Background: Neutralizing anti-drug antibodies (ADA) can greatly reduce the efficacy of biopharmaceuticals used to treat patients with multiple sclerosis (MS). However, the biological factors pre-disposing an individual to develop ADA are poorly characterized. Thus, there is an unmet clinical need for biomarkers to predict the development of immunogenicity, and subsequent treatment failure. Up to 35% of MS patients treated with beta interferons (IFNβ) develop ADA. Here we use machine learning to predict immunogenicity against IFNβ utilizing serum metabolomics data. Methods: Serum samples were collected from 89 MS patients as part of the ABIRISK consortium-a multi-center prospective study of ADA development. Metabolites and ADA were quantified prior to and after IFNβ treatment. Thirty patients became ADA positive during the first year of treatment (ADA+). We tested the efficacy of six binary classification models using 10-fold cross validation; k-nearest neighbors, decision tree, random forest, support vector machine and lasso (Least Absolute Shrinkage and Selection Operator) logistic regression with and without interactions. Results: We were able to predict future immunogenicity from baseline metabolomics data. Lasso logistic regression with/without interactions and support vector machines were the most successful at identifying ADA+ or ADA- cases, respectively. Furthermore, patients who become ADA+ had a distinct metabolic response to IFNβ in the first 3 months, with 29 differentially regulated metabolites. Machine learning algorithms could also predict ADA status based on metabolite concentrations at 3 months. Lasso logistic regressions had the greatest proportion of correct classifications [F1 score (accuracy measure) = 0.808, specificity = 0.913]. Finally, we hypothesized that serum lipids could contribute to ADA development by altering immune-cell lipid rafts. This was supported by experimental evidence demonstrating that, prior to IFNβ exposure, lipid raft-associated lipids were differentially expressed between MS patients who became ADA+ or remained ADA-. Conclusion: Serum metabolites are a promising biomarker for prediction of ADA development in MS patients treated with IFNβ, and could provide novel insight into mechanisms of immunogenicity.

Centre for Cardiometabolic and Vascular Medicine University College London London United Kingdom

Centre for Rheumatology University College London London United Kingdom

Department of Clinical Neuroscience Center for Molecular Medicine Karolinska Institutet Karolinska University Hospital Huddinge Sweden

Department of Neuroinflammation University College London Institute of Neurology and National Hospital of Neurology and Neurosurgery London United Kingdom

Department of Neurology and Centre of Clinical Neuroscience General University Hospital and 1st Faculty of Medicine Charles University Prague Prague Czechia

Scicross AB Skövde Sweden

See more in PubMed

Tintore M, Vidal-Jordana A, Sastre-Garriga J. Treatment of multiple sclerosis — success from bench to bedside. Nat Rev Neurol. (2019) 15:53–8. 10.1038/s41582-018-0082-z PubMed DOI

Sorensen PS, Ross C, Clemmesen KM, Bendtzen K, Frederiksen JL, Jensen K, et al. . Clinical importance of neutralising antibodies against interferon beta in patients with relapsing-remitting multiple sclerosis. Lancet. (2003) 362:1184–91. 10.1016/S0140-6736(03)14541-2 PubMed DOI

Kappos L, Clanet M, Sandberg-Wollheim M, Radue EW, Hartung HP, Hohlfeld R, et al. . Neutralizing antibodies and efficacy of interferon beta-1a: a 4-year controlled study. Neurology. (2005) 65:40–7. 10.1212/01.wnl.0000171747.59767.5c PubMed DOI

Hesse D, Sellebjerg F, Sorensen PS. Absence of MxA induction by interferon beta in patients with MS reflects complete loss of bioactivity. Neurology. (2009) 73:372–7. 10.1212/WNL.0b013e3181b04c98 PubMed DOI

Vennegoor A, Rispens T, Strijbis EM, Seewann A, Uitdehaag BM, Balk LJ, et al. . Clinical relevance of serum natalizumab concentration and anti-natalizumab antibodies in multiple sclerosis. Mult Scler. (2013) 19:593–600. 10.1177/1352458512460604 PubMed DOI

Dubuisson N, Baker D, Kang AS, Pryce G, Marta M, Visser LH, et al. . Alemtuzumab depletion failure can occur in multiple sclerosis. Immunology. (2018) 154:253–60. 10.1111/imm.12879 PubMed DOI PMC

Svenningsson A, Dring AM, Fogdell-Hahn A, Jones I, Engdahl E, Lundkvist M, et al. . Fatal neuroinflammation in a case of multiple sclerosis with anti-natalizumab antibodies. Neurology. (2013) 80:965–7. 10.1212/WNL.0b013e3182840be3 PubMed DOI

Ebers GC. Randomised double-blind placebo-controlled study of interferon β-1a in relapsing/remitting multiple sclerosis. Lancet. (1998) 352:1498–504. 10.1016/S0140-6736(98)03334-0 PubMed DOI

Multiple Sclerosis Society Beta Interferons | Multiple Sclerosis Society UK. (2020). Available online at: www.mssociety.org.uk; https://www.mssociety.org.uk/about-ms/treatments-and-therapies/disease-modifying-therapies/beta-interferons (accessed July 3, 2020).

Bertolotto A, Malucchi S, Sala A, Orefice G, Carrieri PB, Capobianco M, et al. Differential effects of three interferon betas on neutralising antibodies in patients with multiple sclerosis: a follow up study in an independent laboratory. J Neurol Neurosurg Psychiatry. (2002) 73:148–53. 10.1136/jnnp.73.2.148 PubMed DOI PMC

Polman CH, Bertolotto A, Deisenhammer F, Giovannoni G, Hartung H-P, Hemmer B, et al. . Recommendations for clinical use of data on neutralising antibodies to interferon-beta therapy in multiple sclerosis. Lancet Neurol. (2010) 9:740–50. 10.1016/S1474-4422(10)70103-4 PubMed DOI

Sominanda A, Hillert J, Fogdell-Hahn A. In vivo bioactivity of interferon-beta in multiple sclerosis patients with neutralising antibodies is titre-dependent. J Neurol Neurosurg Psychiatry. (2008) 79:57–62. 10.1136/jnnp.2007.122549 PubMed DOI

Sethu S, Govindappa K, Quinn P, Wadhwa M, Stebbings R, Boggild M, et al. . Immunoglobulin G1 and immunoglobulin G4 antibodies in multiple sclerosis patients treated with IFNβ interact with the endogenous cytokine and activate complement. Clin Immunol. (2013) 148:177–85. 10.1016/j.clim.2013.05.008 PubMed DOI PMC

Sorensen P, Koch-Henriksen N, Bendtzen K. Are ex vivo neutralising antibodies against IFN-β always detrimental to therapeutic efficacy in multiple sclerosis? Mult Scler J. (2007) 13:616–21. 10.1177/1352458506072344 PubMed DOI

Comi G, Radaelli M, Soelberg Sørensen P. Evolving concepts in the treatment of relapsing multiple sclerosis. Lancet. (2017) 389:1347–56. 10.1016/S0140-6736(16)32388-1 PubMed DOI

Hoffmann S, Cepok S, Grummel V, Lehmann-Horn K, Hackermueller J, Stadler PF, et al. . HLA-DRB1*0401 and HLA-DRB1*0408 are strongly associated with the development of antibodies against interferon-β therapy in multiple sclerosis. Am J Hum Genet. (2008) 83:219–27. 10.1016/j.ajhg.2008.07.006 PubMed DOI PMC

Weber F, Cepok S, Wolf C, Berthele A, Uhr M, Bettecken T, et al. . Single-nucleotide polymorphisms in HLA- and non-HLA genes associated with the development of antibodies to interferon-β therapy in multiple sclerosis patients. Pharmacogenomics J. (2012) 12:238–45. 10.1038/tpj.2011.14 PubMed DOI

Adriani M, Nytrova P, Mbogning C, Hässler S, Medek K, Jensen PEH, et al. . Monocyte NOTCH2 expression predicts IFN-β immunogenicity in multiple sclerosis patients. JCI Insight. (2018) 3:e99274. 10.1172/jci.insight.99274 PubMed DOI PMC

Zhao Y, Healy BC, Rotstein D, Guttmann CRG, Bakshi R, Weiner HL, et al. . Exploration of machine learning techniques in predicting multiple sclerosis disease course. PLoS ONE. (2017) 12:e0174866. 10.1371/journal.pone.0174866 PubMed DOI PMC

Mateos-Pérez JM, Dadar M, Lacalle-Aurioles M, Iturria-Medina Y, Zeighami Y, Evans AC. Structural neuroimaging as clinical predictor: a review of machine learning applications. NeuroImage Clin. (2018) 20:506–22. 10.1016/j.nicl.2018.08.019 PubMed DOI PMC

Raghavendra U, Acharya UR, Adeli H. Artificial intelligence techniques for automated diagnosis of neurological disorders. Eur Neurol. (2019) 82:41–64. 10.1159/000504292 PubMed DOI

Lötsch J, Schiffmann S, Schmitz K, Brunkhorst R, Lerch F, Ferreiros N, et al. . Machine-learning based lipid mediator serum concentration patterns allow identification of multiple sclerosis patients with high accuracy. Sci Rep. (2018) 8:14884. 10.1038/s41598-018-33077-8 PubMed DOI PMC

Dickens AM, Larkin JR, Griffin JL, Cavey A, Matthews L, Turner MR, et al. . A type 2 biomarker separates relapsing-remitting from secondary progressive multiple sclerosis. Neurology. (2014) 83:1492–9. 10.1212/WNL.0000000000000905 PubMed DOI PMC

Weinstock-Guttman B, Zivadinov R, Horakova D, Havrdova E, Qu J, Shyh G, et al. . Lipid profiles are associated with lesion formation over 24 months in interferon-β treated patients following the first demyelinating event. J Neurol Neurosurg Psychiatry. (2013) 84:1186–91. 10.1136/jnnp-2012-304740 PubMed DOI

van de Kraats C, Killestein J, Popescu V, Rijkers E, Vrenken H, Lütjohann D, et al. . Oxysterols and cholesterol precursors correlate to magnetic resonance imaging measures of neurodegeneration in multiple sclerosis. Mult Scler J. (2014) 20:412–7. 10.1177/1352458513499421 PubMed DOI

Uher T, Fellows K, Horakova D, Zivadinov R, Vaneckova M, Sobisek L, et al. . Serum lipid profile changes predict neurodegeneration in interferon-β1a-treated multiple sclerosis patients. J Lipid Res. (2017) 58:403–11. 10.1194/jlr.M072751 PubMed DOI PMC

Durfinová M, Procházková L, Petrleničová D, Bystrická Z, Orešanská K, Kuračka L, et al. . Cholesterol level correlate with disability score in patients with relapsing-remitting form of multiple sclerosis. Neurosci Lett. (2018) 687:304–7. 10.1016/j.neulet.2018.10.030 PubMed DOI

Gafson AR, Thorne T, McKechnie CIJ, Jimenez B, Nicholas R, Matthews PM. Lipoprotein markers associated with disability from multiple sclerosis. Sci Rep. (2018) 8:17026. 10.1038/s41598-018-35232-7 PubMed DOI PMC

Sorci-Thomas MG, Thomas MJ. High density lipoprotein biogenesis, cholesterol efflux, and immune cell function. Arterioscler Thromb Vasc Biol. (2012) 32:2561–5. 10.1161/ATVBAHA.112.300135 PubMed DOI PMC

Köberlin MS, Snijder B, Heinz LX, Baumann CL, Fauster A, Vladimer GI, et al. . A conserved circular network of coregulated lipids modulates innate immune responses. Cell. (2015) 162:170–83. 10.1016/j.cell.2015.05.051 PubMed DOI PMC

Ito A, Hong C, Oka K, Salazar JV, Diehl C, Witztum JL, et al. . Cholesterol accumulation in CD11c + immune cells is a causal and targetable factor in autoimmune disease. Immunity. (2016) 45:1311–26. 10.1016/j.immuni.2016.11.008 PubMed DOI PMC

Nath AP, Ritchie SC, Byars SG, Fearnley LG, Havulinna AS, Joensuu A, et al. . An interaction map of circulating metabolites, immune gene networks, and their genetic regulation. Genome Biol. (2017) 18:146. 10.1186/s13059-017-1279-y PubMed DOI PMC

Mahadevan S, Shah SL, Marrie TJ, Slupsky CM. Analysis of metabolomic data using support vector machines. Anal Chem. (2008) 80:7562–70. 10.1021/ac800954c PubMed DOI

Trainor PJ, DeFilippis AP, Rai SN. Evaluation of classifier performance for multiclass phenotype discrimination in untargeted metabolomics. Metabolites. (2017) 7:30. 10.3390/metabo7020030 PubMed DOI PMC

Fan TWM, Zhang X, Wang C, Yang Y, Kang W-Y, Arnold S, et al. . Exosomal lipids for classifying early and late stage non-small cell lung cancer. Anal Chim Acta. (2018) 1037:256–64. 10.1016/j.aca.2018.02.051 PubMed DOI PMC

Yuan B, Schafferer S, Tang Q, Scheffler M, Nees J, Heil J, et al. . A plasma metabolite panel as biomarkers for early primary breast cancer detection. Int J Cancer. (2019) 144:2833–42. 10.1002/ijc.31996 PubMed DOI

Polman CH, Reingold SC, Banwell B, Clanet M, Cohen JA, Filippi M, et al. . Diagnostic criteria for multiple sclerosis: 2010 revisions to the McDonald criteria. Ann Neurol. (2011) 69:292–302. 10.1002/ana.22366 PubMed DOI PMC

Ingenhoven K, Kramer D, Jensen PE, Hermanrud C, Ryner M, Deisenhammer F, et al. . Development and validation of an enzyme-linked immunosorbent assay for the detection of binding anti-drug antibodies against interferon beta. Front Neurol. (2017) 8:305. 10.3389/fneur.2017.00305 PubMed DOI PMC

Hermanrud C, Ryner M, Luft T, Jensen PE, Ingenhoven K, Rat D, et al. . Development and validation of cell-based luciferase reporter gene assays for measuring neutralizing anti-drug antibodies against interferon beta. J Immunol Methods. (2016) 430:1–9. 10.1016/j.jim.2016.01.004 PubMed DOI

Jensen PEH, Warnke C, Ingenhoven K, Piccoli L, Gasis M, Hermanrud C, et al. . Detection and kinetics of persistent neutralizing anti-interferon-beta antibodies in patients with multiple sclerosis. Results from the ABIRISK prospective cohort study. J Neuroimmunol. (2019) 326:19–27. 10.1016/j.jneuroim.2018.11.002 PubMed DOI

Soininen P, Kangas AJ, Wurtz P, Tukiainen T, Tynkkynen T, Laatikainen R, et al. . High-throughput serum NMR metabonomics for cost-effective holistic studies on systemic metabolism. Analyst. (2009) 134:1781–5. 10.1039/b910205a PubMed DOI

Würtz P, Kangas AJ, Soininen P, Lawlor DA, Davey Smith G, Ala-Korpela M. Quantitative serum nuclear magnetic resonance metabolomics in large-scale epidemiology: a primer on -omic technologies. Am J Epidemiol. (2017) 186:1084–96. 10.1093/aje/kwx016 PubMed DOI PMC

R Core Team R: A Language and Environment for Statistical Computing (2019).

Janez D, Curk T, Erjavec A, Group C, Hocevar T, Milutinovic M, et al. Orange: data mining toolbox in python. J Mach Learn Res. (2013)14:2349–53.

Bachelet D, Hässler S, Mbogning C, Link J, Ryner M, Ramanujam R, et al. . Occurrence of anti-drug antibodies against interferon-beta and natalizumab in multiple sclerosis: a collaborative cohort analysis. PLoS ONE. (2016) 11:e0162752. 10.1371/journal.pone.0162752 PubMed DOI PMC

Zhang Z. Introduction to machine learning: K-nearest neighbors. Ann Transl Med. (2016) 4:218. 10.21037/atm.2016.03.37 PubMed DOI PMC

Yu W, Liu T, Valdez R, Gwinn M, Khoury MJ. Application of support vector machine modeling for prediction of common diseases: the case of diabetes and pre-diabetes. BMC Med Inform Decis Mak. (2010) 10:16. 10.1186/1472-6947-10-16 PubMed DOI PMC

Meyer D, Dimitriadou E, Hornik K, Weingessel A, Leisch F. e1071: Misc Functions of the Department of Statistics, Probability Theory Group (Formerly: E1071), TU Wien. R package version 1.7-3 (2019). Available online at: https://cran.r-project.org/web/packages/e1071/e1071.pdf (accessed February 26, 2020).

Kingsford C, Salzberg S. What are decision trees? Nat Biotechnol. (2008) 26:1011–3. 10.1038/nbt0908-1011 PubMed DOI PMC

Breiman L. Random forests. Mach Learn. (2001) 45:5–32. 10.1023/A:1010933404324 DOI

Cutler DR, Edwards TC, Beard KH, Cutler A, Hess KT, Gibson J, et al. . Random forests for classification in ecology. Ecology. (2007) 88:2783–92. 10.1890/07-0539.1 PubMed DOI

James G, Witten D, Hastie T, Tibshirani R. An Introduction to Statistical Learning With Applications in R. Springer (2013). Available online at: https://www.springer.com/gp/book/9781461471370#reviews (accessed March 17, 2020).

Breiman L, Cutler A. Breiman and Cutler's Random Forests for Classification and Regression. (2018). Available online at: https://cran.r-project.org/web/packages/randomForest/randomForest.pdf (accessed March 9, 2020).

Ahola-Olli AV, Mustelin L, Kalimeri M, Kettunen J, Jokelainen J, Auvinen J, et al. . Circulating metabolites and the risk of type 2 diabetes: a prospective study of 11,896 young adults from four finnish cohorts. Diabetologia. (2019) 62:2298–309. 10.1007/s00125-019-05001-w PubMed DOI PMC

Jiang H, Fei X, Liu H, Roeder K, Lafferty J, Wasserman L, et al. . Huge: High-Dimensional Undirected Graph Estimation version 1.3.4 from CRAN. (2019). Available online at: https://rdrr.io/cran/huge/ (accessed March 9, 2020)

Liu H, Roeder K, Wasserman L. Stability approach to regularization selection (StARS) for high dimensional graphical models. Adv Neural Inf Process Syst. (2010) 24:1432–40. PubMed PMC

Miguel L, Owen DM, Lim C, Liebig C, Evans J, Magee AI, et al. . Primary human CD4+ T Cells have diverse levels of membrane lipid order that correlate with their function. J Immunol. (2011) 186:3505–16. 10.4049/jimmunol.1002980 PubMed DOI

Smith E, Croca S, Waddington KE, Sofat R, Griffin M, Nicolaides A, et al. . Cross-talk between iNKT cells and monocytes triggers an atheroprotective immune response in SLE patients with asymptomatic plaque. Sci Immunol. (2016) 1:eaah4081. 10.1126/sciimmunol.aah4081 PubMed DOI

Waddington KE, Pineda-Torra I, Jury EC. Analyzing T-Cell plasma membrane lipids by flow cytometry. Methods Mol Biol. (2019) 1951:209–16. 10.1007/978-1-4939-9130-3_16 PubMed DOI

Cocco E, Murgia F, Lorefice L, Barberini L, Poddighe S, Frau J, et al. . (1)H-NMR analysis provides a metabolomic profile of patients with multiple sclerosis. Neurol Neuroimmunol Neuroinflamm. (2016) 3:e185. 10.1212/NXI.0000000000000185 PubMed DOI PMC

Villoslada P, Alonso C, Agirrezabal I, Kotelnikova E, Zubizarreta I, Pulido-Valdeolivas I, et al. . Metabolomic signatures associated with disease severity in multiple sclerosis. Neurol Neuroimmunol Neuroinflamm. (2017) 4:e321. 10.1212/NXI.0000000000000321 PubMed DOI PMC

Klauser AM, Wiebenga OT, Eijlers AJ, Schoonheim MM, Uitdehaag BM, Barkhof F, et al. . Metabolites predict lesion formation and severity in relapsing-remitting multiple sclerosis. Mult Scler J. (2018) 24:491–500. 10.1177/1352458517702534 PubMed DOI

Ehnholm C, Aho K, Huttunen JK, Kostiainen E, Mattila K, Pakkarainen J, et al. . Effect of interferon on plasma lipoproteins and on the activity of postheparin plasma lipases. Arteriosclerosis. (2000) 2:68–73. 10.1161/01.ATV.2.1.68 PubMed DOI

Rosenzweig IB, Wiebe DA, Borden EC, Storer B, Shrago ES. Plasma lipoprotein changes in humans induced by β-interferon. Atherosclerosis. (1987) 67:261–7. 10.1016/0021-9150(87)90287-5 PubMed DOI

Zhu X, Lee J-Y, Timmins JM, Brown JM, Boudyguina E, Mulya A, et al. . Increased cellular free cholesterol in macrophage-specific abca1 knock-out mice enhances pro-inflammatory response of macrophages. J Biol Chem. (2008) 283:22930–41. 10.1074/jbc.M801408200 PubMed DOI PMC

Ito A, Hong C, Rong X, Zhu X, Tarling EJ, Hedde PN, et al. . LXRs link metabolism to inflammation through Abca1-dependent regulation of membrane composition and TLR signaling. Elife. (2015) 4:e08009. 10.7554/eLife.08009.023 PubMed DOI PMC

Yang W, Bai Y, Xiong Y, Zhang J, Chen S, Zheng X, et al. . Potentiating the antitumour response of CD8+ T cells by modulating cholesterol metabolism. Nature. (2016) 531:651–5. 10.1038/nature17412 PubMed DOI PMC

Skeggs JW, Morton RE. LDL and HDL enriched in triglyceride promote abnormal cholesterol transport. J Lipid Res. (2002) 43:1264–74. 10.1194/jlr.M100431-JLR200 PubMed DOI

Girona J, Amigó N, Ibarretxe D, Plana N, Rodríguez-Borjabad C, Heras M, et al. . HDL triglycerides: a new marker of metabolic and cardiovascular risk. Int J Mol Sci. (2019) 20:3151. 10.3390/ijms20133151 PubMed DOI PMC

Pihl-Jensen G, Tsakiri A, Frederiksen JL. Statin treatment in multiple sclerosis: a systematic review and meta-analysis. CNS Drugs. (2015) 29:277–91. 10.1007/s40263-015-0239-x PubMed DOI

Kamm CP, El-Koussy M, Humpert S, Findling O, Burren Y, Schwegler G, et al. . Atorvastatin added to interferon beta for relapsing multiple sclerosis: 12-month treatment extension of the randomized multicenter SWABIMS trial. PLoS ONE. (2014) 9:e86663. 10.1371/journal.pone.0086663 PubMed DOI PMC

Waddington K, Papadaki A, Coelewij L, Adriani M, Nytrova P, Kubala H, et al. Using serum metabolomics to predict development of anti-drug antibodies in multiple sclerosis patients treated with IFNβ. Mendeley Data V1. (2020). 10.17632/jbjh3gmknw.1 PubMed DOI PMC

Dysregulated lipid metabolism networks modulate T-cell function in people with relapsing-remitting multiple sclerosis

Blood metabolomic and transcriptomic signatures stratify patient subgroups in multiple sclerosis according to disease severity

Using Serum Metabolomics to Predict Development of Anti-drug Antibodies in Multiple Sclerosis Patients Treated With IFNβ