of studies on disease mechanisms, which was the primary aim of this study 19 . This is therefore an exploratory study, the p-values were not corrected for multiple testing, and the results need to be reproduced in an independent study. Second, the control group is heterogeneous from a treatment standpoint, which potentially could confound our results. Nevertheless, as the subanalysis with untreated controls (n = 7) demonstrated similar results, and lymphocyte depletion and reduced memory B-cell population in blood are analogue to previous reports 13-16, we believe our findings have merit. While reduction in blood plasmablasts similar to what we observed in CSF have not been extensively described, a recent interim analysis of the PROCLAIM study demonstrated a decline in blood plasmablasts that had a delayed onset compared to memory B cell depletion 28 . This delay, in addition to the relatively low proportions of circulating plasmablasts in blood compared to CSF of MS patients can potentially have hidden this phenomenon in studies with shorter time-frames. Lastly, as our data did not contain markers for IgD, we were unable to differentiate between class-switched and unswitched memory B cells. A recent publication limited to blood found that both populations are affected similarly, with a corresponding increase in CD27 -IgD + naïve B cells 29 . New studies as to how these subpopulations are distributed in the CSF could also enlighten whether the effects of DMF are mainly peripheral or intrathecal.

Conclusion
This is the first study on the effects of DMF treatment on B cells in CSF, suggesting that while memory B cells are reduced in blood, plasmablasts are preferentially depleted in CSF.
These findings, which need to be verified in an independent study, support the hypothesis that DMF mediates positive effects through B-cell modulation.