Relationship between cerebrospinal fluid biomarkers of inflammation and tissue damage in primary progressive multiple sclerosis

Background and Objectives It is unclear to what extent intrathecal inflammation contributes to the pathogenesis in primary progressive multiple sclerosis (PPMS). We conducted an exploratory study to investigate the degree of intrathecal inflammation and its association with biomarkers of disease activity and severity in patients with PPMS.


Abstract Background and Objectives
It is unclear to what extent intrathecal inflammation contributes to the pathogenesis in primary progressive multiple sclerosis (PPMS). We conducted an exploratory study to investigate the degree of intrathecal inflammation and its association with biomarkers of disease activity and severity in patients with PPMS.

Methods
We included patients with PPMS who participated in a randomized controlled trial conducted at the Danish Multiple Sclerosis Center, patients with relapsing-remitting multiple sclerosis (RRMS) and healthy controls. We analyzed concentrations of a panel of cytokines in CSF using electrochemiluminescence assays. We then explored the relationship between cytokines found in increased CSF concentrations in patients with PPMS (compared with healthy controls) with CSF concentrations of neurofilament light chain (NFL) and myelin basic protein (MBP), IgG-index, and magnetic resonance imaging (MRI) metrics (volume, magnetization transfer ratio and diffusion tensor imaging) from lesions, normal-appearing white matter, and cortical grey matter.

Results
We included 59 patients with PPMS, 40 patients with RRMS, and 21 healthy controls. In patients with PPMS, CSF concentrations of CC chemokine ligand 3 (CCL-3), CXC chemokine ligand 8 (CXCL-8), CXCL-10, interleukin (IL)-10, IL-15, and vascular endothelial growth factor (VEGF)-A were increased compared with healthy controls and comparable with CSF concentrations in patients with RRMS. In addition, patients with PPMS had increased CSF concentrations of IL-12p40, IL-17A, tumor necrosis factor (TNF)-α, and lymphotoxin (LT)-α compared with healthy controls, but concentrations of these cytokines were even higher in patients with RRMS. For the remaining seven cytokines (CCL22, interferon-γ, IL-5, IL-7, IL-16, IL-22, IL-27), we found no difference between patients with PPMS and healthy controls. CSF concentrations of NFL and MBP correlated weakly with concentrations of IL-15, while the remaining proinflammatory cytokines were not associated with CSF concentrations of NFL or MBP. The IgG-index correlated with four cytokines (IL-10, IL-12p40, TNF-α, and LT-α). We did not observe any significant associations between MRI metrics and CSF biomarkers of inflammation.

Introduction
Multiple sclerosis (MS) is an immune-mediated, demyelinating disease of the central nervous system (CNS) in which 10-15% of patients experience a slowly progressive course from onset, termed primary progressive MS (PPMS). 1 The disease course in MS can be described using the terms disease activity (relapses, new or enlarging or Gd-enhancing [GEL] magnetic resonance imaging (MRI) lesions) and progression (gradual disability worsening not explained by superimposed relapses). 2 Many disease-modifying treatments (DMTs), specifically targeting immune cells, inhibit relapse and focal MRI activity. 3 The efficacy of DMTs in patients with progressive MS is lower and only B cell-depleting therapy has shown some efficacy in slowing disease progression in patients with PPMS. 4,5 It is still unresolved why the therapeutic benefits of most DMTs on relapses and MRI activity do not translate into an effect on disease progression in PPMS. Intrathecally compartmentalized inflammation suggested as a primary mediator of disease progression in progressive MS may offer part of the explanation. 6 Additionally, it has been proposed that inflammation subsides along with the progressive phase of the disease, and that the mechanisms responsible for propagating progression are independent from inflammatory activity in PPMS. 7 However, the burden of inflammation seems evident especially in patients with secondary progressive MS (SPMS) in whom increased intrathecal concentrations of proinflammatory cytokines were related to worse disease course and extensive cortical inflammatory activity. 8 However, few studies have assessed the degree of inflammation and the relationship with disease activity and progression in PPMS, although the processes applicable in SPMS seem relevant also in PPMS as cerebrospinal fluid (CSF) concentrations of biomarkers of inflammation are comparable in PPMS and SPMS. 9 We recently completed a randomized, placebo-controlled trial of dimethyl fumarate (DMF) treatment in PPMS where CSF concentrations of biomarkers of neuroaxonal damage and demyelination as well as selected biomarkers of inflammation were used as outcome measures. 10,11 In this study we further investigate the extent of inflammation in PPMS compared with untreated patients with early RRMS and healthy controls (HCs). We also characterize the relationship between intrathecal inflammation and disease activity and severity assessed with MRI and CSF biomarkers of demyelination and neuroaxonal damage.

Subjects and CSF collection
We enrolled 59 patients with PPMS. All patients with PPMS were screened for inclusion in a randomized, placebo-controlled trial conducted at the Danish Multiple Sclerosis Center, Copenhagen University Hospital -Rigshospitalet from Dec 2016 to Jan 2019. 11 Primary inclusion criteria were age between 18 and 65 years, no immunomodulatory or immunosuppressive therapy within six months before inclusion, and no steroid treatment within three months before inclusion. The study was approved by the Capital Region Ethics Committee, Denmark (protocol H-16047666, Clinicaltrials.gov number: NCT02959658), and informed consent was obtained from all patients before enrolment. In addition, we collected CSF from 40 untreated patients with RRMS and 21 HCs according to the same procedure as described below. Patients with RRMS were untreated and recruited from the Danish Multiple Sclerosis Center between Feb 2016 and Mar 2021 (ethics committee protocol number H-17005703). All patients with RRMS were diagnosed according to the 2017 McDonald criteria. 12 The HCs were huntingtin gene expansion negative family members of Huntington disease gene expansion carriers from the Neurogenetics Clinic, Copenhagen University Hospital -Rigshospitalet (ethics committee protocol number H2-2011-085 and H-17002606).
We collected 10 to 12 ml of CSF of which three ml was sent for routine diagnostic analysis of cell count, immunoglobulin-G (IgG)-index, albumin quotient, and oligoclonal bands. The remaining CSF was separated from cells by centrifugation and stored at -80°C until analysis.

MRI
We performed MRI on all patients with PPMS at screening and after 48 weeks. Patients were scanned on the same 3T Verio MRI scanner (Siemens Healthcare, Erlangen, Germany) with a 32channel head coil. T1-weighted (T1W), T2-weighted (T2W), and fluid-attenuated inverse recovery (FLAIR) 3D sequences with 1 mm 3 isotropic resolution were acquired with structural imaging. In addition, diffusion tensor imaging (DTI) sequences and data for magnetization transfer ratio (MTR) were acquired. For the present study we analyzed the relationship between CSF biomarkers and lesion volume, MTR in lesions, normal-appearing white matter (NAWM) volume, fractional anisotropy (FA) and mean diffusivity (MD) in NAWM, scaled cortical gray matter (CGM) volume, and MTR in scaled CGM. Finally, we performed longitudinal analysis of new lesions and enlarging lesions developing over 48 weeks. Details regarding the MRI data acquisition and analysis has previously been described in detail. 11

Statistics
Data are presented as mean (SD) and median [Q1, Q3] as appropriate. We compared CSF cytokine concentrations across groups by applying analysis of variance (ANOVA) tests on linear models adjusted for age and sex to assess overall group differences. We used a multiple testing-corrected (Bonferroni) significance value of 0.05. Post-hoc, we conducted age and sex-adjusted linear models with log-transformed concentrations as outcomes to evaluate pairwise differences between the three groups. We performed model control for all generalized linear models. In posthoc analysis we considered a p-value below 0.05 as statistically significant. Cytokines with increased CSF concentrations in patients with PPMS compared with HCs were analyzed for associations using Spearman rank correlation analysis. For all correlation analyses, a two-sided significance level of q<0.05 was considered significant (false discovery rate [FDR] corrected). All analyses and illustrations were conducted with RStudio v 1.2.5 (extension packages tidyverse, ggendro, factoextra, and patchwork). [13][14][15][16][17] Results

Increased cytokine and chemokine concentrations in patients with PPMS
Baseline comparisons revealed differences in sex, age, disease duration, expanded disability status scale (EDSS) and CSF cell count between groups (Table 1). We performed Bonferroni-corrected analysis of variance on linear models adjusted for sex and age to compare concentrations of cytokines between the three groups. We found group differences for the following cytokines We found considerable age differences between groups, and some of the investigated cytokines correlated with age (eTable 2 in the supplement), and we therefore conducted post-hoc test based on a linear model adjusted for age and sex to assess differences between groups for each cytokine (figure 1). We observed increased concentrations of the following cytokines in patients with PPMS compared with HCs (mean fold increase in patients with PPMS patients vs.

Relationships with MRI
We analyzed associations between CSF cytokines and MRI measures of 1-year disease activity (new and enlarging lesions) and cross-sectional MRI measures of disease severity (lesion volume, MTR in lesions, NAWM volume, fractional anisotropy and mean diffusivity in NAWM, CGM volume, and MTR in CGM). We found no significant associations between CSF cytokines and MRIderived metrics of disease activity or severity after correction for multiple comparisons (q<0.05) (Supplementary Data, Figure S1).

Discussion
Our results show that intrathecal inflammation is present in patients with PPMS and, for many cytokines, the relative increase in CSF concentration was comparable to those observed in patients with newly diagnosed RRMS. These results expand our previously published analyses which revealed an increase in CSF concentrations of sCD27, sBCMA, CHI3L1, MBP, and NFL in the same PPMS cohort compared to a group of individuals who had neurologic symptoms but without any objective or paraclinical findings to define a specific neurologic disease. 11 Notably, patients with SPMS with profound meningeal inflammation and gray matter demyelination had elevated mRNA expression and CSF concentrations of proinflammatory and B cell associated cytokines compared with SPMS patients with little meningeal inflammation and gray matter demyelination. 18 Likewise, patients with RRMS with a high cortical lesion load at diagnosis had increased CSF concentrations of proinflammatory and B cell associated chemokines and cytokines (CXCL-8, CXCL-10, IL-10, LT-α, and TNF-α). 18 We also show that proinflammatory and B cell-related biomarkers such as IgG-index, LT-α, and IL12-p40 are increased in patients with PPMS, indicating a pivotal role of B cell activity in the immune pathogenesis of both RRMS and progressive MS. 19 The concentrations of NFL and MBP in CSF have been shown to reflect neuroaxonal damage and demyelination, respectively. 20,21 Surprisingly, patients with PPMS showed only weak and relatively few associations between NFL and MBP levels in CSF and CSF biomarkers of inflammation. Studies with mixed progressive MS patients have found contradictory results for NFL concentrations that were closely associated with several CSF biomarkers of inflammation. 22,23 The patients in the present study were considerably older than the patients in the previous studies, and we hypothesize that age-dependent changes may contribute to the differences between the results of the present and previous studies.
Although we used an extensive and detailed MRI protocol to capture disease-related changes in brain structure, we found no associations between biomarkers previously related to disease activity and severity in patients with SPMS and MRI-metrics of disease activity and disease severity in this cohort of patients with PPMS. 8 However, meningeal inflammation and cortical lesion load were not assessed in this study, and ultra-high field MRI as well as a prospective study of changes in MRI metrics over a longer period could have added valuable information to our findings. 8,24 We chose a conservative method of analysis regarding electrochemiluminescence values, but the low assay sensitivity may influence the results. However, we ensured an acceptable standard curve for all assays, and we excluded assays with a mean CV above 20% to account for high assay variation. The exploratory nature of the study may also induce type I errors, but we chose a conservative definition of significant findings which, on the other hand, increases the risk of type II errors.
All PPMS data were collected from patients participating in a randomized, placebo-controlled trial that showed no difference between patients receiving DMF or placebo. 11,25 The lack of an DMF treatment effect on CSF biomarkers of inflammation in PPMS does not seem to reflect lack of ongoing inflammation as concentrations were increased and comparable to levels in RRMS for many of the biomarkers studied. Histopathological and CSF studies have, indeed, provided strong evidence that there is widespread inflammation in the CNS in patients with PPMS involving both glial and innate immunity in patients with PPMS. 11,24,[26][27][28] The biomarkers of inflammation found to have increased concentrations in CSF in our patient cohort reflect both adaptive and innate immune cell activation.
In conclusion, we explored an extensive panel of biomarkers reflecting CSF inflammation in a cohort of patients with PPMS and found that although there was clear evidence of ongoing intrathecal inflammation, associations with biomarkers of demyelination and neuroaxonal damage were weak, and there were no associations with MRI measures of disease activity and severity. Although many of the analyzed biomarkers have previously been associated with disease severity and neuroaxonal damage in patients with progressive MS (primarily patients with SPMS), this could not be confirmed in the present study Legend (figure 1): Abbreviations: ng/L = nanograms/liter; IL = interleukin; LT = lymphotoxin; VEGF = vascular endothelial growth factor; CCL = chemokine ligand; CXCL = CXCL chemokine ligand; TNF = tumor necrosis factor; PPMS = primary progressive multiple sclerosis; HC = healthy controls; RRMS = relapsing-remitting multiple sclerosis. Boxplots show median and interquartile range, and whiskers show range excluding outliers. *p<0.05, **p<0.005, ***p<0.001, ns = not significant. Pvalues are derived from general linear model with log-transformed concentrations as dependent variable and group, age, and sex as independent variables (covariates). IL-5, IL-16 and IL-22 were not included in this analysis since no difference was found between groups in the pre-conducted ANOVA analysis (see methods).   OCBs; n (%) 52 (88%) 0 (0%) 39 (98%) <0.001