RESEARCH ARTICLE OPEN ACCESS Serum Levels of CXCL13 Are Associated With Teriflunomide Response in Patients With Multiple Sclerosis Nicolás Fissolo, PhD, Agustin Pappolla, MD, Jordi Rio, MD, PhD, Luisa M. Villar, MD, PhD, Correspondence Dr. Fissolo Santiago Perez-Hoyos, BSc, PhD, Alex Sanchez, PhD, Lucı́a Gutierrez, BSc, Xavier Montalban, MD, PhD, and nicolas.fissolo@vhir.org Manuel Comabella, MD, PhD Neurol Neuroimmunol Neuroinflamm 2023;10:e200050. doi:10.1212/NXI.0000000000200050 Abstract Background and Objectives To identify biomarkers associated with treatment response in patients with multiple sclerosis (MS) treated with the oral therapies teriflunomide, dimethyl fumarate (DMF), and fingolimod. Methods Serum levels of IL-6, IL-17, TNF-α, granulocyte-macrophage colony-stimulating factor, IL-10, interferon-gamma (IFN-γ) IL-1β, and chemokine ligand 13 (CXCL13) were measured at baseline and 12 months with single molecule array (Simoa) assays in a cohort of patients with MS treated with teriflunomide (N = 19), DMF (N = 22), and fingolimod (N = 25) and classified into “no evidence of disease activity” (NEDA) and EDA patients after 1 year of treatment. Results Serum CXCL13 and TNF-α levels were significantly decreased after treatment with teriflunomide in NEDA compared with EDA patients after 1 year of treatment (p = 0.008 for both cytokines). These findings were validated in an independent cohort of patients with MS treated with teriflunomide (N = 36) and serum CXCL13, and TNF-α levels were again significantly reduced in NEDA patients (p < 0.0001 for CXCL13 and p = 0.003 for TNF-α). CXCL13, but not TNF-α, showed good performance to classify NEDA and EDA patients according to a cut-off value of 9.64 pg/mL based on the change in CXCL13 levels between baseline and 12 months, with a sensitivity of 75% and specificity of 82% in the original cohort, and sensitivity of 65.4% and specificity of 60% in the validation cohort. Discussion Altogether, these results point to CXCL13 as a treatment response biomarker to teriflunomide in relapsing-remitting patients with MS, and the change in CXCL13 levels during the first year of treatment can be used in clinical practice to identify optimal responders to teriflunomide. From the Servei de Neurologia, Centre d’Esclerosi Múltiple de Catalunya, Institut de Recerca Vall d’Hebron (N.F., A.P., J.R., L.G., X.M., M.C.); Departments of Immunology and Neurology (L.M.V.), Multiple Sclerosis Unit, Hospital Ramon y Cajal, (IRYCIS), Madrid; Statistics and Bioinformatics Unit, Vall d’Hebron Institut de Recerca (VHIR) (S.P.-H., A.S.); and Genetics, Microbiology and Statistics Department (A.S.), Universitat de Barcelona, Spain. Go to Neurology.org/NN for full disclosures. Funding information is provided at the end of the article. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal. Copyright © 2022 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Glossary AUC = area under the ROC curve; CXCL13 = chemokine ligand 13; DMF = dimethyl fumarate; EDA = evidence of disease activity; EDSS = Expanded Disability Status Scale; GM-CSF = granulocyte-macrophage colony-stimulating factor; IFN-γ = interferon-gamma; MS = multiple sclerosis; NEDA = no evidence of disease activity; ROC = Receiver operating characteristic; RRMS = relapsing-remitting MS. Biomarkers are indicators of normal biological or pathogenic (DMF), and fingolimod and classified them according to the processes that can be objectively measured.1 Among the dif- presence or absence of disease activity. For this, we measured ferent types of biomarkers, treatment response biomarkers are the serum levels of a panel of inflammatory biomarkers that those measured in patients receiving therapies to identify those have been proposed at some point to play a role in MS etio- individuals who are at risk for treatment failure and/or serious pathogenesis and included cytokines of the innate immune adverse drug reactions. Multiple sclerosis (MS) is a complex system (IL-1β, IL-6, and TNF-α), T-cell–related cytokines neurologic disorder characterized by a high degree of hetero- (IL-17, interferon-gamma [IFN-γ], IL-10, and granulocyte- geneity regarding several disease aspects including the response macrophage colony-stimulating factor [GM-CSF]), and B-cell to treatment.2 Although numerous studies have been published cytokines (chemokine ligand 13 [CXCL13]).3,5-7 aiming to identify biomarkers associated with the response to treatment in patients with MS, very few treatment response biomarkers are used at present in clinical practice to guide treatment decisions.3,4 This aspect is critical considering the Methods current scenario of MS treatment, with many different options Original Cohort available to treat patients with MS. In this study, we aimed to Patients with relapsing-remitting MS (RRMS) treated with ter- identify treatment response biomarkers in patients with MS iflunomide, DMF, or fingolimod at the outpatient clinic of the receiving the oral therapies teriflunomide, dimethyl fumarate Centre d’Esclerosi Multiple de Catalunya (Cemcat, Barcelona) Table 1 Demographic and Baseline Clinical Characteristics of Patients With MS From the Original Cohort Teriflunomide Dimethyl fumarate Fingolimod Baseline characteristics NEDA EDA p Value NEDA EDA p Value NEDA EDA p Value N 8 11 — 13 9 — 6 19 — Age (y) 44.9 (8.8) 43.0 (8.4) 0.3 45.1 (9.9) 44.5 (9.7) 0.3 37.5 (3.5) 37.7 (7.9) 0.9 Female/male (% women) 3/5 (37.5) 10/1 (90.9) 0.01 10/3 (77.0) 7/2 (77.8) 0.9 4/2 (66.7) 14/5 (73.9) 0.7 Duration of disease (y) 12.1 (6.2) 12.2 (6.2) 0.5 10.8 (12.0) 10.6 (11.5) 0.4 7.9 (5.3) 10.1 (6.4) 0.1 e Naı̈ve/no naı̈ve (% naı̈ve) 1/7 (12.5) 3/8 (27.7) 0.4 8/5 (61.5) 3/6 (33.3) 0.2 1/5 (16.7) 1/18 (5.3) 0.4 EDSSa 1.9 (1.0–2.0) 2.3 (1.5–2.5) 0.3 2.0 (1.0–2.5) 2.6 (1.5–3.5) 0.3 2.7 (1.9–4.0) 2.7 (2.0–3.5) 0.9 No. of relapsesb 0.6 (0.7) 0.7 (0.7) 0.9 1.0 (0.7) 1.1 (0.8) 0.5 2.1 (1.4) 2.2 (1.5) 0.8 c,d No. of T2 lesions [n (%)] 1 5 (62.5) 1 (9.1) 0.04 5 (38.5) 2 (22.2) 0.08 0 (0) 0 (0) 0.7 2 2 (25.0) 5 (45.4) 8 (61.5) 4 (44.5) 2 (33.3) 5 (26.3) 3 1 (12.5) 5 (45.4) 0 (0) 3 (33.3) 4 (66.7) 14 (73.7) c No. of Gd-enhancing lesions 0 0.3 (1.0) 0.3 0.7 (1.2) 0.7 (1.4) 0.5 4.4 (4.7) 3.0 (4.2) 0.4 Abbreviations: EDA = evidence of disease activity; EDSS = Expanded Disability Status Scale; NEDA = no evidence of disease activity. Data are expressed as mean (SD) unless otherwise stated. a Data are expressed as mean (interquartile range). b Refers to the number of relapses in the 2 previous years of blood extraction. c Refers to the number of T2 and gadolinium-enhancing lesions at the time of blood extraction. d T2 lesions were classified into 3 categories according to the number of lesions: 1 = 1–3 lesions, 2 = 4–9 lesions, 3 = ≥10 lesions. e Before teriflunomide, in the NEDA group, 4 patients were treated with platform therapies (interferon-beta or glatiramer acetate) and 3 patients with dimethyl fumarate; in the EDA group, all 8 patients were treated with platform therapies. Before dimethyl fumarate, in the NEDA group, the 5 patients were treated with platform therapies; in the EDA group, 5 patients were treated with platform therapies and 1 patient with teriflunomide. Before fingolimod, in the NEDA group, all 5 patients were treated with platform therapies; in the EDA group, 12 patients were treated with platform therapies, 3 patients with natalizumab, 2 patients with dimethyl fumarate, and 1 patient with siponimod as part of a clinical trial. p-values were obtained after comparisons between NEDA and EDA patients by means of the χ2 test (sex, % naı̈ve patients, and number of T2 lesions), Mann-Whitney U test (EDSS), and Student t test (remaining variables). Significant p-values are shown in bold. 2 Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 Neurology.org/NN Gadolinium-enhancing (Gd+) lesions were counted Table 2 Demographic and Baseline Clinical individually. Characteristics of Patients With MS From the Validation Cohort Determination of Serum Cytokine Levels by Teriflunomide Single Molecule Array Assays Baseline characteristics NEDA EDA p Value Peripheral blood was drawn by standard venipuncture and allowed to clot spontaneously for 30 minutes. Serum was — n 26 10 obtained by centrifugation and stored frozen at −80°C until Age (y) 42.5 (8.7) 42.1 (7.0) 0.5 used. Levels of cytokines were measured in serum samples using commercially available immunoassay kits (Cytokine Female/male (% women) 21/5 (80.8) 5/5 (50.0) 0.07 3-Plex B [IL-6, IL-17A, and TNF-α] cat#101319, IL-10 Duration of disease (y) 10.7 (9.5) 10.0 (8.2) 0.2 cat#101643, IFN-γ cat#103337, GM-CSF cat#102329, Naı̈ve/no naı̈ve (% naı̈ve) e 7/19 (26.9) 4/6 (40.0) 0.4 CXCL13 cat#102635, and IL-1β cat#101605, Quanterix, Billerica, MA) run on the fully automated ultrasensitive Simoa EDSSa 2.1 (1.5–2.5) 2.2 (1.5–2.5) 0.9 HD-1 Analyzer (Quanterix). Samples were run in duplicate in No. of relapsesb 0.7 (0.7) 0.7 (0.8) 0.6 accordance with manufacturers’ instructions with appropriate No. of T2 lesions c,d [n (%)] standards and internal controls. Both the mean intra-assay coefficient of variation of duplicates and the mean interassay 1 5 (19.2) 1 (10.0) 0.8 coefficient of variation were <13% for all assays. 2 16 (61.6) 7 (70.0) Validation Cohort 3 5 (19.2) 2 (20.0) Cytokines associated with the response to treatment were No. of Gd-enhancing lesionsc 0.1 (0.2) 0.1 (0.3) 0.1 measured in an independent cohort of patients with RRMS treated with teriflunomide at the outpatient clinic of the Abbreviations: EDA = evidence of disease activity; EDSS = Expanded Disability Status Scale; NEDA = no evidence of disease activity. Hospital Ramón y Cajal (Madrid). Patients were classified Data are expressed as mean (SD) unless otherwise stated. into responders and nonresponders after 12 months of a Data are expressed as mean (interquartile range). b Refers to the number of relapses in the 2 previous years of blood extraction. treatment according to the same criteria as in the original c Refers to the number of T2 and gadolinium-enhancing lesions at the time of blood extraction. cohort. A total of 36 patients (26 NEDA and 10 EDA) were d T2 lesions were classified into 3 categories according to the number of included in the study. Demographic and baseline clinical lesions: 1 = 1–3 lesions, 2 = 4–9 lesions, 3 = ≥10 lesions. e Before teriflunomide, in the NEDA group, 17 patients were treated with characteristics are summarized in Table 2. Serum levels of platform therapies (interferon-beta or glatiramer acetate) and 2 patients with TNF-α and CXCL13 were measured using the immunoassay dimethyl fumarate; in the EDA group, all 6 patients were treated with platform therapies. p-values were obtained after comparisons between NEDA and EDA kits cat#103337 and cat#102635 (Quanterix), respectively, on patients by means of the χ2 test (sex, % naı̈ve patients, and number of T2 the Simoa HD-1 Analyzer. The intra-assay and interassay lesions), Mann-Whitney U test (EDSS), and Student t test (remaining variables). coefficients of variation were <10% for both assays. were included in the study. Patients were classified into re- Standard Protocol Approvals, Registrations, sponders and nonresponders based on the following criteria after and Patient Consents 1 year of treatment: (1) presence or absence of relapses; (2) This study was approved by the hospital ethics committee progression or lack of progression on neurologic disability; (3) (EPA(AG)57/2013(3834), and patients gave their informed presence or absence of radiologic activity on brain MRI. Re- consent. sponders were patients with no evidence of disease activity (NEDA) defined by the absence of relapses, Expanded Disability Statistical Analysis Status Scale (EDSS) progression, and MRI activity in the first year A Mann-Whitney test was used to test for significant differ- of treatment. Nonresponders were patients with evidence of ences in cytokine levels between NEDA and EDA at baseline disease activity (EDA) defined by the presence during the first and after 1 year of treatment. Comparisons of cytokine levels year of at least 1 relapse, sustained increase of at least 1 point in the before and after treatment with oral therapies in NEDA and EDSS score, and/or presence of active lesions (either new or EDA patients were assessed by a Wilcoxon signed-rank test. enlarging T2 lesions compared with baseline MRI scan or Insomuch as 3 different oral therapies were analyzed in this gadolinium-enhancing lesions). A total of 66 patients were in- study, for the identification of biomarkers associated with the cluded in the study, 19 treated with teriflunomide (8 NEDA and response to treatment, Bonferroni correction was used to 11 EDA), 22 with DMF (13 NEDA and 9 EDA), and 25 with correct the alpha level for multiple comparisons between fingolimod (6 NEDA and 19 EDA). Table 1 summarizes the main cytokine levels and the response outcome (alpha = 0.01). baseline demographic and clinical characteristics of patients. Correlations between cytokine levels and clinical and radio- logic characteristics at baseline were assessed by the Spearman MRI Lesion Analysis rank correlation coefficient. Receiver operating characteristic Total number of T2 lesions was categorized into 3 groups as (ROC) curve analyses and the Youden Index were used to follows: 1 = 1–9 lesions, 2 = 10–50 lesions, 3 = >50 lesions. determine the best cut-off values based on the change of Neurology.org/NN Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 3 Figure 1 Dotplots Showing the Distribution of Serum Cytokine Levels Determined by Simoa in the Original Cohort of Patients With MS at Baseline (T0) and after 12 Months (T12) of Teriflunomide Treatment Each symbol represents an individual, and horizontal bars indicate the median values. p-values were determined using Mann-Whitney U tests (unpaired data) and Wilcoxon signed-rank tests (paired data). For the sake of clarity, only signifi- cant p-values are shown in the graphs. N = 8 for NEDA and N = 11 for EDA. Y-axis segmentations were performed to rep- resent better high and low biomarker levels. CXCL13 = chemokine (C-X-C motif) ligand 13; EDA = evidence of disease activity; GM-CSF = gran- ulocyte-macrophage colony-stimulat- ing factor; IL-1β = interleukin 1 beta; NEDA = no evidence of disease activity; TNF-α = tumor necrosis factor alpha. serum cytokine levels between baseline and 12 months in the in the whole cohort of patients (N = 66) only revealed a original cohort. Sensitivities, specificities, and likelihood ratios significant correlation between serum IL-10 levels and EDSS were calculated for the validation cohort. (rho = −0.350; p = 0.004) (eTable 1, links.lww.com/NXI/ A758). There were no significant correlations between base- Data Availability line serum cytokine levels and radiologic variables such as the The data presented in this study are available on request from number of T2 lesions or the number of contrast-enhancing the corresponding author. lesions (eTable 1). Serum Cytokine Levels and Response to Oral Therapies Results As shown in Figures 1–3 and eFigures 1–3 (links.lww.com/ Patients NXI/A754, links.lww.com/NXI/A755, links.lww.com/NXI/ As shown in Table 1, demographic, clinical, and radiologic A756), serum cytokine levels at baseline were comparable variables were comparable between NEDA and EDA patients between NEDA and EDA patients except for IL-6, whose who will receive DMF or fingolimod treatment (Table 1). For levels were increased in NEDA patients who will receive DMF teriflunomide, a significantly higher proportion of female treatment (p = 0.04; eFigure 2). patients and patients with ≥4 T2 lesions at baseline MRI were observed in EDA compared with NEDA patients (Table 1). To identify biomarkers associated with the response to treat- ment, we focused on significant changes induced by treatment in Serum Cytokine Levels and Clinical and serum cytokine levels between the baseline and the 12-month- Radiologic Variables at Baseline treated time point in NEDA and EDA patients. As shown in Correlations between serum cytokine levels and clinical var- Figure 1, treatment with teriflunomide was associated with sig- iables such as disease duration, number of relapses in the nificant reductions in serum CXCL13 and TNF-α levels in previous 2 years before treatment onset, and EDSS at baseline NEDA patients (p = 0.008 for both cytokines) whereas cytokine 4 Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 Neurology.org/NN Figure 2 Dotplots Showing the Distribution of Serum Levels Measured by Simoa in the Original Cohort of Patients With MS at Baseline (T0) and After 12 Months (T12) of DMF Treatment Each symbol represents an individual, and horizontal bars indicate the me- dian values. p-values were determined using Mann-Whitney U tests (unpaired data) and Wilcoxon signed-rank tests (paired data). For the sake of clarity, only significant p-values are shown in the graphs. N = 13 for NEDA and N = 9 for EDA. Y-axis segmentations were performed to represent better high and low levels. CXCL13 = chemokine (C-X-C motif) ligand 13; EDA = evidence of disease activity; GM-CSF = gran- ulocyte-macrophage colony-stimulat- ing factor; IL-1β = interleukin 1 beta; NEDA = no evidence of disease activity; TNF-α = tumor necrosis factor alpha. levels remained comparable in EDA patients despite treatment Validation of Serum TNF-α and CXCL13 Levels (p = 0.21 and p = 0.07 for CXCL13 and TNF-α, respectively). as Biomarkers Associated With the Response A trend toward increased GM-CSF levels was also observed in to Teriflunomide NEDA patients treated with teriflunomide (p = 0.04 in NEDA vs Based on the results in the original cohort, CXCL13 and p = 0.85 in EDA patients; Figure 1). TNF-α were selected for validation in an independent cohort of patients with RRMS treated with teriflunomide and clas- As shown in Figure 2, in patients receiving DMF, a trend sified into NEDA and EDA after 1 year of treatment according toward higher IL-1β levels was observed in NEDA patients to the same criteria as in the original cohort. As shown in after 12 months of treatment compared with EDA patients Figure 4, both CXCL13 and TNF-α levels were significantly (p = 0.01 in NEDA patients vs p = 0.50 in EDA patients). decreased after 12 months of treatment with teriflunomide in Serum levels of IL-6 and TNF-α were elevated in NEDA NEDA patients whereas no significant changes in cytokine patients treated with DMF at baseline, and the differences levels were observed in the EDA group (p < 0.0001 in NEDA vs were significant after 12 months of treatment compared with p = 0.13 in EDA for CXCL13; p = 0.003 in NEDA vs p = 0.28 in EDA patients (p = 0.003 for IL-6 and p = 0.02 for TNF-α; EDA for TNF-α). eFigure 2, links.lww.com/NXI/A755 and Figure 2). How- ever, IL-6 and TNF-α levels were not modified by the effect We finally analyzed the performance of changes in serum of DMF treatment insomuch as cytokine levels remained levels of CXCL13 and TNF-α between the baseline and the comparable between the baseline and the treated time point 12-month-treated time point to predict response to teri- (eFigure 2 and Figure 2). flunomide. For CXCL13, the area under the ROC curve (AUC) was 77% (95% CI = 54.4%–100%) in the original Treatment with fingolimod was not associated with significant cohort and 73% (95% CI = 54.1%–91.3%) in the validation changes in serum cytokines levels between the baseline and cohort (Figure 5A). A change in CXCL13 levels of 9.64 pg/ the treated time point either in NEDA or EDA patients mL between the baseline and 12 months was identified as the (Figure 3 and eFigure 3, links.lww.com/NXI/A756). best cutoff to classify NEDA and EDA patients in the original Neurology.org/NN Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 5 Figure 3 Dotplots Showing the Distribution of Serum Levels Determined by Simoa in the Original Cohort of Patients With MS at Baseline (T0) and After 12 Months (T12) of Fingolimod Treatment Each symbol represents an individual, and horizontal bars indicate the me- dian values. N = 6 for NEDA and N = 19 for EDA. Y-axis segmentations were performed to represent better high and low levels. CXCL13 = chemokine (C-X-C motif) ligand 13; EDA = evidence of disease activity; GM-CSF = gran- ulocyte-macrophage colony-stimulat- ing factor; IL-1β = interleukin 1 beta; NEDA = no evidence of disease activity; TNF-α: tumor necrosis factor alpha. cohort, with a sensitivity and specificity of 75% and 82%, re- for TNF-α, and the AUC were 66% (95% CI = 39.7%–92.1%) spectively (Figure 5B). When this cutoff was applied to the in the original cohort and 53% (95% CI = 29.0%–77.1%) in the validation cohort, a sensitivity of 65% and specificity of 60% validation cohort (eFigure 4A, links.lww.com/NXI/A757). A were achieved (Figure 5B). A similar analysis was performed change in TNF-α levels of 0.68 pg/mL between baseline and 12 Figure 4 Dotplots Showing the Distribution of Serum Levels Measured by Simoa in the Validation Cohort at Baseline (T0) and After 12 Months (T12) of Teriflunomide Treatment Changes in serum levels of CXCL13 and TNF-α after treatment with teri- flunomide were validated in an in- dependent cohort of patients with MS. p-values were determined using Wilcoxon signed-rank tests (paired data). N = 26 for NEDA and N = 10 for EDA. Y-axis segmentations were per- formed to represent better high and low levels. CXCL13 = chemokine (C-X- C motif) ligand 13; EDA = evidence of disease activity; NEDA = no evidence of disease activity; TNF-α = tumor necrosis factor alpha. 6 Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 Neurology.org/NN Figure 5 Performance of Changes in Serum CXCL13 Levels to Discriminate Between NEDA and EDA Patients (A) Area under the ROC curves with 95% CIs in parentheses for CXCL13. (B) Box- plots showing the classification of NEDA and EDA patients according to a cut-off value of 9.64 pg/mL (dashed line) based on the change in CXCL13 levels between the untreated and treated time points. This cutoff was calculated for the origi- nal cohort and applied to the validation cohort. Sensitivity and specificity values according to this cutoff are shown. In- dividual changes in CXCL13 levels are represented as dots. AUC = area under the curve; CXCL13 = chemokine (C-X-C motif) ligand 13; EDA = evidence of disease activity; NEDA = no evidence of disease activity; Se = sensitivity; Sp = specificity. months resulted in the best cutoff to discriminate between cohort of treated patients. In this study, we report that serum NEDA and EDA patients in the original cohort, with a sensi- CXCL13 and TNF-α levels are decreased after treatment with tivity of 75% and specificity of 63.6% (eFigure 4B). This cutoff teriflunomide in patients classified as treatment responders was associated with a sensitivity of 38.5% and specificity of 50% according to the composite outcome measure NEDA. These in the validation cohort (eFigure 4B). findings were confirmed in an independent cohort of patients with MS treated with teriflunomide. Furthermore, the change in serum CXCL13 levels between the baseline and 12 months showed good performance to discriminate between teri- Discussion flunomide responders and nonresponders. An important gap in the field of biomarkers in MS is the lack of reliable biomarkers that discriminate between patients who Current disease-modifying therapies achieve their effects pri- respond to therapies and patients who continue to have re- marily by blocking the proinflammatory response in a non- lapses, radiologic inflammatory activity, and disease pro- specific manner, inducing changes in the levels of several gression despite treatment. NEDA is a concept that depicts inflammatory molecules.9 Among them, cytokines from the in- the absence of relapses, new MRI T2 lesions, and EDSS nate immune system and T-cell–related and B-cell–related cy- progression during follow-up. Although several criteria have tokines appear as attractive candidate biomarkers associated with been suggested as supportive features to this concept treatment response. In this work, the serum levels of IL-6, IL-17, throughout the past years, such as cognitive decline or brain TNF-α, GM-CSF, IL-10, IFN-γ, IL-1β, and CXCL13 were volume loss, for this study, we decided to use the original 3- measured by Simoa at baseline and after 12 months of treatment domain concept for NEDA because it is the most represen- with the above-mentioned oral therapies, to investigate their tative of daily clinical practice.8 Bearing this in mind, we aimed potential association with the response outcomes. to identify treatment response biomarkers in a cohort of pa- tients with MS receiving the oral therapies teriflunomide, Interestingly, analysis of serum cytokine levels in the whole un- DMF, and fingolimod categorized based on their response to treated cohort revealed a negative correlation between the anti- therapies and further validate the data in an independent inflammatory cytokine IL-10 and EDSS. This finding is in Neurology.org/NN Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 7 agreement with previous reports indicating a negative relationship mechanisms such as an inhibitory effect on cytokines/chemokines between IL-10 levels and EDSS scores.10 In this regard, B cells of production.23 In this context, previous studies have shown a lower patients with MS were found to exhibit deficient production of secretion of TNF-α by T cells after teriflunomide treatment.24 IL-10 compared with B cells of healthy controls.11,12 B cells are major contributors to MS pathogenesis, a role con- At baseline, serum cytokine levels were comparable between firmed with the introduction of B-cell–depleting therapies.25 responders and nonresponders to the 3 oral therapies except Although not specific for B cells, teriflunomide treatment has for IL-6, whose levels were increased in patients who will been shown to reduce the number of effector blood B-cell respond to DMF. Taking into account that those levels were populations.26,27 Considering the important role of CXCL13 in not modified by the effect of DMF treatment, and remained the recruitment of B cells into the CNS, the decrease in serum significantly higher in treatment responders after 1 year, IL-6 CXCL13 levels in teriflunomide responders could be one of the was not considered as a candidate biomarker for validation in mechanisms of action by which teriflunomide exerts its bene- an independent cohort of DMF-treated patients. ficial effects. In this context, other MS therapies such as natali- zumab could be acting by a similar mechanism.13 Comparisons of serum cytokine levels before and after 1 year of treatment revealed that CXCL13 and TNF-α levels were signifi- In conclusion, our approach in serum samples from patients cantly reduced by the effect of treatment in teriflunomide re- treated with different oral therapies and classified into responders sponders and hence were selected for validation. Noteworthy, in the and nonresponders according to NEDA has led to the identifi- validation cohort, serum CXCL13 and TNF-α levels were again cation of CXCL13 and TNF-α as biomarkers associated with the significantly reduced by treatment in teriflunomide responders, response to teriflunomide. In addition, a change in CXCL13 levels while cytokine levels remained unchanged in nonresponders. When above 9.46 pg/mL during the first year of treatment showed good we evaluated the performance of the changes in the levels of both performance to identify teriflunomide treatment responders. Our CXCL13 and TNF-α to predict treatment response to teri- results support the use of longitudinal measurements of serum flunomide, only CXCL13 showed good potential to discriminate CXCL13 levels as a biomarker to monitor treatment response in between responders and nonresponders in both cohorts of patients patients with RRMS receiving teriflunomide. with MS, indicating that a decrease in CXCL13 levels higher than 9.64 pg/mL by the effect of treatment could be used in clinical Acknowledgment practice to identify patients with RRMS having optimal responses This study was funded by Roche. The authors express deep to teriflunomide. By contrast, performance of TNF-α to predict thankfulness to patients for their participation and support teriflunomide treatment response was acceptable in the original and study nurses for their motivated collaboration and cohort but low in the validation cohort, findings that certainly limit recruitment efforts. its use in clinical practice to identify teriflunomide responders. Study Funding CXCL13 is a potent B-cell chemoattractant that plays a role in This study was funded by Roche. Roche had no role in either B-cell recruitment to the CNS during inflammation. Studies of data analysis or drafting of the manuscript. CXCL13 levels showed that concentrations are increased in serum and CSF of patients with active disease, emphasizing that this Disclosure chemokine might be an important biomarker for MS inflammatory The authors have nothing to disclose regarding this manu- activity.6,13,14 Interestingly, CSF CXCL13 levels were found to script. Go to Neurology.org/NN for full disclosures. correlate with intrathecal immunoglobulin production and the number of CSF B cells and plasmablasts.15,16 On the other hand, Publication History TNF-α is a potent cytokine that exerts numerous functions in Received by Neurology: Neuroimmunology & Neuroinflammation immunity, inflammation, and control of cell proliferation, among June 29, 2022. Accepted in final form September 7, 2022. Submitted and externally peer reviewed. The handling editor was Scott S. Zamvil, others.17 This cytokine has long been associated with MS patho- MD, PhD, FAAN. genesis,18 and CSF and serum levels of TNF-α were found to be increased in patients with MS compared with controls and levels correlated with disease severity.19 Teriflunomide is an oral disease- modifying therapy indicated for the treatment of patients with RRMS, which has shown beneficial effects on both clinical and Appendix Authors MRI outcomes in patients with MS.20,21 The main mechanism of Name Location Contribution action relies on the inhibition of enzyme dihydroorotate dehy- Nicolás Servei de Neurologia, Centre Design and drogenase (DHODH) that catalyzes de novo biosynthesis of py- Fissolo, PhD d’Esclerosi Múltiple de conceptualized study; rimidines, exerting a cytostatic effect on proliferating T and Catalunya (Cemcat), acquisition and analysis Institut de Recerca Vall of the data; drafted the B lymphocytes in the periphery and thus inducing both anti- d’Hebron (VHIR), Hospital manuscript for proliferative and anti-inflammatory properties.22 Despite this Universitari Vall d’Hebron; intellectual content Universitat Autònoma de well-documented mechanism of action, it has been reported that Barcelona, Spain teriflunomide may have additional DHODH-independent 8 Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 Neurology.org/NN References Appendix (continued) 1. Biomarkers Definitions Working Group. Biomarkers and surrogate endpoints: pre- ferred definitions and conceptual framework. Clin Pharmacol Ther. 2001;69(3):89-95. Name Location Contribution 2. Bermel RA, You X, Foulds P, et al. Predictors of long-term outcome in multiple sclerosis patients treated with interferon beta. Ann Neurol. 2013;73(1):95-103. Agustı́n Servei de Neurologia, Centre Acquisition and analysis 3. Comabella M, Montalban X. Body fluid biomarkers in multiple sclerosis. Lancet Pappolla, MD d’Esclerosi Múltiple de of the data, and revised Neurol. 2014;13(1):113-126. Catalunya (Cemcat), Institut de the manuscript for 4. Ziemssen T, Akgün K, Brück W. Molecular biomarkers in multiple sclerosis. Recerca Vall d’Hebron (VHIR), intellectual content J Neuroinflammation. 2019;16(1):272. Hospital Universitari Vall 5. Restorick SM, Durant L, Kalra S, et al. CCR6+ Th cells in the cerebrospinal fluid of d’Hebron; Universitat persons with multiple sclerosis are dominated by pathogenic non-classic Th1 cells and Autònoma de Barcelona, Spain GM-CSF-only-secreting Th cells. Brain Behav Immun. 2017;64:71-79. 6. Khademi M, Kockum I, Andersson ML, et al. Cerebrospinal fluid CXCL13 in multiple Jordi Rı́o, MD, Servei de Neurologia, Centre Analysis of the data, and sclerosis: a suggestive prognostic marker for the disease course. Mult Scler. 2011;17(3):335-343. PhD d’Esclerosi Múltiple de revised the manuscript 7. Fernández M, Montalban X, Comabella M. Orchestrating innate immune responses Catalunya (Cemcat), Institut de for intellectual content in multiple sclerosis: molecular players. J Neuroimmunology. 2010;225(1-2):5-12. Recerca Vall d’Hebron (VHIR), 8. Giovannoni G, Tomic D, Bright JR, Havrdová E. No evident disease activity": the use Hospital Universitari Vall of combined assessments in the management of patients with multiple sclerosis. Mult d’Hebron; Universitat Scler. 2017;23(9):1179-1187. Autònoma de Barcelona, Spain 9. Amedei A, Prisco D, D’Elios MM. Multiple sclerosis: the role of cytokines in path- ogenesis and in therapies. Int J Mol Sci. 2012;13(12):13438-13460. Luisa M. Departments of Immunology Acquisition of the data, 10. Farsani ZS, Behmanesh M, Sahraian MA. Interleukin-10 but not transforming growth Villar, MD, and Neurology, Multiple and revised the factor-β1 gene expression is up-regulated by vitamin D treatment in multiple sclerosis PhD Sclerosis Unit, Hospital Ramon manuscript for patients. J Neurol Sci. 2015;350(1-2):18-23. y Cajal, (IRYCIS), Madrid, Spain intellectual content 11. Correale J, Farez M, Razzitte G. Helminth infections associated with multiple sclerosis induce regulatory B cells. Ann Neurol. 2008;64(2):187-199. Santiago Statistics and Bioinformatics Acquisition of the data, 12. Duddy M, Niino M, Adatia F, et al. Distinct effector cytokine profiles of memory and Perez-Hoyos, Unit, Vall d’Hebron Institut de and revised the naive human B cell subsets and implication in multiple sclerosis. J Immunol. 2007; BSc, PhD Recerca (VHIR), Barcelona, manuscript for 178(10):6092-6099. Spain intellectual content 13. Sellebjerg F, Börnsen L, Khademi M, et al. Increased cerebrospinal fluid concentra- tions of the chemokine CXCL13 in active MS. Neurology. 2009;73(23):2003-2010. Alex Statistics and Bioinformatics Acquisition of the data, 14. Festa ED, Hankiewicz K, Kim S, et al. Serum levels of CXCL13 are elevated in active Sanchez, PhD Unit, Vall d’Hebron Institut de and revised the multiple sclerosis. Mult Scler. 2009;15(11):1271-1279. Recerca (VHIR), Barcelona, manuscript for 15. Kowarik MC, Cepok S, Sellner J, et al. CXCL13 is the major determinant for B cell Spain Genetics, Microbiology intellectual content recruitment to the CSF during neuroinflammation. J Neuroinflammation. 2012;9(1):93. and Statistics Department, 16. Krumbholz M, Theil D, Cepok S, et al. Chemokines in multiple sclerosis: CXCL12 Universitat de Barcelona, Spain and CXCL13 up-regulation is differentially linked to CNS immune cell recruitment. Brain. 2006;129(1):200-211. Lucı́a Servei de Neurologia, Centre Acquisition of the data, 17. Wallach D, Varfolomeev EE, Malinin NL, Goltsev YV, Kovalenko AV, Boldin MP. Gutierrez, d’Esclerosi Múltiple de and revised the Tumor necrosis factor receptor and Fas signaling mechanisms. Annu Rev Immunol. BSc Catalunya (Cemcat), Institut de manuscript for 1999;17(1):331-367. Recerca Vall d’Hebron (VHIR), intellectual content 18. Caminero A, Comabella M, Montalban X. Tumor necrosis factor alpha (TNF-α), anti- Hospital Universitari Vall TNF-α and demyelination revisited: an ongoing story. J Neuroimmunology. 2011; d’Hebron; Universitat 234(1-2):1-6. Autònoma de Barcelona, Spain 19. Sharief MK, Hentges R. Association between tumor necrosis factor-alpha and disease progression in patients with multiple sclerosis. N Engl J Med. 1991;325(7):467-472. Xavier Servei de Neurologia, Centre Acquisition of the data, 20. Confavreux C, O’Connor P, Comi G, et al. Oral teriflunomide for patients with Montalban, d’Esclerosi Múltiple de and revised the relapsing multiple sclerosis (TOWER): a randomised, double-blind, placebo- MD, PhD Catalunya (Cemcat), Institut de manuscript for controlled, phase 3 trial. Lancet Neurol. 2014;13:247-256. Recerca Vall d’Hebron (VHIR), intellectual content 21. O’Connor P, Wolinsky JS, Confavreux C, et al. Randomized trial of oral teriflunomide Hospital Universitari Vall for relapsing multiple sclerosis. N Engl J Med. 2011;365(14):1293-1303. d’Hebron; Universitat 22. Bar-Or A, Pachner A, Menguy-Vacheron F, Kaplan J, Wiendl H. Teriflunomide and its Autònoma de mechanism of action in multiple sclerosis. Drugs. 2014;74(6):659-674. Barcelona, Spain 23. Li L, Liu J, Delohery T, Zhang D, Arendt C, Jones C. The effects of teriflunomide on lymphocyte subpopulations in human peripheral blood mononuclear cells in vitro. Manuel Servei de Neurologia, Centre Design and J Neuroimmunology. 2013;265(1-2):82-90. Comabella, d’Esclerosi Múltiple de conceptualized study; 24. Tilly G, Cadoux M, Garcia A, et al. Teriflunomide treatment of multiple sclerosis MD, PhD Catalunya (Cemcat), Institut de analyzed the data; selectively modulates CD8 memory T cells. Front Immunol. 2021;12:730342. Recerca Vall d’Hebron (VHIR), drafted the manuscript 25. Sabatino JJ, Zamvil SS, Hauser SL. B-cell therapies in multiple sclerosis. Cold Spring Hospital Universitari Vall for intellectual content Harb Perspect Med. 2019;9(2):a032037. d’Hebron; Universitat 26. Gandoglia I, Ivaldi F, Laroni A, et al. Teriflunomide treatment reduces B cells in Autònoma de patients with MS. Neurol Neuroimmunol Neuroinflamm. 2017;4(6):e403. Barcelona, Spain 27. Medina S, Sainz de la Maza S, Villarrubia N, et al. Teriflunomide induces a tolerogenic bias in blood immune cells of MS patients. Ann Clin Transl Neurol. 2019;6(2):355-363. Neurology.org/NN Neurology: Neuroimmunology & Neuroinflammation | Volume 10, Number 1 | January 2023 9