ARTICLE OPEN ACCESS CSF Chitinase 3–Like 2 Is Associated With Long-term Disability Progression in Patients With Progressive Multiple Sclerosis Manuel Comabella, MD, PhD, Jaume Sastre-Garriga, MD, PhD, Eva Borras, PhD, Luisa M. Villar, MD, PhD, Correspondence Dr. Comabella Albert Saiz, MD, PhD, Sergio Martı́nez-Yélamos, MD, PhD, Juan Antonio Garcı́a-Merino, MD, PhD, manuel.comabella@vhir.org Rucsanda Pinteac, MSc, Nicolas Fissolo, PhD, Antonio J. Sánchez López, PhD, Lucienne Costa-Frossard, MD, Yolanda Blanco, MD, Sara Llufriu, MD, PhD, Angela Vidal-Jordana, MD, PhD, Eduard Sabidó, PhD, and Xavier Montalban, MD, PhD Neurol Neuroimmunol Neuroinflamm 2021;8:e1082. doi:10.1212/NXI.0000000000001082 Abstract MORE ONLINE Objective Class of Evidence This study aimed to identify long-term prognostic protein biomarkers associated with disease Criteria for rating progression in patients with progressive multiple sclerosis (MS). therapeutic and diagnostic studies Methods NPub.org/coe CSF samples were collected from a discovery cohort of 28 patients with progressive MS who participated in a clinical trial with interferon beta. Patients were classified into high and low disability progression phenotypes according to numeric progression rates (NPR) and step- based progression rates (SPR) after a mean follow-up time of 12 years. Protein abundance was measured by shotgun proteomics. Selected proteins from the discovery cohort were quantified by parallel reaction monitoring in CSF samples from an independent validation cohort of 41 patients with progressive MS classified also into high and low disability progression phenotypes after a mean follow-up time of 7 years. Results Of 2,548 CSF proteins identified in the discovery cohort, 10 were selected for validation based on their association with long-term disability progression: SPATS2-like protein, chitinase 3–like 2 (CHI3L2), plasma serine protease inhibitor, metallothionein-3, phospholipase D4, beta-hexosaminidase, neurexophilin-1, adipocyte enhancer-binding protein 1, cathepsin L1, and lipopolysaccharide-binding protein. Only CHI3L2 was validated, and patients with high disability progression exhibited significantly higher CSF protein levels compared with patients with low disability progression (p = 0.03 for NPR and p = 0.02 for SPR). CHI3L2 levels showed good performance to discriminate between high and low disability progression in patients with progressive MS (area under the curve 0.73; sensitivity 90% and specificity 63%). Conclusions Although further confirmatory studies are needed, we propose CSF CHI3L2 as a prognostic protein biomarker associated with long-term disability progression in patients with progressive MS. From the Unitat de Neuroimmunologia Clı́nica (M.C., J.S.-G., R.P., N.F., A.V.-J., X.M.), Hospital Universitari Vall d´Hebron; Eva Borràs (E.B., E.S.), Proteomics Unit, Universitat Pompeu Fabra, Barcelona; Departments of Neurology and Immunology (L.M.V., L.C.-F.), Hospital Universitario Ramón y Cajal, Instituto Ramón y Cajal de Investigacion Sanitaria, Madrid; Service of Neurology (A.S., Y.B., S.L.), Hospital Clinic and Institut d’Investigacions Biomèdiques August Pi Sunyer, University of Barcelona; Department of Neurology (S.M.-Y.), Bellvitge University Hospital, Barcelona; and Neuroimmunology Unit (J.A.G.M., A.J.S.L.), Hospital Universitario Puerta de Hierro, Madrid, Spain. Go to Neurology.org/NN for full disclosures. Funding information is provided at the end of the article. The Article Processing Charge was funded by the authors. The authors report no conflict of interest. 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 © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Glossary AEBP1 = adipocyte enhancer-binding protein 1; AGC = auto gain control; CHI3L2 = chitinase 3–like 2; CIS = clinically isolated syndrome; CTSL = cathepsin L1; EDSS = Expanded Disability Status Scale; HEXB = hexosaminidase; IFNβ = interferon beta-1b; MS = multiple sclerosis; MT3 = metallothionein-3; NFL = neurofilament light chain; NPR = numeric progression rates; NXPH1 = neurexophilin-1; PLD4 = phospholipase D4; PPMS = progressive MS; PRM = parallel reaction monitoring; ROC = receiver operating characteristic; RRMS = relapsing-remitting course; SERPINA5 = plasma serine protease inhibitor; SPATS2L = SPATS2-like protein; SPR = step progression rates; UPLC = ultra-performance liquid chromatography. Classification of Evidence This study provides Class II evidence that high CSF CHI3L2 levels identified higher disability progression in patients with progressive MS. Multiple sclerosis (MS) is a neurodegenerative immune- Thirteen patients (46.4%) received IFNβ during the 2-year mediated disorder of the CNS and an important cause of trial. Except for 1 patient treated with IFNβ during the trial disability among young adults.1 Although the etiology of MS who received mitoxantrone for 1 year after trial completion is unknown, it is assumed that both a complex genetic back- and then was treated with IFNβ for 11 additional years, the ground and environmental triggers contribute to disease remaining 27 patients included in this study were un- manifestation.2-6 In approximately 80%–85% of patients, MS treated during the whole follow-up period. Progression starts with acute episodes of neurologic dysfunction and fol- rate variables were created by dividing Expanded Disability lows a relapsing-remitting course (RRMS).1 However, over Status Scale (EDSS) changes, confirmed after 6 months, by time, a significant number of these patients will enter a sec- the time on follow-up. Because many patients were in the ondary progressive phase of the disease which is characterized range of EDSS where a change in 0.5 points corresponds to by maintained neurologic deterioration with or without fur- 1 EDSS step increase, 2 different variables were computed ther relapses.7 The remaining 15%–20% of patients with MS using these 2 different approaches: numeric progression follow a progressive course from disease onset, and they are rates (NPR) and step-based progression rates (SPR). referred to as primary progressive MS (PPMS).8 Whereas Progression rates were computed between the time of trial currently available therapies in MS are highly effective to onset and the last visit, and between the time after trial suppress the predominantly inflammatory component ob- completion and the last visit. High disability progression served in patients with RRMS, these therapies have proven to was defined by the presence of NPR or SPR above the 75th be ineffective in patients with progressive forms of MS in percentile of disability progression. Patients with pro- whom a neurodegenerative component dominates.9,10 In this gression rates below the 75th percentile were defined as setting, a more in-depth understanding of the prognostic low disability progression. Demographic and clinical factors associated with disability progression will be critical to characteristics of patients with high and low disability stratify patients with progressive MS according to disease progression are summarized in eTable, links.lww.com/ evolution and ultimately may set the rationale for the design NXI/A581. of more specific therapeutic approaches to prevent disease progression. In this study, we aimed to identify long-term Chromatographic and Mass Spectrometric prognostic biomarkers predictive of heightened disability Analysis in the Discovery Cohort progression in patients with progressive MS. CSF samples were collected by lumbar puncture during the trial both in placebo and IFNβ-treated patients (mean time [range] from trial onset: 13 months [8.8–17.9]). CSF samples were centrifuged to remove cells, and the remaining Methods volume was aliquoted and conserved at −80°C until used. Discovery Cohort For each sample, 120 μL of CSF were used for protein Twenty-eight patients with progressive MS (17 patients with precipitation in cold acetone overnight at 4°C. Pellets were primary progressive forms [PPMS]11 and 11 patients with then solubilized in 6 M urea in 200 mM ammonium bi- transitional progressive forms8) who participated in a 2-year carbonate, reduced with 100 mM dithiothreitol, and alky- phase II randomized, double-blind, placebo-controlled trial of lated with 200 mM iodoacetamide. Proteins were then interferon beta (IFNβ)-1b12 were included in this study. Se- digested with LysC 1:10 ratio (w:w; enzyme:substrate) at lection of patients was performed based on the availability of 37°C overnight followed by trypsin 1:10 ratio (w:w; enzyme: CSF samples during the trial, and as shown in Table 1, this substrate) at 37°C for 8 hours. Tryptic peptide mixtures were cohort was representative of the entire cohort that partici- desalted using a C18 UltraMicroSpin column (The Nest pated in the trial for demographics and clinical characteristics. Group). 2 Neurology: Neuroimmunology & Neuroinflammation | Volume 8, Number 6 | November 2021 Neurology.org/NN Peptide mixtures were analyzed in an Orbitrap Fusion Lumos mass tolerance was set to 0.5 Da, methionine oxidation was with an EASY-Spray source coupled to a nano-ultra-performance set as variable modification, and cysteine carbamidomethyla- liquid chromatography (UPLC) system (EASY-nanoLC 1000 tion was set as fixed modification. The identified peptides liquid chromatography) equipped with a reverse-phase chro- were filtered by 5% false discovery rate. Peptide areas were matography 25-cm column with an inner diameter of 75 μm, obtained using the “precursor ions area detector” module in packed with 1.9 μm C18 particles (Nikkyo Technos, Japan). the Proteome Discoverer software suite (v2.0, Thermo Fisher Chromatographic gradients started at 7% buffer B with a flow Scientific). Protein abundance in each condition was esti- rate of 250 nL/minute and gradually increased to 35% in 120 mated using the log-transformed average of the 3 most intense minutes. After each run, the column was washed for 15 minutes peptides per protein group. with 90% buffer B (buffer A: 0.1% formic acid in water and buffer B: 0.1% formic acid in acetonitrile). The mass spectrometer was Selection of Candidate Proteins Associated operated in data-dependent acquisition mode, with full MS scans With Disability Progression for Validation over a mass range of m/z 350–1,500 with detection in the Selection of proteins for validation in an independent cohort of Orbitrap (120K resolution) and with auto gain control (AGC) patients with progressive MS was performed based on the dif- set to 100,000. In each cycle of data-dependent acquisition ferences in CSF protein levels between patients with high and low analysis, after each survey scan, the most intense ions above a disability progression according to NPR and SPR at trial onset and threshold ion count of 10,000 were selected for fragmentation after trial completion, and correlations between CSF protein levels with HCD at normalized collision energy of 28%. The number and numeric and step progression rates at trial onset and after trial of selected precursor ions for fragmentation was determined by completion. Those proteins exhibiting a statistically significant the “top speed” acquisition algorithm (maximum cycle time of 3 abundance change (p value <0.05) or significant correlation (p seconds), and a dynamic exclusion of 60 seconds was set. value <0.05) in most (n ≥ 5) of the disability progression criteria Fragment ion spectra were acquired in the ion trap with an AGC reported in Table 3 were selected for further validation. of 4,000 and a maximum injection time of 300 ms. Validation Cohort Acquired data were analyzed using the Proteome Discoverer An independent cohort of 41 patients with PPMS was in- software suite (v2.0, Thermo Fisher Scientific), and peptides cluded in this study to validate selected proteins from the were identified using the Mascot search engine (v1.6.0.16, targeted quantitative proteomics approach. Patients were Matrix Science). Data were searched against a human protein recruited from 4 MS centers (Barcelona—Hospital de Bell- database derived from the SwissProt database (as in April vitge [n = 7], Barcelona—Hospital Clinic [n = 11], 2015) plus the most common contaminants. The precursor Madrid—Puerta de Hierro [n = 6], and Madrid—Ramón y ion mass tolerance was 7 ppm at the MS1 level, and up to 3 Cajal [n = 17]), and CSF sampling was performed as de- missed cleavages for trypsin were allowed. The fragment ion scribed above. NPR and SPR were calculated between the Table 1 Demographic and Clinical Characteristics of Patients With Progressive MS Belonging to the Discovery Cohort Characteristics Discovery cohort Whole cohort N 28 72 a Age (y) 49.6 (9.4) 49.3 (8.2) Female/male (% women) 12/16 (42.9) 37/35 (50.7) PP/PT (% PP) 17/11 (61.0) 48/24 (66.0) b Follow-up time onset/off (y) 12.5 (4.6)/10.9 (4.0) 12.1 (4.6)/10.3 (4.4) EDSS onset/lastc 6.0 (4.0–6.0)/8.0 (6.5–8.5) 6.0 (4.0–6.0)/8.0 (6.5–8.8) Numeric progression rate onset/offd 0.3 (0.7)/0.2 (0.1) 0.3 (0.5)/0.2 (0.3) d Step progression rate onset/off 0.6 (1.3)/0.3 (1.3) 0.5 (1.0)/0.4 (0.4) e Patients treated with IFNβ (%) 13 (46.4) 35 (47.9) Abbreviations: EDSS = Expanded Disability Status Scale, IFNβ = interferon beta-1b. All data are expressed as mean SD unless otherwise stated. EDSS is expressed as median (interquartile range). PP/PT refers to the number of patients with primary progressive and transitional progressive MS. a Refers to age at the time of trial onset. b Refers to time from trial onset to the last visit (onset) and from the time after trial completion to the last visit (off). c Refers to EDSS at trial onset (onset) and at the time of last visit (last). d Refers to numeric and step-based progression rates from trial onset to the last visit (onset) and from the time after trial completion to the last visit (off). e Refers to patients under IFNβ treatment at the time of CSF collection. Whole cohort refers to all the patients who participated in the double-blind, placebo- controlled trial of interferon beta. Neurology.org/NN Neurology: Neuroimmunology & Neuroinflammation | Volume 8, Number 6 | November 2021 3 time of CSF collection performed at each recruiting center for over a mass range of m/z 300–2,000, with detection in the diagnostic purposes and the time of last visit. As in the discovery Orbitrap mass analyzer at a 30K resolution. MS2 fragmenta- cohort, patients in the validation cohort were classified according tion was performed using HCD fragmentation at 30 nor- to the 75th percentiles of confirmed disability progression based malized collision energy; the AGC was set at 1E5 and the on NPR and SPR. Patients were untreated for the whole follow- maximum injection time at 54 ms. The size of the scheduled up period. A summary of demographics and clinical character- window was 6 minutes, and the maximum cycle time was 1 istics of the validation cohort is given in Table 2. seconds. All data were acquired using XCalibur software. Standard Protocol Approvals, Registrations, Product ion chromatographic traces corresponding to the and Patient Consents targeted precursor peptides were evaluated using Skyline Written informed consent was obtained from each partici- software v.20.1 based on (1) co-elution of endogenous and pant. This study was approved by the corresponding local internal standard peptides, (2) the number of detected traces, Ethics Committees, and written informed consent was (3) correlation of the trace relative intensities between en- obtained from each participant. dogenous and internal standard peptides, and (4) expected retention time. Normalization was performed based on the Quantification of CSF Levels of Selected median abundance of the internal standard peptides across all Proteins in the Validation Cohort runs. Protein abundance estimates were performed using the Levels of selected proteins were measured after a targeted software package MSstats 3.8.2. proteomics approach with parallel reaction monitoring (PRM). For each patient, 80 μL of CSF were used for protein pre- The mass spectrometry proteomics data have been deposited to cipitation in cold acetone overnight at 4°C. Pellets were then the ProteomeXchange Consortium (proteomecentral.proteo- solubilized in 6 M urea in 200 mM ammonium bicarbonate, mexchange.org) through the Proteomics Identifications Database reduced with 100 mM dithiothreitol, alkylated with 200 mM partner repository13 with the data set identifier PXD022958. iodoacetamide, and digested with LysC 1:10 ratio (w:w; en- zyme:substrate) at 37°C overnight followed by trypsin 1:10 Quantification of Neurofilament Light Chain ratio (w:w; enzyme:substrate) at 37°C for 8 hours. Tryptic Levels in CSF by Single-Molecule Array peptide mixtures were desalted using a C18 UltraMicroSpin Levels of neurofilament light chain (NFL) were measured in column (The Nest Group, Inc). For each targeted peptide, an CSF samples from 36 patients with PPMS belonging to the isotopically labeled counterpart (13C6,15N2-Lys and 13C6,15N4- validation cohort using commercially available NFL immu- Arg, Pepotec Peptides; Thermo Fisher Scientific) was spiked in noassay kits (Quanterix, Billerica, MA, cat no. 103186) and the peptide mixtures and used as internal standard for quan- run on a single-molecule array (Simoa) HD-1 Analyzer tification purposes. Up to 2 unique peptides per protein were (Quanterix). Samples were run in duplicate in line with selected for targeted protein quantification. For each selected manufacturers’ instructions using appropriate standards and peptide, an isotopically labeled peptide (13C6,15N4-Arg and 13 C6,15N2-Lys) was spiked in the samples and used as internal standard for quantification by PRM. The amount of internal Table 2 Demographic and Clinical Characteristics of standard peptide to be spiked in each sample was evaluated Patients With Progressive MS Belonging to the using dilution curves, and the final concentration was chosen Validation Cohort based on the following criteria: (1) to be within the concen- tration range in which a linear response of the peptide was Characteristics Validation cohort observed and (2) to have an area as close to the endogenous N 41 peptide area as possible. Age (y)a 51.9 (9.1) Digested samples were analyzed by PRM using an Orbitrap Female/male (% women) 28/13 (68.3) Eclipse coupled to an EASY-nanoLC 1200 UPLC system Follow-up time (y) b 7.3 (5.1) (Thermo Fisher Scientific) with a 50-cm C18 chromato- EDSS at CSF collection 3.0 (2.5–4.0) graphic column (Easy-Spray Column, PepMap RSLC C18). Peptide mixes were separated with a chromatographic gradi- EDSS at last visit 6.0 (3.5–6.8) ent starting at 5% B with a flow rate of 300 nL/minute and c Numeric progression rate 0.4 (0.5) going up to 22% B in 79 minutes and to 32% B in 11 minutes c (buffer A: 0.1% formic acid in water and buffer B: 0.1% formic Step progression rate 0.4 (0.6) acid in acetonitrile). The Orbitrap Eclipse was operated in Abbreviation: EDSS = Expanded Disability Status Scale. positive ionization mode with an EASY-Spray nanosource at All data are expressed as mean SD unless otherwise stated. EDSS is 1.4 kV and at a source temperature of 275°C. expressed as median (interquartile range). a Refers to age at the time of CSF collection. b Refers to time from CSF collection to the last visit. c Refers to numeric and step-based progression rates from the time of CSF A scheduled PRM method was used for data acquisition with a collection to the last visit. quadrupole isolation window set to 1.4 m/z and MS2 scans 4 Neurology: Neuroimmunology & Neuroinflammation | Volume 8, Number 6 | November 2021 Neurology.org/NN internal controls. The intra-assay coefficient of variation more abundant in the CSF of patients with low disability was 6%. progression (Table 3). Only NXPH1 showed significant changes in CSF protein levels in patients receiving IFNβ, Classification of Evidence although differences stayed significant after adjusting by Our primary research question was to identify CSF bio- treatment. The remaining detected proteins showed either a markers associated with disability progression in patients with lack of statistically significant association with disability pro- progressive MS. The classification of evidence assigned to this gression or were significantly associated with disability pro- question is Class II. gression only in a minority (<5) of the classification criteria (eTable 3, links.lww.com/NXI/A583). Statistical Analysis Differences in CSF protein abundances were assessed CSF CHI3L2 Is Validated as Long-term Prognostic among patients with high and low disability progression Biomarker Associated With Disability Progression according to NPR and SPR using an analysis of variance in Patients With PPMS model. Linear association between CSF protein abundances To validate the results obtained in the discovery cohort, levels and NPR and SPR during the follow-up was evaluated using of the 10 selected proteins were determined by targeted the Spearman rank correlation coefficient. Considering that proteomics in CSF samples from an independent cohort of a proportion of patients were receiving IFNβ at the time of patients with progressive MS. Although the discovery cohort CSF collection, the potential effect of IFNβ in CSF protein included patients with transitional progressive MS and PPMS, levels was also evaluated including treatment as covariate in no statistically significant differences were observed in the the analysis. Differences in progression rates between pa- CSF levels of the 10 selected proteins between both groups of tients from the discovery and validation cohorts were patients (data not shown), results that supported the use of a assessed using a Mann-Whitney U test. Performance of CSF validation cohort consisting exclusively of patients with chitinase 3–like 2 (CHI3L2) levels to discriminate between PPMS. Follow-up time of patients with PPMS in the valida- patients with high and low disability progression was tion cohort was shorter and EDSS scores at onset lower assessed using receiver operating characteristic (ROC) curve compared with patients belonging to the discovery cohort analysis. (eTable 1, links.lww.com/NXI/A581 and eTable 2, links.lww. com/NXI/A582). However, there were no statistically sig- Data Availability nificant differences in NPR and SPR between patients from All data analyzed during this study will be shared anonymized the discovery and validation cohorts (0.3 [0.7] and 0.2 [0.1] by request of a qualified investigator to the corresponding for NPR at onset and off trial, respectively, in the discovery author. cohort vs 0.4 [0.5] in the validation cohort; 0.6 [1.3] and 0.3 [0.2] for SPR at onset and off trial, respectively, in the dis- covery cohort vs 0.4 [0.6] in the validation cohort; p > 0.05 for Results all comparisons). CSF Proteomic Analysis in the Discovery Cohort of Patients With Progressive MS Identifies Only CHI3L2 was validated, and similar to the findings in the Proteins Associated With Long-term discovery cohort (Table 3), CSF CHI3L2 levels were signif- Disability Progression icantly higher in patients with PPMS who progressed faster A schematic flowchart summarizing the main steps of the during the follow-up based on both NPR and SPR compared study design is represented in Figure 1. To identify proteins with those patients having lower disability progression (p = associated with disability progression, we first performed a 0.03 and p = 0.02, respectively) (Figure 2A). CSF CHI3L2 quantitative proteomic analysis in CSF samples from a dis- levels significantly correlated with NPR (r: 0.34, p = 0.03), covery cohort of 28 patients with progressive MS classified whereas a trend for significant correlation was observed be- according to numeric and step progression rates. A total of tween CHI3L2 levels and SPR progression rates (r: 0.30, p = 2,548 proteins were identified in CSF samples by mass 0.05) (Figure 2a). spectrometric analysis (the full list of identified proteins is provided in eTable, links.lww.com/NXI/A582). Of these Figure 2B shows the potential for CSF CHI3L2 levels to identified proteins, 10 proteins were prioritized for further discriminate between high and low disability progression in validation because of their potential association with disability patients with PPMS. The area under the ROC curve (AUC) progression, as described in the Methods section: SPATS2- was 73% (p = 0.03) and a CSF CHI3L2 value of 17.1 (protein like protein (SPATS2L), CHI3L2, plasma serine protease abundance estimate) resulted in the best cutoff to classify inhibitor (SERPINA5), metallothionein-3 (MT3), phospho- patients with PPMS with high and low disability progression lipase D4 (PLD4), and beta-hexosaminidase (HEXB), which according to both NPR and SPR, with a sensitivity of 90% and were more abundant in the CSF of patients with high dis- specificity of 63%. ability progression, and neurexophilin-1 (NXPH1), adipocyte enhancer-binding protein 1 (AEBP1), cathepsin L1 (CTSL), Partial correlations adjusting for age did not reveal statistically and lipopolysaccharide-binding protein (LBP), which were significant associations between CSF CHI3L2 and NFL levels Neurology.org/NN Neurology: Neuroimmunology & Neuroinflammation | Volume 8, Number 6 | November 2021 5 (r = 0.02, p = 0.91) or between CSF NFL levels and numeric this in mind, we aimed to identify long-term prognostic bio- progression rates (r = 0.08, p = 0.62) or SPR (r = 0.09, markers in the CSF of patients with progressive MS by con- p = 0.60). ducting a 2-phase study. An initial or discovery cohort included patients with progressive MS with a follow-up longer In the validation cohort, except for SPATS2L which could not than 10 years who participated in a single-center trial of IFNβ- be detected, the mean CSF levels of the remaining proteins 1b.12 By quantitative proteomics analysis of the CSF samples did not significantly differ between high and low disability collected during the trial, we identified a number of candidate progression using NPR or SPR in patients with PPMS, and CSF proteins that were associated with long-term disability hence, they were not validated as biomarkers associated with progression in patients classified according to 2 related pro- disability progression in MS (Table 4; eFigure 1, links.lww. gression rate indexes, which were applied at different time com/NXI/A579 and eFigure 2, links.lww.com/NXI/A580). points during the trial. Although almost half of the patients were receiving IFNβ at the time of CSF collection, treatment did not seem to have a major impact on CSF protein levels. Discussion A second or validation cohort included exclusively untreated Our understanding of disease progression is limited and still patients with PPMS whose CSF samples were collected from remains as one of the unmet needs in MS.10 In this context, different MS centers. Although in these patients follow-up time there is a lack of prognostic factors associated with disability was shorter and EDSS scores were lower, the progression rates, progression in patients with progressive forms of MS. Bearing which relate EDSS changes to time on follow-up, were Figure 1 Flowchart Summarizing the Different Steps Undertaken in the Study NPR and SPR were calculated in a discovery cohort of 28 patients with progressive MS who participated in a 2-year placebo-controlled trial of IFNβ-1b. Progression rates were computed between the time of trial onset and the last visit (NPR/SPR at trial onset), and between the time after trial completion and the last visit (NPR/SPR off trial). Based on these progression rates, patients with progressive MS were classified into high and low disability progression phenotypes. By means of a quantitative proteomic approach, a total of 2,548 proteins were identified in CSF samples from patients belonging to the discovery cohort, of which 10 proteins were found to be associated with disability progression in most of the selection criteria shown in the figure. Protein levels for these 10 proteins were measured by PRM in CSF samples from a validation cohort of 41 patients with progressive MS classified into high and low disability progression phenotypes according to NPR and SPR computed between the time of CSF collection and the time of the last visit. 6 Neurology: Neuroimmunology & Neuroinflammation | Volume 8, Number 6 | November 2021 Neurology.org/NN Neurology.org/NN Table 3 Selected CSF Proteins in the Discovery Cohort for Validation NPR “onset”a SPR “onset”a NPR “off”b SPR “off”b Uniprot Symbol Levels H/L (p) Levels H/L (p) Levels H/L (p) Levels H/L (p) NPR “onset”a rho (p) SPR “onset”a rho (p) NPR “off”b rho (p) SPR “off”b rho (p) H/Lc Nd P58417 NXPH1e 25.3/26.8 (0.007) 25.4/26.8 (0.018) 25.7/26.9 (0.037) 25.3/26.8 (0.012) −0.52 (0.042) −0.62 (0.023) −0.76 (0.001) −0.79 (0.008) Down 8 Q9NUQ6 SPATS2L 25.4/23.8 (0.035) 25.6/23.8 (0.041) 25.4/23.6 (0.017) 25.8/23.7 (0.016) 0.54 (0.019) 0.47 (0.045) 0.44 (0.066) 0.49 (0.041) Up 7 Q8IUX7 AEBP1 24.2/25.0 (0.003) 24.2/25.1 (0.002) 24.5/25.0 (0.066) 24.5/25.0 (0.066) −0.56 (0.014) −0.62 (0.005) −0.58 (0.008) −0.61 (0.006) Down 6 Q15782 CHI3L2 26.3/25.4 (0.030) 26.4/25.3 (0.012) 25.8/25.3 (0.202) 26.1/25.3 (0.044) 0.38 (0.111) 0.52 (0.022) 0.49 (0.024) 0.50 (0.021) Up 6 P05154 SERPINA5 26.2/25.7 (0.073) 26.2/25.7 (0.051) 26.2/25.7 (0.018) 26.3/25.7 (0.013) 0.46 (0.023) 0.45 (0.025) 0.68 (0.001) 0.63 (0.001) Up 6 P07711 CTSL 27.0/27.5 (0.090) 26.8/27.6 (0.009) 27.1/27.5 (0.206) 26.9/27.5 (0.047) −0.37 (0.033) −0.38 (0.032) −0.32 (0.073) −0.42 (0.025) Down 5 Neurology: Neuroimmunology & Neuroinflammation | Volume 8, Number 6 | November 2021 P25713 MT3 21.5/20.9 (0.010) 21.5/20.9 (0.010) 21.2/20.8 (0.051) 21.2/20.9 (0.267) 0.69 (0.007) 0.79 (0.001) 0.49 (0.101) 0.60 (0.047) Up 5 Q96BZ4 PLD4 25.9/24.2 (0.002) 25.7/24.3 (0.038) 25.7/24.1 (0.017) 25.6/24.3 (0.083) 0.66 (0.028) 0.59 (0.056) 0.75 (0.016) 0.61 (0.061) Up 5 P18428 LBP 24.6/25.4 (0.008) 24.7/25.3 (0.070) 24.7/25.5 (0.011) 24.8/25.3 (0.192) −0.62 (0.012) −0.66 (0.011) −0.59 (0.043) −0.57 (0.064) Down 5 P07686 HEXB 26.7/26.0 (0.078) 26.3/26.2 (0.796) 26.7/26.0 (0.036) 26.5/26.1 (0.302) 0.45 (0.025) 0.40 (0.047) 0.50 (0.012) 0.46 (0.025) Up 5 Abbreviations: AEBP1 = adipocyte enhancer-binding protein 1; CHI3L2 = chitinase 3-like 2; CTSL = cathepsin L1; HEXB = hexosaminidase; IFNβ = interferon beta-1b; LBP = lipopolysaccharide-binding protein; MT3 = metallothionein-3; NPR = numeric progression rates; NXPH1 = neurexophilin-1; PLD4 = phospholipase D4; SERPINA5 = plasma serine protease inhibitor; SPATS2L = SPATS2-like protein; SPR = step-based progression rates. rho: Spearman rank correlation coefficient. a Refers to NPR and SPR calculated from trial onset to the last visit. b Refers to NPR and SPR calculated from the time after trial completion to the last visit. c Refers to the direction in protein levels: up/down = increased/decreased levels in patients with high/low disability progression, respectively. d Refers to the total number of statistically significant associations observed after applying the different classification methods. Levels H/L refers to mean protein levels in patients with high and low disability progression for each of the different indexes calculated from peptide intensities at log-scale (arbitrary units). Significant p values (p) are shown in bold. e NXPH1 was the only selected protein showing significant changes in CSF abundance in patients treated with IFNβ, but differences remained significant after adjusting by this covariate. 7 Figure 2 Performance of CHI3L2 as a Biomarker of Disability Progression in Patients With Progressive MS (A) Graphs comparing CSF CHI3L2 lev- els between high and low disability progression patients with progressive MS according to numeric and step- based progression rates, and correla- tion plots evaluating linear association between CSF CHI3L2 levels and pro- gression rates, both in the validation cohort. (B) Performance of CSF CHI3L2 levels to discriminate between patients with high and low disability pro- gression based on numeric (B.a) and step-based progression rates (B.b). AU = protein abundance estimate; AUC = area under the receiver operating characteristic curve. Numbers in pa- rentheses represent 95% confidence intervals of the AUC; CHI3L2 = chitinase 3-like 2. comparable among patients with progressive MS belonging to of CHI3L2 in the CNS. CSF CHI3L2 levels have been both study cohorts. Selected candidate CSF biomarkers iden- reported to be elevated in a number of neuroinflammatory tified in the initial cohort were measured in the validation conditions.17-21 In patients with clinically isolated syndrome cohort by means of PRM, a well-known specific and re- (CIS), high CSF CHI3L2 levels were associated with an in- producible mass spectrometric technique for protein quanti- creased risk for MS and predicted cognitive impairment eval- tation.14 Only CHI3L2 was found to be significantly associated uated by the paced auditory serial addition test.18 In another with disability progression in both cohorts of patients. study, CSF CHI3L2 levels were found to be significantly in- creased in patients with RRMS compared with patients with CHI3L2, also known as YKL39, is a member of the 18 glycosyl CIS, patients with progressive MS, and controls.17 hydrolase family that can bind chitin with high affinity but lacks enzymatic activity.15 CHI3L2 was first discovered in condi- In our study, CSF abundance for CHI3L2 was higher in pa- tioned medium from primary cultures of human articular car- tients with progressive MS who later progressed faster com- tilage chondrocytes15, and its function has been related to the pared with those patients with lower progression rates. In immune response and tissue remodeling. In this context, addition, CSF CHI3L2 levels showed good potential to dis- CHI3L2 was expressed by M2 polarized macrophages and was criminate between high and low disability progression in pa- found to act as a pro-angiogenic and monocyte recruiting fac- tients with progressive MS after long-term follow-up, with tor.16 Little is known about the physiologic and pathologic roles excellent sensitivity and acceptable specificity. Although the 8 Neurology: Neuroimmunology & Neuroinflammation | Volume 8, Number 6 | November 2021 Neurology.org/NN Table 4 CSF Proteins Levels of Selected Proteins in the Validation Cohort NPR SPR Uniprot Symbol Levels H/L p Value Levels H/L p Value P58417 NXPH1 19.7 (19.2–20.1)/19.7 (19.2–20.2) 1.000 19.7 (19.2–20.1)/19.7 (19.2–20.2) 0.964 Q8IUX7 AEBP1 20.2 (19.5–20.9)/20.3 (19.7–20.6) 0.754 20.3 (20.0–20.9)/20.3 (19.6–20.6) 0.360 P05154 SERPINA5 20.1 (19.1–20.8)/19.7 (18.7–20.3) 0.376 20.2 (19.5–20.8)/19.7 (18.6–20.3) 0.112 P07711 CTSL 21.0 (20.8–21.2)/20.9 (20.6–21.3) 0.777 21.0 (20.8–21.2)/20.9 (20.6–21.3) 0.709 P25713 MT3 20.9 (20.5–21.3)/20.9 (20.1–21.4) 0.622 20.7 (20.4–21.1)/20.9 (20.1–21.4) 0.665 Q96BZ4 PLD4 20.4 (20.1–20.9)/20.4 (19.9–20.9) 0.777 20.5 (20.3–21.2)/20.3 (19.8–20.9) 0.235 P18428 LBP 17.9 (16.5–18.5)/17.8 (16.9–18.9) 0.893 18.1 (17.2.18.6)/17.8 (16.6–18.9) 0.622 P07686 HEXB 19.9 (19.7–20.2)/19.9 (19.5–20.3) 0.687 19.9 (19.7–20.2)/19.9 (19.5–20.3) 0.665 Abbreviations: AEBP1 = adipocyte enhancer-binding protein 1; CTSL = cathepsin L1; HEXB = hexosaminidase; LBP = lipopolysaccharide-binding protein; MT3 = metallothionein-3; NPR = numeric progression rates; SPR = step-based progression rates; NXPH1 = neurexophilin-1; PLD4 = phospholipase D4; SERPINA5 = plasma serine protease inhibitor. Levels H/L refers to median protein levels (interquartile range) in patients with high and low disability progression for each of the different indexes. underlying molecular mechanisms of the association observed scores in the validation cohort. However, as mentioned before, between CSF CHI3L2 levels and disease progression are un- some of these differential factors were considered in the anal- known, a previous study revealed CHI3L2 expression in as- ysis, and also the progression rates that were used to classify trocytes from white matter plaques and from normal appearing patients into high and low disability progression phenotypes white matter.17 In this context, several lines of evidence have were comparable among patients from the discovery and vali- pointed to astrocytes as important players in MS pathogenesis dation cohorts. Other factors may relate to differences in the and suggested a contribution of these cells to the neurode- proteomic techniques used to determine protein abundance in generative component of the disease.22-27 This astrocytic origin the CSF of patients with MS. However, based on the results in of CHI3L2 may explain the lack of correlation observed in our the validation cohort, except for CHI3L2, the remaining can- study between CSF CHI3L2 abundance and CSF NFL levels, didate CSF proteins identified in the first phase of the study can as a representative biomarker of neuronal damage.28 probably be considered as false-positive results. CSF proteins that were significantly associated with long-term In conclusion, a proteomic approach in CSF samples from disability progression in a lower number of classification patients with progressive MS classified into high and low methods are also interesting candidate biomarkers that re- disability progression phenotypes has led to the identification quire further validation in independent cohorts of patients and validation of CHI3L2 as a biomarker associated with with progressive MS. Of note, one of these candidates is long-term disability progression. Further mechanistic studies CHI3L1, a protein closely related to CHI3L2 in size and are needed to better understand the relationship between sequence, although with structural differences,29 which has CSF CHI3L2 and disability progression in MS. been extensively implicated in MS.30 In contrast to CHI3L2, CHI3L1 was found significantly increased in the CSF of pa- Acknowledgment tients with progressive MS compared with patients with The authors want to particularly acknowledge the patients and the RRMS,17 and high CSF CHI3L1 levels were identified as an Biobank HUB-ICO-IDIBELL (PT17/0015/0024) integrated in independent predictor of disability worsening in patients with the Spanish Biobank Network for their collaboration. The MS.31 authors also wish to thank the Biobank Hospital Universitario Puerta de Hierro Majadahonda (HUPHM)/Instituto de Inves- The reasons for the lack of validation of the remaining proteins tigación Sanitaria Puerta de Hierro-Segovia de Arana (IDIPHISA) selected from the first phase of the study are unknown and (PT17/0015/0020 in the Spanish National Biobanks Network) probably manifold. They may be related to differences in the for the human specimens used in this study. clinical characteristics of patients included in both cohorts such as the clinical form (patients with transitional progressive and Study Funding PPMS in the discovery cohort vs patients with purely PPMS in This work was funded by grants from the International Pro- the validation cohort), treatment with IFNβ in a subgroup of gressive MS Alliance (grant no. PA0020), Asociación Escle- patients from the discovery cohort vs untreated patients in the rosis Múltiple (EME)—Red Española de Esclerosi múltiple validation cohort, and shorter follow-up and lower disability (REEM), REEM (RD16/0015/002 and RD16/0015/003) Neurology.org/NN Neurology: Neuroimmunology & Neuroinflammation | Volume 8, Number 6 | November 2021 9 cofunded by the Instituto de Salud Carlos III and Fondo Europeo de Desarrollo Regional (FEDER, Otra manera de Appendix (continued) hacer Europa). Name Location Contribution Disclosure Nicolas Servei de Neurologia- Data acquisition and revised Fissolo, PhD Neuroimmunologia. Centre the article for intellectual The authors report no disclosures relevant to the manuscript. d’Esclerosi Múltiple de content Go to Neurology.org/NN for full disclosures. Catalunya (Cemcat). Institut de Recerca Vall d’Hebron (VHIR). Hospital Universitari Publication History Vall d’Hebron. Universitat Autònoma de Barcelona, Received by Neurology: Neuroimmunology & Neuroinflammation Spain February 26, 2021. Accepted in final form July 15, 2021. Antonio J. Neuroimmunology Unit, Data acquisition and revised Sánchez- Hospital Universitario the article for intellectual López, PhD Puerta de Hierro, Madrid, content Spain Appendix Authors Lucienne Departments of Neurology Data acquisition and revised Name Location Contribution Costa- and Immunology, Hospital the article for intellectual Frossard, MD Universitario Ramón y Cajal, content Manuel Servei de Neurologia- Design and conceptualized Instituto Ramón y Cajal de Comabella, Neuroimmunologia. Centre study, analyzed the data, Investigacion Sanitaria, MD, PhD d’Esclerosi Múltiple de and drafted the article for Madrid, Spain Catalunya (Cemcat). Institut intellectual content de Recerca Vall d’Hebron Yolanda Center of Data acquisition and revised (VHIR). Hospital Universitari Blanco, MD Neuroimmunology, Service the article for intellectual Vall d’Hebron. Universitat of Neurology, Hospital Clinic content Autònoma de Barcelona, and Institut d’Investigacions Spain Biomèdiques August Pi Sunyer (IDIBAPS), University Jaume Sastre- Servei de Neurologia- Study concept and design of Barcelona, Spain Garriga, MD, Neuroimmunologia. Centre and revised the article for PhD d’Esclerosi Múltiple de intellectual content Sara Llufriu, Center of Data acquisition and revised Catalunya (Cemcat). MD, PhD Neuroimmunology, Service the article for intellectual Hospital Universitari Vall of Neurology, Hospital Clinic content d’Hebron. Universitat and Institut d’Investigacions Autònoma de Barcelona, Biomèdiques August Pi Spain Sunyer (IDIBAPS), University of Barcelona, Spain Eva Borras, Proteomics Unit, Universitat Study concept and design, PhD Pompeu Fabra (UPF), analyzed the data, and Angela Vidal- Servei de Neurologia- Data acquisition and revised Barcelona, Spain revised the article for Jordana, MD, Neuroimmunologia. Centre the article for intellectual intellectual content PhD d’Esclerosi Múltiple de content Catalunya (Cemcat). Luisa M. Departments of Neurology Data acquisition and revised Hospital Universitari Vall Villar, MD, and Immunology, Hospital the article for intellectual d’Hebron. Universitat PhD Universitario Ramón y Cajal, content Autònoma de Barcelona, Instituto Ramón y Cajal de Spain Investigacion Sanitaria, Madrid, Spain Eduard Proteomics Unit, Centre de Study concept and design, Sabidó, PhD Regulació Genòmica (CRG), analyzed the data, and Albert Saiz, Center of Data acquisition and revised Barcelona Institute of revised the article for MD, PhD Neuroimmunology, Service the article for intellectual Science and Technology intellectual content of Neurology, Hospital Clinic content (BIST), Barcelona, Spain and Institut d’Investigacions Biomèdiques August Pi Xavier Servei de Neurologia- Study concept and design Sunyer (IDIBAPS), University Montalban, Neuroimmunologia. Centre and revised the article for of Barcelona, Spain MD, PhD d’Esclerosi Múltiple de intellectual content Catalunya (Cemcat). Sergio Department of Neurology, Data acquisition and revised Hospital Universitari Vall Martı́nez- Bellvitge University the article for intellectual d’Hebron. Universitat Yélamos, MD, Hospital, Barcelona, Spain content Autònoma de Barcelona, PhD Spain Juan Antonio Neuroimmunology Unit, Data acquisition and revised Garcı́a- Hospital Universitario the article for intellectual References Merino, MD, Puerta de Hierro, Madrid, content 1. Thompson AJ, Baranzini SE, Geurts J, et al. Multiple sclerosis. Lancet. 2018; PhD Spain 391(10130):1622-1636. 2. International Multiple Sclerosis Genetics Consortium, Wellcome Trust Case Control Rucsanda Servei de Neurologia- Data acquisition and revised Consortium 2, , Sawcer S, et al. Genetic risk and a primary role for cell-mediated Pinteac, MSc Neuroimmunologia. Centre the article for intellectual immune mechanisms in multiple sclerosis. Nature. 2011;476(7359):214-219. d’Esclerosi Múltiple de content 3. International Multiple Sclerosis Genetics Consortium, Beecham AH, Patsopoulos Catalunya (Cemcat). Institut NA, et al. Analysis of immune-related loci identifies 48 new susceptibility variants for de Recerca Vall d’Hebron multiple sclerosis. Nature Genet. 2013;45(11):1353-1360. (VHIR). Hospital Universitari 4. International Multiple Sclerosis Genetics Consortium. Multiple sclerosis genomic Vall d’Hebron. Universitat map implicates peripheral immune cells and microglia in susceptibility. Science. 2019; Autònoma de Barcelona, 365(6460):eaav7188. Spain 5. Ascherio A, Munger KL, Lünemann JD. The initiation and prevention of multiple sclerosis. Nat Rev Neurol. 2012;8(11):602-612. 10 Neurology: Neuroimmunology & Neuroinflammation | Volume 8, Number 6 | November 2021 Neurology.org/NN 6. Waubant E, Lucas R, Mowry E, et al. Environmental and genetic risk factors for 19. Sanfilippo C, Nunnari G, Calcagno A, et al. The chitinases expression is related to MS(XX): an integrated review. Ann Clin Transl Neurol. 2019;6(9):1905-1922. simian immunodeficiency virus encephalitis (SIVE) and in HIV encephalitis (HIVE). 7. Confavreux C, Vukusic S. Natural history of multiple sclerosis: a unifying concept. Virus Res. 2017;227:220-230. Brain. 2006;129(Pt 3):606-616. 20. Sanfilippo C, Longo A, Lazzara F, et al. CHI3L1 and CHI3L2 overexpression in motor 8. Thompson AJ, Polman C, Miller CH, et al. Primary progressive multiple sclerosis. cortex and spinal cord of sALS patients. Mol Cell Neurosci. 2017;85:162-169. Brain. 1997;120(Pt 6):1085-1096. 21. Thompson AG, Gray E, Bampton A, et al. CSF chitinase proteins in amyotrophic 9. Tintore M, Vidal-Jordana A, Sastre-Garriga J. Treatment of multiple sclerosis - success lateral sclerosis. J Neurol Neurosurg Psychiatry. 2019;90(11):1215-1220. from bench to bedside. Nat Rev Neurol. 2019;15(1):53-58. 22. Wang D, Ayers MA, Catmull DV, et al. Astrocyte-associated axonal damage in pre-onset 10. Thompson A, Ciccarelli O. Towards treating progressive multiple sclerosis. Nat Rev stages of experimental autoimmune encephalomyelitis. Glia. 2005;51(3):235-240. Neurol. 2020;16(11):589-590. 23. Brambilla R, Persaud T, Hu X, et al. Transgenic inhibition of astroglial NF-kappa B 11. Lublin FD, Reingold SC. Defining the clinical course of multiple sclerosis: results of an improves functional outcome in experimental autoimmune encephalomyelitis by sup- international survey. Neurology. 1996;46(4):907-911. pressing chronic central nervous system inflammation. J Immunol. 2009;182(5):2628-2640. 12. Montalban X, Sastre-Garriga J, Tintoré M, et al. A single-center, randomized, double- 24. Nair A, Frederick TJ, Miller SD. Astrocytes in multiple sclerosis: a product of their blind, placebo-controlled study of interferon beta-1b on primary progressive and environment. Cell Mol Life Sci. 2008;65(17):2702-2720. transitional multiple sclerosis. Mult Scler. 2009;15(10):1195-1205. 25. Kim RY, Hoffman AS, Itoh N, et al. Astrocyte CCL2 sustains immune cell infiltration in 13. Perez-Riverol Y, Csordas A, Bai J, et al. The PRIDE database and related tools and chronic experimental autoimmune encephalomyelitis. J Neuroimmunol. 2014;274(1-2):53-61. resources in 2019: improving support for quantification data. Nucleic Acids Res. 2019; 26. Farez MF, Quintana FJ, Gandhi R, et al. Toll-like receptor 2 and poly(ADP-ribose) 47(D1):D442-D450. polymerase 1 promote central nervous system neuroinflammation in progressive EAE. 14. Lesur A, Domon B. Advances in high-resolution accurate mass spectrometry appli- Nat Immunol. 2009;10(9):958-964. cation to targeted proteomics. Proteomics. 2015;15(5-6):880-890. 27. Mayo L, Trauger SA, Blain M, et al. Regulation of astrocyte activation by glycolipids 15. Guan S-P, Mok Y-K, Koo K-N, et al. Chitinases: biomarkers for human diseases. drives chronic CNS inflammation. Nat Med. 2014;20(10):1147-1156. Protein Pept Lett. 2009;16(5):490-498. 28. Khalil M, Teunissen CE, Otto M, et al. Neurofilaments as biomarkers in neurological 16. Liu T, Larionova I, Litviakov N, et al. Tumor-associated macrophages in human breast disorders. Nat Rev Neurol. 2018;14(10):577-589. cancer produce new monocyte attracting and pro-angiogenic factor YKL-39 indicative 29. Ranok A, Wongsantichon J, Robinson RC, Suginta W. Structural and thermodynamic for increased metastasis after neoadjuvant chemotherapy. Oncoimmunology. 2018; insights into chitooligosaccharide binding to human cartilage chitinase 3-like protein 2 7(6):e1436922. (CHI3L2 or YKL-39). J Biol Chem. 2015;290(5):2617-2629. 17. Hinsinger G, Galéotti N, Nabholz N, et al. Chitinase 3-like proteins as diagnostic and 30. Pinteac R, Montalban X, Comabella M. Chitinases and chitinase-like proteins as bio- prognostic biomarkers of multiple sclerosis. Mult Scler. 2015;21(10):1251-1261. markers in neurological disorders. Neurol Neuroimmunol Neuroinflamm. 2020;8(1):e921. 18. Møllgaard M, Degn M, Sellebjerg F, et al. Cerebrospinal fluid chitinase-3-like 2 and 31. Gil-Perotin S, Castillo-Villalba J, Cubas-Nuñez L, et al. Combined cerebrospinal fluid chitotriosidase are potential prognostic biomarkers in early multiple sclerosis. Eur J neurofilament light chain protein and chitinase-3 like-1 levels in defining disease Neurol. 2016;23(5):898-905. course and Prognosis in multiple sclerosis. Front Neurol. 2019;10:1008. Neurology.org/NN Neurology: Neuroimmunology & Neuroinflammation | Volume 8, Number 6 | November 2021 11 CSF Chitinase 3−Like 2 Is Associated With Long-term Disability Progression in Patients With Progressive Multiple Sclerosis Manuel Comabella, Jaume Sastre-Garriga, Eva Borras, et al. Neurol Neuroimmunol Neuroinflamm 2021;8; DOI 10.1212/NXI.0000000000001082 This information is current as of September 8, 2021 Updated Information & including high resolution figures, can be found at: Services http://nn.neurology.org/content/8/6/e1082.full.html References This article cites 31 articles, 5 of which you can access for free at: http://nn.neurology.org/content/8/6/e1082.full.html##ref-list-1 Subspecialty Collections This article, along with others on similar topics, appears in the following collection(s): Class II http://nn.neurology.org//cgi/collection/class_ii Multiple sclerosis http://nn.neurology.org//cgi/collection/multiple_sclerosis Permissions & Licensing Information about reproducing this article in parts (figures,tables) or in its entirety can be found online at: http://nn.neurology.org/misc/about.xhtml#permissions Reprints Information about ordering reprints can be found online: http://nn.neurology.org/misc/addir.xhtml#reprintsus Neurol Neuroimmunol Neuroinflamm is an official journal of the American Academy of Neurology. Published since April 2014, it is an open-access, online-only, continuous publication journal. Copyright Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology.. All rights reserved. Online ISSN: 2332-7812.