EditorJosef S Smolen (Austria)Associate EditorsFrancis Berenbaum (France)Dimitrios Boumpas (Greece)Gerd Burmester (Germany)Mary Crow (USA)Iain McInnes (UK)Thomas Pap (Germany)David Pisetsky (USA)Désirée van der Heijde (The Netherlands)Kazuhiko Yamamoto (Japan)Methodological and Statistical AdvisorStian Lydersen (Norway)Social Media AdvisorsAlessia Alunno (Italy)Mary Canavan (Ireland)Meghna Jani (UK) Elena Nikiphorou (UK)Caroline Ospelt (Switzerland)Christophe Richez (France)Paul Studenic (Austria)

Guidelines for Authors and ReviewersFull instructions are available online at http://ard.bmj.com/ pages/authors. Articles must be submitted electronically at http://mc.manuscriptcentral. com/ard. Authors retain copyright but are required to grant ARD an exclusive licence to publish. (http://authors.bmj.com/ policies/copyright-and-authors-rights).

Annals of the Rheumatic Diseases publishes original work on all aspects of rheumatology and disorders of connective tissue. Laboratory and clinical studies are equally welcomeEditorial Board

Chairman of Advisory CommitteeJohannes Bijlsma (The Netherlands)

Advisory CommitteeFerry Breedveld (The Netherlands)Marco Matucci Cerinic (Italy)Michael Doherty (UK)Maxime Dougados (France)Paul Emery (UK)Daniel Furst (USA) Steffen Gay (Switzerland) Marc Hochberg (USA)Joachim Kalden (Germany)Edward Keystone (Canada)Lars Klareskog (Sweden)Tore Kvien (Norway)

Zhan-guo Li (China)Peter Lipsky (USA)Sir Ravinder Maini (UK)Emilio Martín-Mola (Spain)Haralampos Moutsopoulos(Greece)Karel Pavelka (Czech Republic)Yehuda Shoenfeld (Israel)Leo van de Putte (The Netherlands) Frank Wollheim (Sweden)Anthony Woolf (UK)

Daniel Aletaha (Austria) Johan Askling (Sweden) Sang-Cheol Bae (Korea) Xenofon Baraliakos (Germany) Anne Barton (UK) Maarten Boers (The Netherlands) Matthew Brown (Australia) Maya Buch (UK) Loreto Carmona (Spain) Carlo Chizzolini (Switzerland) Bernard Combe (France) Philip Conaghan (UK) Maurizio Cutolo (Italy) José da Silva (Portugal) Nicola Dalbeth (Australia) Christian Dejaco (Austria) Oliver Distler (Switzerland) Thomas Dörner (Germany) Dirk Elewaut (Belgium) Axel Finckh (Switzerland) Roy Fleischmann (USA) Mary Goldring (USA) Juan Gomez-Reino (Spain) Laure Gossec (France) Walter Grassi (Italy) Ahmet Gül (Turkey) Frederic Houssiau (Belgium) Tom Huizinga (The Netherlands) Arthur Kavanaugh (USA)

Robert Landewé (The Netherlands) Rik Lories (Belgium) Ingrid Lundberg (Sweden) Gary MacFarlane (UK) Xavier Mariette (France) Alberto Martini (Italy) Dennis McGonagle (UK) Fred Miller (USA) Peter Nash (Australia) Michael Nurmohamed (The Netherlands) Caroline Ospelt (Switzerland) Monika Østensen (Norway) Constatino Pitzalis (UK) Jane Salmon (USA) Georg Schett (Germany) Hendrik Schulze-Koops (Germany) Nan Shen (China) Alexander So (Switzerland) Zoltan Syekanecz (Hungary) Hiroshi Takayanagi (Japan) Tsutomu Takeuchi (Japan) Yoshiya Tanaka (Japan) Ronald van Vollenhoven (Sweden) Dimitrios Vassilopoulos (Greece) Douglas Veale (Ireland) Jiri Vencovsky (Czech Republic) Erwin Wagner (Spain) Michael Ward (USA) Kevin Winthrop (USA)

Contact DetailsEditorial OfficeAnnals of the Rheumatic DiseasesBMJ Journals, BMA House, Tavistock SquareLondon WCIH 9JR,UKE: ard@bmj.com

Production EditorTeresa JobsonE: production.ard@bmj.com

EULAREULAR Executive SecretariatSeestrasse 240, 8802 Kilchberg, SwitzerlandE: eular@eular.orgwww.eular.org

Customer supportFor general queries and support with existing and new subscriptions:W: support.bmj.comT: +44 (0)20 7111 1105E: support@bmj.com

Self-archiving and permissionsW: bmj.com/company/products-services/rights-and-licensing/E: bmj.permissions@bmj.com

AdvertisingW: bmj.com/company/for-advertisers- and-sponsor/

Display Advertising ROWSophie FitzsimmonsT: +44 (0)20 3655 5612E: sfitzsimmons@bmj.com

Online Advertising ROWMarc CliffordT: +44 (0) 20 3655 5610E: mclifford@bmj.com

Display & Online Advertising AmericasAmerican Medical Communications (AMC) T: +1 973 214 4374E: rgordon@americanmedicalcomm.com

Reprints

Author ReprintsBMJ Reprints TeamE: admin.reprints@bmj.com

Commercial Reprints ROWNadia Gurney-RandallM: +44 07866 262344E: ngurneyrandall@bmj.com

Commercial Reprints AmericasRay ThibodeauT: +1 267 895 1758M: +1 215 933 8484E: ray.thibodeau@contentednet.com

For all other journal contactsard.bmj.com/contact-us

Subscription Information

ARD is published monthly; subscribers receive all supplements ISSN 0003-4967 (print); 1468-2060 (online)

Institutional Rates 2019Print£1036

OnlineSite licences are priced on FTE basis and allow access by the whole institution. Details available online at http://journals.bmj.com/content/subscribers or contact the Subscription Manager in the UK (see above right)

Personal Rates 2019Print (includes online access at no additional cost)£428

Online only£182

EULAR congress delegatesDelegates receive a Continuous Professional Development package that includes a 12 month complimentary subscription to ARD in print and/or online

Personal print or online only and institutional print subscriptions may be purchased online at http://journals.bmj.com/content/ subscribers (payment by Visa/Mastercard only)

Residents of some EC countries must pay VAT; for details, call us or visit http://journals.bmj.com/content/subscribers

For more information on subscription rates or to subscribe online please visit ard/bmj.com/pages/contact-us/

ARDThe EULAR Journal

Editorial1699 Therapeutic innovation in adult-onset

Still’s disease (and other rare inflammatory disorders): how to secure evidence-based medicine?P Guilpain, A Le Quellec, A T J Maria

Views on news1702 Extraintestinal translocation of microbes and

tissue specificity in rheumatic musculoskeletal disease (RMD): its more than a gut feelingT R D J Radstake

Clinical and epidemiological research1705 Characteristics of difficult-to-treat rheumatoid

arthritis: results of an international surveyN M T Roodenrijs, M J H de Hair, M C van der Goes, J W G Jacobs, P M J Welsing, D van der Heijde, D Aletaha, M Dougados, K L Hyrich, I B McInnes, U Mueller-Ladner, L Senolt, Z Szekanecz, J M van Laar, G Nagy, On behalf of the whole EULAR Task Force on development of EULAR recommendations for the comprehensive management of difficult-to-treat rheumatoid arthritis

1710 Canakinumab in patients with systemic juvenile idiopathic arthritis and active systemic features: results from the 5-year long-term extension of the phase III pivotal trialsN Ruperto, H I Brunner, P Quartier, T Constantin, N M Wulffraat, G Horneff, O Kasapcopur, R Schneider, J Anton, J Barash, R Berner, F Corona, R Cuttica, M Fouillet-desjonqueres, M Fischbach, H E Foster, D Foell, S C Radominski, A V Ramanan, R Trauzeddel, E Unsal, J Levy, E Vritzali, A Martini, D J Lovell, On behalf of the Paediatric Rheumatology International Trials Organisation (PRINTO) and the Pediatric Rheumatology Collaborative Study Group (PRCSG)

1720 Tocilizumab in patients with adult-onset still’s disease refractory to glucocorticoid treatment: a randomised, double-blind, placebo-controlled phase III trialY Kaneko, H Kameda, K Ikeda, T Ishii, K Murakami, H Takamatsu, Y Tanaka, T Abe, T Takeuchi

1730 Reliability of a consensus-based ultrasound definition and scoring for enthesitis in spondyloarthritis and psoriatic arthritis: an OMERACT US initiativeP V Balint, L Terslev, P Aegerter, G A W Bruyn, I Chary-Valckenaere, F Gandjbakhch, A Iagnocco, S Jousse-Joulin, I Möller, E Naredo, W A Schmidt, R J Wakefield, M-A D'Agostino, on behalf of the OMERACT Ultrasound Task Force members

1736 Can disease activity in patients with psoriatic arthritis be adequately assessed by a modified Disease Activity index for PSoriatic Arthritis (DAPSA) based on 28 joints?B Michelsen, J Sexton, J S Smolen, D Aletaha, N S Krogh, D van der Heijde, T K Kvien, M L Hetland

1742 Effect of in utero hydroxychloroquine exposure on the development of cutaneous neonatal lupus erythematosusJ Barsalou, N Costedoat-Chalumeau, A Berhanu, C Fors-Nieves, U Shah, P Brown, C A Laskin, N Morel, K Levesque, J P Buyon, E D Silverman, P M Izmirly

1750 Incidence, prevalence and treatment burden of polymyalgia rheumatica in the UK over two decades: a population-based studyR J Partington, S Muller, T Helliwell, C D Mallen, A Abdul Sultan

1757 Etanercept in patients with inflammatory hand osteoarthritis (EHOA): a multicentre, randomised, double-blind, placebo-controlled trialM Kloppenburg, R Ramonda, K Bobacz, W-Y Kwok, D Elewaut, T W J Huizinga, F P B Kroon, L Punzi, J S Smolen, B Vander Cruyssen, R Wolterbeek, G Verbruggen, R Wittoek

Basic and translational research1765 Ectopic RASGRP2 (CalDAG-GEFI) expression

in rheumatoid synovium contributes to the development of destructive arthritisH Nakamura, S Shimamura, S Yasuda, M Kono, M Kono, Y Fujieda, M Kato, K Oku, T Bohgaki, T Shimizu, N Iwasaki, T Atsumi

Contents Volume 77 Issue 12 | ARD December 2018ARDThe Eular Journal

EditorJosef S Smolen

Associate EditorsFrancis BerenbaumDimitrios BoumpasGerd BurmesterMary CrowIain McInnesThomas PapDavid PisetskyDésirée van der HeijdeKazuhiko Yamamoto

Editorial officeAnnals of the Rheumatic DiseasesBMJ Publishing Group LtdBMA HouseTavistock SquareLondon WCIH 9JR,UKT: +44 (0)20 3655 5889E: ard@bmj.comTwitter: @ARD_BMJISSN: 0003-4967 (print)ISSN: 1468-2060 (online)Impact Factor: 12.350

Disclaimer: ARD is owned and published by BMJ Publishing Group Ltd (a wholly owned subsidiary of the British Medical Association) and the European League Against Rheumatism. The owners grant editorial freedom tothe Editor of ARD. ARD follows guidelines on editorial independence produced by the World Association of Medical Editors and the code on good publication practice of the Committee on Publication Ethics.ARD is intended for medical professionals and is provided without warranty, express or implied. Statements in the journal are the responsibility of their authors and advertisers and not authors’ institutions the BMJ Publishing Group, the European League Against Rheumatism or the BMA unless otherwise specified or determined by law. Acceptance of advertising does not imply endorsement.To the fullest extent permitted by law, the BMJ Publishing Group shall not be liable for any loss, injury or damage resulting from the use of ARD or any information in it whether based on contract, tort, or otherwise. Readers are advised to verify any information they choose to rely on.Copyright: © 2018 BMJ Publishing Group and European League Against Rheumatism. All rights reserved; no part of this publication may be reproduced stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying recording, or otherwise without prior permission

ARD is published by BMJ Publishing Group Ltd typeset by Exeter Premedia Services Private Ltd, Chennai, India and printed in the UK on acid-free paper.

Annals of the Rheumatic Diseases (ISSN No: 0003-4967) is published monthly by BMJ Publishing Group and distributed in the USA by Air Business Ltd. Periodicals postage paid at Jamaica NY 11431 POSTMASTER: send address changes to Annals of the Rheumatic Diseases, Air Business Ltd, c/o Worldnet Shipping Inc., 156-15, 146th Avenue, 2nd Floor, Jamaica, NY 11434, USA.

December 2018 Volume 77 Issue 12ImpactFactor

12.350

MORE CONTENTS

This journal is a member of and subscribes to the principles of the Committee on Publication Ethicshttp://publicationethics.org/

This article has been made freely available online under the BMJ Journals open access scheme. See http://authors.bmj.com/open-access/

This article has been chosen by the Editor to be of special interest or importance and is freely available online.

Member since 2008 JM00004

Contents Volume 77 Issue 12 | ARD December 2018

1773 Mast cells in early rheumatoid arthritis associate with disease severity and support B cell autoantibody productionF Rivellese, D Mauro, A Nerviani, S Pagani, L Fossati-Jimack, T Messemaker, F A S Kurreeman, R E M Toes, A Ramming, S Rauber, G Schett, G W Jones, S A Jones, F W Rossi, A de Paulis, G Marone, M E M El Shikh, F Humby, C Pitzalis

1782 Intrarenal activation of adaptive immune effectors is associated with tubular damage and impaired renal function in lupus nephritisC Pamfil, Z Makowska, A De Groof, G Tilman, S Babaei, C Galant, P Montigny, N Demoulin, M Jadoul, S Aydin, R Lesche, F McDonald, F A Houssiau, B R Lauwerys

1790 Cleaved N-terminal histone tails distinguish between NADPH oxidase (NOX)-dependent and NOX-independent pathways of neutrophil extracellular trap formationE Pieterse, N Rother, C Yanginlar, J Gerretsen, S Boeltz, L E Munoz, M Herrmann, P Pickkers, L B Hilbrands, J van der Vlag

1799 Increased autophagy is cytoprotective against podocyte injury induced by antibody and interferon-α in lupus nephritisY Qi, X Zhou, F Cheng, P Hou, Y Ren, S Wang, M Zhao, L Yang, J Martinez, H Zhang

1810 Galectin-9 is an easy to measure biomarker for the interferon signature in systemic lupus erythematosus and antiphospholipid syndromeL L van den Hoogen, J A G van Roon, J S Mertens, J Wienke, A P Lopes, W de Jager, M Rossato, A Pandit, C G K Wichers, F van Wijk, R D E Fritsch-Stork, T R D J Radstake

1815 Interleukin 12 and interleukin 23 play key pathogenic roles in inflammatory and proliferative pathways in giant cell arteritisR Conway, L O'Neill, G M McCarthy, C C Murphy, A Fabre, S Kennedy, D J Veale, S M Wade, U Fearon, E S Molloy

1825 Dysregulated neutrophil responses and neutrophil extracellular trap formation and degradation in PAPA syndromeP Mistry, C Carmona-Rivera, A K Ombrello, P Hoffmann, N L Seto, A Jones, D L Stone, F Naz, P Carlucci, S Dell’Orso, G Gutierrez-Cruz, H-W Sun, D L Kastner, I Aksentijevich, M J Kaplan

Letters1835 Frequency of tumour necrosis factor alpha

receptor superfamily 1A multiple sclerosis-associated variants in patients with rheumatoid arthritis with anti-tumour necrosis factor therapy-related demyelinating complicationsS Bitoun, C Miceli-Richard, C Verstuyft, P A Juge, P Dieudé, J-M Berthelot, C Richez, C Cauquil, C Sordet, S Melac-Ducamp, L Gossec, B Bouvard, E Dernis, E Houvenagel, M-A Boutry-Bacle, X Mariette, R Seror

1836 In RA, becoming seronegative over the first year of treatment does not translate to better chances of drug-free remissionE C de Moel, V F A M Derksen, L A Trouw, H Bang, Y P M Goekoop-Ruiterman, G M Steup-Beekman, T W J Huizinga, C F Allaart, R E M Toes, D van der Woude

1838 Short-term and low-dose IL-2 therapy restores the Th17/Treg balance in the peripheral blood of patients with primary Sjögren’s syndromeM Miao, Z Hao, Y Guo, X Zhang, S Zhang, J Luo, X Zhao, C Zhang, X Liu, X Wu, D Xu, J Zhao, X Lu, C Gao, X Li

1840 Clinical associations and expression pattern of the autoimmunity susceptibility factor DIORA-1 in patients with primary Sjögren’s syndromeL A Aqrawi, L Mentlein, L Meneghel, A Björk, G E Thorlacius, M Ivanchenko, J I Ramírez Sepúlveda, K Skarstein, M Kvarnström, S Brauner, A Espinosa, M Wahren-Herlenius

1842 Increase in circulating cells coexpressing M1 and M2 macrophage surface markers in patients with systemic sclerosisS Soldano, A C Trombetta, P Contini, V Tomatis, B Ruaro, R Brizzolara, P Montagna, A Sulli, S Paolino, C Pizzorni, V Smith, M Cutolo

1845 Association of circulating CXCL10 and CXCL11 with systemic sclerosisC Crescioli, C Corinaldesi, V Riccieri, V Raparelli, M Vasile, F Del Galdo, G Valesini, A Lenzi, S Basili, C Antinozzi

Electronic pagese84 Calprotectin is not independent from baseline

erosion in predicting radiological progression in early rheumatoid arthritis. Comment on ‘Calprotectin as a marker of inflammation in patients with early rheumatoid arthritis’ by Jonsson et alM Chevreau, M-H Paclet, X Romand, J-L Quesada, O Vittecoq, P Dieudé, B Toussaint, P Gaudin, A Baillet

e85 Response to: ‘Calprotectin is not independent from baseline erosion in predicting radiological progression in early rheumatoid arthritis’ by Chevreau et alM K Jonsson, N P Sundlisæter, H H Nordal, H B Hammer, A-B Aga, D van der Heijde, T K Kvien, B-T S Fevang, S Lillegraven, E A Haavardsholm

e86 Simultaneous inhibition of α4/β7 integrin and tumour necrosis factor-α in concomitant spondyloarthritis and inflammatory bowel diseaseN Richard, E M Hazel, N Haroon, R D Inman

Contents Volume 77 Issue 12 | ARD December 2018

e87 Response to: ‘Simultaneous inhibition of α4/β7 integrin and tumor necrosis factor-α in concomitant spondyloarthritis and inflammatory bowel disease’ by Richard et alG Varkas, D Elewaut

e88 Standard dose of Ustekinumab for childhood-onset deficiency of interleukin–36 receptor antagonistB Cherqaoui, L Rossi-Semerano, M Piram, I Koné-Paut

e89 Response to: 'Standard dose of ustekinumab for childhood-onset deficiency of interleukin-36 receptor antagonist’ by Cherqaoui et alN Bonekamp, R Caorsi, J Frenkel, M Gattorno

e90 Performance of the 2017 European League Against Rheumatism/American College of Rheumatology classification criteria for adult and juvenile idiopathic inflammatory myopathies in clinical practiceA Hočevar, Z Rotar, M Krosel, S Čučnik, M Tomšič

e91 Response to: ‘Performance of the 2017 European League Against Rheumatism/American College of Rheumatology classification criteria for adult and juvenile idiopathic inflammatory myopathies in clinical practice’ by Hočevar et alI E Lundberg, M Bottai, A Tjärnlund

1699Guilpain P, et al. Ann Rheum Dis December 2018 Vol 77 No 12

Therapeutic innovation in adult-onset Still’s disease (and other rare inflammatory disorders): how to secure evidence-based medicine?Philippe Guilpain,1,2,3 Alain Le Quellec,1,2 Alexandre Thibault Jacques Maria1,2,3

Adult-onset Still’s disease (AOSD) is a rare inflammatory disorder with heteroge-neous clinical presentation and unspecific features (spiking fever, pharyngitis, arthritis, skin rash with elevated acute phase reactants).1 Its diagnosis is one of exclusion, and necessitates ruling out many other conditions, notably neoplastic and infectious ones. Based on clinical experience and literature, a ‘dichotomous view’ of AOSD is emerging, with two distinct AOSD clinical subtypes, poten-tially requiring distinct treatments1 2: a non-Mendelian autoinflammatory disease with pre-eminent systemic symptoms and intense inflammatory state, sometimes associated with life-threatening complica-tions such as reactive haemophagocytic lymphohistiocytosis, and a rheumatic disease with pre-eminent chronic polyar-thritis and possibly destructive polyar-thritis with lower inflammatory state. Currently, this ‘dichotomous view’ is not only supported by clinical observations, but also by findings on cytokine profiles and responses to biotherapies in some refractory patients.2 Indeed, inter-leukin  (IL)-1β,  IL-6  and  IL-18 would  be associated with ‘systemic AOSD’, whereas tumour  necrosis  factor  alpha  (TNF-β), interferon gamma (IFN-β) and IL-8 would be of greater involvement in ‘rheumatic AOSD’.3–6 Of note, this dichotomous approach remains controversial and the choice of biologics (targeting ‘appropriate’ cytokines) is still empirical in refractory patients. One could hope that cytokine

monitoring provides information towards a personalised approach. However, such immunobiological tools are technically difficult to develop for daily practice and remains phantasmagorias even though they would be very useful at the time when several biotherapies are already success-fully used or are under evaluation in clin-ical trials (https:// clinicaltrials. gov/).

In Annals of the Rheumatic Diseases, Gabay et al report the results of a multi-centre, open-label study evaluating the safety and efficacy of the recombinant IL-18-binding  protein  (IL-18BP),  tade-kinig alfa, in patients with difficult-to-treat AOSD.7  IL-18 was  indeed  reported increased in patients with AOSD and systemic onset juvenile idiopathic arthritis. This work provides the first demonstra-tion of the therapeutic benefits obtained with IL-18BP in AOSD. Hence, this study represents a proof-of-concept suggesting that  IL-18  inhibition  could  be  a  thera-peutic option in this systemic inflamma-tory disease, with a reasonably acceptable safety profile. Actually, the drug appears mildly efficient, in a subgroup of incom-pletely controlled and moderately severe AOSD. So, at this stage, tadekinig may represent an interesting tool for some patients with AOSD. One could hope that  IL-18  blockade  would  provide greater benefit in a subgroup of patients with systemic form of AOSD rather than articular forms (herein recruited for this study). Thus, further trials are required to establish  IL-18BP at  its  true place within therapeutic armamentarium.

Undoubtedly, the current study suffers from several methodological limitations (including lack of control group, small number of included subjects and unbal-anced groups of patients). These limita-tions are linked both to the open-label, uncontrolled, design of the study and also to the rarity and heterogeneity of AOSD. This is usual in AOSD and should not limit the scientific interest of physi-cians for new biologics in this condition.

Efficacy of anti-IL-1 agents (ie, anakinra8 and canakinumab9) was demonstrated in randomised controlled trials (RCT) in systemic-onset juvenile idiopathic arthritis, which is usually considered as the juvenile equivalent of AOSD. Since then, canakinumab10–12 and anakinra13–19 accumulated demonstration of their effi-cacy in case reports or series on AOSD, but not in RCT. At the end of a chase, both biologics obtained authorisation, which allows more easily their beneficial use for some patients. At this point of discussion, the therapeutic innovation and the great interest exhibited by pharmaceu-tical industry in this rare disease should be commended. However, rare diseases (such as AOSD) may also be economical niches for industrials, and this represents a risk for trials (and drug development) of being insufficiently relevant to the goals of patient’s management.

Medical necessities and economical strategies should not be opposing players and must act together to promote thera-peutic innovation, a real challenge, partic-ularly in rare diseases.20 21 Of course, the development of a new drug requires considering economical standpoints. In this complex interplay, the governing law of clinical trials should be carefully handled since it may represent a double-edged sword for physicians. On the one hand, physicians should be in a position to provide precise and contemporary definitions of both disease and treatment goals, but on the other hand, they should not blindly and suddenly abandon their conception of disease (definitions of the disease, remission, refractory forms and so on) and change their practice without taking a critical distance. An example may be found within another inflamma-tory disease, giant cell arteritis (GCA). Since decades, physicians are in search of steroid-sparing agents and also new drugs for refractory forms of GCA. Responding to this medical necessity, pharmaceutical industry recently provided new drugs such  as  tocilizumab,  an  IL-6  inhibitor.22 Thus, tocilizumab proved to be effective in reducing vascular inflammation in GCA and should be useful in so many patients. Nevertheless, physicians should not forget the initial goal of therapy in this disease and ask themselves: how have we been managing steroids tapering in GCA since decades? Meeting how many real fail-ures? And which risk for each individual patient?  Anti-IL-6  agents  may  be  great molecules, but do they represent a genuine revolution for every patient with GCA or only a subgroup? So, the question would be: which subgroup(s) to target?23–25 One

1Medical School, Montpellier University, Montpellier, France2Department of Internal Medicine – Multi-Organic Diseases, Local Referral Center for Autoimmune Diseases, University of Montpellier, Saint-Eloi Hospital, Montpellier, France3Institute for Regenerative Medicine and Biotherapy (IRMB), Inserm, U1183, Saint-Eloi Hospital, Montpellier, France

Correspondence to Dr Philippe Guilpain, Department of internal medicine, Multi-Organic Diseases, St Eloi Hospital, CHRU de Montpellier, Montpellier, F-34295, France; p- guilpain@ chu- montpellier. fr

Editorial

1700 Guilpain P, et al. Ann Rheum Dis December 2018 Vol 77 No 12

Editorial

shall keep this perspective in mind when attempting to manage such diseases and before changing practice or the whole paradigm. Generally speaking, we should reconcile medical and economical necessi-ties, in order to benefit from therapeutic innovation without overlooking the lessons from the past.

Indeed, an emerging problem could be a discrepancy between economical stand-points and scientific objectives serving patients’ needs. So, the emerging question may become: ‘what really drives thera-peutic innovation?’ The unpredictability and versatility of pharmaceutical industry in the last decades, mentioned by some authors,26 appears quite disconcerting both for physicians and patients and may contribute to a loss of confidence and eventually a distrust of industrial partners. In this context, the physicians, as well as the health authorities and managers, should be very careful with the results of trials and their applications in everyday practice.

This careful consideration of ‘evidence-based medicine’ (EBM) is even more complex in AOSD, since recommenda-tions are lacking and merely impossible to establish because of methodological limita-tions inherent to this rare and heteroge-neous disease.27 In order to summarise the peculiar challenge regarding AOSD, here are some issues: first, the definition of AOSD is vague, since it is not based on definite histopathological demonstra-tion of tissue lesions, and leads to some uncertainty in diagnosis; second, in daily practice, diagnosis is based on criteria,2829 which were developed for classifica-tion; third, strong biomarkers are not available and even glycosylated ferritin appears a disappointing marker in prac-tice30; fourth, prognostic scores (such as Pouchot’s  ‘systemic  score’)  remain  to  be validated in larger cohorts.31 32 So, what is the goal of therapy in AOSD and how to define a primary end-point in clinical trials? Since AOSD is very heterogeneous both in presentation and disease course, which subgroup of patients should be targeted? More precisely, we think it is very important to figure out which patients are treated in such trials. This may be quite difficult for readers and reviewers to find this information in the published studies. For instance, is the population made of ‘rheumatoid’ or ‘systemic’ AOSD? What is the disease course: first flare? polycyclic or chronic articular disease? These questions are not anecdotal. A third of patients may experience one self-limited disease flare, which may sometimes be easily managed by non-steroidal anti-inflammatory drugs

or corticosteroids.31 So, is it legitimate to develop new approaches for these patients? Should not we focus on refrac-tory patients with AOSD, the one with true unmet medical need? Obviously, all these considerations may affect the rele-vance of the results obtained from clin-ical trials and one can easily imagine the hardness to transpose such results into everyday practice.

Anyhow, we physicians have to promote research and innovation. We also have to convince health authorities of the neces-sity of innovative drugs. Trials are useful and provide demonstration. Considering ethical aspects, we also owe patients to develop innovative strategies to cure the most refractory forms of AOSD. In that sense,  IL-18BP  may  be  such  an  alterna-tive therapeutic option for some patients with difficult-to-treat disease. So, even if the study by Gabay et al is not perfect and may appear methodologically controver-sial, we should be delighted by the devel-opment of new effective drugs in AOSD. However, considering the above-men-tioned comments, it is very uncertain that the required further studies testing IL-18BP  in  AOSD  will  be  conducted by industrials. Targeting a subgroup of patients within AOSD niche may not be a good strategy for them. We physicians do not know much about that. Industrials and we stem from very different places, but we are partners, acting for the devel-opment of therapies. As already claimed by many others, we should work together while keeping in mind the potential sponsorship bias associated with indus-trial funded trials. Beyond this common assumption, we could obviously provide something very useful for the relevance of future trials in AOSD, that is, method-ological recommendations. Consequently, we should combine efforts at national and international levels to develop inde-pendent institutional research in AOSD and better apprehend this multifaceted disease. Only one such advance will allow us to improve the relevance of trials and further secure EBM in this rare entity, and thus refine the use of innovative drugs in everyday practice.

Handling editor Josef S Smolen

Funding The author has not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

Competing interests PG is a medical expert for LFB (Laboratoire Français du Biofractionnement) and has received fees from Abbvie, Actelion, Boehringer Ingelheim France, Bouchara-Recordati, Novartis, Pfizer, and Roche in the last 5 years. ALeQ has received fees from Actelion, Bouchara-Recordati, GlaxoSmithKline, Intermune, LFB, Novartis, Pfizer, Roche, Shire and

Zambon in the last 5 years. ATJM has received fees from Abbvie, Actelion, CSL Behring, Experf, Novartis and Shire in the last 5 years.

Provenance and peer review Commissioned; externally peer reviewed.

© Article author(s) (or their employer(s) unless otherwise stated in the text of the article) 2018. All rights reserved. No commercial use is permitted unless otherwise expressly granted.

To cite Guilpain P, Le Quellec A, Maria ATJ. Ann Rheum Dis 2018;77:1699.

► http:// dx. doi. org/ 10. 1136/ annrheumdis- 2017- 212608

Ann Rheum Dis 2018;77:1699.doi:10.1136/annrheumdis-2018-213106

RefeRences 1 Gerfaud-Valentin M, Jamilloux Y, Iwaz J, et al. Adult-

onset Still’s disease. Autoimmun Rev 2014;13:708–22. 2 Maria AT, Le Quellec A, Jorgensen C, et al. Adult onset

Still’s disease (AOSD) in the era of biologic therapies: dichotomous view for cytokine and clinical expressions. Autoimmun Rev 2014;13:1149–59.

3 Efthimiou P, Moorthy LN, Mavragani CP, et al. Adult Onset Still’s Disease and Autoinflammation. Int J Inflam 2012;2012:1–4.

4 Fujii T, Nojima T, Yasuoka H, et al. Cytokine and immunogenetic profiles in Japanese patients with adult Still’s disease. Association with chronic articular disease. Rheumatology 2001;40:1398–404.

5 Ichida H, Kawaguchi Y, Sugiura T, et al. Clinical manifestations of Adult-onset Still’s disease presenting with erosive arthritis: Association with low levels of ferritin and Interleukin-18. Arthritis Care Res 2014;66:642–6.

6 Saiki O, Uda H, Nishimoto N, et al. Adult Still’s disease reflects a Th2 rather than a Th1 cytokine profile. Clin Immunol 2004;112:120–5.

7 Gabay C, Fautrel B, Rech J, et al. Open-label, multicentre, dose-escalating phase II clinical trial on the safety and efficacy of tadekinig alfa (IL-18BP) in adult-onset Still’s disease. Ann Rheum Dis 2018;77:840–7.

8 Quartier P, Allantaz F, Cimaz R, et al. A multicentre, randomised, double-blind, placebo-controlled trial with the interleukin-1 receptor antagonist anakinra in patients with systemic-onset juvenile idiopathic arthritis (ANAJIS trial). Ann Rheum Dis 2011;70:747–54.

9 Ruperto N, Brunner HI, Quartier P, et al. Two randomized trials of canakinumab in systemic juvenile idiopathic arthritis. N Engl J Med 2012;367:2396–406.

10 Barsotti S, Neri R, Iacopetti V, et al. Successful treatment of refractory adult-onset still disease with canakinumab: a case report. J Clin Rheumatol 2014;20:121.

11 Kontzias A, Efthimiou P. The use of Canakinumab, a novel IL-1β long-acting inhibitor, in refractory adult-onset Still’s disease. Semin Arthritis Rheum 2012;42:201–5.

12 Lo Gullo A, Caruso A, Pipitone N, et al. Canakinumab in a case of adult onset still’s disease: efficacy only on systemic manifestations. Joint Bone Spine 2014;81:376–7.

13 Kalliolias GD, Georgiou PE, Antonopoulos IA, et al. Anakinra treatment in patients with adult-onset Still’s disease is fast, effective, safe and steroid sparing:

1701Guilpain P, et al. Ann Rheum Dis December 2018 Vol 77 No 12

Editorial

experience from an uncontrolled trial. Ann Rheum Dis 2007;66:842–3.

14 Lequerré T, Quartier P, Rosellini D, et al. Interleukin-1 receptor antagonist (anakinra) treatment in patients with systemic-onset juvenile idiopathic arthritis or adult onset Still disease: preliminary experience in France. Ann Rheum Dis 2008;67:302–8.

15 Laskari K, Tzioufas AG, Moutsopoulos HM. Efficacy and long-term follow-up of IL-1R inhibitor anakinra in adults with Still’s disease: a case-series study. Arthritis Res Ther 2011;13:R91.

16 Nordström D, Knight A, Luukkainen R, et al. Beneficial effect of interleukin 1 inhibition with anakinra in adult-onset Still’s disease. An open, randomized, multicenter study. J Rheumatol 2012;39:2008–11.

17 Giampietro C, Ridene M, Lequerre T, et al. Anakinra in adult-onset Still’s disease: long-term treatment in patients resistant to conventional therapy. Arthritis Care Res 2013;65:822–6.

18 Hong D, Yang Z, Han S, et al. Interleukin 1 inhibition with anakinra in adult-onset Still disease: a meta-analysis of its efficacy and safety. Drug Des Devel Ther 2014;8:2345–57.

19 Ortiz-Sanjuán F, Blanco R, Riancho-Zarrabeitia L, et al. Efficacy of Anakinra in Refractory Adult-Onset Still’s Disease: Multicenter Study of 41 Patients and Literature Review. Medicine 2015;94:e1554.

20 Smith BP. Challenges and Opportunities in Rare Disease Drug Development. Clin Pharmacol Ther 2016;100:312–4.

21 Harari S, Humbert M, Blasi F, et al. Rare pulmonary diseases and orphan drugs: where do we stand and where are we going to? Eur Respir Rev 2015;24:375–7.

22 Stone JH, Tuckwell K, Dimonaco S, et al. Trial of Tocilizumab in Giant-Cell Arteritis. N Engl J Med 2017;377:317–28.

23 Akiyama M. Trial of Tocilizumab in Giant-Cell Arteritis. N Engl J Med 2017;377:1493–4.

24 Leuchten N, Aringer M. Tocilizumab in the treatment of giant cell arteritis. Immunotherapy 2018;10:465–72.

25 Samson M, Bonnotte B. [Does tocilizumab mean the end of glucocorticoids in giant cell arteritis?]. Rev Med Interne 2018;39:75–7.

26 Honig P, Huang SM. Intelligent pharmaceuticals: beyond the tipping point. Clin Pharmacol Ther 2014;95:455–9.

27 Guilpain P, Le Quellec A. About the complexity of adult onset Still’s disease… and advances still required for its management. BMC Med 2017;15:5.

28 Yamaguchi M, Ohta A, Tsunematsu T, et al. Preliminary criteria for classification of adult Still’s disease. J Rheumatol 1992;19:424–30.

29 Fautrel B, Zing E, Golmard JL, et al. Proposal for a new set of classification criteria for adult-onset still disease. Medicine 2002;81:194–200.

30 Lambotte O, Cacoub P, Costedoat N, et al. High ferritin and low glycosylated ferritin may also be a marker of excessive macrophage activation. J Rheumatol 2003;30:1027–8.

31 Pouchot J, Sampalis JS, Beaudet F, et al. Adult Still’s disease: manifestations, disease course, and outcome in 62 patients. Medicine 1991;70:118–36.

32 Ruscitti P, Cipriani P, Masedu F, et al. Adult-onset Still’s disease: evaluation of prognostic tools and validation of the systemic score by analysis of 100 cases from three centers. BMC Med 2016;14:194.

1702 Radstake TRDJ. Ann Rheum Dis 2018;77:1702–1704. doi:10.1136/annrheumdis-2018-213974

Extraintestinal translocation of microbes and tissue specificity in rheumatic musculoskeletal disease (RMD): its more than a gut feeling

Millions of micro-organisms populate our orifices from the moment we are born until we die. The microbiome represents the totality of micro-organisms that coexist with us rather peacefully and which we acquire purposefully from our environment. Health is associated with a balance between commensal and pathogenic microbes, which control the intes-tinal barrier integrity through the production of mucus and lipid metabolites, such as short-chain fatty acids. In the case of a disbalance (dysbiosis), leakage of these microbial products negatively affects the immune system, thereby contributing to a chronic proinflammatory state or worse, cancer. Accu-mulating evidence implicates the microbiota to a plethora of diseases among which rheumatological musculoskeletal disease (RMD).

For instance, the idea that the microbiota is associated with the development of rheumatoid arthritis (RA) has been put forward by several investigators, many of those who have sought the link to intestinal microbes.1–3 On a different take, the oral microbiota, more specifically the presence of Porphyromonas gingivalis, in patients with RA with accompanying periodontitis has been linked to the presence of anticitrullinated protein antibodies.4 In addi-tion, dendritic cells from patients with RA were reported to have a blunted response towards Porphyromonas gingi-valis further supporting the notion of an altered systemic immune response to microbiota in RA.5 More recently, the proposed relation between the microbiota and RMD has been extended to systemic sclerosis (SSc) and primary Sjogrens syndrome (pSS).6 7

Hitherto, all these studies associate gut and/or oral dysbi-osis to pathology—most likely due to the altered produc-tion of metabolites associated by the altered microbiota. However, this puts us with a conundrum, as such postulate does not explain why patients with RA, SSc or pSS present with symmetrical arthritis, organ fibrosis and exocrine gland dysfunction, respectively.

The paradox mentioned above is, however, challenged by a recent study by Manfredo Vieira et al published in Science.8 In this study, the authors hypothesised that the loss of the gut barrier function leads to translocation of pathobionts to organs, thereby leading to autoimmunity. The authors focused on systemic lupus erythematosus, given the rela-tionship between disease pathology, excessive signalling of RNA-sensing Toll-like receptor (TLR7) and the presence of a type I IFN signature in the majority of patients. Lupus prone (NZW × BXSB)F1 mice, at 16 weeks of age, showed marked bacterial growth in the mesenteric veins, mesenteric lymph nodes and liver, followed by the presence in the spleen as well 2 weeks later. Subsequent 16S sequencing revealed transloca-tion of Enterococcus gallinarum to the livers of these mice. Mortality, lupus-related autoantibodies and autoimmune manifestations were relieved in these mice after oral admin-istration or vancomycin or ampicillin. The authors provide multiple molecular mechanisms on how E. gallinarum induces pathology in these mice, for instance via upregulation of Enpp3, a protein known to increase plasmacytoid dendritic

cells considered key cells in lupus via the production of type I IFNs9 10 as well as activation of the aryl hydrocarbon receptor, a known innate antimicrobial defence mechanism potentially via the induction of IL-17 producing T cells.11 Importantly, longitudinal stool samples from patients with systemic lupus erythematosus (SLE) showed increased levels of albumin and calprotectin, reflective of impaired gut barriers function as seen in the lupus prone mice. Moreover, E. gallinarum was readily detected in the livers of patients with SLE and patients with autoimmune hepatitis sharing autoimmune features with the former, and most of these patients showed increased serum levels of antibody titres directed against E. gallinarum. In line with these observa-tions, the same group published a concurrent study in Science Translational Medicine showing that commensal orthologs of the autoantigen Ro60 trigger autoimmunity as observed in SLE.12

Taken together, these studies provide strong rational for a change in the existing paradox where intestinal/oral microbiota elicit distal immunological effects by metabo-lites such as short chain fatty acids. In contrast, these studies indicate that certain microbes might migrate to organs and tissues thereby potentially leading to immune reactions and possibly tissue-specific breach of tolerance or, in other words tissue-specific autoimmunity. In fact, this is in corroboration with earlier studies highlighting the presence of bacterial DNA in RA synovial tissues suggestive for early migration of microbes to the synovial compartment perhaps far before actual disease onset.13 14 There are, however, potential caveats to consider. For instance, the field of human development has witnessed a fundamental paradigm shift in the ‘sterile womb paradigm’ when several studies applied molecular techniques suggesting bacterial communities in the placenta and amniotic fluid.15 16 Today, several studies report evidence for this latter as extremely weak due to issues with regard to contamination of materials with bacteria, use of infe-rior molecular techniques and lack of appropriate controls (reviewed in Ref.17). Moreover, although the experimental models in the papers discussed provide strong evidence for pathobiont translocation, the attempts to find support for this in men is merely suggestive as it does not deal with the big question to whether pathobiont translocation is cause or consequence. Also, the paradox proposed by the work of Manfredo Vieira et al focused primarily on the liver and does not discuss how E. gallinarum would get to the kidney and/or skin, more usual locations or lupus involvement nor does it deal with the fact that in case of a general disturbance of barrier function species other than E. gallinarum would be involved. Last, several studies now provide evidence for the efficacy of faecal microbiota transplant (FMT) in multiple conditions. The FOCUS trial has reported a strong effect in patients with ulcerative colitis (n=81), 27% of the treated patients who received steroid-free remission compared with 8% of the patients treated with placebo.18 It is thought that this is effect is mainly sorted through changes in the gut microbiome leading to changed local immunology and barrier function as well as a change in circulating immune-metab-olites. It is harder to understand that the translocation of pathobionts that have led to autoimmunity is affected by FMT.

Taking these potential caveats into account, it is still tempting to speculate that migration of microbiota to organs and tissues is disease-specific and dependent on host factors, including genetic susceptibility, environment

Views on news

1703Radstake TRDJ. Ann Rheum Dis 2018;77:1702–1704. doi:10.1136/annrheumdis-2018-213974

Views on news

and smoking, thereby contributing to disease heteroge-neity (figure 1). Such considerations are important for the rheumatologist, as many argue that shaping of the micro-biome in disease is worth pursuing. Maybe one has to start in the diseased tissue rather than the gut—it is just a gut feeling.

Timothy R D J Radstake

Correspondence to Dr Timothy R D J Radstake, Department of Rheumatology, Clinical Immunology and laboratory of Translational Immunology, University Medical Center Utrecht, Utrecht, The Netherlands; T.R.D.J.Radstake@umcutrecht.nl

Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

Competing interests None declared.

Patient consent Not required.

Provenance and peer review Not commissioned; internally peer reviewed.

Data sharing statement No additional data are available.

© Author(s) (or their employer(s)) 2018. No commercial re-use. See rights and permissions. Published by BMJ.

To cite Radstake TRDJ. Ann Rheum Dis 2018;77:1702–1704.

Received 21 June 2018Revised 19 September 2018Accepted 22 September 2018Published Online First 8 October 2018

RefeRences 1 Eerola E, Möttönen T, Hannonen P, et al. Intestinal flora in early rheumatoid arthritis.

Br J Rheumatol 1994;33:1030–8.

2 Warden CC. The toxemic factor in rheumatoid arthritis. Cal State J Med 1909;7:299–301.

3 Scher JU, Sczesnak A, Longman RS, et al. Expansion of intestinal Prevotella copri correlates with enhanced susceptibility to arthritis. Elife 2013;2: e01202.

4 Hitchon CA, Chandad F, Ferucci ED, et al. Antibodies to porphyromonas gingivalis are associated with anticitrullinated protein antibodies in patients with rheumatoid arthritis and their relatives. J Rheumatol 2010;37:1105–12.

5 Santegoets KC, Wenink MH, Braga FA, et al. Impaired Porphyromonas gingivalis-Induced Tumor Necrosis Factor Production by Dendritic Cells Typifies Patients With Rheumatoid Arthritis. Arthritis Rheumatol 2016;68:795–804.

6 Volkmann ER, Chang YL, Barroso N, et al. Systemic sclerosis is associated with a unique colonic microbial consortium. Arthritis Rheum 2016;68: 1483–92.

7 Mandl T, Marsal J, Olsson P, et al. Severe intestinal dysbiosis is prevalent in primary Sjögren’s syndrome and is associated with systemic disease activity. Arthritis Res Ther 2017;19:237.

8 Manfredo Vieira S, Hiltensperger M, Kumar V, et al. Translocation of a gut pathobiont drives autoimmunity in mice and humans. Science 2018;359:1156–61.

9 Furuta Y, Tsai SH, Kinoshita M, et al. E-NPP3 controls plasmacytoid dendritic cell numbers in the small intestine. PLoS One 2017;12:e0172509.

10 Crow MK. Type I interferon in lupus. J Immunol 2014;192:5459–68. 11 Schiering C, Wincent E, Metidji A, et al. Feedback control of AHR signalling regulates

intestinal immunity. Nature 2017;542:242–5. 12 Greiling TM, Dehner C, Chen X, et al. Commensal orthologs of the human autoantigen

Ro60 as triggers of autoimmunity in lupus. Sci Transl Med 2018;10: eaan2306.

13 van der Heijden IM, Wilbrink B, Tchetverikov I, et al. Presence of bacterial DNA and bacterial peptidoglycans in joints of patients with rheumatoid arthritis and other arthritides. Arthritis Rheum 2000;43:593–8.

14 Chen T, Rimpiläinen M, Luukkainen R, et al. Bacterial components in the synovial tissue of patients with advanced rheumatoid arthritis or osteoarthritis: analysis with gas chromatography-mass spectrometry and pan-bacterial polymerase chain reaction. Arthritis Rheum 2003;49:328–34.

15 Aagaard K, MA J, Antony KM, et al. The placenta harbors a unique microbiome. Sci Transl Med 2014;6:237.

Figure 1 Microbes populate the intestine, oral cavity and skin. Based on the observations by Manfredo Vieira et al8 it is tempting to speculate that certain events that disturb epithelial barrier defence (eg, infection, trauma) could lead to pathobiont translocation to specific sites and/or organs. Several host factors might favour the translocation of specific microbiota to certain organs thereby eliciting organ/tissue-specific immune reactions and/or autoimmunity.

1704 Radstake TRDJ. Ann Rheum Dis 2018;77:1702–1704. doi:10.1136/annrheumdis-2018-213974

Views on news

16 Collado MC, Rautava S, Aakko J, et al. Human gut colonisation may be initiated in utero by distinct microbial communities in the placenta and amniotic fluid. Sci Rep 2016;6:23129.

17 Perez-Muñoz ME, Arrieta MC, Ramer-Tait AE, et al. A critical assessment of the "sterile womb" and "in utero colonization" hypotheses: implications for research

on the pioneer infant microbiome. Microbiome 2017;5:48.

18 Moayyedi P, Surette MG, Kim PT, et al. Fecal microbiota transplantation induces remission in patients with active ulcerative colitis in a randomized controlled trial. Gastroenterology 2015;149:102–9.

1705Roodenrijs NMT, et al. Ann Rheum Dis 2018;77:1705–1709. doi:10.1136/annrheumdis-2018-213687

Clinical and epidemiological research

ConCise report

Characteristics of difficult-to-treat rheumatoid arthritis: results of an international surveynadia M t roodenrijs,1 Maria J H de Hair,1 Marlies C van der Goes,1 Johannes W G Jacobs,1 paco M J Welsing,1 Désirée van der Heijde,2 Daniel Aletaha,3 Maxime Dougados,4 Kimme L Hyrich,5 iain B Mcinnes,6 Ulf Mueller-Ladner,7 Ladislav senolt,8 Zoltan szekanecz,9 Jacob M van Laar,1 György nagy,10,11 on behalf of the whole eULAr task Force on development of eULAr recommendations for the comprehensive management of difficult-to-treat rheumatoid arthritis

To cite: roodenrijs nMt, de Hair MJH, van der Goes MC, et al. Ann Rheum Dis 2018;77:1705–1709.

Handling editor Gerd r Burmester

► Additional material is published online only. to view please visit the journal online (http:// dx. doi. org/ 10. 1136/ annrheumdis- 2018- 213687).

For numbered affiliations see end of article.

Correspondence tonadia M t roodenrijs, Department of rheumatology & Clinical immunology, University Medical Center Utrecht, Utrecht 3508 GA, the netherlands; n. M. t. roodenrijs@ umcutrecht. nl

this manuscript has been previously presented at eULAr 2018 (roodenrijs nMt, de Hair MJH, Jacobs JWG, et al. op0139 Characteristics of difficult-to-treat rheumatoid arthritis: results of an international survey (abstract). Ann Rheum Dis 2018;77:120.)

received 1 May 2018revised 3 July 2018Accepted 6 August 2018published online First 7 september 2018

© Author(s) (or their employer(s)) 2018. no commercial re-use. see rights and permissions. published by BMJ.

AbsTrACTObjectives patients with difficult-to-treat rheumatoid arthritis (rA) remain symptomatic despite treatment according to current european League Against rheumatism (eULAr) management recommendations. these focus on early phases of the disease and pharmacological management. We aimed to identify characteristics of difficult-to-treat rA and issues to be addressed in its workup and management that are not covered by current management recommendations.Methods An international survey was conducted among rheumatologists with multiple-choice questions on disease characteristics of difficult-to-treat rA. Using open questions, additional items to be addressed and items missing in current management recommendations were identified.results 410 respondents completed the survey: 50% selected disease activity score assessing 28 joints >3.2 or presence of signs suggestive of active disease as characteristics of difficult-to-treat rA; 42% selected fatigue; 48% selected failure to ≥2 conventional synthetic disease-modifying antirheumatic drugs (DMArDs) AnD ≥2 biological/targeted synthetic DMArDs; 89% selected inability to taper glucocorticoids below 5 mg or 10 mg prednisone equivalent daily. interfering comorbidities, extra-articular manifestations and polypharmacy were identified as important issues missing in current management recommendations.Conclusions there is wide variation in concepts of difficult-to-treat rA. several important issues regarding these patients are not addressed by current eULAr recommendations.

InTrOduCTIOnThe European League Against Rheumatism (EULAR) recommendations and the American College of Rheumatology guidelines on manage-ment of rheumatoid arthritis (RA) focus on early phases of the disease and on pharmacological management.1 2 These recommendations suggest intensifying the disease-modifying antirheumatic drug (DMARD) strategy, if improvement or the treatment target is not achieved within 3 or 6 months, respectively. Nevertheless, a significant proportion of patients remains symptomatic after several cycles of treatment, which makes them

difficult to treat; this is a significant clinical problem in daily practice.3

A wide array of potential factors contributes to difficult-to-treat RA, such as DMARD resistance or intolerance, adverse reactions, treatment non-ad-herence and limited drug options due to comorbid-ities. Importantly, patients with RA may also remain symptomatic due to non-inflammatory factors, such as secondary osteoarthritis, pain syndromes, social and occupational decline and coping difficulties. All these may (in different combinations) play a role in individual patients and require specific manage-ment approaches, which should be addressed in management recommendations.

Currently, different concepts exist on difficult-to-treat RA, such as refractory, multidrug resistant or persistent RA, or concepts based on number of failed DMARDs and failed treatment goals.4–7 Depending on the criteria used, the estimated prev-alence of difficult-to-treat RA ranges from 5% to 20%.6

We aimed to identify characteristics of the concept of difficult-to-treat RA and to explore issues to be addressed in its workup and management that are not covered by current EULAR management recommendations.

MeTHOdsOnline survey among rheumatologistsAn online survey (online supplementary file 1, set up by DvdH, GN, JWGJ, JMvL, MJHdH and PMJW) was conducted among rheumatologists (including rheumatologists-in-training). The survey was distributed by email in the network of the authors and by Emerging EULAR Network (EMEUNET) and it was asked to additionally forward it to other rheumatologists. The survey consisted of two ques-tions regarding the background of the respondents (Where do you work? How many RA patients do you treat?).

Four multiple-choice questions addressed the perceived necessity of incorporating the following items, and their cut-offs, into the concept of diffi-cult-to-treat RA: disease activity level; presence of fatigue; number of DMARDs that failed; and inability to taper glucocorticoid (GC) treatment. Only one response option could be selected at

1706 Roodenrijs NMT, et al. Ann Rheum Dis 2018;77:1705–1709. doi:10.1136/annrheumdis-2018-213687

Clinical and epidemiological research

Figure 1 Number of respondents per country. Less than 4 (not shown): Austria, Belarus, Bulgaria, Denmark, Egypt, Estonia, Iceland, Israel, Kenya, Pakistan, Russia, Serbia, Slovakia, Slovenia, Tunisia and Turkey.

each multiple-choice item, which were selected as - according to expert opinion - being among the most frequent and relevant characteristics in clinical practice. Fatigue was selected as one of the most relevant patient reported outcomes in RA.8

Three open questions were: ‘Please define any additional characteristics and suggested criteria for difficult-to-treat RA’; ‘Please mention additional clinical issues or comorbidities to be addressed in the workup and management of these patients’; and ‘Please mention any clinically relevant situations which are not covered by the current RA EULAR recommendations’.

Qualification and quantification of the responses to the open questionsNMTR and MJHdH independently classified the responses to the open questions into categories (online supplementary file 1). This enabled summarising and quantifying. Categories were defined based on the responses that were given to the open ques-tions. One response could fit multiple categories. ‘Other’ was used to classify characteristics that did not fit into one of the categories. Discrepancies in classification between NMTR and MJHdH were resolved by consensus.

statistical analysesAll responses were evaluated using descriptive statistics, performed using IBM SPSS Statistics V.21 software.

resulTsrespondentsFour hundred and ten respondents from 33 countries completed the survey between July 2017 and March 2018. Of the 385 respondents who filled out the name of their country, 96% was European (figure 1). Twenty-five per cent of the respondents (n=7 missing) treated <100 unique patients with RA, 42% 100–300 patients and 32% >300 patients.

selected difficult-to-treat rA disease characteristicsFifty per cent of respondents selected ‘disease activity score assessing 28 joints using erythrocyte sedimentation rate (DAS28-ESR) >3.2 OR presence of signs suggestive of active inflammatory disease activity with a DAS28-ESR ≤3.2’ as characteristics (figure 2A). Forty-two per cent included fatigue (figure 2B). Forty-eight per cent selected ‘≥2 conventional

synthetic (cs) DMARDs AND ≥2 biological (b) DMARDs or targeted synthetic (ts) DMARDs with different modes of action’ for how many insufficiently effective antirheumatic drugs should at least have been applied (figure 2C). Eighty-nine per cent selected inability to taper GCs <5 mg (43%) or 10 mg (46%) prednisone or equivalent daily for more than 1 year, irrespec-tive of DMARD treatment (figure 2D), as difficult-to-treat RA characteristic.

Additional difficult-to-treat rA characteristicsTwo hundred and forty-three additional characteristics of diffi-cult-to-treat RA were given by 169 respondents (figure 2E), most frequently categorised into ‘interfering comorbidities’ and ‘extra-articular manifestations’. Examples are cardiovas-cular risk, malignancies, interstitial lung disease and vasculitis. A diversity of ‘other’ responses was given, for example, inflamma-tion on MRI, morning stiffness and patient dissatisfaction.

Interfering clinical issues and items missing in current eulAr recommendations important to manage difficult-to-treat rAFor interfering issues to be addressed in the workup and manage-ment of difficult-to-treat RA, 396 suggestions were given by 170 respondents (figure 3A), most frequently cardiovascular disease and extra-articular manifestations. Other interfering clinical issues were drug intolerance, smoking and chronic liver disease.

For issues not covered by the current EULAR recommenda-tions, 64 were mentioned by 54 respondents (figure 3B). These were most frequently classified as interfering comorbidities and extra-articular manifestations. Also issues regarding pharmaco-logical management (eg, tapering regimen, adverse events and polypharmacy), pain syndromes and pregnancy and lactation were mentioned. Other items were ongoing joint destruction, coping problems and persistent single joint involvement.

dIsCussIOnOur results show a wide variety in concepts of difficult-to-treat RA; active disease, failure to DMARD treatment and inability to taper GCs are considered main characteristics. Additional diffi-cult-to-treat RA characteristics were mostly related to extra-ar-ticular manifestations and interfering comorbidities that may hamper assessment of disease activity or limit treatment possi-bilities. As items missing in current RA EULAR management

1707Roodenrijs NMT, et al. Ann Rheum Dis 2018;77:1705–1709. doi:10.1136/annrheumdis-2018-213687

Clinical and epidemiological research

Figure 2 Difficult-to-treat RA characteristics. b/tsDMARDs, biological/targeted synthetic disease-modifying antirheumatic drugs; csDMARDs, conventional synthetic DMARDs; DAS28-ESR, disease activity score assessing 28 joints using erythrocyte sedimentation rate; RA, rheumatoid arthritis; US, ultrasonography. * with different modes of action; ° or equivalent daily for more than 1 year, irrespective of DMARD treatment.

recommendations, interfering comorbidities (especially cardio-vascular disease, infection and malignancy), extra-articular mani-festations, pharmacological management (eg, tapering strategies, adverse events and polypharmacy) and pain syndromes were mentioned most frequently.

Of the factors mentioned as contributing to difficult-to-treat RA in this survey, for example treatment non-adherence, adverse events and coping strategies, exact prevalences are unknown. These should be determined in future research for an indication of their need to be included in management recommendations.

Our results mainly reflect how difficult-to-treat RA is experi-enced in European countries. Additional contributing factors to difficult-to-treat RA in countries outside Europe might be limited access to diagnostic tests, to rheumatologists and to DMARDs. These should be addressed in management recommendations as well.

Our study has limitations. The survey was distributed via email, and it was asked to forward it to other rheumatologists to increase the number of respondents. As a drawback, the total number of rheumatologists who received it is unknown.

The four multiple-choice questions had prespecified response options, limiting input to these questions but enabling the responses to be easily summarised and quantified. The open questions required a classification system for the responses; some responses were classified into two categories, and there was a number of responses that was classified as ‘other’. Addi-tionally, the prespecified multiple-choice questions may have biased the results of the open questions. However, by these open questions, we received a large inventory of issues that may need to be addressed in clinical practice.

The strengths of this study are the large number of respondents and of European countries represented by the respondents; the

1708 Roodenrijs NMT, et al. Ann Rheum Dis 2018;77:1705–1709. doi:10.1136/annrheumdis-2018-213687

Clinical and epidemiological research

Figure 3 Interfering clinical issues and items missing in the current EULAR recommendations important to manage difficult-to-treat RA. EULAR, European League Against Rheumatism; RA, rheumatoid arthritis. *For example, tapering strategies, adverse events and polypharmacy.

many suggestions of items that are not covered by the current EULAR RA management recommendations underline the unmet clinical need for this subpopulation of patients with RA.

Recently, a EULAR Task Force has been initiated on the devel-opment of recommendations for the comprehensive manage-ment of difficult-to-treat RA. The results of this survey will fuel discussions on items to include in the management recommenda-tions of difficult-to-treat RA.

In conclusion, the results of this survey underscore the diffi-culty in establishing an unambiguous concept of difficult-to-treat RA, which is seen as a heterogeneous condition not fully covered by current EULAR recommendations. The recently established EULAR Task Force will explore the management of difficult-to-treat RA further.

Author affiliations1Department of rheumatology & Clinical immunology, University Medical Center Utrecht, Utrecht University, Utrecht, the netherlands2Department of rheumatology, Leiden University Medical Center, Leiden, the netherlands3Department of internal Medicine iii, Division of rheumatology, Medical University of Vienna, Vienna, Austria4Department of rheumatology, Cochin Hospital, Assistance publique-Hôpitaux de paris, inserM (U1153): Clinical epidemiology and Biostatistics, pres sorbonne paris-Cité, paris Descartes University, paris, France5niHr Manchester Musculoskeletal Biomedical research Unit, Central Manchester nHs Foundation trust, Manchester Academic Health science Centre, Manchester, UK6institute of infection, immunity and inflammation, University of Glasgow, Glasgow, UK7Department of rheumatology and Clinical immunology, Justus-Liebig University Giessen, Kerckhoff Clinic Bad nauheim, Bad nauheim, Germany8Department of rheumatology, 1st Faculty of Medicine, Charles University and rheumatology institute, prague, Czech republic9Department of rheumatology, Faculty of Medicine, University of Debrecen, Debrecen, Hungary

10Department of Genetics, Cell and immunobiology, semmelweis University, Budapest, Hungary11Department of rheumatology, 3rd Department of internal Medicine, semmelweis University, Budapest, Hungary

Contributors nMtr contributed to the data analysis, interpretation of data and manuscript preparation. MJHdH and JWGJ contributed to the design of the study, data analysis, interpretation of data and manuscript preparation. pMJW, DvdH and JMvL contributed to the design of the study, interpretation of data and manuscript preparation. MCVdG, DA, MD, KLH, iBM, UM-L, Ls and Zs contributed to the acquisition of data and manuscript preparation. Gn contributed to the design of the study, acquisition of data, interpretation of data and manuscript preparation. All authors read and approved the final manuscript.

Funding the authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

Competing interests nMtr, MJHdH, MCvdG, JWGJ, pMJW, DA, MD, KLH, iBM, UM-L and Zs declare to have no competing interests. DvdH received consulting fees from AbbVie, Amgen, Astellas, AstraZeneca, BMs, Boehringer ingelheim, Celgene, Daiichi, eli-Lilly, Galapagos, Gilead, Glaxo-smith-Kline, Janssen, Merck, novartis, pfizer, regeneron, roche, sanofi, takeda and UCB. Ls received fees from AbbVie, BMs, Celgene Corporation, eli Lilly, Merck sharp and Dohme, novartis, pfizer, roche, takeda and UCB. JMvL received fees from Arthrogene, MsD, pfizer, eli Lilly and BMs and research grants from Astra Zeneca and roche-Genentech. Gn received fees from Amgen, AbbVie, BMs, KrKA, MsD, pfizer, roche and UCB and research grants from pfizer and AbbVie.

Patient consent not required.

Provenance and peer review not commissioned; externally peer reviewed.

RefeRences 1 smolen Js, Landewé r, Bijlsma J, et al. eULAr recommendations for the management

of rheumatoid arthritis with synthetic and biological disease-modifying antirheumatic drugs: 2016 update. Ann Rheum Dis 2017;76:960–77.

2 singh JA, saag KG, Bridges sL, et al. 2015 American College of rheumatology guideline for the treatment of rheumatoid Arthritis. Arthritis Rheumatol 2016;68:1–26.

3 de Hair MJH, Jacobs JWG, schoneveld JLM, et al. Difficult-to-treat rheumatoid arthritis: an area of unmet clinical need. Rheumatology 2017.

1709Roodenrijs NMT, et al. Ann Rheum Dis 2018;77:1705–1709. doi:10.1136/annrheumdis-2018-213687

Clinical and epidemiological research

4 Unger M, Alasti F, supp G. the good, the bad and the ugly – refractory rheumatoid arthritis. Arthritis Rheumatol 2016;68.

5 Miossec p. introduction: ’Why is there persistent disease despite aggressive therapy of rheumatoid arthritis?’. Arthritis Res Ther 2014;16:113.

6 Kearsley-Fleet L, Davies r, De Cooket al. Biologic refractory disease in rheumatoid arthritis: results from the British society for rheumatology Biologics register for

rheumatoid Arthritis. Ann Rheum Dis 2018. doi: 10.1136/annrheumdis-2018-213378. [epub ahead of print].

7 Buch MH. Defining refractory rheumatoid arthritis. Ann Rheum Dis 2018;77:966–9. 8 taylor pC, Moore A, Vasilescu r, et al. A structured literature review of the burden of

illness and unmet needs in patients with rheumatoid arthritis: a current perspective. Rheumatol Int 2016;36:685–95.

1710 Ruperto N, et al. Ann Rheum Dis 2018;77:1710–1719. doi:10.1136/annrheumdis-2018-213150

Clinical and epidemiological research

ExtEndEd rEport

Canakinumab in patients with systemic juvenile idiopathic arthritis and active systemic features: results from the 5-year long-term extension of the phase III pivotal trialsnicolino ruperto,1 Hermine I Brunner,2 pierre Quartier,3 tamàs Constantin,4 nico M Wulffraat,5 Gerd Horneff,6,7 ozgur Kasapcopur,8 rayfel Schneider,9 Jordi Anton,10 Judith Barash,11 reinhard Berner,12 Fabrizia Corona,13 ruben Cuttica,14 Marine Fouillet-desjonqueres,15 Michel Fischbach,16 Helen E Foster,17 dirk Foell,18 Sebastião C radominski,19 Athimalaipet V ramanan,20 ralf trauzeddel,21 Erbil Unsal,22 Jérémy Levy,23 Eleni Vritzali,24 Alberto Martini,25 daniel J Lovell,2 on behalf of the paediatric rheumatology International trials organisation (prInto) and the pediatric rheumatology Collaborative Study Group (prCSG)

To cite: ruperto n, Brunner HI, Quartier p, et al. Ann Rheum Dis 2018;77:1710–1719.

Handling editor Josef S Smolen

► Additional material is published online only. to view please visit the journal online (http:// dx. doi. org/ 10. 1136/ annrheumdis- 2018- 213150).

For numbered affiliations see end of article.

Correspondence todr nicolino ruperto, Clinica pediatrica e reumatologia, prInto, Istituto G. Gaslini, Genova 16147, Italy; nicolaruperto@ gaslini. org

nr and HIB contributed equally.

received 1 February 2018revised 20 July 2018Accepted 6 August 2018published online First 29 September 2018

© Author(s) (or their employer(s)) 2018. re-use permitted under CC BY-nC. no commercial re-use. See rights and permissions. published by BMJ.

AbsTrACTObjectives to evaluate the long-term efficacy and safety of canakinumab in patients with active systemic juvenile idiopathic arthritis (JIA).Methods patients (2–19 years) entered two phase III studies and continued in the long-term extension (LtE) study. Efficacy assessments were performed every 3 months, including adapted JIA American College of rheumatology (aJIA-ACr) criteria, Juvenile Arthritis disease Activity Score (JAdAS) and ACr clinical remission on medication criteria (CrACr). Efficacy analyses are reported as per the intent-to-treat population.results 144 of the 177 patients (81%) enrolled in the core study entered the LtE. overall, 75 patients (42%) completed and 102 (58%) discontinued mainly for inefficacy (63/102, 62%), with higher discontinuation rates noted in the late responders group (n=25/31, 81%) versus early responders (n=11/38, 29%). At 2 years, aJIA-ACr 50/70/90 response rates were 62%, 61% and 54%, respectively. CrACr was achieved by 20% of patients at month 6; 32% at 2 years. A JAdAS low disease activity score was achieved by 49% of patients at 2 years. Efficacy results were maintained up to 5 years. of the 128/177 (72.3%) patients on glucocorticoids, 20 (15.6%) discontinued and 28 (22%) tapered to 0.150 mg/kg/day. Seven patients discontinued canakinumab due to Cr. there were 13 macrophage activation syndrome (three previously reported) and no additional deaths (three previously reported). no new safety findings were observed.Conclusion response to canakinumab treatment was sustained and associated with substantial glucocorticoid dose reduction or discontinuation and a relatively low retention-on-treatment rate. no new safety findings were observed on long-term use of canakinumab.Trial registration numbers nCt00886769, nCt00889863, nCt00426218 and nCt00891046.

InTrOduCTIOnCurrently available therapies for systemic juvenile idiopathic arthritis (sJIA) include non-steroidal

Key messages

What is already known about this subject? ► The key role of IL1 in the pathogenesis of sJIA and the therapeutic implications from its blocade.

► Canakinumab, a fully human monoclonal Ab which selectively blocks IL1 beta in patients with sJIA, has proved its efficacy and safety during a phase II and phase III clinical program.

What does this study add? ► The study provides with long term (up to 5 years) safety and drug survival data on a pooled population from canakinumab’s clinical program. Canakinumab’s effect on systemic features and joints proved to be maintained in the long term particularly in the early responders patients. For the first time early response has been shown to be linked to canakinumab’s long term survival rendering it as an easy identifiable clinical predictor factor of long term maintenace of remission/low disease activity.

How might this impact on clinical practice or future developments?

► As long as achievement, as much as maintenance of remission or alternatively low disease activity constitute the ultimate therapeutic target in order to prevent future organ damages and disease related comorbidities, time to response will facilitate physicians in their decision making to keep or switch canakinumab to another treatment in a timely manner.

1711Ruperto N, et al. Ann Rheum Dis 2018;77:1710–1719. doi:10.1136/annrheumdis-2018-213150

Clinical and epidemiological research

anti-inflammatory drugs, glucocorticoids, synthetic disease-mod-ifying antirheumatic drugs (DMARD) and biologic DMARDs that inhibit primarily interleukin (IL)-6 and IL-11–7; tumour necrosis factor blockers and CTL4-Ig for sJIA without systemic features.8–13 Management of sJIA is aimed at achieving and maintaining clinical remission (CR). Minimising glucocorti-coid exposure, ideally up to discontinuation, is also of foremost importance in an effort to prevent the inhibition of growth.14 15

IL-1 plays a key role in the pathogenesis of sJIA.16 Several reports have suggested that inhibition of IL-1 provides clin-ical benefit in sJIA.1 17 Canakinumab is a fully human mono-clonal antibody that selectively binds to IL-1β, inactivating its downstream signalling cascade. Previous phase II–III trials have demonstrated the efficacy and safety of canakinumab in patients with sJIA.4 5

Here, we report the long-term efficacy and safety of canak-inumab in patients with sJIA with active systemic features and arthritis at baseline, who were enrolled from the previously reported pivotal phase III studies,5 and followed for up to 5 years.

MeTHOdsstudy designThe study design of the two pivotal phase III trials has been previ-ously reported.5 Briefly, in trial 1 of 1-month duration, a single canakinumab dose or placebo was administered. Patients from trial 1 could enter the two-part trial 2 where canakinumab-naïve patients and patients from a phase II trial4 were additionally enrolled. Trial 2 was a randomised withdrawal study,18 with an open-label lead in part I up to 32 weeks. Glucocorticoid tapering was permitted, and monitored by the Paediatric Rheumatology International Trials Organisation (PRINTO) and the Pediatric Rheumatology Collaborative Study Group (PRCSG) coordinating centres,19 based on disease activity level5 (achieved at least an American College of Rheumatology (ACR) 50 with no fever and C-reactive protein (CRP) <10 mg/L; further details in (online supplementary file 1). Part I of trial 2 was followed by a randomised, double-blind, place-bo-controlled, event-driven withdrawal part II in which adapted JIA American College of Rheumatology (aJIA-ACR) 30 responders able to taper/discontinue glucocorticoids were randomised to receive placebo or continue canakinumab until the end of trial 2 or a flare of sJIA had occurred (part II; average total study dura-tion of 29.5 weeks).20 Patients from trial 2 (figure 1) were allowed to enter the open-label, long-term extension (LTE) study where patients were planned to be followed for a minimum of 96 weeks, with further glucocorticoid tapering as per physician’s decision. Patients received canakinumab 4 mg/kg subcutaneously every 4 weeks (maximum dose: 300 mg); canakinumab dose was tapered in the LTE to 2 mg/kg every 4 weeks in patients who were gluco-corticoid free as per physicians’ judgement.

PatientsPatients were followed up in the LTE study between 6 July 2009 and 5 December 2014 at 63 centres of PRINTO/PRCSG in 21 countries. Eligibility criteria for the phase III trials have been described previously.5 In brief, eligible children (2–19 years old) with confirmed sJIA as per the International League Against Rheumatism classification criteria, active systemic features of sJIA, at least two active joints, CRP level >30 mg/L (normal range: 0–10 mg/L) and being treated with a prednisone equiva-lent of ≤1.0 mg/kg/day were included. Major exclusion criteria included macrophage activation syndrome (MAS) within the last 6 months, active infections, malignancies and concurrent use of other biologics.

AssessmentsEfficacy assessments were performed at least every 3 months to assess the levels of improvement using various composite vali-dated measures: aJIA-ACR 50/70/90, based on the JIA core set variables,21–24 plus the absence of fever (defined as temperature ≤38°C in the preceding 7 days); clinically inactive disease (CID)/clinical remission on medication (CR), defined as at least 6 months of CID, which were evaluated by either the ACR criteria (CIDACR/CRACR)25 26 or by the Juvenile Arthritis Disease ActivityScore 71-CRP (JADAS; CIDJADAS/CRJADAS).

27 Disease activity wasmeasured by JADAS score with the following cut-offs: CIDJADAS score ≤1; low disease activity (LDA) score ≤3.8; moderate disease activity score 3.9–10.5; and high disease activity (HDA) score >10.527–29; systemic features were reflected in the physi-cian global evaluation of disease activity measured on a visual analogue scale.

Safety and tolerability of canakinumab were assessed in terms of adverse events (AE), serious AEs (SAE) and clinical and labo-ratory assessments from first injection until the last available observation. Serious infections, malignancies and cases of MAS were adjudicated by independent committees.30–33

A three-tiered approach was used to measure anti-canaki-numab antibodies (anti-drug antibody (ADAs)), consisting of a screening, a confirmatory and a titration assay, respectively. Serum canakinumab concentrations (pharmacokinetics, PK) were determined to assess the relationships between canaki-numab exposure and the immunogenicity data.34 35

statistical analysisThe European League Against Rheumatism (EULAR) recom-mendation for reporting LTE studies36 and the Consolidated Standards of Reporting Trials statement37 38 were followed. Categorical variables were summarised by absolute frequencies and percentages, while continuous variables were summarised by median and lower and upper quartiles. The aJIA-ACR criteria used the starting day of canakinumab as baseline. Efficacy anal-yses were performed in two ways: (1) for primary analysis: in the intent-to-treat (ITT) population based on observed data with all discontinuations at different time points counted as missing (n=177; patients enrolled in trial 2) and (2) with missing data imputed using last observation carried forward (LOCF) (online supplementary appendix).39

In exploratory analysis, the hypothesis was tested that early responders achieve more substantial sJIA control as compared with late responders of canakinumab. For the purpose of this analysis, early responders were defined as patients who had successfully completed the glucocorticoid tapering in part I of trial 2 as per protocol and who were randomised to the with-drawal part; late responders were defined as patients who moved directly from the open-label part of trial 2 because they failed to taper glucocorticoids in part I. Mixed model analyses were used to quantify the difference in efficacy responses for early and late responders. Subgroup exploratory analyses were also consid-ered by methotrexate (MTX) coadministration and prior use of biologics using the generalised mixed model framework. All mixed models were adjusted for repeated measures, and include a centre-related random effect and the following covariates: time in study, prior use of biologics, prior use of MTX, prior use of corticosteroid, number of joints with limitation of motion (LOM) and number of active joints.

AEs were summarised by presenting the number of events and percentage of patients and time-adjusted frequencies of expo-sure as per the primary system organ class based on the Medical

1712 Ruperto N, et al. Ann Rheum Dis 2018;77:1710–1719. doi:10.1136/annrheumdis-2018-213150

Clinical and epidemiological research

Figure 1 Flow chart with patient disposition. *One death occurred during part I; patient died due to MAS. †A patient in the placebo group died due to MAS 2 days after discontinuing the part II phase due to MAS. ‡One patient died from disease progression 3 months after discontinuation from the long-term extension phase due to unsatisfactory therapeutic effect. The grey box represents the patients who discontinued the part I or part II of trial 2 and entered the long-term extension study. Patients who entered the LTE are divided into two subgroups: (1) early responders, defined as patients who had successfully completed the glucocorticoid tapering in part I of trial 2 as per protocol and who were randomised to the withdrawal part; (2) late responders, defined as patients who moved directly from the open-label part of trial 2 and who failed to taper glucocorticoids in part I. LTE, long-term extension; MAS, macrophage activation syndrome.

Dictionary for Regulatory Activities (MedDRA, Version 17.1) preferred term.

resulTsbaseline characteristics and patient disposition over timeDemographics and baseline disease characteristics of the 177 patients at baseline as well as the 144 patients rolling over to the LTE study are presented in table 1. At baseline, as previously reported, 99.4% of patients had HDA (median JADAS=32.5).

Of the 177 patients (per ITT, denominator is 177 unless other-wise stated) enrolled in trial 2, a total of 144 (81%) continued

in the LTE study (figure 1); 122 (69%) stayed for a maximum of 271 weeks. Overall, 75 (42%) patients completed the LTE study.

As shown in table 1, among the 144 patients who entered LTE, there were 96 (54.2%) early responders with a median JADAS of 1.85 denoting an LDA status (62% JADAS LDA and 47.9% CIDJADAS) at baseline of the LTE study, and 48 (27.1%) late responders with a median JADAS of 18.8 denoting HDA status (69% with JADAS HDA at baseline of the LTE). Compared with early responders, the late group had a higher frequency of prior use of biologics (79.2% vs 57.3%) and required higher gluco-corticoid doses (0.36 mg/kg/day vs 0.17 mg/kg/day). Similar

1713Ruperto N, et al. Ann Rheum Dis 2018;77:1710–1719. doi:10.1136/annrheumdis-2018-213150

Clinical and epidemiological research

Table 1 Demographics and clinical characteristics of patients at baseline (date of first canakinumab administration and at the time of entry into the LTE study)

Characteristics

baseline lTe study*early responders (lTe study entry data)*

late responders (lTe study entry data)*

(n=177†) (n=144†) (n=96†) (n=48†)

Female 98 (55%) 79 (55%) 53 (55%) 26 (54%)

Age (years) 8.0 (5.0–12.0) 9.0 (6.0–13.0) 9.0 (6.0–13.0) 9.0 (5.5–11.5)

Body weight (kg) n=16725.8 (17.8–42.9)

n=14229.6 (20.1–46.0)

n=9431.4 (21.0–51.2)

26.3 (18.1–38.5)

Disease duration (years) n=1242.1 (0.8–4.3)

n=1012.3 (0.9–4.4)

n=712.1 (0.8–4.3)

n=302.8 (1.3–4.8)

Physician’s global assessment of disease activity (VAS) (mm) 70.0 (55–80) 7.0 (0–33.0) 1.0 (0–19.0) 29.0 (9.5–47.5)

Parent or patient’s assessment of overall well-being (VAS) (mm)

n=17663.5 (45.0–80.0)

6.0 (1.0–44.5) 2.0 (0–26.5) 29.0 (8.0–52.0)

Number of joints with active arthritis‡ 10.0 (4.0–22.0) 1.0 (0–5.0) 0 (0–2.5) 3.5 (1.5–12.5)

Number of joints with limited range of motion§ 9.0 (4.0–23.0) 1.5 (0–5.0) 0 (0–3.0) 5.0 (1.5–15.5)

CHAQ score 1.8 (1.1–2.3) 0.3 (0–1.1) 0 (0–0.9) 0.8 (0.2–1.6)

C-reactive protein (mg/L) 160.0 (88–271.0) 16.0 (3.30–87.6) 6.6 (2.0–42.5) 83.1 (22.8–133.2)

Fever 161 (91%) 25 (17.4%) 11 (11.5%) 14 (29.2%)

JADAS score n=17632.5 (26.0–43.3)

7.85 (0.3–19.6) 1.85 (0.1–12.2) 18.85 (8.2–28.3)

JADAS high disease activity (>10.5) 175 (99.4%) 63 (43.8%) 30 (31.3%) 33 (68.8%)

JADAS low disease activity (≤3.8) 0 67 (46.5%) 59 (61.5%) 8 (16.7%)

JADAS CID (≤1) 0 48 (33.3%) 46 (47.9%) 2 (4.2%)

Use of methotrexate at baseline 94 (53%) 76 (52.8%) 49 (51.0%) 27 (56.3%)

Prior use of biologics¶ 116 (66%) 93 (64.6%)** 55 (57.3%) 38 (79.2%)

Anakinra 83 (47%) 65 (45.1%)** 44 (45.8%) 21 (43.8%)

Tocilizumab 10 (6%) 7 (4.9%)** 5 (5.2%) 2 (4.2%)

Anti-TNF agent or other biologic agent 62 (35%) 50 (34.7%)** 25 (26.0%) 25 (52.1%)

Prednisone therapy at baseline 128 (72%) 63 (43.8%) 20 (20.8%) 43 (89.6%)

Prednisone equivalent dose†† (mg/kg/day) n=1280.27 (0.17–0.54)

n=630.23 (0.16–0.49)‡‡

n=200.17 (0.08–0.20)

n=430.36 (0.21–0.55)

Data are n (%) or medians (first to third quartiles).Patients entering the LTE are divided into those coming from the double-blind placebo-controlled withdrawal part (early responders who were patients randomised in the withdrawal part) or from the open-label part of trial 2 (late responders who failed to achieve an adapted juvenile idiopathic arthritis American College of Rheumatology (aJIA-ACR) 30 response or to taper glucocorticoids in part I).*Baseline characteristics are carried forward from last non-missing values of previous study.†n is indicated in the rows only if the total number is different from the heading column.‡The range of possible values for number of joints with active arthritis was 0–73.§The range of possible values for number of joints with limited range of motion was 0–69.¶A patient could have received one or more biologic agents previously.**Patients were not allowed to take any biologics during treatment. These were patients with prior use of biologics before the core study.††Median dose provided for the patients who were on steroids at study entry.‡‡Median dose of the subgroups from part I and part II with great difference in the median steroid doses as demonstrated in the adjacent columns.CHAQ, Childhood Health Assessment Questionnaire; CID, clinically inactive disease; JADAS, Juvenile Arthritis Disease Activity Score; LTE, long-term extension; n, number of observations; TNF, tumour necrosis factor; VAS, visual analogue scale.

differences were noted at baseline (day of the first canakinumab administration) in the two groups.

A total of 102 (58%) patients discontinued canakinumab: 33 during trial 2 as previously reported5 and the remaining 69 during the LTE study. Non-response (n=26/102, 25%) and loss of response over time (n=36/102, 35%) were the main reasons for discontinuation with majority coming from the late responders group (n=25/36, 69%), followed by intolerance (n=19/102, 19%). Seven patients (n=7/102, 7%) discontinued canakinumab as per the physicians’/family decision no longer requiring canakinumab.

Canakinumab tapering/discontinuationForty-four (25%) patients received at least three consecutive reduced doses of canakinumab of 2 mg/kg. Of these, 26 (59%) remained on a sustained reduced dose until study end with a

median time of follow-up of 25 months. Of these 26 patients, five discontinued canakinumab (no longer requiring study drug) and two discontinued the study (consent withdrawal, n=1; lymphadenitis, n=1). Conversely, of the remaining 18 patients who flared after tapering of canakinumab, 15 regained sJIA control on uptitration of canakinumab to 4 mg/kg/month while three patients did not. In two of those patients disease flare was associated with MAS.

sJIA disease control and response to canakinumabFigure 2 shows the change in JADAS score over time in the ITT population for the two major subgroups (early and late responders) who entered the LTE from part I and part II of trial 2.

For the patients who entered the LTE from the open-label part of trial 2 (n=48), a decrease in disease activity by a median of

1714 Ruperto N, et al. Ann Rheum Dis 2018;77:1710–1719. doi:10.1136/annrheumdis-2018-213150

Clinical and epidemiological research

Figure 2 JADAS scores over time in the ITT population for the two major subgroups (early and late responders) who entered the LTE from part I and part II of trial 2 (observed data). n denotes the number of patients available at that time point; horizontal lines represent the cut-offs for JADAS HDA>10.5, LDA≤3.8 and ID≤1. Denominator is equal to 177 patients per the ITT principle. The upper and lower error bars represent the third (q3) and first (q1) quartiles, respectively. Patients who entered the LTE are divided into two subgroups: (1) early responders, defined as patients who had successfully completed the glucocorticoid tapering in part I of trial 2 as per protocol and who were randomised to the withdrawal part; (2) late responders, defined as patients who moved directly from the open-label part of trial 2 and who failed to taper glucocorticoids in part I. All patients who belonged to the full analysis set in trial 2 were taken into account. Only the last available assessment within the given interval was taken into account. Only patients with a value at both BSL and the respective post-BSL time point were included. BSL, baseline; CRP, C-reactive protein; HDA, high disease activity; ID, inactive disease; ITT, intent-to-treat; JADAS, Juvenile Arthritis Disease Activity Score; LDA, low disease activity.

64.6% (n=48) (median change: −24.9 (Q1 −32.7; Q3 −14.9)) and 93.5% (n=25) (median change: −31.8 (Q1 −40.3; Q3 −24.8)) was observed within the 6 months and 2 years of canak-inumab treatment, respectively. However, despite improvement on canakinumab, this group of patients continued to remain in JADAS HDA status (JADAS >10.5).

Early responders (n=96) achieved a greater decrease in the JADAS during the study as compared with late responders (mixed model; p<0.01). Notably, in early responders, JADAS improve-ment by a median of 96.4% (n=95) (median change: −25.1 (Q1 −33.1; Q3 −20.2)) occurred within the first 6 months of canakinumab initiation, and was maintained over at least 2 years (median decrease of 99.2%, n=87) (median change: −27.6 (Q1 −34.3; Q3 −20.6) at 2 years). Overall, from the entire cohort, a total of 79/177 (44.6%) achieved JADAS-LDA/inactive disease (ID) over the first 6 months of canakinumab exposure which increased to 48.6% at 2 years and was sustained until the end of the study.

As shown in figure 3, aJIA-ACR 50/70/90 response rates in the ITT population at 6 months since initiation of canakinumab were 73.4%, 65.5% and 52.0%, respectively. This level of response was maintained up to 3 years with 54.8%, 53.7% and 49.7% of the patients achieving aJIA-ACR 50/70/90 response rates, respectively. ClDACR since initiation of canakinumab was achieved by 32.8% of patients at 6 months, which increased up to 39.5% at 2 years and maintained through (36.7%) at year 3 and progressively decreased to 12.4% in 5 years. The CIDJADAS criteria use yielded similar results.

As per figure 3, CRACR criteria were achieved by 18.6% (33/177) of patients within the first 6 months of canakinumab therapy. In these 33 early responding patients, CRACR was reached at a median time of 29 days. The number of patients reaching CRACR further increased to 31.6% after 24 months. During the first 2 years, 71 patients had achieved CRACR on canakinumab with the median time to CR being 213 days. During the study, 80 patients had achieved CRACR (median time=254 days). Consis-tently similar rates were observed based on CRJADAS (rates are shown in figure 3).

Efficacy analyses were also performed by means of LOCF. (online supplementary figures S1 and S2) show the corre-sponding LOCF data confirming a trend towards higher level of response over time.

differences between the ITT and lOCF analysesBoth ITT and LOCF analyses were conducted for the entire study duration of 5 years and showed about 30%–40% achieving ID at year 3, the overall planned study duration. Because the study lasted longer in some countries (until year 5), patients who completed the study as planned at year 3 were counted as discontinuations in the ITT analysis. This resulted in the dissociation observed in the remission rates from year 3 to year 5 between the two methods. While maintenance of response up to year 5 has been demonstrated using LOCF, remission percentages started fading from year 3 onwards in the ITT analysis.

1715Ruperto N, et al. Ann Rheum Dis 2018;77:1710–1719. doi:10.1136/annrheumdis-2018-213150

Clinical and epidemiological research

Figure 3 aJIA-ACR responses, CIDACR/CIDJADAS rates and clinical remission on medication rates as per CRACR or CRJADAS over time (observed data). n denotes the number of patients available at that time point; denominator was equal to 177 patients as per the ITT principle in the pivotal study. Clinical remission on medication, defined as CID (ACR or JADAS) for 6 continuous months. aJIA-ACR, adapted juvenile idiopathic arthritis by American (online supplementary appendix) College of Rheumatology criteria; BSL, baseline; CIDACR, clinically inactive disease by ACR criteria; CIDJADAS, clinically inactive disease by Juvenile Arthritis Disease Activity Score criteria; CR, clinical remission; CRACR, CR by American College of Rheumatology criteria; CRJADAS, CR by Juvenile Arthritis Disease Activity Score criteria; ITT, intent-to-treat.

efficacy analysis of subgroupsA more pronounced decrease in disease activity as measured by the JADAS scores was observed in biologic naïve patients compared with biologic exposed patients at 6 months (92.7% (n=61); median change: −23.0 (Q1 −31.3; Q3 −14.3) vs 75.6% (n=113); median change: −23.4 (Q1 −31.8; Q3 −16.4)) (online supplementary figure 3). Similar trends were observed for the aJIA-ACR response criteria and CIDACR (online supplementary figure 4).

Patients treated with canakinumab while on MTX background therapy and those treated with canakinumab alone showed similar JADAS and aJIA-ACR response rates and CIDACR (online supplementary figures S5 and S6).

Glucocorticoid taperingAmong 128/177 (72.3%) patients who were on glucocorticoids at the beginning of trial 2 (median dose of 0.27 mg/kg/day; Q1–Q3: 0.16–0.53 mg/kg/day), 38/128 (29.7%), 51/128 (39.8%) and 20/128 (15.6%) discontinued glucocorticoid therapy by 6 months, 2 years and 5 years of canakinumab therapy, respec-tively. At the end of the study, 72 patients (56%) remained on glucocorticoid treatment at a median dose of 0.25 mg/kg/day (Q1–Q3: 0.16–0.46 mg/kg/day), with 28 patients (22%) having the dose below 0.2 mg/kg/day level.

As shown in table 1, patients who entered LTE from the double-blind withdrawal part were on lower dose of glucocor-ticoids (0.17 mg/kg/day) compared with patients who moved from the open-label part I of trial 2 (median dose of 0.36 mg/kg/day). Of the 38 patients who failed to discontinue glucocor-ticoids at part I of trial 2, ten became steroid free, with another nine patients reaching the below 0.2 mg/kg/day level.

safetyOverall, the median duration of exposure (canakinumab and placebo) in the study was 3.5 years (Q1, 0.6; Q3, 4.4) corre-sponding to 476.530 patient-years. The exposure-adjusted inci-dence rate of AEs was 796.69/100 patient-years (table 2).

The incidence rate of SAEs was 40.68/100 patient-years. Most common SAEs included sJIA flare (5.24/100 patient-years), MAS events (histiocytosis haematophagic in MedDRA terms) and fever either due to infection or sJIA flare (table 3).

All MAS cases were adjudicated by an independent adjudi-cation committee for a total of 13 events (2.726/100 patient-years) in 12 patients leading to nine patients discontinuing the study. Three of these were reported earlier5 and the other 10 MAS cases were reported as SAEs in the LTE study. One of the reported MAS events was complicated by pulmonary hyper-tension and interstitial pneumonia, resulting in patient death in the pivotal study as previously reported,5 and another event was complicated by transfusion-related acute lung injury; acute interstitial pneumonitis, following blood transfusion products for MAS in the LTE study. The second event resolved and the patient discontinued the study. No malignancies, anaphylaxis or anaphylactoid reactions were reported. There were no addi-tional deaths besides the three previously reported.5

InfectionsThe incidence of serious infections was 10.28/100 patient-years of exposure and all resolved without sequelae. Most common serious infections were gastroenteritis (1.05/100 patient-years), pneumonia (0.84/100 patient-years), and varicella, subcutaneous abscess, gastrointestinal viral infection, septic shock and strepto-coccal tonsillitis (0.42/100 patient-years each).

1716 Ruperto N, et al. Ann Rheum Dis 2018;77:1710–1719. doi:10.1136/annrheumdis-2018-213150

Clinical and epidemiological research

Table 2 Incidence rates of adverse events, by system organ class

system organ class

n=177 rate per 100 patient-years n (%) events

Infections and infestations 136 (76.8) 1036 217.26

Gastrointestinal disorders 99 (55.9) 513 107.58

Musculoskeletal and connective tissue disorders

97 (54.8) 467 97.94

Respiratory, thoracic and mediastinal disorders

82 (46.3) 391 82.0

General disorders and administration site conditions

77 (43.5) 226 47.39

Skin and subcutaneous tissue disorders 76 (42.9) 245 51.38

Nervous system disorders 61 (34.5) 223 46.77

Injury, poisoning and procedural complications

58 (32.8) 164 34.39

Investigations 55 (31.1) 166 34.81

Blood and lymphatic system disorders 34 (19.2) 74 15.52

Eye disorders 30 (16.9) 54 11.32

Neoplasms benign, malignant and unspecified* (including cysts and polyps)

24 (13.6) 29 6.08

Psychiatric disorders 21 (11.9) 29 6.08

Hepatobiliary disorders 19 (10.7) 26 5.45

Ear and labyrinth disorders 17 (9.6) 39 8.18

Immune system disorders 16 (9.0) 24 5.03

Metabolism and nutrition disorders 16 (9.0) 26 5.45

Reproductive system and breast disorders 16 (9.0) 19 3.98

Vascular disorders 12 (6.8) 21 4.40

Renal and urinary disorders 9 (5.1) 12 2.52

Cardiac disorders 8 (4.5) 11 2.31

Endocrine disorders 2 (1.1) 2 0.42

Congenital, familial and genetic disorders*† 1 (0.6) 1 0.21

Social circumstances‡ 1 (0.6) 1 0.21

N denotes the total number of patients; a patient with multiple occurrences of an adverse event (AE) under one category is counted only once in the AE category.*Neoplasms reported here are benign and unspecified.†MedDRA PT: keratosis follicular.‡MedDRA PT: dental prosthesis.MedDRA PT, Medical Dictionary for Regulatory Activities preferred term.

Table 3 Incidence rates of serious adverse events by preferred term (≥2 events)

n=177 rate per 100 patient-yearsn (%) events

Serious adverse events 64 (36.2) 194 40.68

Preferred term

Juvenile idiopathic arthritis 17 (9.6) 25 5.24

Histiocytosis haematophagic (MAS)* 10 (5.6) 17 3.56

Fever 8 (4.5) 8 1.68

Gastroenteritis 5 (2.8) 5 1.05

Abdominal pain 4 (2.3) 4 0.84

Pneumonia 3 (1.7) 4 0.84

Hepatitis 2 (1.1) 3 0.63

Hepatic enzyme increased 3 (1.7) 3 0.63

Septic shock 2 (1.1) 2 0.42

Arthralgia 2 (1.1) 2 0.42

Lymphadenopathy 2 (1.1) 2 0.42

Gastrointestinal viral infection 2 (1.1) 2 0.42

Subcutaneous abscess 2 (1.1) 2 0.42

Tonsillitis streptococcal 2 (1.1) 2 0.42

Musculoskeletal chest pain 2 (1.1) 2 0.42

Varicella 2 (1.1) 2 0.42

Vomiting 2 (1.1) 2 0.42

Drug reaction with eosinophilia and systemic symptoms

2 (1.1) 2 0.42

C-reactive protein increased 2 (1.1) 2 0.42

Serum ferritin increased 2 (1.1) 2 0.42

Paraesthesia 2 (1.1) 2 0.42

Anxiety 2 (1.1) 2 0.42

Traumatic fracture 2 (1.1) 2 0.42

Rash 2 (1.1) 2 0.42

N denotes the total number of patients; serious adverse events occurring after pivotal study baseline are presented in this table.*MAS: macrophage activation syndrome is the terminology used in the literature.30–33

There were four opportunistic infections (toxoplasmosis, cyto-megalovirus infection, Salmonella gastroenteritis and adenovirus infection) in one patient, each was adjudicated as such by the adjudication committee; two of these events were suspected to be related to canakinumab by the investigator. All events were moderate in severity and resolved following treatment. No tuberculosis cases were reported.

Laboratory abnormalitiesTransient neutropenia was reported in 18 patients (grade 3 (n=17); grade 4 (n=1)). Eleven neutropenia events occurred in patients on MTX background and three were associated with mild infections (pharyngitis, molluscum contagiosum, naso-pharyngitis and otitis media), respectively. None of the patients discontinued canakinumab due to neutropenia but two patients discontinued MTX.

Besides laboratory testing that reflect the anti-inflammatory effects of canakinumab, other clinical chemistry parameters remained largely unchanged. CRP and fibrinogen decreased markedly by week 2 of canakinumab therapy and low levels were maintained over time.

ImmunogenicityPost-treatment ADAs were detected in five patients. ADAs were all non-neutralising and their presence had no effect on PK of canak-inumab; the observed trough canakinumab concentrations in pres-ence of ADAs were comparable to those without ADAs. All but one of the five patients were on background treatment with MTX.

dIsCussIOnHere, we report the long-term efficacy and safety data of canak-inumab in patients with sJIA with active systemic features and arthritis enrolled in the pivotal phase III trials.5 There was a marked, rapid improvement of sJIA activity at 6 months, which was main-tained for up to 5 years and allowed for the marked reduction or even discontinuation of glucocorticoids in the majority of patients.

Due to the long duration of the study, canakinumab efficacy seems best described by the level of disease control as measured by the JADAS. Exploratory analysis suggested that early response to canakinumab leads to a better long-term favourable outcome. As such, patients who entered the LTE study from the double-blind, placebo-controlled part because they responded to canak-inumab quickly and successfully tapered glucocorticoids, fared better than late responders, that is, patients who moved directly from part I of trial 2 because they failed to respond to canaki-numab initially or were unable to taper glucocorticoids as per protocol. These data were further corroborated by the CR

1717Ruperto N, et al. Ann Rheum Dis 2018;77:1710–1719. doi:10.1136/annrheumdis-2018-213150

Clinical and epidemiological research

data, which showed a trend towards better improvement in the subgroup of patients who were naïve to biologics.

Glucocorticoid discontinuation was possible for 44% of the patients, some rapidly and some in the long term, suggesting a continuous glucocorticoid tapering effect of canakinumab.

The limited therapeutic benefits of MTX in sJIA were confirmed by similar response rates to canakinumab irrespective of MTX background therapy. Thus, canakinumab/MTX combi-nation therapy is unlikely expected to improve sJIA control versus using canakinumab alone.

Most children who reached CID/LDA level after canakinumab initiation who were randomised to placebo and flared in the double-blinded part of trial 2 regained the LDA status on retreatment with canakinumab. This observation may suggest that canakinumab ther-apeutic benefits can be recaptured after withdrawal or interruption of medication in children who previously responded well to canaki-numab. Based on the time to flare in the double-blinded part of the study, a CID/LDA state would be expected to be maintained longer than the half-life of the drug. Further, these data also support that the withdrawal study design with an event-driven approach as was chosen for trial 2 did not expose children switched to placebo to worse long-term outcomes than children who were randomised to continue canakinumab.

The long-term use of canakinumab was well tolerated and consistent with the safety profile that has been reported in other canakinumab trials.5 Infections were the most common AEs. Despite disease control, new MAS events occurred while on canakinumab therapy. This is consistent with the previous reports,33 which suggested that canakinumab treatment does not seem to alter the risk of MAS in patients with sJIA, regard-less of the response of sJIA to canakinumab. Of note, the MAS adjudication committee was established before the EULAR/ACR/PRINTO MAS classification criteria were published.33

A limitation of this study may be the overall low retention rate with 58% of the patients discontinuing canakinumab over the 5 years, which is typically expected in a study with such long follow-up. However, LOCF was carried out in addition to the observed data to address this shortcoming and possible artificial increasing percentages of response rates over time. The higher discontinuation rate in the late responders group as opposed to the early responders group further supports the notion of early response being used as a predictor factor of long-term outcome by the physi-cians when considering canakinumab change to another treatment. In addition, this LTE study included patients with a wide range of canakinumab treatment durations due to the adaptive design chosen for trial 2 to limit placebo and corticosteroid exposures to patients. To address the limitations, another cohort of sJIA is being studied, where canakinumab was started in patients with and without fever at baseline (analysis in progress). Further, a dedicated dose reduc-tion/dose interval prolongation study in patients with sJIA who are canakinumab responders is ongoing (NCT02296424).

In conclusion, response to canakinumab treatment was sustained or improved up to 5 years in patients with sJIA with active systemic features and arthritis and was associated with glucocor-ticoid discontinuation. Early response seemed to be a predictive factor of long-term outcome enabling physicians to incorporate in their decision-making the time to response in the consideration of changing canakinumab to another treatment. No new safety find-ings were observed on long-term use of canakinumab.

Author affiliations1Clinica pediatrica e reumatologia, prInto, IrCCS Istituto Giannina Gaslini, Genoa, Italy

2division of rheumatology, Cincinnati Children’s Hospital Medical Center, Cincinnati, ohio, USA3Centre de référence national pour les maladies inflammatoires rhumatologiques et auto-immunes systémiques rares de l’enfant (rAISE), Unité d’Immunologie, Hématologie et rhumatologie pediatrique, Université paris-descartes, IMAGInE Institute, Hôpital necker-Enfants Malades, paris, France42nd department of pediatrics, Unit of pediatric rheumatology-Immunology, Semmelweis University, Budapest, Hungary5department of pediatric Immunology and rheumatology, Wilhelmina Children’s Hospital, Utrecht, the netherlands6department of General paediatrics, Asklepios Klinik Sankt Augustin, Sankt Augustin, Germany7department of paediatric and Adolescents Medicine, Medical Faculty, University Hospital of Cologne, Cologne, Germany8department of pediatric rheumatology, Cerrahpasa Medical School, Istanbul University, Istanbul, turkey9pediatric rheumatology, the University of toronto and Hospital for Sick Children, toronto, ontario, Canada10pediatric rheumatology, Unidad de reumatología pediátrica, Esplugues de Llobregat, Hospital Sant Joan de déu, Universitat de Barcelona, Barcelona, Spain11pediatrics, Kaplan Medical Center, rehovot, Israel12Klinik und poliklinik für Kinder- und Jugendmedizin, Universitätsklinikum Carl Gustav Carus, technische Universität dresden, dresden, Germany13Clinica pediatrica de Marchi, Fondazione IrCCS Ca’ Granda-ospedale Maggiore policlinico, Milano, Italy14Unidad de reumatología, Hospital General de niños pedro de Elizalde, Buenos Aires, Argentina15Service de néphrologie et rhumatologie pédiatrique, Hôpital Universitaire femme mère enfant, Groupement Hospitalier Est, Bron (Lyon), France16pédiatrie I, Hôpital Universitaire Hautepierre, Strasbourg, France17newcastle University and newcastle Hospitals nHS Foundation trust, Great north Children’s Hospital, newcastle Upon tyne, UK18Universitätsklinikum Münster, Klinik für pädiatrische rheumatologie und Immunologie, Münster, Germany19Centro de estudos em terapias inovadoras, Hospital de Clínicas da UFpr, Curitiba, Brazil20department of pediatric rheumatology Berlin, Helios Kliniken Berlin-Buch, Children’s Hospital, Berlin, Germany21department of paediatric rheumatology, Bristol royal Hospital for Children & royal national Hospital for rheumatic diseases, Bristol, UK22department of pediatrics, division of pediatric rheumatology, dokuz Eylül University Medical Faculty, Balcova, Izmir, turkey23Biometrical practice BIop, Basel, Switzerland24Global Clinical development, Immunology and dermatology, novartis pharma AG, Basel, Switzerland25direzione Scientifica, IrCCS Istituto Giannina Gaslini, Genoa, Italy

Acknowledgements the authors thank all additional investigators for their participation in the study: Wolfgang Emminger, Wien, Austria; Laurence Goffin, Laeken, Belgium; rik Joos, Gent, Belgium; Bernard Lauwerys, Bruxelles, Belgium; Carine Wouters, Leuven, Belgium; Claudio Len, Sao paulo, Brazil; Maria Hilário, São paulo, Brazil; Sheila oliveira, rio de Janeiro, Brazil; Flavio Sztajnbok, rio de Janeiro, Brazil; Elie Haddad, Montreal, QC, Canada; Lori tucker, Vancouver, Canada; Kristin Houghton, Vancouver, BC, Canada; Brigitte Bader Meunier, paris, France; richard Mouy, paris, France; Isabelle Kone-paut, Le Kremlin-Bicêtre, Bicêtre, France; Angelika thon, Hannover, Germany; Elisabeth Weibarth-riedel, Hamburg, Germany; Hartwig Lehmann, Gießen, Germany; Ivan Foeldvari, Hamburg, Germany; Jasmin Kummerle-deschner, tübingen, Germany; Johannes-peter Haas, Germany; Michael Frosch, Muenster, Germany; nikolay tzaribachev, Germany; thomas Lutz, Heidelberg, Germany; tilmann Kallinich, Berlin, Germany; Giorgios Chrousos, Athens, Greece; Maria trachana, thessaloniki, Greece; olga Vougiouka, Ampelokipi, Greece; riva Brik, Haifa, Israel; Liora Harel, petach-tikvah, Israel; Shay padeh, ramat Gan, Israel; Yosef Uziel, Kfar-Sava, Israel; Yackov Berkun, ramat Gan, Israel; Maria Alessio, napoli, Italy; rolando Cimaz, Firenze, Italy; Valeria Gerloni, Milano, Italy; Berit Flato, oslo, norway; Manuel Ferrandiz, Breña, peru; Lidia rutkowska-Sak, Warszawa, poland; Ekaterina Alekseeva, Moscow, russia; Evgeny nasonov, Moscow, russia; Vyacheslav Chasnyk, Saint-petersburg, russia; Inmaculada Calvo, Valencia, Spain; Maria Luz Gamir, Madrid, Spain; Juan Carlos Lopez robledillo, Madrid, Spain; Bo Magnusson, Stockholm, Sweden; Michael Hofer, Lausanne, Switzerland; Huri ozdogan, Istanbul, turkey; Muferet Erguven, Istanbul, turkey; Seza ozen, Ankara, turkey; Alice Chieng, Manchester, UK; Liza McCann, Liverpool, UK; Clarissa pilkington, London, UK; nicholas Wilkinson, oxford, UK; patricia Woo, London, UK; taunton Southwood, Birmingham, UK; Gloria Higgins, Columbus, oH, USA; daniel Kingsbury, portland, or, USA; Jorge Lopez-Benitez, Boston, MA, USA; Katherine Marzan, Los Angeles, CA, USA; paula Morris, Little rock, Ar, USA; Kenneth Schikler, Louisville, KY, USA. Medical writing and editorial assistance was provided by divya Chandrasekhar, phd, of novartis Healthcare, Hyderabad, India, which was funded by novartis pharma, Basel, Switzerland.

1718 Ruperto N, et al. Ann Rheum Dis 2018;77:1710–1719. doi:10.1136/annrheumdis-2018-213150

Clinical and epidemiological research

Contributors the study was designed jointly by academic authors (nr, HIB, AM and dJL) and novartis, with data collected by prInto/prCSG investigators. the first and subsequent versions of the manuscript were written by nr and HIB, edited by AM and dJL and revised critically by all remaining coauthors. All authors attest to the completeness and veracity of data and data analyses. Consistency in reporting the study data to healthcare authorities and institutional review boards was ensured by novartis. All authors had full access to study data, reviewed and revised the manuscript and approved the final version to be published. All authors were involved in the decision to submit the manuscript for publication and had the right to accept or reject comments or suggestions.

Funding this study was funded by novartis pharma.

Competing interests nr: consultant and speaker’s bureaus from AbbVie, Ablynx, Amgen, AstraZeneca, Baxalta Biosimilars, Biogen Idec, Boehringer Ingelheim, Bristol-Myers Squibb, Celgene, Eli-Lilly, EMd Serono, Gilead Sciences, Janssen, MedImmune, novartis, pfizer, r-pharm, roche, Sanofi, Servier, takeda; nr works as a full-time public employee of the public hospital Istituto Giannina Gaslini, which has received contributions from Bristol-Myers Squibb, Hoffman-La roche, Janssen, novartis, pfizer, Sobi for the coordination activity of the prInto network. HIB: consultant: AstraZeneca, Bristol-Myers Squibb, Genentech, Janssen, novartis, pfizer, Sanofi, takeda; speaker’s bureaus: Genentech, novartis pierre Quartier; has received consultant fees from novimmune, novartis and SoBI; has received speaker’s bureaus from Abbvie, BMS, Chugai-roche, Lilly, novartis, pfizer and Sobi; has acted as coordinator or investigator in clinical trials for Abbvie, BMS, Chugai-roche, novartis, Sanofi, Sobi; has acted as member of a data monitoring board committee for Sanofi. nMW has received grant/research support from AbbVie; has received consultant fees from AbbVie, Sobi and novartis. GH has received speaker’s bureaus from AbbVie, Boehringer Ingelheim, Chugai, MSd, novartis, pfizer, roche and Sobi; has received scientific grants from AbbVie, Chugai, MSd, novartis, pfizer and roche. oK has received speaker’s bureaus from novartis, roche, pfizer and Abbvie. rS has received consultant fees from novartis, Sobi and novimmune. JLA has received consultant fees and speaker’s bureaus from AbbVie, Gebro, novartis, pfizer, roche, Sanofi and Sobi. rC has received consultant fees and speaker’s bureaus from Bristol-Myers Squibb, Janssen, novartis, pfizer, Sanofi, roche, Lilly and GlaxoSmithKline. HEF has acted as member of advisory boards for AbbVie, novartis, pfizer, Sanofi and Sobi; has received unrestricted educational bursaries from pfizer, Genzyme, BioMarin and Sobi to develop educational resources for healthcare professionals. dF has received grant/research support from pfizer and novartis; has received consultant fees from novartis, pfizer, Chugai-roche and Sobi; has received speaker’s bureaus from novartis. SCr has received grants for clinical research from novartis. AVr has received honoraria and participated in Advisory Boards for novartis. rt has received lecture fees from pfizer and Bristol-Myers Squibb. JL received fees from novartis for conducting the statistical analysis. EV is an employee of novartis pharma, Basel, Switzerland. AM has no conflicts of interest to declare since March 2016 when he became the Scientific director of the Istituto Giannina Gaslini, because this role does not allow him to render private consultancy resulting in personal income; consultant on behalf of the Istituto Giannina Gaslini: AbbVie, Boehringer, novartis, r-pharm. Istituto Giannina Gaslini has received contributions from Bristol-Myers Squibb, Hoffman-La roche, Janssen, novartis, pfizer, Sobi for the coordination activity of the prInto network. dJL has received grant/research support from AbbVie, Bristol-Myers Squibb, nIH, pfizer, roche; speaker’s bureaus from Genentech; consultant of AbbVie, Boehringer Ingelheim, Bristol-Myers Squibb, Celgene, Genentech, GlaxoSmithKline, Janssen, Johnson & Johnson, novartis, takeda, UCB. EU, FC, JB, MF, MF-d, rB and tC have nothing to disclose.

Patient consent parental/guardian consent obtained.

ethics approval the study was conducted in accordance with the declaration of Helsinki, the International Conference on Harmonisation Guidelines for Good Clinical practice and local regulations. An institutional review board or independent ethics committee at each site approved the protocol, consent form and any other written information provided to patients or their legal representatives.

Provenance and peer review not commissioned; externally peer reviewed.

data sharing statement All data generated or analysed are included in this article and the supplementary information files.

Open access this is an open access article distributed in accordance with the Creative Commons Attribution non Commercial (CC BY-nC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http:// creativecommons. org/ licenses/ by- nc/ 4. 0/

RefeRences 1 Quartier p, Allantaz F, Cimaz r, et al. A multicentre, randomised, double-blind,

placebo-controlled trial with the interleukin-1 receptor antagonist anakinra in patients with systemic-onset juvenile idiopathic arthritis (AnAJIS trial). Ann Rheum Dis 2011;70:747–54.

2 nigrovic pA, Mannion M, prince FH, et al. Anakinra as first-line disease-modifying therapy in systemic juvenile idiopathic arthritis: report of forty-six patients from an international multicenter series. Arthritis Rheum 2011;63:545–55.

3 de Benedetti F, Brunner HI, ruperto n, et al. randomized trial of tocilizumab in systemic juvenile idiopathic arthritis. N Engl J Med 2012;367:2385–95.

4 ruperto n, Quartier p, Wulffraat n, et al. A phase II, multicenter, open-label study evaluating dosing and preliminary safety and efficacy of canakinumab in systemic juvenile idiopathic arthritis with active systemic features. Arthritis Rheum 2012;64:557–67.

5 ruperto n, Brunner HI, Quartier p, et al. two randomized trials of canakinumab in systemic juvenile idiopathic arthritis. N Engl J Med 2012;367:2396–406.

6 Lovell dJ, Giannini EH, reiff Ao, et al. Long-term safety and efficacy of rilonacept in patients with systemic juvenile idiopathic arthritis. Arthritis Rheum 2013;65:2486–96.

7 Ilowite nt, prather K, Lokhnygina Y, et al. randomized, double-blind, placebo-controlled trial of the efficacy and safety of rilonacept in the treatment of systemic juvenile idiopathic arthritis. Arthritis Rheumatol 2014;66:2570–9.

8 Lovell dJ, Giannini EH, reiff A, et al. Etanercept in children with polyarticular juvenile rheumatoid arthritis. pediatric rheumatology Collaborative Study Group. N Engl J Med 2000;342:763–9.

9 ruperto n, Lovell dJ, Cuttica r, et al. A randomized, placebo-controlled trial of infliximab plus methotrexate for the treatment of polyarticular-course juvenile rheumatoid arthritis. Arthritis Rheum 2007;56:3096–106.

10 ruperto n, Lovell dJ, Quartier p, et al. Abatacept in children with juvenile idiopathic arthritis: a randomised, double-blind, placebo-controlled withdrawal trial. Lancet 2008;372:383–91.

11 Lovell dJ, ruperto n, Goodman S, et al. Adalimumab with or without methotrexate in juvenile rheumatoid arthritis. N Engl J Med 2008;359:810–20.

12 Brunner HI, ruperto n, Zuber Z, et al. Efficacy and safety of tocilizumab in patients with polyarticular-course juvenile idiopathic arthritis: results from a phase 3, randomised, double-blind withdrawal trial. Ann Rheum Dis 2015;74:1110–7.

13 Brunner HI, ruperto n, tzaribachev n, et al. Subcutaneous golimumab for children with active polyarticular-course juvenile idiopathic arthritis: results of a multicentre, double-blind, randomised-withdrawal trial. Ann Rheum Dis 2018;77:21-29–.

14 deWitt EM, Kimura Y, Beukelman t, et al. Consensus treatment plans for new-onset systemic juvenile idiopathic arthritis. Arthritis Care Res 2012;64:n/a–10.

15 de Benedetti F, Brunner H, ruperto n, et al. Catch-up growth during tocilizumab therapy for systemic juvenile idiopathic arthritis: results from a phase III trial. Arthritis Rheumatol 2015;67:840–8.

16 pascual V, Allantaz F, Arce E, et al. role of interleukin-1 (IL-1) in the pathogenesis of systemic onset juvenile idiopathic arthritis and clinical response to IL-1 blockade. J Exp Med 2005;201:1479–86.

17 Gattorno M, pelagatti MA, Meini A, et al. persistent efficacy of anakinra in patients with tumor necrosis factor receptor-associated periodic syndrome. Arthritis Rheum 2008;58:1516–20.

18 Giannini EH, Lovell dJ, Silverman Ed, et al. Intravenous immunoglobulin in the treatment of polyarticular juvenile rheumatoid arthritis: a phase I/II study. pediatric rheumatology Collaborative Study Group. J Rheumatol 1996;23:919–24.

19 ruperto n, Martini A. networking in paediatrics: the example of the paediatric rheumatology International trials organisation (prInto). Arch Dis Child 2011;96:596–601.

20 Brunner HI, Lovell dJ, Finck BK, et al. preliminary definition of disease flare in juvenile rheumatoid arthritis. J Rheumatol 2002;29:1058–64.

21 Giannini EH, ruperto n, ravelli A, et al. preliminary definition of improvement in juvenile arthritis. Arthritis Rheum 1997;40:1202–9.

22 Singh G, Athreya BH, Fries JF, et al. Measurement of health status in children with juvenile rheumatoid arthritis. Arthritis Rheum 1994;37:1761–9.

23 paediatric rheumatology International trials organisation (prInto). Quality of life in juvenile idiopathic arthritis patients compared to healthy children. Clin Exp Rheumatol 2001;1923(Suppl):S1–172.

24 ruperto n, ravelli A, pistorio A, et al. Cross-cultural adaptation and psychometric evaluation of the Childhood Health Assessment Questionnaire (CHAQ) and the Child Health Questionnaire (CHQ) in 32 countries. review of the general methodology. Clin Exp Rheumatol 2001;19:S1–S9.

25 Wallace CA, Giannini EH, Huang B, et al. Childhood Arthritis rheumatology research Alliance; pediatric rheumatology Collaborative Study Group; paediatric rheumatology International trials organisation. Arthritis Care Res 2011;63:929–36.

26 Wallace CA, ruperto n, Giannini E, et al. preliminary criteria for clinical remission for select categories of juvenile idiopathic arthritis. J Rheumatol 2004;31:2290–4.

27 Consolaro A, ruperto n, Bracciolini G, et al. defining criteria for high disease activity in juvenile idiopathic arthritis based on the juvenile arthritis disease activity score. Ann Rheum Dis 2014;73:1380–3.

28 Consolaro A, ruperto n, Bazso A, et al. development and validation of a composite disease activity score for juvenile idiopathic arthritis. Arthritis Rheum 2009;61:658–66.

29 Consolaro A, Bracciolini G, ruperto n, et al. remission, minimal disease activity and acceptable symptom state in juvenile idiopathic arthritis. Arthritis Rheum 2012;64:2366–74.

1719Ruperto N, et al. Ann Rheum Dis 2018;77:1710–1719. doi:10.1136/annrheumdis-2018-213150

Clinical and epidemiological research

30 ravelli A, Minoia F, davì S, et al. 2016 Classification criteria for macrophage activation syndrome complicating systemic juvenile idiopathic arthritis: A european league against rheumatism/american college of rheumatology/paediatric rheumatology international trials organisation collaborative initiative. Ann Rheum Dis 2016;75:481–9.

31 ravelli A, Minoia F, davì S, et al. 2016 Classification criteria for macrophage activation syndrome complicating systemic juvenile idiopathic arthritis: A european league against rheumatism/american college of rheumatology/paediatric rheumatology international trials organisation collaborative initiative. Arthritis Rheumatol 2016;68:566–76.

32 ravelli A, Minoia F, davì S, et al. Expert consensus on dynamics of laboratory tests for diagnosis of macrophage activation syndrome complicating systemic juvenile idiopathic arthritis. RMD Open 2016;2:e000161.

33 Grom AA, Ilowite nt, pascual V, et al. rate and clinical presentation of macrophage activation syndrome in patients with systemic juvenile idiopathic arthritis treated with canakinumab. Arthritis Rheumatol 2016;68:218–28.

34 Sun H, Van LM, Floch d, et al. pharmacokinetics and pharmacodynamics of canakinumab in patients with systemic juvenile idiopathic arthritis. J Clin Pharmacol 2016;56:1516–27.

35 European Medicines Agency (EMEA)Committee for Medicinal products for Human Use (CHMp). 2017.Guideline on immunogenicity assessment of therapeutic proteins http://www. ema. europa. eu/ docs/ en_ GB/ document_ library/ Scientific_ guideline/ 2017/ 06/ WC500228861. pdf (accessed 18 dec 2017).

36 Buch MH, Aletaha d, Emery p, et al. reporting of long-term extension studies: lack of consistency calls for consensus. Ann Rheum Dis 2011;70:886–90.

37 Schulz KF, Altman dG, Moher d, et al. ConSort 2010 statement: updated guidelines for reporting parallel group randomised trials. BMJ 2010;340:c332.

38 Moher d, Schulz KF, Altman dG. the ConSort statement: revised recommendations for improving the quality of reports of parallel-group randomised trials. Lancet 2001;357:1191–4.

39 xu H. 2009.Biogenidec LoCF method and application in clinical data analysis, Inc., nESUG http://www. lexjansen. com/ nesug/ nesug09/ po/ po12. pdf

1720 Kaneko Y, et al. Ann Rheum Dis 2018;77:1720–1729. doi:10.1136/annrheumdis-2018-213920

Clinical and epidemiological research

EXTENDED REPORT

Tocilizumab in patients with adult-onset still’s disease refractory to glucocorticoid treatment: a randomised, double-blind, placebo-controlled phase III trialYuko Kaneko,1 Hideto Kameda,1,2 Kei Ikeda,3 Tomonoti Ishii,4 Kosaku Murakami,5 Hyota Takamatsu,6 Yoshiya Tanaka,7 Takayuki Abe,8 Tsutomu Takeuchi1

To cite: Kaneko Y, Kameda H, Ikeda K, et al. Ann Rheum Dis 2018;77:1720–1729.

Handling editor Josef S Smolen

► Additional material is published online only. To view please visit the journal online (http:// dx. doi. org/ 10. 1136/ 10. 1136/ annrheumdis- 2018- 213920).

For numbered affiliations see end of article.

Correspondence toTsutomu Takeuchi, Division of Rheumatology, Department of Internal Medicine, Keio University School of Medicine, Tokyo 160-8582, Japan; tsutake@ z5. keio. jp

Received 11 June 2018Accepted 8 August 2018Published Online First 2 October 2018

© Author(s) (or their employer(s)) 2018. No commercial re-use. See rights and permissions. Published by BMJ.

Key messages

What is already known about this subject? ► Suppression of pro-inflammatory cytokines has been reported to be effective in many case reports with adult-onset Still's disease, however, there has been no placebo-controlled trial on these drugs.

What does this study add? ► This is the first double-blinded, randomised placebo-controlled trial conducted in patients with adult-onset Still's disease that suggests tocilizumab, an anti-interleukin-6 receptor monoclonal antibody, is effective in improving systemic symptoms and decreasing glucocorticoid dose.

How might this impact on clinical practice or future developments?

► Our findings provide useful insights in the management of adult-onset Still’s disease and in designing and conducting future clinical trials on adult-onset Still’s disease.

AbsTrACTObjective To evaluate the efficacy and safety of tocilizumab, an interleukin-6 receptor antibody, in patients with adult-onset Still’s disease.Methods In this double-blind, randomised, placebo-controlled phase III trial, 27 patients with adult-onset Still’s disease refractory to glucocorticoids were randomised to tocilizumab at a dose of 8 mg/kg or placebo given intravenously every 2 weeks during the 12-week, double-blind phase. Patients received open-label tocilizumab for 40 weeks subsequently. The primary outcome was American College of Rheumatology (ACR) 50 response at week 4. The secondary outcomes included ACR 20/50/70, systemic feature score, glucocorticoid dose and adverse events at each point.results In the full analysis set, ACR50 response at week 4 was achieved in 61.5% (95% CI 31.6 to 86.1) in the tocilizumab group and 30.8% (95% CI 9.1 to 61.4) in the placebo group (p=0.24). The least squares means for change in systemic feature score at week 12 were –4.1 in the tocilizumab group and –2.3 in the placebo group (p=0.003). The dose of glucocorticoids at week 12 decreased by 46.2% in the tocilizumab group and 21.0% in the placebo group (p=0.017). At week 52, the rates of ACR20, ACR50 and ACR70 were 84.6%, 84.6% and 61.5%, respectively, in both groups. Serious adverse events in all participants who received one dose of tocilizumab were infections, aseptic necrosis in the hips, exacerbation of adult-onset Still’s disease, drug eruption and anaphylactic shock.Conclusion The study suggests that tocilizumab is effective in adult-onset Still’s disease, although the primary endpoint was not met and solid conclusion was not drawn.

InTrOduCTIOnAdult-onset Still’s disease is a rare, systemic inflam-matory disorder of unknown aetiology character-ised by high spiking fever, evanescent rash and polyarthritis.1 In addition to these major symptoms, other features simultaneously occur with multiple organ involvement, including sore throat, lymph-adenopathy, hepatosplenomegaly, and elevated serum liver enzymes and ferritin. Glucocorticoids are the first-line treatment for this disease, and the initial response to glucocorticoids is generally good despite intensive systemic inflammation. Overall treatment, however, remains challenging because high-dose glucocorticoids sometimes fail to cause remission, with occasionally fatal consequences, and dependence on glucocorticoids is frequently observed with a relapse of symptoms along with

dose tapering or discontinuation, which leads to organ damage accrual and long-term side effects.2 3

Immunosuppressive agents, such as methotrexate and ciclosporin, have been used as a steroid-sparing drug, but their effectiveness is limited.4 5 Progress in the understanding of the critical role of proinflam-matory cytokines in the pathogenesis of adult-onset Still’s disease has led to pilot use of anticytokine agents, resulting in an increasing number of successful case reports in patients who were unresponsive to conventional treatments.6–10 Furthermore, systemic juvenile idiopathic arthritis, previously known as Still’s disease, has responded significantly better to canakinumab, an anti-interleukin-1β monoclonal antibody, and tocilizumab, an anti-interleukin-6 receptor monoclonal antibody, than to placebo in several randomised controlled trials.11–16 Since adult-onset Still’s disease closely resembles systemic juve-nile idiopathic arthritis in terms of pathogenesis and cytokine profiles, these anticytokine inhibitors are promising treatments for adult-onset Still’s disease. However, there have been no randomised, place-bo-controlled trials on these anti-interleukin-1β or anti-interleukin-6 treatments in patients with adult-onset Still’s disease.

1721Kaneko Y, et al. Ann Rheum Dis 2018;77:1720–1729. doi:10.1136/annrheumdis-2018-213920

Clinical and epidemiological research

This phase III trial was conducted to evaluate the efficacy and safety of tocilizumab in refractory adult-onset Still’s disease. This article reports the results from the 1-year, multicentre, randomised, double-blind, placebo-controlled trial.

subjeCTs And MeTHOdsstudy designThis multicentre, randomised, double-blind, placebo-controlled phase III trial was conducted at eight centres in Japan as an investigator-initiated phase III trial and consisted of three parts: part 1, a randomised, double-blind, placebo-controlled, 4-week phase with a fixed dose of glucocorticoids; part 2, continued double-blind, placebo-controlled, 8-week phase with a condi-tional decrease in glucocorticoid dose; and part 3, a single, open-label, 40-week phase. This study is registered with UMIN (trial registration number: UMIN000012987).

Patients were recruited from 31 January 2014 to 31 July 2016 and randomly assigned (1:1) by a central automatic system to receive intravenous tocilizumab at a dose of 8 mg/kg or placebo every 2 weeks during parts 1 and 2. All patients received intrave-nous tocilizumab at the same dose and interval during part 3. The tocilizumab dose was determined in reference to trials in patients with juvenile idiopathic arthritis.13 14 Patients and all medical staff, including doctors, nurses and pharmacists, were unaware of the identity of the trial drugs at all times. Use of concomitant disease-modifying antirheumatic drugs or immunosuppressive agents was prohibited during the study. Doses of glucocorticoids were tapered during parts 2 and 3 according to the predefined rules when patients met the American College of Rheumatology (ACR) 50 criteria without fever: glucocorticoid doses equivalent to prednisolone at least 30 mg/day were tapered by 5–10 mg/day every week, 15–29 mg/day prednisolone by 2.5–5 mg/day every 2 weeks, 7.5–14 mg/day prednisolone by 1.5–2.5 mg/day every 4 weeks, and 5–7.4 mg/day prednisolone by 0.5–1.5 mg/day every 4 weeks. Tapering to prednisolone less than 5 mg/day was at the discretion of the investigators. Glucocorticoids were increased by about 50% within the initial dose (the exact dose was at the investigator’s discretion) when patients did not meet the ACR50 criteria without fever on two consecutive visits. Patients were omitted from the rest of part 2 and were entered into part 3 (escape) when they did not meet the criteria for ACR20 achievement without fever at the beginning of part 2 or ACR50 achievement without fever under the initial dose of glucocorticoids during part 2.

In part 3, the interval of tocilizumab administration could be prolonged in patients who could reduce their glucocorticoids equivalent to or less than 5 mg/day of prednisolone and main-tained in stable disease for 4 weeks or more. The prolongation of the interval should be conducted by 1 week. An interval of more than 5 weeks was not allowed. The interval of tocili-zumab administration could be shortened to every week at the discretion of the investigators if the response to tocilizumab was considered insufficient. The dosing interval was determined in reference to the trial in juvenile idiopathic arthritis.13

PatientsEligible patients were 20 years or older with disease onset at more than 15 years old who had been diagnosed with adult-onset Still’s disease with Yamaguchi criteria,17 and had shown inadequate response to more than 2 weeks of glucocorticoids equivalent to at least 0.5 mg/kg/day of prednisolone, with a stable dose of current glucocorticoids equivalent to 10 mg/day of prednisolone for at least 2 weeks. Patients had at least two

swollen joints and two tender joints and had ≥1 clinical point of systemic feature score (at least one of the following should be present: fever, rash, lymphadenopathy, hepatosplenomegaly or serositis).18 Patients’ blood tests showed erythrocyte sedimenta-tion rate of at least 20 mm/hour or C reactive protein of at least 1.0 mg/dL.

For 2 weeks before the study drug was first given, treatment with immunosuppressive drugs and disease-modifying anti-rheumatic drugs was prohibited. Treatment with infliximab, golimumab, adalimumab, abatacept, certolizumab-pegol and leflunomide was not allowed for 12 weeks and etanercept for 4 weeks before patients were given the study drug. Patients were excluded if they had important concurrent medical or surgical disorders, including leucocytopaenia (<3.0×109/L), neutrocy-topaenia (<1.0×109/L), lymphocytopaenia (<0.5×109/L) or thrombocytopaenia (<50×109/L), elevated alanine transaminase (>5 times of an upper limit of each site), elevated total bilirubin (>3 times of an upper limit of each site), serious disease assessed by investigators as inappropriate and a history of malignancy within 5 years. All patients were examined for active infections, especially viral hepatitis and tuberculosis.

Assessments and outcomesBaseline information included age, sex, disease duration, body weight and immunological profile. Clinical and laboratory assessments were done, including fever, skin rash, lymphade-nopathy, hepatosplenomegaly, serositis, number of swollen joints and tender joints, patient’s global assessment using a 100 mm Visual Analogue Scale, patient’s pain using Visual Analogue Scale, physician’s global assessment using Visual Analogue Scale, physical function using the Health Assessment Questionnaire Disability Index, complete blood cell counts, erythrocyte sedi-mentation rate, C reactive protein, liver enzymes and ferritin, every 2 weeks in parts 1 and 2 and every 4 weeks in part 3. The ACR core set and systemic feature score were evaluated.18 19 ACR20, ACR50 and ACR70 responses were defined as both improvement of 20%, 50% and 70% in the number of tender and number of swollen joints and 20%, 50% and 70% improvement in at least three of the remaining five variables of the ACR core set, respectively. Systemic feature score consisted of five clinical and five laboratory assessments. Clinical features included fever, rash, lymphadenopathy, hepatosplenomegaly and serositis, and the laboratory features were erythrocyte sedimentation rate, C reactive protein, leucocyte count, haemoglobin level and platelet count. Each clinical feature was assigned a score of 1 (present) or 0 (absent). At baseline, each laboratory feature was assigned a score of 1 when erythrocyte sedimentation rate was ≥20 mm/hour, C reactive protein was ≥1.0 mg/dL, white cell count was ≥12×103/μL, haemoglobin was ≤11 g/dL and platelet count was ≥400×103/μL. At follow-up visits, laboratory parameters were scored as follows: for erythrocyte sedimentation rate, 0 if <20 mm/hour or if decreased by ≥30% compared with base-line, or 1 if it increased or decreased by <30% compared with baseline; for C reactive protein, 0 if <1.0 mg/L or if decreased by ≥30% compared with baseline, or 1 if it increased or decreased by <30%; for white cell count, 0 if ≥12×103/μL or if decreased by 20% compared with baseline, or 1 if it increased or decreased by <20%; for haemoglobin level, 0 if ≥11 g/dL or if increased by ≥20% compared with baseline, or 1 if it decreased or increased by <20%; and for platelet count, 0 if ≥400×103/μL or if decreased by ≥20% compared with baseline, or 1 if it increased or decreased by <20%.

1722 Kaneko Y, et al. Ann Rheum Dis 2018;77:1720–1729. doi:10.1136/annrheumdis-2018-213920

Clinical and epidemiological research

Figure 1 Trial profile. *The underlying disease of one participant was found to be malignant lymphoma, rather than adult-onset Still’s disease (AOSD); therefore, this participant was excluded from the efficacy analyses.

One of the two coprimary outcomes was the proportion of patients who achieved ACR50 response at week 4, and the other primary outcome was the proportion of patients who had an ACR50 response at week 12 (the primary time point was week 4). We recognised that 4 weeks would be too short for a primary outcome, but at the same time had a great concern about a risk for critically uncontrollable exacerbation of adult-onset Still’s disease in the placebo group. Therefore, we predefined that the primary time point of the primary outcomes was week 4. We selected ACR50 as a primary outcome measure instead of ACR20 in reference to the results from the trial in patients with juvenile idiopathic arthritis,16 in which JIA ACR50 response showed greater effect size than juvenile idiopathic arthritis ACR30. The secondary endpoints were changes in glucocorti-coid dose, ACR20, ACR50, ACR70, systemic feature score, and presence of fever and skin rash at each time point. Patients were monitored for safety throughout the 52-week period.

statistical analysesIn the protocol, sample size calculation was based on the first of the coprimary endpoints, the proportion of patients who achieved ACR50 at week 4. The ACR50 response rate was assumed to be 65% in the tocilizumab group and 15% in the placebo group, in reference to the trials in patients with juve-nile idiopathic arthritis.13 14 Under these assumptions, 17 patients were needed in each group to have power of 80% with

a two-tailed significance level of 5%. During the clinical trial, we noted that recruitment was proceeding at a slower pace than anticipated because of the difficulty in the enrolment of eligible patients. In May 2016, we performed a blind review to assess whether a reduction in the number of patients was possible. The calculation predicted an ACR50 response in the tocilizumab group to be 80%, assuming that ACR50 in the placebo group was 15%, suggesting that 12 patients per group yielded a power of 80%. We had to accept the risk that our trial would fail to demonstrate significance in the primary outcome despite a true difference, and decided to close recruitment after at least 24 patients were enrolled.

We used descriptive statistics to summarise demographic and baseline disease characteristics by treatment group. The primary analysis population for efficacy evaluation in the double-blind phase (parts 1 and 2) was full analysis set, which included patients who received at least one dose of the study medication. All analyses, including those for the secondary endpoints, were prespecified in the statistical analysis plan before unblinding. The coprimary outcomes in this trial were the ACR50 response rates at weeks 4 and 12 with a predefinition that a test for week 12 would only be evaluated if between-group comparison at week 4 was significant with an alpha=0.05 (two-tailed). We used Fish-er’s exact test for dichotomous response outcomes; patients were categorised as either having a response or not having a response. Patients who did not complete the study, escaped from part 2

1723Kaneko Y, et al. Ann Rheum Dis 2018;77:1720–1729. doi:10.1136/annrheumdis-2018-213920

Clinical and epidemiological research

Table 1 Baseline patient characteristics

Placebo (n=13) Tocilizumab (n=13)

Age, years 55.5 (17.9) 51.3 (20.3)

Female, n (%) 10 (76.9) 10 (76.9)

Disease duration*, years 0.1 (0.0–10.8) 0.5 (0.1–23.8)

Body weight, kg 53.8 (9.9) 56.1 (9.6)

Swollen joint count 5.7 (3.3) 4.2 (2.7)

Tender joint count 5.5 (3.3) 4.2 (3.6)

Patient pain VAS, mm 37.2 (26.3) 33.8 (30.2)

Patient global VAS, mm 44.5 (29.2) 33.7 (28.8)

Physician global VAS, mm 42.3 (20.1) 38.4 (24.5)

HAQ–DI 1.0 (1.0) 0.7 (0.9)

CRP, mg/dL 4.7 (4.5) 4.2 (4.1)

Ferritin, ng/mL 3869 (6272) 2920 (1376)

SFS 5.1 (1.4) 4.6 (1.7)

Fever, n (%) 6 (46.2) 6 (46.2)

Skin rash, n (%) 7 (53.8) 8 (61.5)

ANA positivity, n (%) 5 (38.5) 8 (61.5)

RF positivity, n (%) 0 (0.0) 1 (7.7)

Anti-CCP positivity, n (%) 0 (0.0) 0 (0.0)

PSL dose, mg/day 32.5 (20.4) 23.0 (16.2)

Data are mean (SD) unless otherwise stated.*Median (IQR).ANA, antinuclear antibody; CRP, C reactive protein; HAQ–DI, Health Assessment Questionnaire Disability Index; PSL, prednisolone; RF, rheumatoid factor; SFS, systemic feature score; VAS, Visual Analogue Scale; anti-CCP, anticyclic citrullinated peptide antibody.

to part 3, or received weekly intravenous tocilizumab because of inadequate response to biweekly tocilizumab were classified as a non-responder. We used the Clopper-Pearson method to calculate the 95% CIs of proportions. For repeated measure-ments in part 1, we used mixed-effects model for repeated measurements with unstructured covariance matrix to compare the mean change from baseline between the two groups. For outcomes based on changes from baseline, we used an analysis of covariance model, which contains treatment group as a factor and baseline data as a covariate. The least squares means for the change from baseline were estimated by treatment group with the analysis of covariance model. For missing data in parts 2 and 3, we used the last observation carried forward method. We used the Kenward-Roger method to adjust the df for error. Safety data included full-exposure data for each patient. We defined serious adverse events in accordance with the definition in the Interna-tional Conference on Harmonisation guidelines.20 We used SAS V.9.4 for all analyses.

resulTsPatient flow and characteristicsTwenty-seven patients were enrolled, of whom 14 were randomly assigned to placebo and 13 to tocilizumab. One patient who had been allocated to the placebo group was subsequently diagnosed with malignant lymphoma rather than adult-onset Still’s disease during part 3; therefore, we excluded this patient from efficacy analyses (figure 1). Eight patients who received placebo and three patients who received tocilizumab met the criteria for escape; they did not complete part 2 and moved to part 3. Five patients withdrew from the study for various reasons. Baseline demo-graphic and disease characteristics were balanced between the groups (table 1), but patients who received placebo had slightly worse disease activity and took higher doses of glucocorticoids.

efficacy in 12-week, double-blind phaseFigure 2 shows the time course transition in ACR20, ACR50 and ACR70 response rates during the study. At week 4, 61.5% (8 of 13, 95% CI 31.6 to 86.1) of patients who received tocili-zumab had an ACR50 response, compared with 30.8% (4 of 13, 9.1–61.4) of patients who received placebo (p=0.238, two-tailed Fisher’s exact test). The proportion of patients with an ACR20 response was 76.9% (10 of 13) in the tocilizumab group and 38.5% (5 of 13) in the placebo group (p=0.111), whereas 38.5% (5 of 13) in the tocilizumab group and 30.8% (4 of 13) in the placebo group had an ACR70 response (p=1.000). At week 12, the proportion of patients with an ACR20 response was 61.5% (8 of 13) in the tocilizumab group, compared with 30.8% (4 of 13) in the placebo group (p=0.238). The propor-tion of patients with an ACR50 response was 61.5% (8 of 13) in the tocilizumab group and 30.8% (4 of 13) in the placebo group (p=0.238), whereas 46.2% (6 of 13) in the tocilizumab group and 30.8% (4 of 13) in the placebo group had an ACR70 response (p=0.688).

Systemic symptoms significantly improved with tocilizumab (figure 3A). The decrease in systemic feature score at week 4 from baseline in patients who received tocilizumab was –4.1, compared with –2.7 (p=0.080) in patients who received placebo. The decrease in systemic feature score at week 12 from base-line was –4.1 in the tocilizumab group and –2.3 in the placebo group (p=0.003). The results of each component of the systemic feature score are shown in online supplementary figure 1.

Tocilizumab had a significantly stronger effect on glucocorti-coid sparing (figure 3B). The dose of prednisolone was decreased from baseline by 46.2% in the tocilizumab group and by 21.0% in the placebo group at week 12 (p=0.017).

For secondary endpoints at week 4, no statistical significance in between-group comparisons was observed (figure 4A–G). For secondary endpoints at week 12, significant or borderline significant differences in swollen joint count (p=0.054), Physi-cian Visual Analogue Scale (p=0.013) and C reactive protein (p=0.030) were observed between the tocilizumab and placebo groups (figure 4A, E and G). The improvements in the other variables in the ACR core set (tender joint count, Pain Visual Analogue Scale, Patient Visual Analogue Scale and Health Assess-ment Questionnaire Disability Index) were not different between the groups (figure 4B, C, D and F). Serum ferritin concentra-tions decreased in both treatment groups (online supplementary figure 2).

At week 4, no patients in the tocilizumab group had fever, compared with 1 (7.7%) of 13 patients in the placebo group (p=1.000). At week 12, no patients in the tocilizumab group and 2 (15.4 %) of 13 patients in the placebo group had fever (p=0.480). Skin rash was observed in two (15.4%) patients in the tocilizumab group and five (38.5%) patients in the placebo group (p=0.378) at weeks 4 and 12. The prednisolone dose was 20.9 (±14.7) mg/day in the tocilizumab group and 29.0 (±17.4) mg/day in the placebo group at week 4, and 9.4 (±3.4) mg/day in the tocilizumab group and 16.3 (±6.8) mg/day in the placebo group at week 12.

efficacy in 40-week, open-label phaseIn part 3, the open-label phase, patients who were randomly assigned to receive tocilizumab continued to receive it and patients who were randomly assigned to receive placebo were given tocilizumab at week 12. The proportion of patients with an ACR20, ACR50 or ACR70 response (figure 2A–C) and systemic feature score in patients who switched placebo

1724 Kaneko Y, et al. Ann Rheum Dis 2018;77:1720–1729. doi:10.1136/annrheumdis-2018-213920

Clinical and epidemiological research

Figure 2 Time course of American College of Rheumatology (ACR) responses. Percentage of patients achieving (A) ACR20, (B) ACR50 and (C) ACR70. Bars are 95% CI.

1725Kaneko Y, et al. Ann Rheum Dis 2018;77:1720–1729. doi:10.1136/annrheumdis-2018-213920

Clinical and epidemiological research

Figure 3 Clinical efficacy on systemic inflammation and sparing glucocorticoid dose. (A) Changes in systemic feature score (SFS) from baseline. (B) Decrease in glucocorticoid dose from baseline. Bars are 95% CI. *P<0.05.

to tocilizumab improved at week 16 (figure 3A). The dose of glucocorticoids was also reduced (figure 3B) after the switch from placebo to tocilizumab. At week 52, the proportion of patients with an ACR20, ACR50 and ACR70 was 84.6% (11 of 13), 84.6% (11 of 13) and 61.5% (8 of 13), in both groups who received tocilizumab from baseline and those who received placebo and subsequently tocilizumab (figure 2A–C).

The systemic feature score was 0.6±0.8 in the tocilizumab group and 0.5±0.9 in the placebo group (figure 3A). The results of each component of the systemic feature score are shown in online supplementary figure 1. The doses of pred-nisolone were 6.4±5.1 mg/day in the tocilizumab group with cessation in two patients and 3.3±2.4 mg/day with no cessa-tion in the placebo group.

1726 Kaneko Y, et al. Ann Rheum Dis 2018;77:1720–1729. doi:10.1136/annrheumdis-2018-213920

Clinical and epidemiological research

Figure 4 Changes in other variables from baseline: (A) swollen joint count (SJC), (B) tender joint count (TJC), (C) Pain VAS, (D) Patient VAS, (E) Physician VAS, (F) Health Assessment Questionnaire Disability Index (HAQ–DI) and (G) C reactive protein (CRP). Bars are 95% CI. *P<0.05. VAS, Visual Analogue Scale.

The change in each variable in the ACR core set became compa-rable between the two groups or numerically better in patients who initially received placebo owing to the baseline condition, which was slightly worse in patients receiving placebo (figure 4). The variables at week 52 were as follows: swollen joint count, 0.7±1.3 (tocilizumab) and 0.3±0.6 (placebo); tender joint count, 0.5±0.8 (tocilizumab) and 0.5±1.0 (placebo); Pain Visual Analogue Scale, 12.2±21.4 mm (tocilizumab) and 7.5±11.9 mm (placebo); Patient Visual Analogue Scale, 10.4±18.8 mm (tocili-zumab) and 8.0±12.3 mm (placebo); Physician Visual Analogue Scale, 6.9±5.7 mm (tocilizumab) and 3.6±3.7 mm (placebo); Health Assessment Questionnaire Disability Index, 0.1±0.2 (tocilizumab) and 0.2±0.5 (placebo); and C reactive protein concentration, 0.0±0.0 mg/dL (tocilizumab) and 0.0±0.0 mg/dL (placebo). The time course change in ferritin concentrations is shown in online supplementary figure 2.

In part 3, six patients (three in each group) received tocili-zumab at an interval of less than 10 days. The level of C reactive protein in these patients decreased from 3.3 mg/dL to 0.20 mg/dL, systemic feature score decreased from 2.3 to 0.8, and three of four patients who did not achieve ACR50 before the interval shortening fulfilled the ACR50 criteria (but were regarded as non-responders in the analysis).

safetyIn the double-blind phase, more patients in the tocilizumab group (11 patients) than in the placebo group (8 patients) had adverse events, with the most frequent adverse event being nasopharyngitis (table 2). No serious adverse events were observed. Cumulatively, 11 serious adverse events occurred in seven patients. Three serious adverse events were not related to tocilizumab: malignant lymphoma (the patient was rediagnosed with lymphoma as an underlying disease, not adult-onset Still’s disease), anaphylactic shock to contrast enhancement in CT, and drug eruption to a pain killer. Instances of serious infection

included one patient with subcutaneous abscess, cellulitis and spleen abscess, one patient with cellulitis, and one patient with pneumonia. One patient was hospitalised because of adult-onset Still’s disease exacerbation, which was ameliorated by the shortening of tocilizumab interval. Aseptic necrosis in the hips of one patient was highly likely due to high-dose gluco-corticoids. Anaphylactic shock to tocilizumab was observed in one patient. Infusion reaction occurred in five patients, of whom four received tocilizumab and one received placebo. Adverse events led to tocilizumab cessation in two patients. Antitocili-zumab antibody was found in two patients. No death or macro-phage-activated syndrome was reported. None of the patients who received tocilizumab infusion with a shorter interval than 2 weeks developed serious adverse events.

dIsCussIOnIn this double-blind, randomised, placebo-controlled phase III trial, we tried to prove the efficacy of tocilizumab in patients with adult-onset Still’s disease, but the primary endpoint of ACR50 response at week 4 did not show statistical superiority of tocilizumab to placebo because of limitations in patient accrual, study design and appropriateness of endpoints. It showed a significantly greater improvement in systemic feature score and larger decrease in glucocorticoid dose in patients who received tocilizumab than in patients who received placebo, although the validity of the statistical differences is uncertain because the study was underpowered. Despite these limitations, this has been the first randomised, placebo-controlled trial ever conducted in patients with adult-onset Still’s disease and has suggested the efficacy of anti-interleukin-6 treatment on the disease.

Although almost twice the number of patients who received tocilizumab achieved ACR50 through the double-blind phases, the primary endpoint of ACR50 response rate 4 weeks after the start of tocilizumab or placebo treatment was not statisti-cally met. Three factors could be related to the non-significance

1727Kaneko Y, et al. Ann Rheum Dis 2018;77:1720–1729. doi:10.1136/annrheumdis-2018-213920

Clinical and epidemiological research

Table 2 Adverse events by group

double-blind phase Cumulative data*

Placebo(n=14)

Tocilizumab(n=13)

Tocilizumab(n=27)

Number of adverse events 12 38 182

Number of patients 8 (57.1) 11 (84.6) 27 (100.0)

Most frequently reported events†

Nasopharyngitis 0 (0.0) 3 (23.1) 11 (40.7)

Constipation 0 (0.0) 2 (15.4) 4 (14.8)

Anaemia 0 (0.0) 0 (0.0) 4 (14.8)

Insomnia 1 (7.1) 0 (0.0) 4 (14.8)

Dyslipidaemia 1 (7.1) 1 (7.7) 5 (18.5)

Drug eruption 0 (0.0) 0 (0.0) 4 (14.8)

Number of serious adverse events

0 0 11‡

Subcutaneous abscess 0 (0.0) 0 (0.0) 1 (3.7)

Cellulitis 0 (0.0) 0 (0.0) 2 (7.4)

Spleen abscess 0 (0.0) 0 (0.0) 1 (3.7)

Pneumonia 0 (0.0) 0 (0.0) 1 (3.7)

Adult-onset Still’s disease exacerbation

0 (0.0) 0 (0.0) 1 (3.7)

Anaphylaxis shock 0 (0.0) 0 (0.0) 2 (7.4)§

Drug eruption 0 (0.0) 0 (0.0) 1 (3.7)

Aseptic necrosis 0 (0.0) 0 (0.0) 1 (3.7)

Malignant lymphoma 0 (0.0) 0 (0.0) 1 (3.7)¶

Macrophage activation syndrome

0 (0.0) 0 (0.0) 0 (0.0)

Death 0 (0.0) 0 (0.0) 0 (0.0)

Infusion reaction 1 (7.1) 3 (23.1) 5 (18.5)

Antitocilizumab antibody – 1 (7.7) 2 (7.4)

Multiple occurrences in one patient were counted. Data are n (%) unless otherwise stated.*Cumulative data included safety profile for all patients while receiving tocilizumab.†Only adverse events that occurred in three or more participants in either group in the double-blind phase or in four or more participants in all participants while receiving tocilizumab are presented.‡11 serious adverse events in seven patients were observed.§One was attributed to tocilizumab and one to contrast enhancement during a CT.¶The underlying disease of the participant in the placebo group was found to be malignant lymphoma, rather than adult-onset Still’s disease.

of ACR50. One was a lack of statistical power because of the insufficient number of cases. We found it challenging to recruit patients with rare adult-onset Still’s disease who were refractory to glucocorticoids but in a non-life-threatening state who would tolerate placebo treatment, which reduced the sample size. The initial assumption of placebo achieving ACR50 of 15% was an underestimation, which also decreased the statistical power. We referred to the trial in patients with juvenile idiopathic arthritis14 to assume the 15%; however, the baseline prednisolone dose was much greater in our trial, which might partly have caused the high ACR50 achievement rate in the placebo group. Another potential reason was that tocilizumab could be more effective on systemic features than on arthritis despite our primary endpoint of ACR50, which focused on improvement in joint symp-toms. Several case reports have suggested dichotomous thera-peutic responses dependent on cytokines that are suppressed by biological agents, for example, articular symptoms with tumour necrosis factor antagonists, and systemic organ involvement with interleukin-1β or interleukin-6 antagonists.10 21–25 Our results are consistent with this notion, showing significant superiority to placebo on suppressing systemic inflammation. We should note,

however, that although the systemic feature score we adopted in our trial significantly separated tocilizumab from placebo, assessment for systemic features in adult-onset Still’s disease has not been established yet, which is a problem to be solved in the future. One further possibility for the unmet primary endpoint is that tocilizumab dose higher than biweekly intravenous 8 mg/kg adopted in our study might be necessary. Indeed, 6 of 26 patients in our study needed tocilizumab infusion with a shorter interval to suppress inflammation. The optimal dose of tocili-zumab might be different among patients, possibly depending on the soluble concentrations of interleukin-6 receptor before tocilizumab administration, as was observed in patients with rheumatoid arthritis.26

Direct inhibition of inflammatory cytokines has shown promising outcomes in the management of systemic inflam-matory manifestations of adult-onset Still’s disease. Although the pathogenic mechanisms of adult-onset Still’s disease are not fully understood, the pivotal role of innate immunity involving macrophage cell activation and Th1 cytokines, such as inter-leukin-1β, interferon-γ, tumour necrosis factor-α, interleukin-6, interleukin-8 and interleukin-18, has been demonstrated.27–33 Most data on the benefits of suppressing cytokines have stemmed from empirical observations through small case series, with patients who were treated with currently available biolog-ical agents targeting tumour necrosis factor, interleukin-1β and interleukin-6.10 34–36 Interleukin-6 is a pleiotropic cytokine involved in both innate and acquired immunity.37 Overproduc-tion of interleukin-6 can explain most symptoms of adult-onset Still’s disease, and high concentrations of interleukin-6 have been observed in the blood of patients with adult-onset Still’s disease. Our trial investigated tocilizumab, a monoclonal anti-body targeting interleukin-6 receptor, and suggested its poten-tial efficacy on systemic features of adult-onset Still’s disease and glucocorticoid dose reduction.

Another notable implication of this trial is that a domi-nant cytokine can vary according to patients. In several randomised controlled trials for systemic juvenile idiopathic arthritis conducted on tocilizumab, canakinumab, anakinra and rilonacept, some patients with systemic juvenile idiopathic arthritis responded to one drug but not to another type of drug.11–16 38 39 Moreover, there is a hypothesis that dominant cytokines may change from interleukin-1β at the beginning to interleukin-6 and subsequently to tumour necrosis factor in an individual.11 40–42 Likewise, adult-onset Still’s disease has compli-cated cytokine profiles. One study reported that patients with adult-onset Still’s disease could be divided into two types of disease according to distinct interleukin-6-based and interleu-kin-18-based cytokine profiles.32 As such, some patients with adult-onset Still’s disease might be well controlled by suppressing interleukin-18, if available. A recent, non-controlled study inves-tigating an interleukin-18 inhibitor showed somewhat effective-ness on adult-onset Still’s disease.42

Although glucocorticoids are drugs of great importance in the treatment of adult-onset Still’s disease, its toxicity causes iatro-genic problems. The decrease in glucocorticoids dose during the disease course frequently induces flares and subsequent increases in dose, which lead to the inevitable disadvantages of glucocor-ticoids.2 3 In the present trial, the reduction in oral glucocorti-coid dose was more than twice as much with tocilizumab as with placebo. Clinically relevant dose reductions of glucocorticoids in addition to greater decreases in systemic feature score in patients receiving tocilizumab than those receiving placebo are suggestive of the usefulness of tocilizumab in the management of adult-onset Still’s disease.

1728 Kaneko Y, et al. Ann Rheum Dis 2018;77:1720–1729. doi:10.1136/annrheumdis-2018-213920

Clinical and epidemiological research

A total of 11 serious adverse events were documented during the treatment of tocilizumab. Three serious adverse events were not related to tocilizumab: malignant lymphoma (the patient was rediagnosed with lymphoma as an underlying disease, not adult-onset Still’s disease), anaphylactic shock to contrast enhancement in CT and drug eruption to a pain killer. The aseptic necrosis of the hip joints was highly likely due to high-dose glucocorticoids, and the exacerbation of adult-onset Still’s disease was ameliorated by the shortening of tocilizumab interval. Instances of serious infection, the most concerning of adverse events, included combined skin abscess, cellulitis and spleen abscess in one patient, pneumonia in one patient, and cellulitis in one patient, with the former two patients withdrawn from the trial. No death or macrophage-activated syndrome was observed. We believe that biweekly intravenous tocilizumab at 8 mg/kg in patients with adult-onset Still’s disease can be justi-fied from the benefit–risk point of view, although careful use is essential.

Our study has some limitations. First, the sample size was small and underpowered, which led to the low statistical power resulting in the uncertainty of the results. Also, in such a study with small sample size, the imbalance in baseline characteristics might not be negligible, and the influence of one person on ratio calculations could be large, which might have caused both under-estimation and overestimation. Second, the primary endpoint should have included systemic feature assessment rather than simple ACR50 to prove the efficacy of tocilizumab on systemic inflammation. Third, the double-blind period without rescue of open-label tocilizumab might be short because there was a concern about continuing placebo treatment for a longer time in patients with active adult-onset Still’s disease. Fourth, although we did not set the upper limit of disease activity, the exclusion criteria regarding thrombocytopaenia and transaminase eleva-tion might have excluded eligible but severely ill patients. In addition, investigators must have selected patients at enrolment who they thought would tolerate placebo treatment. These could have led to selection bias. Fifth, the high response to placebo might be due to the short duration of prior glucocorticoid treat-ment and the baseline glucocorticoid dose. Sixth, the optimal usage of glucocorticoids at tocilizumab initiation is unclear. Patients in the placebo group received higher doses of baseline glucocorticoids and a longer period of glucocorticoid treatment without dose reduction at tocilizumab initiation, and in this context it appears that higher ACR responses were obtained in the placebo group than in the tocilizumab group in the open-label phase. Despite these limitations, this is the first double-blind, randomised, placebo-controlled trial demonstrating the efficacy of tocilizumab on relieving systemic inflammation and glucocorticoid sparing, and our findings provide useful insights into the management of adult-onset Still’s disease.

In conclusion, we suggest that inhibiting interleukin-6 with tocilizumab is effective in patients with adult-onset Still’s disease refractory to glucocorticoid therapy on suppressing systemic features and glucocorticoid dose reduction. However, the efficacy was not proven in our trial because of limitations in patient accrual, study design and appropriateness of endpoints. A well-designed, properly powered study will substantiate our preliminary findings.

Author affiliations1Division of Rheumatology, Department of Internal Medicine, Keio University School of Medicine, Tokyo, Japan2Division of Rheumatology, Department of Internal Medicine, Toho University Ohashi Medical Center, Tokyo, Japan

3Department of Allergy and Clinical Immunology, Chiba University Hospital, Chiba, Japan4Department of Hematology and Rheumatology, Tohoku University Graduate School of Medicine, Sendai, Japan5Department of Rheumatology and Clinical Immunology, Kyoto University, Kyoto, Japan6Department of Respiratory Medicine and Clinical Immunology, Graduate School of Medicine, Osaka University, Osaka, Japan7The First Department of Internal Medicine, School of Medicine, University of Occupational and Environmental Health, Kitakyushu, Japan8Department of Preventive Medicine and Public Health, Biostatistics at Clinical and Translational Research Center, Keio University School of Medicine, Tokyo, Japan

Acknowledgements The authors acknowledge and thank all investigators, participants and staff in this study. Shinsuke Yasuda (Hokkaido University, Sapporo, Japan) and Takayuki Sumida (Tsukuba University, Tsukuba, Japan) are other investigators of this trial.

Contributors YK, HK, TA and TT designed the study in collaboration with the Japanese Pharmaceuticals and Medical Devices Agency. YK, KI, TI, KM, HT, YT and TT were involved in patient recruitment, collecting data and managing their clinical research sites. All authors were involved in writing the manuscript and approved the final version. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Funding This study was supported by the Clinical Trial on Development of New Drugs and Medical Devices from the Japanese Ministry of Health, Labour and Welfare under grant number H23-実用化(臨床)-指定-005, and the Japan Agency for Medical Research and Development under grant number JP15lk0103004. Treatment drugs were provided by Chugai Pharmaceutical. The funders had no role in the study design, data collection, data analysis, data interpretation or writing of the report.

Competing interests YK has received grants or speaking fees from AbbVie, Astellas, Ayumi, Bristol-Myers Squibb, Chugai, Eisai, Eli Lilly, Hisamitsu, Janssen, Kissei, Pfizer, Sanofi, Takeda, Mitsubishi-Tanabe and UCB. HK has received grants or speaking fees from AbbVie, Ayumi, Chugai, Mitsubishi-Tanabe, Astellas, Takeda, Eisai, Novartis, Sanofi, Eli Lilly, GlaxoSmithKline, Bristol-Myers Squibb, Pfizer and Janssen. KI has received grants or speaking fees from Tanabe-Mitsubishi Pharma, AbbVie, Bristol-Myers KK, Eisai, UCB, Pfizer Japan, Takeda Pharmaceutical, Eli Lilly, Astellas Pharma, Chugai Pharmaceutical and Kyowa Hakko Kirin. IT has received grants or speaking fees from Chugai, Ono, Pfizer, Mitsubishi-Tanabe and Astellas. KM has received grants or speaking fees from AbbVie, Chugai, Mitsubishi-Tanabe, Astellas, Takeda, Eisai, UCB, Daiichi Sankyo, Eli Lilly, GlaxoSmithKline, Bristol-Myers Squibb, Pfizer and Janssen. HT has received grants or speaking fees from Pfizer, Daiichi Sankyo and Mitsubishi-Tanabe. YT has received grants or speaking fees from Mitsubishi-Tanabe, Takeda, Bristol-Myers, Chugai Pharmaceutical, Astellas Pharma, AbbVie, MSD, Eli Lilly, YL Biologics, Daiichi Sankyo, Sanofi, Janssen, Pfizer, Kyowa Kirin, Eisai and Ono. HT and TA have nothing to declare. TT has received research grants or speaking fees from Astellas Pharma, Bristol-Myers KK, Chugai Pharmaceutical, Daiichi Sankyo, Takeda Pharmaceutical, Teijin Pharma, AbbVie GK, Asahi Kasei Pharma, Mitsubishi-Tanabe Pharma, AstraZeneca KK, Eli Lilly Japan KK, Novartis Pharma KK, AbbVie GK, Nippon Kayaku, Janssen Pharmaceutical KK, Taiho Pharmaceutical and Pfizer Japan.

Patient consent Obtained.

ethics approval The institutional review board at each site approved the study.

Provenance and peer review Not commissioned; externally peer reviewed.

references1. CastañedaS,BlancoR,González-GayMA.Adult-onset Still’s disease: Advances in the

treatment.Best Pract Res Clin Rheumatol2016;30:222–38.2. NishinaN,KanekoY,KamedaH,et al.The effect of tocilizumab on preventing

relapses in adult-onset Still’s disease: A retrospective, single-center study.Mod Rheumatol2015;25:401–4.

3. AarntzenEH,van RielPL,BarreraP.Refractory adult onset Still’s disease and hypersensitivity to non-steroidal anti-inflammatory drugs and cyclo-oxygenase-2inhibitors:are biological agents the solution?Ann Rheum Dis2005;64:1523–4.

4. FujiiT,AkizukiM,KamedaH,et al.Methotrexate treatment in patients with adult onset Still’s disease--retrospective study of 13 Japanese cases.Ann Rheum Dis1997;56:144–8.

5. FautrelB,BorgetC,RozenbergS,et al.Corticosteroid sparing effect of low dose methotrexate treatment in adult Still’s disease.J Rheumatol1999;26:373–8.

6. KötterI,WackerA,KochS,et al.Anakinra in patients with treatment-resistant adult-onset Still’s disease: four case reports with serial cytokinemeasurementsand a review of the literature.Semin Arthritis Rheum2007;37:189–97.

7. NaumannL,FeistE,NatuschA,et al.IL1-receptor antagonist anakinra provides long-lasting efficacy in the treatment of refractory adult-onset Still’sdisease.Ann Rheum Dis2010;69:466–7.

1729Kaneko Y, et al. Ann Rheum Dis 2018;77:1720–1729. doi:10.1136/annrheumdis-2018-213920

Clinical and epidemiological research

8. MaeshimaK,IshiiK,IwakuraM,et al.Adult-onset Still’s disease with macrophage activation syndrome successfully treated with a combination of methotrexateandetanercept.Mod Rheumatol2012;22:137–41.

9. CiprianiP,RuscittiP,CarubbiF,et al.Tocilizumab for the treatment of adult-onset Still’s disease: results from a case series.Clin Rheumatol2014;33:49–55.

10. SuematsuR,OhtaA,MatsuuraE,et al.Therapeutic response of patients with adult Still’s disease to biologic agents: multicenter results in Japan.Mod Rheumatol2012;22:712–9.

11. WoernerA,UettwillerF,MelkiI,et al.Biological treatment in systemic juvenile idiopathic arthritis: achievement of inactive disease or clinical remissionona first, second or third biological agent.RMD Open2015;1:e000036.

12. YokotaS,ImagawaT,MoriM,et al.Long-term treatment of systemic juvenile idiopathic arthritis with tocilizumab: results of an open-label extensionstudyin Japan.Ann Rheum Dis2013;72:627–8.

13. YokotaS,ImagawaT,MoriM,et al.Efficacy and safety of tocilizumab in patients with systemic-onset juvenile idiopathic arthritis: a randomised, double-blind,placebo-controlled,withdrawal phase III trial.Lancet2008;371:998–1006.

14. De BenedettiF,BrunnerHI,RupertoN,et al.Randomized trial of tocilizumab in systemic juvenile idiopathic arthritis.N Engl J Med2012;367:2385–95.

15. QuartierP,AllantazF,CimazR,et al.A multicentre, randomised, double-blind, placebo-controlled trial with the interleukin-1 receptor antagonist anakinrainpatients with systemic-onset juvenile idiopathic arthritis (ANAJIS trial).Ann Rheum Dis2011;70:747–54.

16. RupertoN,BrunnerHI,QuartierP,et al.Two randomized trials of canakinumab in systemic juvenile idiopathic arthritis.N Engl J Med2012;367:2396–406.

17. YamaguchiM,OhtaA,TsunematsuT,et al.Preliminary criteria for classification of adult Still’s disease.J Rheumatol1992;19:424–30.

18. VojinovicJ,DamjanovN,D’UrzoC,et al.Safety and efficacy of an oral histone deacetylase inhibitor in systemic-onset juvenile idiopathic arthritis.Arthritis Rheum2011;63:1452–8.

19. FelsonDT,AndersonJJ,BoersM,et al.American College of Rheumatology. Preliminary definition of improvement in rheumatoid arthritis.Arthritis Rheum1995;38:727–35.

20. European Medicines Agency,2009.ICH Topic E 2 A: clinical safety data management: definitions and standards for expedited reporting. Availablefrom:http://www. ema. europa. eu/ docs/ en_ GB/ document_ library/ Scientific_ guideline/ 2009/ 09/ WC500002749. pdf

21. SavageE,WazirT,DrakeM,et al.Fulminant myocarditis and macrophage activation syndrome secondary to adult-onset Still’s disease successfully treatedwithtocilizumab.Rheumatology2014;53:1352–3.

22. SumidaK,UbaraY,HoshinoJ,et al.Etanercept-refractory adult-onset Still’s disease with thrombotic thrombocytopenic purpura successfully treated withtocilizumab.Clin Rheumatol2010;29:1191–4.

23. ElkayamO,JiriesN,DranitzkiZ,et al.Tocilizumab in adult-onset Still’s disease: the Israeli experience.J Rheumatol2014;41:244–7.

24. Ortiz-SanjuánF,BlancoR,Calvo-RioV,et al.Efficacy of tocilizumab in conventional treatment-refractory adult-onset Still’s disease: multicenter retrospectiveopen-labelstudy of thirty-four patients.Arthritis Rheumatol2014;66:1659–65.

25. Lo GulloA,CarusoA,PipitoneN,et al.Canakinumab in a case of adult onset still’s disease: efficacy only on systemic manifestations.Joint Bone Spine2014;81:376–7.

26. NishinaN,KikuchiJ,HashizumeM,et al.Baseline levels of soluble interleukin-6 receptor predict clinical remission in patients with rheumatoid arthritistreatedwithtocilizumab: implications for molecular targeted therapy.Ann Rheum Dis2014;73:945–7.

27. MariaAT,Le QuellecA,JorgensenC,et al.Adult onset Still’s disease (AOSD) in the era of biologic therapies: dichotomous view for cytokine and clinical expressions.Autoimmun Rev2014;13:1149–59.

28. ChenDY,LanJL,LinFJ,et al.Proinflammatory cytokine profiles in sera and pathological tissues of patients with active untreated adult onsetStill’sdisease.J Rheumatol2004;31:2189–98.

29. FujiiT,NojimaT,YasuokaH,et al.Cytokine and immunogenetic profiles in Japanese patients with adult still’s disease. association with chronic articular disease.Rheumatology2001;40:1398–404.

30. IchidaH,KawaguchiY,SugiuraT,et al.Clinical manifestations of Adult-onset Still’s disease presenting with erosive arthritis: Association with low levelsofferritin and Interleukin-18.Arthritis Care Res2014;66:642–6.

31. ChoiJH,SuhCH,LeeYM,et al.Serum cytokine profiles in patients with adult onset Still’s disease.J Rheumatol2003;30:2422–7.

32. InoueN,ShimizuM,TsunodaS,et al.Cytokine profile in adult-onset Still’s disease: Comparison with systemic juvenile idiopathic arthritis.Clin Immunol2016;169:8–13.

33. RauM,SchillerM,KrienkeS,et al.Clinical manifestations but not cytokine profiles differentiate adult-onset Still’s disease and sepsis.J Rheumatol2010;37:2369–76.

34. FautrelB,SibiliaJ,MarietteX,et al.Tumour necrosis factor alpha blocking agents in refractory adult Still’s disease: an observational study of 20 cases.Ann Rheum Dis2005;64:262–6.

35. LequerréT,QuartierP,RoselliniD.Société Francophone pour la Rhumatologie et les Maladies Inflammatoires en Pédiatrie (SOFREMIP); Club RhumatismesetInflammation (CRI). Interleukin-1 receptor antagonist (anakinra) treatment in patients with systemic-onsetjuvenileidiopathic arthritis or adult onset Still disease: preliminary experience in France.Ann Rheum Dis2008;67:302–8.

36. NordströmD,KnightA,LuukkainenR,et al.Beneficial effect of interleukin 1 inhibition with anakinra in adult-onset Still’s disease. An open, randomized,multicenterstudy.J Rheumatol2012;39:2008–11.

37. NishimotoN,KishimotoT.Interleukin 6: from bench to bedside.Nat Clin Pract Rheumatol2006;2:619–26.

38. LovellDJ,GianniniEH,ReiffAO,et al.Long-term safety and efficacy of rilonacept in patients with systemic juvenile idiopathic arthritis.Arthritis Rheum2013;65:2486–96.

39. IlowiteNT,PratherK,LokhnyginaY,et al.Randomized, double-blind, placebo-controlled trial of the efficacy and safety of rilonacept in the treatment of systemicjuvenileidiopathic arthritis.Arthritis Rheumatol2014;66:2570–9.

40. ReiffA.Treatment of systemic juvenile idiopathic arthritis with tocilizumab - the role of anti-interleukin-6 therapy afteradecade of treatment.Biol Ther2012;2:1.

41. GrevichS,ShenoiS.Update on the management of systemic juvenile idiopathic arthritis and role of IL-1 and IL-6 inhibition.Adolesc Health Med Ther2017;8:125–35.

42 Gabay C, Fautrel B, Rech J, et al. Open-label, multicentre, dose-escalating phase II clinical trial on the safety and efficacy of tadekinig alfa (IL-18BP) in adult-onset Still’s disease. Ann Rheum Dis 2018;77:840–7.

1730 Balint PV, et al. Ann Rheum Dis 2018;77:1730–1735. doi:10.1136/annrheumdis-2018-213609

Clinical and epidemiological research

ExtEndEd rEport

Reliability of a consensus-based ultrasound definition and scoring for enthesitis in spondyloarthritis and psoriatic arthritis: an OMERACT US initiativepeter V Balint,1 Lene terslev,2 philippe Aegerter,3 George Arthur Willem Bruyn,4 Isabelle Chary-Valckenaere,5 Frederique Gandjbakhch,6 Annamaria Iagnocco,7 Sandrine Jousse-Joulin,8 Ingrid Möller,9 Esperanza naredo,10 Wolfgang A Schmidt,11 richard J Wakefield,12 Maria-Antonietta d’Agostino,13 on behalf of the oMErACt Ultrasound task Force members

To cite: Balint pV, terslev L, Aegerter p, et al. Ann Rheum Dis 2018;77:1730–1735.

Handling editor Josef S Smolen

For numbered affiliations see end of article.

Correspondence toprofessor Maria-Antonietta d’Agostino, rheumatology department, Ambroise paré Hospital, Boulogne-Billancourt 92100, France; maria- antonietta. dagostino@ apr. aphp. fr

received 17 April 2018revised 21 July 2018Accepted 23 July 2018published online First 3 August 2018

© Author(s) (or their employer(s)) 2018. no commercial re-use. See rights and permissions. published by BMJ.

AbsTrACTObjectives to evaluate the reliability of consensus-based ultrasound (US) definitions of elementary components of enthesitis in spondyloarthritis (SpA) and psoriatic arthritis (psA) and to evaluate which of them had the highest contribution to defining and scoring enthesitis.Methods Eleven sonographers evaluated 40 entheses from five patients with SpA/psA at four bilateral sites. nine US elementary lesions were binary-scored: hypoechogenicity, thickened insertion, enthesophytes, calcifications, erosions, bone irregularities, bursitis and doppler signal inside and around enthesis. Kappa statistics were used to evaluate reliability. Sonographers were also asked to state which lesions can be considered as inflammatory or structural and should be included in the final definition of enthesitis. only the lesions, scored as present in at least 75% of the entheses considered as having an enthesitis, were included in the final definition. results the prevalence of detected lesions was quite low except for enthesophytes (55%) and bone irregularities (54%). reliability ranged from poor to good (the lowest for thickened enthesis (kappa 0.1 (95% CI 0 to 0.7)) and the highest for enthesophytes (kappa 0.6 (95% CI 0.5 to 0.7)). When adjusted for low prevalence, kappa values increased for all lesions, with the best result observed for detecting doppler signal at insertion (0.9) and for bursitis (0.8). the US components included in the final definition were hypoechogenicity, increased thickness at enthesis, erosions and calcifications/enthesophytes and doppler signal at insertion. Conclusion By using a consensus-based stepwise approach, a final reliable US score and definition of enthesitis in SpA/psA were produced. Further studies are sought for implementing this score in clinical trials and practice.

Enthesitis is typically defined as an inflammation of the insertion of tendons, ligaments, aponeurosis and capsules into the bone, and it is considered a pathological, clinical and imaging hallmark of the spondyloarthritis (SpA) group, including psoriatic arthritis (PsA).1–3 Assessments in Spondyloarthritis International Working Group (ASAS) and the Group for Research and Assessment of Psoriasis and Psoriatic Arthritis (GRAPPA) have recommended

enthesitis as one of the outcome domains for assessing disease activity and response in SpA (axial and peripheral) and PsA.4 5

The European League Against Rheumatism and GRAPPA also have treatment and management recommendations for predominant entheseal disease in PsA; although different in their recom-mendations, ultrasound (US) evaluation is an accepted method for detecting enthesitis in both.5–7

Since 1994, US has been used for evaluating peripheral enthesitis in patients with SpA in both lower and upper limb entheses.8–11 B-mode and Doppler US (colour and power) both depict the structure and vascularity of the enthesis. Together they are helpful in detecting subclinical enthesitis, measuring disease activity and evaluating treat-ment effect.12–14 In recent years, several enthesitis scoring systems have been published; however, as each is different by incorporating different US elementary lesions, comparisons across studies and the use of US as outcome measurement instru-ment of enthesitis, in multicentre studies, are problematic.15

Within the Outcome Measures in Rheumatology (OMERACT) US Working Group, a subtask force for enthesitis was created in order to produce a standardised, agreed definition and reliable scoring system. The group previously published a systematic literature review highlighting the great variability in the definition of enthesitis applied in US studies since 1994, and in particular the great variability of the definitions of its constituent elementary compo-nents.16 Following a standardised stepwise consen-sus-based approach, the group initially agreed on the definitions of each potential elementary component and then decided which of them should be included in the definition of a US-detected enthesitis for SpA/PsA.17 The reliability for detecting these elementary components was subsequently tested in a web-based exercise on static images showing a good reliability among sonographers.17 The aim of the final step presented here was to evaluate, in a live scanning exercise on patients with SpA and PsA, the reli-ability of the acquisition of the consensus-based definitions of elementary components in order to identify which of them were scored as present when a US enthesitis was detected.

1731Balint PV, et al. Ann Rheum Dis 2018;77:1730–1735. doi:10.1136/annrheumdis-2018-213609

Clinical and epidemiological research

Table 1 Consensus-based definitions of ultrasound (US) elementary components

Us modeElementary component Consensus-based Us definition

B-mode Bone irregularity Bone profile changes not including definite enthesophytes nor bone erosions.

B-mode Calcification Hyperechoic foci, with or without acoustic shadow, detected at the enthesis (<2 mm from the cortical bone).

B-mode Enthesophyte Enthesophyte was defined as a step-up of bony prominence, seen in two perpendicular planes at the end of the bone contour of the enthesis.

B-mode Erosion Cortical break with a step-down contour defect, seen in two perpendicular planes, at the insertion of the enthesis.

B-mode Hypoechogenicity at the enthesis

Lack of the homogeneous fibrillar pattern in the enthesis (<2 mm from the cortical bone) with loss of the tightly packed echogenic lines after correcting for anisotropy.

B-mode Thickened enthesis Increased thickness of the tendon insertion into the bone (<2 mm from the cortical bone) as compared with the body of tendon, with or without blurring of the tendon margins.

Doppler mode Doppler signal at enthesis

Doppler signal seen at bone insertion (<2 mm from the cortical bone), different from reflecting surface artefact or nutrition vessel signal, with or without cortical irregularities, erosions or enthesophytes.

Doppler mode Doppler signal outside the enthesis

Doppler signal far from the enthesis (ie, >2 mm from the cortical bone, within the body of tendon) and clearly different from nutrition vessel signals.

B-mode and Doppler

Bursitis An enlargement (ie, increase in diameter of the bursa), with well-defined, anechoic or hypoechoic area inside, with or without Doppler signal.

METHOdsstudy design and settingA 2-day workshop was conducted in Paris, France, in order to evaluate the intraobserver and interobserver reliability of the detection of each elementary component and of a global defi-nition of enthesitis, by performing a live scanning exercise on patients with SpA and PsA. Before the start of the study, the group met to agree on the scanning protocol and on the proce-dure to use for deciding which elementary component should be included in the final definition of a US-detected enthesitis, as well as which of them should be defined as inflammatory or structural lesions. Each day was divided into a morning and an afternoon session (which was aimed at evaluating the intraob-server reliability).

Patient selectionTo ensure that all elementary lesions would be represented, five voluntary patients with SpA or PsA were recruited from the outpatient clinic of the Rheumatology Department of the Hôpital Ambroise Paré, Boulogne-Billancourt, France, and pres-canned by an independent investigator. The participant sonog-raphers were blinded to clinical information, and they were informed that some of the patients may be healthy subjects.

UltrasonographersEleven rheumatologists from eight countries (Denmark, France, Germany, Hungary, Italy, The Netherlands, Spain and UK) with extensive experience in US (>10 years and European Federation of Societies for Ultrasound in Medicine and Biology (EFSUMB) level 3) participated in the present study.

Us evaluationEach patient was assigned to a single machine and the ultrasonog-raphers then moved from one patient to the next in a predefined random sequence. Maximum scanning time was 20 min per patient, including time to fill out the prepared scoring sheet. The data were collected immediately after the session to ensure no communication between examiners.

Each patient underwent bilateral dynamic B-mode and power Doppler (PD) US examination of four entheseal sites: lateral condyle of the elbow, superior and inferior poles of the patella, and calcaneal insertion of the Achilles tendon.

The US examination of the lateral epicondyle at the common extensor tendon insertion was performed with the elbows flexed and the palms in a prayer’s position. The examination of the superior pole of the patella (quadriceps tendon insertion) and the inferior pole of the patella (patellar tendon origin) was performed with the patient in supine position with the knee flexed at 30°–45° for B-mode and extended knee position for PD examination. The Achilles entheses were examined with the patient in prone position with the feet overhanging the edge of the examination table in neutral position.

US was performed using Esaote MyLab 70 XVG machines equipped with 6–18 MHz transducers and identical B-mode and Doppler settings, which were optimised and fixed for all exam-inations. Doppler parameters were pulse repetition frequency of 500 Hz and Doppler frequency of 10.1 MHz. During the examination each enthesis was examined first in the longitudinal axis and then at 90° in short axis in order to evaluate the whole structure.

Us definitions and measurementsAn enthesis was defined as the 2 mm zone of soft tissue adja-cent to the bone cortex, based on published data18 and following the consensus session performed prior to the practical exercise. Each enthesis was evaluated for the following B-mode and PD elementary components: (1) hypoechogenicity, (2) increased thickness, (3) enthesophyte, (4) calcification, (5) bone erosion, (6) bone irregularity and (7) Doppler signal, which were defined as already published.17 In addition, the presence of Doppler signal outside the enthesis (8) as well as the presence of an adja-cent bursitis (9) at the Achilles tendon insertion were also eval-uated. Each of the above-mentioned components was defined as present or absent. Sonographers were also asked (1) to give an overall final judgement on the presence of enthesitis (yes or no); (2) to state which elementary component should be classified as inflammatory or structural lesion; and (3) which of them should be included in the final definition of a US-detected enthesitis. Table 1 shows the definition of each elementary lesion (including bursitis) used during the exercise.

statistical analysisIntraobserver and interobserver reliability was estimated based on measuring all lesions’ prevalence, and calculating observed agreement and kappa values with 95% CI. These coefficients were interpreted according to Landis and Koch.19

1732 Balint PV, et al. Ann Rheum Dis 2018;77:1730–1735. doi:10.1136/annrheumdis-2018-213609

Clinical and epidemiological research

As a result of our study using a fully crossed design, with repeated measures across 40 entheses (5 patients) and 11 rheu-matologists, Cohen ƙ for measuring agreement between 2 raters had to be extended for use with multiple raters (Light’s kappa).20 21 Thus, we used a crossed design in which all raters1–11 evaluated all patients in duplicates (two tests), in all anatomical positions (four positions: Achilles insertion, inferior pole and superior pole of patella, and lateral epicondyle insertion) and on both sides of the body (right, left). CIs of kappa were obtained after 1000 bootstrap samples. In order to take into account the observed low prevalence of the lesions, or possible discrepancies in prevalence across observers, a prevalence and bias adjusted kappa (PABAK) was additionally used to evaluate the reli-ability.22 To get a synthetic measure for interobserver observed agreement, we applied the same calculation (mean of all possible pairs) as for Light’s kappa.

In order to define which elementary components best contrib-uted to the presence of a US-detected enthesitis, participants were asked at the end of the US evaluation to give an overall judge-ment on the presence or absence of a US-detected enthesitis, and to indicate which elementary components influenced such judge-ment, and therefore included in the final definition of enthesitis. However, only the components detected in more than 75% of the entheseal sites with a final judgement of enthesitis, and which obtained an acceptable intraobserver and interobserver reliability (>0.6), were included in the final definition. In addi-tion, sonographers were asked to confirm which findings were characterised as inflammatory or as structural lesions.

rEsUlTsAll patients were women and fulfilled the ASAS classification criteria for axial and/or peripheral SpA (two of them also fulfilled the CASPAR, Classification Criteria for Psoriatic Arthritis). The mean Bath Ankylosis Spondylitis Activity Index (BASDAI) was 4.5/10, the mean C-reactive protein (CRP) was 5 mg/L, all patients were HLA B27-positive, and all were treated with non-steroidal anti-inflammatory drugs (NSAIDs) and two of them also with antitumour necrosis factor-alpha inhibitors.

The intraobserver and interobserver kappa values, as well as the prevalence of the observed findings, are shown in table 2.

Prevalence and agreementThe mean prevalence of observed findings was variable across examiners. The highest median prevalence of positive findings, associated with the lowest variability across sonographers, was observed for enthesophytes (44%, range 36%–54%), while bone irregularities exhibited larger variability (median 46%, range 23%–71%). The lowest median prevalence was observed for bursitis (median 2%, range 0%–12%), Doppler signal at inser-tion (4%) and erosions (5%).Therefore, except for entheso-phytes, which was not prone to low prevalence bias, all other elementary components were also evaluated using PABAK. Regarding the observed agreement: the mean inter observer agreement was moderate to good: 70 % for enthesophytes, 60 % for calcifications, 60 % for bone irregularities and 60 % Doppler signal inside the enthesis. Very high intraobserver observed agree-ments (70%–100%) were found for all components (table 2).

Intraobserver reliabilityThe best intraobserver reliability without adjustment was seen for erosion, hypoechogenicity at the enthesis, Doppler at the enthesis and outside, and for bursitis (for all components, the kappa ranged from 0 to 1.0, 95% CI (0 to 0) ~ (1.0 to 1.0). The

lowest reliability was seen for bone irregularities, calcifications and enthesophytes (kappa: −0.1 to 0.5, 95% CI (−0.1 to 0.4) ~ (0.3 to 0.7); kappa: 0.1–0.5, 95% CI (−0.1 to 0.4) ~ (0.3 to 0.7); and 0.1–0.5, 95% CI (−0.2 to 0.4) ~ (0 to 1.0), respec-tively). The use of PABAK improved the reliability of all compo-nents (median PABAK was 0.8 or over). The complete results are shown in table 2.

Interobserver reliabilityThe best interobserver reliability without adjustment was seen for enthesophytes (kappa 0.6 (95% CI 0.5 to 0.7), followed by Doppler signal at insertion (0.4 (95% CI 0.3 to 0.8)). The lowest reliability was found for thickened enthesis with a poor kappa of 0.1 (95% CI 0 to 0.7).

The best PABAK values were seen for detecting Doppler signal at insertion and for bursitis with values of 0.9 and 0.8, respec-tively, while the lowest PABAK was obtained for bone irregulari-ties with a PABAK of 0.3. As expected the use of PABAK improved the crude in case of low prevalence or discordance on observed prevalence (table 2). Table 3 shows which US components were the most frequently used for the final scoring of enthesitis and therefore included in the final definition. The classification of each elementary component as inflammatory or structural lesion resulted in a perfect agreement (100%), as shown in table 3. The elementary components with (1) an acceptable intraobserver and interobserver reliability, (2) which were scored as present in more than 75% of the sites with enthesitis, and (3) which were indicated by the sonographer as important for the final definition of enthesitis were the following: (1) Doppler at the enthesis, (2) increased in thickness at the enthesis, (3) calcifi-cations, (4) erosions and (5) enthesophytes. However, despite hypoechogenicity at enthesis insertion not reaching acceptable kappa (0.5), it was always scored positively when enthesitis was defined. Therefore, despite the low reliability, it was included in the final definition. Enthesophytes and calcifications were felt difficult to differentiate as they were very close to the cortical bone, but important for the final definition; therefore, it was agreed to score them as a unique element (enthesophyte/calcifi-cation) and to score present if at least one of the two would be detected. The final agreed definition of a US-detected enthesitis in patients with SpA and PsA was the following: hypoechoic and/or thickened insertion of the tendon close to the bone (within 2 mm from the bony cortex) which exhibits Doppler signal if active and which may show erosions and enthesophytes/calcifi-cations as a sign of structural damage (table 3). Sixteen out of 40 examined entheses were scored as having an enthesitis.

dIsCUssIOnIn order to validate US as a possible measurement instrument for enthesitis in SpA and PsA, we have followed a standardised step-wise validation procedure based on the analysis of each compo-nent that can contribute to the global definition of a US-detected enthesitis. The prevalence of each observed elementary compo-nent was variable across participants, probably due to the disease spectrum, despite the effort made for including a wide range of lesions, but also explained by the difficulty to differentiate some elementary components (ie, calcifications and enthesophytes) which can sometimes have the same US appearance. The lower prevalence of inflammatory components was probably due to the fact that the studied patients were all under treatment (NSAIDs and/or biologic treatment). We cannot exclude an influence of the gender on the prevalence and distribution of the elementary

1733Balint PV, et al. Ann Rheum Dis 2018;77:1730–1735. doi:10.1136/annrheumdis-2018-213609

Clinical and epidemiological research

Tabl

e 2

Sum

mar

y da

ta o

f int

raob

serv

er a

nd in

tero

bser

ver r

esul

ts a

naly

sing

B-m

ode

stru

ctur

al a

nd in

flam

mat

ory

and

Dopp

ler u

ltras

ound

com

pone

nts

Intr

aobs

erve

rIn

tero

bser

ver

Prev

alen

ce*

med

ian,

ra

nge

(%)

Obs

erve

d ag

reem

ent

med

ian,

ran

ge (%

)Ka

ppa

med

ian,

ran

ge(9

5% C

I)PA

bAK

med

ian,

ra

nge

Prev

alen

ce†

mea

n (%

)O

bser

ved

agre

emen

t m

ean

(%)

ligh

t’s k

appa

(95%

CI)

PAbA

K

B-m

ode

stru

ctur

al c

ompo

nent

s

Enth

esop

hyte

44, 3

8–54

70, 5

6–76

0.4,

0.1

–0.5

(−0.

1 to

0.4

) ~ (0

.3 to

0.7

)0.

4, 0

.1–0

.555

700.

6 (0

.5 to

0.7

)0.

6

Calc

ifica

tion

17, 4

–40

82, 6

0–96

0.3,

0.1

–0.5

(−0.

2 to

0.4

) ~ (0

to 1

.0)

0.6,

0.2

–0.9

2060

0.3

(0.2

to 0

.4)

0.6

Bone

irre

gula

rity

46, 2

3–71

66, 4

4–78

0.2,

−0.

1 to

0.5

(−0.

1 to

0.4

) ~ (0

.3 to

0.7

)0.

3, 0

.1–0

.854

600.

3 (0

.2 to

0.4

)0.

3

Eros

ion

5, 2

–26

94, 7

2–10

00.

4, 0

–1.0

(0 to

0) ~

(1.0

to 1

.0)

0.9,

0.4

–1.0

940

0.2

(0.2

to 0

.9)

0.7

B-m

ode

infla

mm

ator

y co

mpo

nent

s

Thic

kene

d en

thes

is9,

0–1

988

, 82–

100

0.4,

−0.

1 to

1.0

(−0.

1 to

0) ~

(1.0

to 1

.0)

0.8,

0.6

–1.0

1210

0.1

(0 to

0.7

)0.

6

Hy

poec

hoge

nici

ty15

, 4–3

182

, 66–

100

0.3,

0–1

.0 (0

to 0

) (N

A to

NA)

0.6,

0.3

–0.8

1840

0.2

(0.1

to 0

.8)

0.5

Dopp

ler c

ompo

nent

s

Dopp

ler s

igna

l at i

nser

tion

4, 2

–14

96, 8

8–10

00.

5, 0

–1.0

(0 to

0) ~

(1.0

to 1

.0)

0.9,

0.7

–1.0

660

0.4

(0.3

to 0

.8)

0.9

Dopp

ler s

igna

l out

side

8, 2

–88

90, 6

8–10

00.

5, 0

–1.0

(0 to

0) ~

(1 to

1)

0.8,

0.4

–1.0

1850

0.3

(0.1

to 0

.5)

0.6

Extr

a en

thes

eal c

ompo

nent

Bu

rsiti

s2,

0–1

296

, 4–1

000,

−0.

1 to

0.7

(−0.

1 to

0) ~

(0.2

to 1

.0)

0.9,

0.7

–1.0

650

0.2

(0.2

to 0

.3)

0.8

Fina

l dia

gnos

is

Enth

esiti

s as

glo

bal

12, 2

–44

96, 6

0–10

00.

6, 0

.1–1

.00

(0 to

0.4

) ~ (1

.0 to

1.0

)0.

9, 0

.2–1

.015

600.

4 (0

.2 to

0.8

)0.

6

*Mea

n pr

eval

ence

of p

ositi

ve fi

ndin

gs b

y so

nogr

aphe

r alo

ng th

e tw

o se

ssio

ns.

†Mea

n pr

eval

ence

of p

ositi

ve fi

ndin

gs a

mon

g so

nogr

aphe

rs a

t the

firs

t ses

sion

.N

A, n

ot a

pplic

able

; PAB

AK, p

reva

lenc

e and

bia

s ad

just

ed k

appa

.

1734 Balint PV, et al. Ann Rheum Dis 2018;77:1730–1735. doi:10.1136/annrheumdis-2018-213609

Clinical and epidemiological research

Table 3 Accepted inflammatory and structural components for ultrasound enthesitis definition

Accepted inflammatory components (100% of agreement)

Accepted structural components (100% of agreement)

Doppler signal at enthesis (figure 1A). Calcification/enthesophyte at enthesis (figure 1C).

Hypoechogenicity at enthesis (figure 1B). Erosion at enthesis (figure 1A).

Thickened enthesis (figure 1B).

Final definition of enthesitis

Hypoechoic and/or thickened insertion of the tendon close to the bone (within 2 mm from the bony cortex) which exhibits Doppler signal if active and which may show erosions and enthesophytes/calcifications as a sign of structural damage.

Figure 1 Ultrasound elementary components of enthesitis (A,B,C). Longitudinal scans of Achilles (A and C) and patellar (B) entheseal insertions, in B-mode and power Doppler ultrasound. (A) Achilles enthesis: the arrow points to the Doppler signal at the enthesis, while the open arrow points to the Doppler signal outside of the enthesis at the retrocalcaneal bursa. A, Achilles tendon; C, calcaneal bone surface; E, erosion; the dashed line represents 2 mm distance around the bony surface of the enthesis. (B) Patellar enthesis at the lower pole of the patella: the two arrows show the increased thickness of the enthesis, while the asterisk shows the hypoechogenicity at the enthesis insertion. P, patella bone surface; T, patellar tendon; the dashed line represents 2 mm distance around the bony surface of the enthesis. (C) Achilles enthesis: the arrow points to the calcification, and the open arrow points to the enthesophyte at the enthesis. A, Achilles tendon; C, calcaneal bone surface; colour dots: Doppler signals; the dashed line represents 2 mm distance around the bony surface of the enthesis.

components and on the inflammatory findings, as all patients examined were women.

When the reliability was adjusted for the prevalence (PABAK), a moderate to excellent interobserver reliability was observed espe-cially for Doppler signal, enthesophyte and erosion. However, the intraobserver reliability for enthesophyte and calcification remained quite poor. This phenomenon was already observed when we scored static images.17 When the ‘contribution’ of each component to the final definition of a US-detected enthesitis was evaluated, enthesophyte and calcification were always present. The same was observed for thickened insertion and the hypoechoge-nicity of the insertion. Therefore, it was decided to include them. Doppler signal at insertion and erosion was felt as really important for the definition of enthesitis and was unanimously included. Interestingly, the presence of bursitis was never mentioned as a single elementary component of enthesitis, as well as the presence of Doppler signal outside the enthesis (ie, tendonitis).

Therefore, bursitis and tendonitis were not included in the final definition of enthesitis. However, both components may be involved in more severe/extensive cases when the inflammation extends outside the defined margins of the enthesis (ie, <2 mm from the bony cortex), and it may be important to look for them

during US examination, but to define them as separate patho-logical findings and not as part of the enthesitis definition as observed in older studies.23 24

As a result of this study we were able to agree on which compo-nents belong to a structural and an inflammatory group. Doppler signal at the enthesis, hypoechogenicity and increased thickness of the enthesis were grouped as inflammatory components and erosions and calcifications/enthesophytes as structural components.

This study confirms the ability of US to detect the components that describe an enthesitis and the good reliability (according to the observed prevalence of the lesions) in scoring them. The combination of both structural and inflammatory lesions for defining enthesitis, as well as the decision to merge some components in a unique ‘US entity’ (such as calcifications and enthesophytes), is an innovative approach from our group to improve reliability and insure the easier application of this technique for evaluating the enthesis involvement in SpA and PsA. For the first time a standardised definition of a US-detected enthesitis has been proposed with a unanimous agreement. This definition clearly states that enthesitis by US should be searched for at the anatomical enthesis (ie, within 2 mm from the cortical bone insertion). The identified elementary components are not new since they were already proposed in previously developed scoring systems15; however, they were not all present in each scoring system at the same time and their definition was not always clearly established. For example, in the Spanish Enthesitis Index and in the Glasgow Ultrasound Enthesitis Scoring System, no Doppler activity was included10 25; in contrast, Doppler activity is part of both the Madrid Scoring Enthesitis Index and the French scoring system.13 26 The clear differentiation between lesions at the enthesis (ie, within 2 mm from the cortical bone) and lesions outside the enthesis (ie, in the tendon or into the bursa) proposed by our group will insure a greater reproduc-ibility and homogeneity of the US evaluation of enthesis across centres. The decision to score each component as present or absent will also facilitate the detection of an enthesitis.

1735Balint PV, et al. Ann Rheum Dis 2018;77:1730–1735. doi:10.1136/annrheumdis-2018-213609

Clinical and epidemiological research

At the moment the scoring system does not weigh either the extent of the enthesitis (ie, severity) or its activity (ie, inflamma-tory findings). In this scoring system the inflammatory compo-nents in B-mode or Doppler have the same weight. The future next steps will be to try to weigh these findings and to test their sensitivity to change.

With this study the OMERACT group finalises the process of development of a reliable and unanimously defined definition of enthesitis by US, including each elementary component. This is an important step towards ensuring a higher degree of homoge-neity between studies and daily clinical work, thereby facilitating the comparability of results.

Author affiliations13rd rheumatology department, national Institute of rheumatology and physiotherapy, Budapest, Hungary2Center for rheumatology and Spine diseases, Copenhagen University Hospital, rigshospitalet, Copenhagen, denmark3department of public Health and Biostatistics, UMr 1168 InSErM Université Versailles St-Quentin en Yvelines, GIrCI IdF, paris, France4department of rheumatology, MC Groep Hospitals, Lelystad, the netherlands5department of rheumatology, University Hospital of nancy, nancy, France6rheumatology department, University of paris VI, CHU pitié Salpétriére, paris, France7dipartimento di Scienze Cliniche e Biologiche, University of turin, turin, Italy8rheumatology department, Cavale Blanche Hospital and Brest occidentale University, Brest, France9department of rheumatology, Instituto poal de reumatología and University of Barcelona, Barcelona, Spain10department of rheumatology, Joint and Bone research Unit, Hospital Universitario Fundación Jiménez díaz and Autónoma University, Madrid, Spain11Immanuel Krankenhaus Berlin, Medical Centre for rheumatology Berlin-Buch, Berlin, Germany12Leeds Institute of rheumatic and Musculoskeletal Medicine, University of Leeds and nIHr Leeds Biomedical research Centre, Leeds, UK13rheumatology department, ApHp, Hôpital Ambroise paré, InSErM U1173, Labex Inflamex, Université Versailles St-Quentin en Yvelines, Boulogne-Billancourt, France

Acknowledgements We thank the patients who participated in the reliability session. We are grateful to Esaote SpA France for providing the ultrasound machines.

Collaborators oMErACt Ultrasound task Force members: Sibel Z Aydin, Artur Bachta, Marina Backhaus, Fred Joshua, david Bong, paz Collado, Eugenio de Miguel, Emilio Filippucci, Jane E Freeston, Walter Grassi, Marwin Gutierrez, Hilde B Hammer, david Kane, Helen I Keen, damien Loeuille, peter Mandl, Carlos pineda, Marcin Szkudlarek.

Contributors Study design: M-Ad’A, pA. Acquisition of data: AI, En, FG, GAWB, IC-V, IM, Lt, M-Ad’A, pVB, rJW, SJ-J, WAS. Analysis and interpretation of data: M-Ad’A, pA, Lt, pVB. Manuscript: M-Ad’A, Lt, pVB. Statistical analysis: M-Ad’A, pA.

Funding Unrestricted grant from AbbVie France.

Competing interests none declared.

Patient consent not required.

Ethics approval the study was performed according to a pre-specified protocol and conducted in accordance with the declaration of Helsinki and each patient gave written informed consent to participate.

Provenance and peer review not commissioned; externally peer reviewed.

RefeRences 1 Ball J. Enthesopathy of rheumatoid and ankylosing spondylitis. Ann Rheum Dis

1971;30:213–23.

2 Mander M, Simpson JM, McLellan A, et al. Studies with an enthesis index as a method of clinical assessment in ankylosing spondylitis. Ann Rheum Dis 1987;46:197–202.

3 Mandl p, navarro-Compán V, terslev L, et al. EULAr recommendations for the use of imaging in the diagnosis and management of spondyloarthritis in clinical practice. Ann Rheum Dis 2015;74:1327–39.

4 van der Heijde d, van der Linden S, dougados M, et al Ankylosing spondylitis: plenary discussion and results of voting on selection of domains and some specific instruments. J Rheumatol 1999;26:1003–5.

5 ritchlin Ct, Kavanaugh A, Gladman dd, et al. treatment recommendations for psoriatic arthritis. Ann Rheum Dis 2009;68:1387–94.

6 Gossec L, Smolen JS, ramiro S, et al. European League Against rheumatism (EULAr) recommendations for the management of psoriatic arthritis with pharmacological therapies: 2015 update. Ann Rheum Dis 2016;75:499–510.

7 Gossec L, Coates LC, de Wit M, et al. Management of psoriatic arthritis in 2016: a comparison of EULAr and GrAppA recommendations. Nat Rev Rheumatol 2016;12:743–50.

8 Lehtinen A, taavitsainen M, Leirisalo-repo M. Sonographic analysis of enthesopathy in the lower extremities of patients with spondylarthropathy. Clin Exp Rheumatol 1994;12:143–8.

9 Lehtinen A, Leirisalo-repo M, taavitsainen M. persistence of enthesopathic changes in patients with spondylarthropathy during a 6-month follow-up. Clin Exp Rheumatol 1995;13:733–6.

10 Balint pV, Kane d, Wilson H, et al. Ultrasonography of entheseal insertions in the lower limb in spondyloarthropathy. Ann Rheum Dis 2002;61:905–10.

11 Spadaro A, Iagnocco A, perrotta FM, et al. Clinical and ultrasonography assessment of peripheral enthesitis in ankylosing spondylitis. Rheumatology 2011;50:2080–6.

12 d’Agostino MA, Aegerter p, Bechara K, et al. How to diagnose spondyloarthritis early? Accuracy of peripheral enthesitis detection by power doppler ultrasonography. Ann Rheum Dis 2011;70:1433–40.

13 d’Agostino MA, Said-nahal r, Hacquard-Bouder C, et al. Assessment of peripheral enthesitis in the spondylarthropathies by ultrasonography combined with power doppler: a cross-sectional study. Arthritis Rheum 2003;48:523–33.

14 naredo E, Batlle-Gualda E, García-Vivar ML, et al. power doppler ultrasonography assessment of entheses in spondyloarthropathies: response to therapy of entheseal abnormalities. J Rheumatol 2010;37:2110–7.

15 d’Agostino MA, terslev L. Imaging Evaluation of the Entheses: Ultrasonography, MrI, and Scoring of Evaluation. Rheum Dis Clin North Am 2016;42:679–93.

16 Gandjbakhch F, terslev L, Joshua F, et al. Ultrasound in the evaluation of enthesitis: status and perspectives. Arthritis Res Ther 2011;13:r188.

17 terslev L, naredo E, Iagnocco A, et al. defining enthesitis in spondyloarthritis by ultrasound: results of a delphi process and of a reliability reading exercise. Arthritis Care Res 2014;66:741–8.

18 Aydin SZ, Bas E, Basci o, et al. Validation of ultrasound imaging for Achilles entheseal fibrocartilage in bovines and description of changes in humans with spondyloarthritis. Ann Rheum Dis 2010;69:2165–8.

19 Landis Jr, Koch GG. the measurement of observer agreement for categorical data. Biometrics 1977;33:159–74.

20 Fleiss JL. Measuring nominal scale agreement among many raters. Psychol Bull 1971;76:378–82.

21 Light rJ. Measures of response agreement for qualitative data: Some generalizations and alternatives. Psychol Bull 1971;76:365–77.

22 Byrt t, Bishop J, Carlin JB. Bias, prevalence and kappa. J Clin Epidemiol 1993;46:423–9.

23 Gisondi p, tinazzi I, El-dalati G, et al. Lower limb enthesopathy in patients with psoriasis without clinical signs of arthropathy: a hospital-based case-control study. Ann Rheum Dis 2008;67:26–30.

24 Feydy A, Lavie-Brion MC, Gossec L, et al. Comparative study of MrI and power doppler ultrasonography of the heel in patients with spondyloarthritis with and without heel pain and in controls. Ann Rheum Dis 2012;71:498–503.

25 Alcalde M, Acebes JC, Cruz M, et al. A sonographic enthesitic index of lower limbs is a valuable tool in the assessment of ankylosing spondylitis. Ann Rheum Dis 2007;66:1015–9.

26 de Miguel E, Cobo t, Muñoz-Fernández S, et al. Validity of enthesis ultrasound assessment in spondyloarthropathy. Ann Rheum Dis 2009;68:169–74.

1736 Michelsen B, et al. Ann Rheum Dis 2018;77:1736–1741. doi:10.1136/annrheumdis-2018-213463

Clinical and epidemiological research

ExtEndEd rEport

Can disease activity in patients with psoriatic arthritis be adequately assessed by a modified Disease Activity index for PSoriatic Arthritis (DAPSA) based on 28 joints?Brigitte Michelsen,1,2 Joseph Sexton,1 Josef S Smolen,3 daniel Aletaha,3 niels Steen Krogh,4 désirée van der Heijde,1,5 tore K Kvien,1 Merete Lund Hetland6,7

To cite: Michelsen B, Sexton J, Smolen JS, et al. Ann Rheum Dis 2018;77:1736–1741.

Handling editor Francis Berenbaum

► Additional material is published online only. to view please visit the journal online (http:// dx. doi. org/ 10. 1136/ annrheumdis- 2018- 213463)

For numbered affiliations see end of article.

Correspondence todr Brigitte Michelsen, department of rheumatology, department of Medicine, Hospital of Southern norway trust, Kristiansand 4604, norway; brigitte_ michelsen@ yahoo. no

received 24 March 2018revised 24 July 2018Accepted 17 August 2018published online First 20 September 2018

© Author(s) (or their employer(s)) 2018. no commercial re-use. See rights and permissions. published by BMJ.

Key messages

What is already known about this subject? ► Many databases and registries solely collect 28 joint counts and not the comprehensive 66/68 joint counts, not only in RA but also in PsA patients. Disease Activity index for PSoriatic Arthritis (DAPSA) including 66/68 joint counts has been validated in PsA, whereas a simplified DAPSA based on 28 joints has not been compared with the original version.

What does this study add? ► We used the Danish national quality registry DANBIO to explore psychometric performance of a modified DAPSA based on 28 rather than 66/68 joint counts. DAPSA28 showed overall good criterion, correlational and construct validity as well as sensitivity to change.

How might this impact on clinical practice or future developments?

► We recommend that the original DAPSA should be preferred and that 66/68 joint counts should be performed in patients with PsA to compute DAPSA. However, our study suggests that data sets with only 28-joint counts available can be used to calculate a modified DAPSA28.

AbsTrACTObjective to test the psychometric performance of a modified disease Activity index for pSoriatic Arthritis (dApSA) using 28 instead of 66 swollen/68 tender joint counts (SJC/tJC).Methods We included patients with psoriatic arthritis (psA) from the danish national quality registry dAnBIo, divided into examination (n=3157 patients, 23987 visits) and validation cohorts (n=3154 patients, 24160 visits). We defined dApSA28 = (28tJC × conversion factor1) + (28SJC × conversion factor2) + patient global (0–10VAS) + pain (0–10VAS) + C reactive protein (Crp) (mg/dL). Identification of the conversion factors was performed by generalised estimating equations in the examination cohort and evaluation of criterion, correlational and construct validity in the validation cohort.results We estimated dApSA28 = (28tJC × 1.6) + (28SJC × 1.6) + patient global (0–10VAS) + pain (0–10VAS) + Crp (mg/dL). Criterion validity: dApSA/dApSA28 had comparable discriminative power expressed as standardised mean difference (dApSA, 0.90; dApSA28, 0.93) to distinguish between patients in high and low disease activity. Kappa with quadratic weighting of dApSA/dApSA28 disease activity states was high: 0.92 (95% CI 0.92 to 0.92). Standardised response means for dApSA/dApSA28 were –0.96/–0.92 for visits after biological dMArd-initiation. Correlational validity: Baseline dApSA/dApSA28 had high correlation with 28-joint disease activity score with Crp (r=0.87/r=0.93), simplified disease activity index (r=0.92/r=0.99), p<0.001. Bland-Altman plot showed better agreement between dApSA/dApSA28 for low than high disease activity. Construct validity: dApSA/dApSA28 were similarly correlated to Health Assessment Questionnaire; r=0.60/0.62, p<0.001. dApSA/dApSA28 discriminated patients reporting their symptom state as acceptable versus not acceptable equally well: mean (Sd) 9.1 (8.7)/8.4 (8.0) and 24.2 (14.9)/22.5 (13.8), respectively.Conclusion our study suggests that data sets with only 28-joint counts available can be used to calculate dApSA28, especially in patients with low disease activity. dApSA28 showed good criterion, correlational and construct validity and sensitivity to change. Still, our results support that 66/68 joint count should be performed and the original dApSA should be preferred in psA.

bACKgrOundThe Disease Activity index for PSoriatic Arthritis (DAPSA), originally developed for reactive arthritis,1 was validated for psoriatic arthritis (PsA) in 2010 and is based on the simple summation of visual analogue scales (0–10 cm) of patient’s global and pain assessments, 66 swollen joint count (66SJC), 68 tender joint count (68TJC) and C reac-tive protein (CRP, mg/dL).2 These variables had previously been identified in principal component analysis as most reflective of PsA disease activity.3

In 2015, cut-off values for DAPSA remission (≤4), low (>4≤14), moderate (>14≤28) and high (>28) disease activity were proposed as well as debated.4–6

By virtue of joint involvement in PsA, which often comprises swelling and tenderness of distal interphalangeal, ankle and foot joints, the 66/68 joint counts are traditionally used to assess patients

1737Michelsen B, et al. Ann Rheum Dis 2018;77:1736–1741. doi:10.1136/annrheumdis-2018-213463

Clinical and epidemiological research

Table 1 Demographic characteristics and baseline variables

Examination cohort(n=3157 with 23 987 visits)

Validation cohort(n=3154 with 24 160 visits) P values

Age (years), mean (SD) 52.0 (13.8) 52.0 (13.8) 0.96

Female, n (%) 1746 (55.3) 1686 (53.5) 0.14

66 swollen joint count, mean (SD)/median (25th,75th percentile) 2.3 (4.1)/0 (0,3) (n=1449) 2.4 (4.4)/0 (0,3) (n=1450) 0.55

28 swollen joint count, mean (SD)/median (25th, 75th percentile) 1.3 (2.5)/0 (0,2) (n=3157) 1.3 (2.4)/0 (0,2) (n=3154) 0.53

68 tender joint count, mean (SD)/median (25th, 75th percentile) 6.7 (9.2)/3 (0,9) (n=1510) 6.8 (9.3)/4 (0,9) (n=1506) 0.42

28 tender joint count, mean (SD)/median (25th, 75th percentile) 3.4 (5.2)/1 (0,5) (n=3157) 3.3 (5.0)/1 (0,4) (n=3154) 0.45

Patient’s global, 0–100 VAS scale, mean (SD)/median (25th, 75th percentile)

46.4 (29.6)/48 (20, 71) (n=2732) 45.9 (29.0)/46 (20, 70) (n=2728) 0.53

Evaluator’s global, 0–100 VAS scale, mean (SD)/median (25th, 75th percentile)

16.7 (15.8)/12 (5, 24) (n=2792) 16.8 (15.6)/12 (5, 24) (n=2806) 0.83

Patient’s pain, 0–100 VAS scale, mean (SD)/median (25th, 75th percentile)

41.0 (26.9)/ 38 (19, 63) (n=2664) 40.3 (26.4)/ 37 (18, 61) (n=2670) 0.68

CRP, mg/L, mean (SD)/ median (25th,75th percentile) 8.6 (16.0)/4 (2, 9) (n=2780) 8.4 (13.0)/4 (2, 10) (n=2754) 0.73

SDAI, mean (SD)/median (25th, 75th percentile) 11.9 (9.9)/9.4 (4.5, 16.3) (n=2200) 11.5 (9.7)/9 (2, 16) (n=2189) 0.22

CDAI, mean (SD)/median (25th, 75th percentile) 11.0 (9.5)/8.7 (4.0, 15.3) (n=2427) 10.7 (9.2)/8.3 (4.0, 14.9) (n=2436) 0.33

DAS28CRP, mean (SD)/median (25th, 75th percentile) 3.1 (1.3)/3.0 (2.1, 4.0) (n=2483) 3.1 (1.3)/2.9 (2.1, 4.0) (n=2455) 0.16

HAQ, mean (SD)/median (25th, 75th percentile) 0.7 (0.6)/0.6 (0.1, 1.1) (n=2610) 0.7 (0.6)/0.6 (0.1, 1.0) (n=2593) 0.30

CDAI, clinical disease activity index; CRP, C reactive protein; DAS28CRP, 28-joint disease activity score with CRP; HAQ, Health Assessment Questionnaire; SDAI, simplified disease activity index; VAS, visual analogue scale.

with PsA during clinical examinations. However, many data-bases and registries solely collect 28 joint counts and not the comprehensive 66/68 joint counts, not only in patients with RA but also in patients with PsA, despite the lack of face validity of reduced joint counts outside of RA. In the present study, we assessed whether reduced joint counts may be used posthoc also for PsA or if they have to be regarded as insufficient and thus ‘lost’ for that purpose. The background for this study was also that 28-joint counts have been found to be sufficiently valid in comparison to more comprehensive joint counts in patients with RA.7–9

Thus, the objective of this study was to compute and test the potential validity of a simplified DAPSA score including 28 instead of the original 66/68 joint counts in patients with PsA from the Danish national quality registry DANBIO.

METHOdsdAnbIODANBIO is a Danish national quality registry established in year 2000 which provides data on the disease course of patients with RA, PsA and ankylosing spondylitis, regardless of treatment.10 From year 2006 registration of patients treated with biolog-ical disease-modifying anti-rheumatic drug (bDMARDs) has been mandatory. The completeness of DANBIO regarding the overall use of bDMARDs in Denmark has been estimated to be ≥92%.10 Patients with PsA with peripheral disease are routinely monitored by 28-joint disease activity score (DAS28; four vari-ables, CRP). The 66/68 joint counts are optional. Information on subgroups of PsA (eg, oligoarthritis, polyarthritis, arthritis mutilans) is not recorded. Fully anonymised DANBIO data on patients with ICD-10 code 07.03A (PsA) were collected. The study was approved by the DANBIO steering committee, the RKKP (Regionernes Kliniske Kvalitetsudviklings program) as well as the data authorities. Patients’ informed consent and research ethical approval are not necessary for this kind of study according to Danish law.

statistical analysesStatistical analyses were performed according to a predefined statistical analysis plan.

Descriptive statistics were used to summarise patients’ demographic and baseline variables. Quantitative results were compared using Mann-Whitney U test or independent t-test as appropriate and proportions using χ² test. The cohort was divided into one examination and one validation cohort according to odd or even ID numbers.

We defined: DAPSA28 = (28TJC x conversion factor1) + (28SJC x conversion factor2) + patient’s global assessment (0–10VAS) + patient’s pain assessment (0–10VAS)+CRP (mg/dL). Identification of the conversion factors was performed by generalised estimating equations (GEE, multiple visits per patient) with 68TJC/66SJC as dependent and 28TJC/28SJC as independent variables in the examination cohort. All further analyses were performed in the validation cohort.

Evaluation of criterion validity was performed through (1) assessment of Pearson correlation between physician’s global assessment and DAPSA/DAPSA28, (2) evaluation of the discrim-inative power of DAPSA/DAPSA28 to distinguish between groups with high (defined as starting bDMARD treatment) or low (defined as not starting or changing synthetic/biological DMARD (s/bDMARD) treatment for ≥60 days) disease activity, including assessment of standardised mean differences (mean differences divided by pooled SD), (3) Agreement between DAPSA/DAPSA28 disease activity categories, including kappa coefficients with linear and quadratic weighting, (4) standardised response means (SRMs).

Evaluation of correlational validity was performed through (1) correlations between changes from start of s/bDMARD treat-ment to follow-up visits in 28SJC and 66SJC and in 28TJC and 68TJC, (2) scatterplots to visualise correlations between DAPSA and DAPSA28, (3) correlations between simplified disease activity index (SDAI), clinical disease activity index (CDAI), DAS28CRP and DAPSA/DAPSA28, (4) Bland-Altman plot11 of DAPSA and DAPSA28.

1738 Michelsen B, et al. Ann Rheum Dis 2018;77:1736–1741. doi:10.1136/annrheumdis-2018-213463

Clinical and epidemiological research

Table 2 Discriminative power to distinguish between patients in high and low disease activity

Patients starting bdMArd treatment

Patients not starting or changing s/bdMArd treatment for ≥60 days

standardised mean difference*

Number of visits/patients 391/290 1946/774

DAPSA, mean (SD)/median (25th, 75th percentile) 29.1 (17.4)/26.1 (18.0, 37.5) 15.1 (13.7)/12.1 (5.1, 20.4) 0.90

DAPSA28, mean (SD)/median (25th, 75th percentile) 26.9 (15.7)/24.5 (16.1, 34.4) 13.8 (12.5)/10.9 (4.6, 18.7) 0.93

*Standardised mean difference=mean of DAPSA or DAPSA28 in patients starting bDMARD subtracted by the mean of DAPSA or DAPSA28 in patients not starting or changing s/bDMARD) divided by pooled SD of both treatment groups.DAPSA, Disease Activity index for PSoriatic Arthritis; DMARD, disease-modifying anti-rheumatic drug;bDMARD, biological DMARD; s/bDMARD, synthetic/biological DMARD.

Table 3 Agreement between DAPSA and DAPSA28 disease activity states

dAPsA (n=6421)

dAPsA28 (n=6421)

remission;dAPsA≤4(n=1153)

Low disease activity;dAPsA>4 and ≤14(n=2265)

Moderate disease activity;dAPsA>14 and ≤28(n=1938)

High disease activity;dAPsA>28(n=1065)

remission; dAPsA28 ≤4(n=1265)

1135(98.4%)/(89.7%)

126(5.6%)/(10%)

4(0.2%)/(0.3%)

0(0%)/(0%)

Low disease activity;dAPsA28>4 and ≤14 (n=2450)

18(1.6%)/(0.7%)

2073(91.5%)/(84.6%)

345(17.8%)/(14.1%)

14(1.3%)/(0.6%)

Moderate disease activity;dAPsA28>14 and ≤28 (n=1823)

0(0%)/(0%)

66(2.9%)/(3.6%)

1509(77.9%)/(82.8%)

248(23.3%)/ (13.6%)

High disease activity;dAPsA28>28 (n=883)

0(0%)/(0%)

0(0%)/0%)

80(4.1%)/(9.1%)

803(75.4%)/(90.9%)

n (%) within DAPSA disease activity states/(%) within DAPSA28 disease activity states.DAPSA, Disease Activity index for PSoriatic Arthritis.

Evaluation of construct validity was performed through (1) comparison of mean (SD) DAPSA/DAPSA28 according to the change and state scores/external anchors: (i) patient’s evaluation of disease activity (much worse, worse, little worse, unchanged, little better, better or much better), (ii) patient’s acceptable symptom state and (iii) physician’s global assessment with a cut-off <10 (0–100 scale), (2) correlations of DAPSA/DAPSA28 with Health Assessment Questionnaire (HAQ) at several points in time and comparison of mean (SD) HAQ levels for patients in different DAPSA/DAPSA28 disease activity states.

We defined the strength of Pearson correlation coefficients as; none: <0.2, low: 0.2–0.3, moderate: >0.3–<0.5, substantial: 0.5–0.7 and high: >0.7. Statistical tests were performed using SPSS for Windows V.23.0.

rEsuLTsDemographics and baseline disease activity measures were similar in the examination and validation cohorts (table 1).

Transformation to dAPsA28By means of GEE, we estimated a conversion factor of 1.6 for both TJC (95% CI 1.6 to 1.7) (8179 visits included) and for SJC (95% CI 1.5 to 1.6) (7701 visits included), leading to:

DAPSA28 = (28TJC× 1.6) + (28SJC× 1.6)+patient′s global assessment [0− 10 VAS]+patient′s pain assessement [0− 10 VAS]+CRP [mg/dL]

DAPSA and DAPSA28 scores were similar in the examination (n=1949 patients with 5995 visits) versus validation cohorts (1975 patients with 6421 visits); Mean (SD)/median (25th–75th percentiles) DAPSA: 16.4 (14.5)/12.9 (5.6–22.9) vs 16.4 (14.5)/13.0 (5.8–22.2), p=0.90; DAPSA28: 14.9 (13.7)/11.5 (5.1–20.2) vs 14.9 (13.3)/11.7 (5.2–20.0), p=0.99, respectively.

All further analyses were performed in the validation cohort.

Boxplot showed similar medians (25th–75th percentiles), but some more extreme values for DAPSA (online supplementary figure S1).

Criterion validityPhysician’s global assessment and DAPSA/DAPSA28 were overall substantially and similarly correlated (r=0.63/r=0.61), p<0.001.

DAPSA and DAPSA28 had comparable discriminative power, expressed as standardised mean difference, to distinguish between patients in high and low disease activity (table 2).

Agreements between DAPSA and DAPSA28 disease activity states,4 using same cut-offs for DAPSA28 as DAPSA, were best for patients in remission and low disease activity (table 3). DAPSA and DAPSA28 disease activity states were in agreement in 5520 of 6421 (86%) visits, had one grade difference in 887 of 6421 (13.8%) visits and two grade difference in 18 (0.2%) visits.

Kappa with linear/quadratic weighting between DAPSA/DAPSA28 disease activity states was 0.87, 95% CI (0.86 to 0.87)/ 0.92 95% CI (0.92 to 0.92) indicating very good agreement. However, table 3 indicates that the agreement differed according to disease activity states. Kappa with linear/quadratic weighting between DAPSA/DAPSA28 remission was 0.93 (0.91–0.94)/0.93 (0.92–0.94).

SRM for very first follow-up visit after start of bDMARD was −0.79 for DAPSA and −0.74 for DAPSA28 (n=232). SRM for first follow-up visit with available SRM value after start of bDMARD was −0.96 for DAPSA and −0.92 for DAPSA28 (n=572).

Correlational validityCorrelations were high between differences in 28SJC and 66SJC (r=0.84, p<0.001 (n=5758)) and between 28TJC and 68TJC

1739Michelsen B, et al. Ann Rheum Dis 2018;77:1736–1741. doi:10.1136/annrheumdis-2018-213463

Clinical and epidemiological research

Figure 1 Scatterplot of DAPSA and DAPSA28: (A) all visits, n=6421, (B) baseline visits, n=1033. DAPSA, Disease Activity index for PSoriatic Arthritis.

Table 4 Correlational validity; Pearson correlations at baseline and of all visits

baseline/ All visits (grey background)

dAPsA dAPsA28 dAs28CrP sdAI CdAI

dAPsA 1 0.93*(n=1033)

0.87*(n=1033)

0.92*(n=947)

0.92*(n=947)

dAPsA28 0.94*(n=6421)

1 0.93*(n=2399)

0.99*(n=2148)

0.99*(n=2148)

dAs28CrP 0.89*(n=6421)

0.94*(n=20 406)

1 0.94*(n=2189)

0.92*(n=2189)

sdAI 0.94*(n=5888)

0.99*(n=18 565)

0.94*(n=18 819)

1 0.99*(n=2189)

CdAI 0.94*(n=5888)

0.99*(n=18 565)

0.92*(n=18 819)

0.99*(n=18 819)

1

*P<0.001 (p found by GEE when multiple registrations per patient).CDAI, clinical disease activity index;CRP, C reactive protein;DAS28CRP, 28-joint disease activity score with CRP;GEE, generalised estimating equation;SDAI, simplified disease activity index.

Figure 2 Bland-Altman plot of DAPSA and DAPSA28: (A) all visits, n=6421, (B) baseline visits, n=1033. DAPSA, Disease Activity index for PSoriatic Arthritis.

(r=0.85, p<0.001 (n=6215)) from start of s/bDMARD treat-ment to follow-up visits.

Scatterplots showed high correlation between DAPSA and DAPSA28 with higher spread for higher disease activity levels and more scores in the upper half (figure 1). Scatterplots of base-line DAPSA vs DAPSA28 stratified according to DAPSA disease activity states showed high correlation for patients in remis-sion, low and high disease activity and substantial correlation for patients in moderate disease activity (online supplementary figure S2A–D).

DAPSA/DAPSA28 had both high correlations with DAS28CRP, SDAI and CDAI at baseline as well as for all registered visits (table 4).

Bland-Altman plots showed better agreement between DAPSA and DAPSA28 for low than high disease activity levels (figure 2).

The differences between DAPSA and DAPSA28 increased with an increasing average of DAPSA and DAPSA28.11 This difference between DAPSA and DAPSA28 was skewed and log transforma-tion (which is the only transformation recommended by Bland

1740 Michelsen B, et al. Ann Rheum Dis 2018;77:1736–1741. doi:10.1136/annrheumdis-2018-213463

Clinical and epidemiological research

and Altman for this plot) did not improve the skewness, hence limits of agreement are not given as they would be invalid.11

Construct validityDAPSA/DAPSA28 were both substantially correlated to HAQ; all visits with DAPSA (n=6204) r=0.60/0.62; baseline visits with DAPSA (n=994) r=0.53/r=0.56; starting bDMARD (n=387) r=0.55/r=0.59; not starting bDMARD (n=5817) r=0.59/r=0.61, respectively, all p<0.001.

Mean (SD) HAQ was similar for different DAPSA versus DAPSA28 categories; remission, 0.2 (0.3)/0.2 (0.3); low disease activity 0.6 (0.5)/ 0.6 (0.5); moderate disease activity 1.0 (0.6)/ 1.1 (0.6); high disease activity, 1.5 (0.6)/1.5 (0.6), respectively.

Mean (SD) DAPSA and DAPSA28 were comparable for different change and state scores used as external anchors, but mean values of DAPSA were consistently higher than for DAPSA28 (online supplementary table S1).

dIsCussIOnFor full assessment of disease activity in PsA, there is general agreement about the need to assess a full joint count with 68 tender and 66 swollen joints. However, many databases and registries only collect 28 joint counts. We used the Danish national quality registry DANBIO to explore psychometric performance of a modified DAPSA based on 28 rather than 66/68 joint counts.

DAPSA28 showed overall good criterion, correlational and construct validity as well as sensitivity to change, but the agreement between DAPSA and DAPSA28 was better for low than high disease activity levels. Psychometric performance was also poorer in patients with moderate and high disease activity.

Most patients were classified as having the same disease activity levels, regardless of the use of DAPSA or DAPSA28. When discrepancies occurred, these were moderate in magni-tude and more frequent for patients in high disease activity. Kappa with quadratic weighting was high, indicating very good agreement between DAPSA and DAPSA28.

Correlational validity was good, including high correlations between DAPSA/DAPSA28 and DAS28, SDAI and CDAI. All these scores are joint focused and include many of the same variables. DAS28 has been validated for use in PsA in clin-ical trials of biological therapies and SDAI in an observational study on patients with PsA.12–14 Of note, longitudinal observa-tional studies on patients with PsA comparing the performance of DAPSA, DAS28, SDAI and CDAI are lacking in the current literature.

The main objection against the use of DAPSA28 in PsA is the 28 joint counts. It is undisputable that a full joint count is preferable to the 28 joint counts in PsA. However, as previously mentioned, in many registries and real life observational studies, a full joint count is not available.

Further general objections against DAPSA, as well as DAPSA28, are the need for assessment of multiple domains in PsA.5 6 However, we believe that in addition to the more exten-sive composite scores of disease activity in PsA like PASDAS15 or CPDAI,15 there may be a need for more feasible composite scores like DAPSA/DAPSA28. This issue is under constant debate4–6 15 and a task force has recently recommended DAPSA for use in PsA.15

To our knowledge, DAPSA28 is the first 28 joint disease activity score developed and validated in PsA. It is easy to calculate and

may pose an important substitute to DAS28, which probably is the mostly used 28-joint composite score in PsA, even if origi-nally developed for RA.16 Still, we do not recommend the use of DAPSA28 in clinical practice, as the 66/68 joint counts should be used in PsA whenever feasible. Of note, none of the joint counts distinguishes between the potentially different clinical relevance of which joints being affected.

The psychometric performance of DAPSA28 in patients with moderate and high disease activity was not as good as in patients in remission or with low disease activity. This observation may not be surprising since the distal interphalangeal joints and foot joints may be assumed to be more frequently inflamed in patients with increasingly overall inflammatory activity.

Limitations of the study include non-standardised times of visits (observational database) as well as lack of radiographic data, which could have been useful for additional criterion vali-dation. The results may be influenced by choice of study popula-tion, as patients in, for example, low-income countries may have higher disease activity and different cultural background, which possibly might influence the identification of conversion factors. The major strength of the study is the large study population from the national Danish longitudinal observational database DANBIO, including large numbers of patients with 66/68 joint counts and extensive clinical data as well as patients with both early and established disease.

In conclusion, we recommend that the original DAPSA should be preferred and that 66/68 joint counts should be performed in patients with PsA to compute DAPSA. However, our study suggests that data sets with only 28-joint counts available can be used to calculate a modified DAPSA28, as a valid tool for database or outcomes research in the field of PsA, especially in patients with low disease activity.

Author affiliations1department of rheumatology, diakonhjemmet Hospital, oslo, norway2division of rheumatology, department of Medicine, Hospital of Southern norway trust, Kristiansand, norway3division of rheumatology, department of Internal Medicine III, Medical University of Vienna and Lainz Hospital, Vienna, Austria4ZiteLab ApS, Frederiksberg, denmark5department of rheumatology, Leiden University Medical Center, Leiden, the netherlands6dAnBIo, Copenhagen Center for Arthritis research (CopECArE), Center for rheumatology and Spine diseases, Centre of Head and orthopaedics, rigshospitalet, Copenhagen, denmark7department of Clinical Medicine, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, denmark

Acknowledgements thanks to the departments of rheumatology in Aalborg, Aarhus, Esbjerg, Fredericia, Frederiksberg, Gentofte, Glostrup, Gråsten, Hjørring, Holbæk, Holstebro, Horsens, Helsingør, Kolding, Køge, næstved/nykøbing Falster, odense, randers, rigshospitalet, roskilde, Silkeborg, Slagelse, Svendborg, Vejle, Viborg, denmark for reporting to the dAnBIo registry. Manuscripts based on work previously presented at a conference and published as a conference abstract: Michelsen B et al. Ann rheum dis 2018;77:343.

Contributors nSK and MLH were responsible for data acquisition. BM analysed the data and wrote the manuscript. All authors were responsible for study design, critical revision of the manuscript and approval of the final version.

Funding dAnBIo is supported by unrestricted grants from AbbVie A/S, Biogen (denmark) A/S, Bristol-Myers Squibb danmark, Eli Lilly denmark, MSd danmark ApS, novartis danmark A/S, pfizer danmark ApS, roche A/S, UCB nordic A/S. these entities have no influence on dAnBIo’s work, including user interface, data collection, statistical analyses or research projects. BM has received clinical research fellowships from diakonhjemmet Hospital and the Hospital of Southern norway trust.

Competing interests tKK has received fees for speaking and/or consulting from AbbVie, Biogen, BMS, Boehringer Ingelheim, Celgene, Celltrion, Eli Lilly, Epirus, Hospira, Merck-Serono, MSd, Mundipharma, novartis, oktal, orion pharma, Hospira/pfizer, roche, Sandoz and UCB and received research funding to diakonhjemmet Hospital from AbbVie, BMS, MSd, pfizer, roche and UCB. MLH has received research

1741Michelsen B, et al. Ann Rheum Dis 2018;77:1736–1741. doi:10.1136/annrheumdis-2018-213463

Clinical and epidemiological research

funding/consultancy/speakers fee from Abbvie, Biogen, BMS, Celltrion, MSd, novartis, orion, pfizer, Samsung, UCB. All other authors have declared that no competing interests exist.

Patient consent not required.

Provenance and peer review not commissioned; externally peer reviewed.

data sharing statement All relevant data are within the text.

RefeRences 1 Eberl G, Studnicka-Benke A, Hitzelhammer H, et al. development of a disease

activity index for the assessment of reactive arthritis (dArEA). Rheumatology. 2000;39:148–55.

2 Schoels M, Aletaha d, Funovits J, et al. Application of the dArEA/dApSA score for assessment of disease activity in psoriatic arthritis. Ann Rheum Dis 2010;69:1441–7.

3 nell-duxneuner Vp, Stamm tA, Machold Kp, et al. Evaluation of the appropriateness of composite disease activity measures for assessment of psoriatic arthritis. Ann Rheum Dis 2010;69:546–9.

4 Schoels MM, Aletaha d, Alasti F, et al. disease activity in psoriatic arthritis (psA): defining remission and treatment success using the dApSA score. Ann Rheum Dis 2016;75:811–8.

5 Helliwell pS, Coates LC. the definition of remission in psoriatic arthritis: can this be accurate without assessment of multiple domains? Ann Rheum Dis 2015;74:e66.

6 Schoels MM, Aletaha d, Smolen JS. defining remission and treatment success using the dApSA score: response to letter by Helliwell and Coates. Ann Rheum Dis 2015;74:e67.

7 Fuchs HA, Brooks rH, Callahan LF, et al. A simplified twenty-eight-joint quantitative articular index in rheumatoid arthritis. Arthritis Rheum 1989;32:531–7.

8 Aletaha d, Martinez-Avila J, Kvien tK, et al. definition of treatment response in rheumatoid arthritis based on the simplified and the clinical disease activity index. Ann Rheum Dis 2012;71:1190–6.

9 Smolen JS, Breedveld FC, Eberl G, et al. Validity and reliability of the twenty-eight-joint count for the assessment of rheumatoid arthritis activity. Arthritis Rheum 1995;38:38–43.

10 Hetland ML. dAnBIo--powerful research database and electronic patient record. Rheumatology. 2011;50:69–77.

11 Bland JM, Altman dG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1:307–10.

12 Fransen J, Antoni C, Mease pJ, et al. performance of response criteria for assessing peripheral arthritis in patients with psoriatic arthritis: analysis of data from randomised controlled trials of two tumour necrosis factor inhibitors. Ann Rheum Dis 2006;65:1373–8.

13 Salaffi F, Ciapetti A, Carotti M, et al. disease activity in psoriatic arthritis: comparison of the discriminative capacity and construct validity of six composite indices in a real world. Biomed Res Int. 2014;2014:1–12.

14 Saber tp, ng Ct, renard G, et al. remission in psoriatic arthritis: is it possible and how can it be predicted? Arthritis Res Ther 2010;12:r94.

15 Smolen JS, Schöls M, Braun J, et al. treating axial spondyloarthritis and peripheral spondyloarthritis, especially psoriatic arthritis, to target: 2017 update of recommendations by an international task force. Ann Rheum Dis 2018;77:3–17.

16 Helliwell pS, Fitzgerald o, Mease pJ. development of composite measures for psoriatic arthritis: a report from the GrAppA 2010 annual meeting. J Rheumatol 2012;39:398–403.

1742 Barsalou J, et al. Ann Rheum Dis 2018;77:1742–1749. doi:10.1136/annrheumdis-2018-213718

Clinical and epidemiological research

ExtEndEd rEport

Effect of in utero hydroxychloroquine exposure on the development of cutaneous neonatal lupus erythematosusJulie Barsalou,1 nathalie Costedoat-Chalumeau,2 Adey Berhanu,3 Cesar Fors-nieves,4 Ummara Shah,5 patrick Brown,6 Carl A Laskin,7 nathalie Morel,2 Kateri Levesque,2 Jill p Buyon,4 Earl d Silverman,1 peter M Izmirly4

To cite: Barsalou J, Costedoat-Chalumeau n, Berhanu A, et al. Ann Rheum Dis 2018;77:1742–1749.

Handling editor Josef S Smolen

For numbered affiliations see end of article.

Correspondence todr Julie Barsalou, division of rheumatology and Immunology, department of pediatrics, CHU Sainte-Justine, Université de Montréal, Montreal, Quebec, Canada; juliebarsalou@ hotmail. com

received 6 May 2018revised 7 August 2018Accepted 7 August 2018published online First 8 october 2018

© Author(s) (or their employer(s)) 2018. no commercial re-use. See rights and permissions. published by BMJ.

AbsTrACTObjective Cutaneous neonatal lupus (cnL) occurs in possibly 5%–16% of anti-ro±anti-La antibody–exposed infants. data suggest in utero exposure to hydroxychloroquine (HCQ) may prevent cardiac nL. the aim was to assess whether in utero exposure to HCQ decreases the risk of cnL and/or delays onset.Methods A multicentre case–control study was performed with 122 cnL cases and 434 controls born to women with a rheumatological disease who had documentation of maternal anti-ro±anti-La antibodies at pregnancy and confirmation of medication use and the child’s outcome. A secondary analysis was performed on 262 cnL cases, irrespective of maternal diagnosis, to determine if HCQ delayed time to cnL onset.results twenty (16%) cnL cases were exposed to HCQ compared with 146 (34%) controls (or 0.4 (95% CI 0.2 to 0.6); p<0.01). Exposure to HCQ was associated with a reduced risk of cnL; exposure to anti-La antibody and female gender were associated with an increased risk of cnL. Exposure to HCQ remained significantly associated with a reduced cnL risk in the analyses limited to mothers with systemic lupus erythematosus and those who developed rash ≤1 month. When analysing all 262 cnL cases, HCQ-exposed infants were older (6.0 (95% CI 5.7 to 6.3) weeks) at cnL onset versus HCQ-non-exposed infants (4.4 (95% CI 3.9 to 5.0) weeks), but the difference was not statistically significant (p=0.21).Conclusion Exposure to HCQ was associated with a reduced risk of cnL. Among cnL cases, those exposed to HCQ tend to have later onset of rash. Both findings suggest a protective effect of HCQ on cnL.

InTrOduCTIOnNeonatal lupus (NL) is an autoimmune disease associated with the transplacental passage of maternal anti-Ro±anti-La antibodies. Cutaneous involvement is one of the most common non-car-diac manifestations of NL, affecting possibly 5%–16% of anti-Ro±anti-La antibody–exposed infants1 2 and the recurrence rate of cutaneous NL (cNL) approaches 23%.3 Biopsy specimens of affected areas usually show interface derma-titis.4 cNL is a transient condition that generally heals spontaneously as maternal autoantibodies are cleared from the child’s circulation,5 although permanent scarring can develop in cases where the lesions are extensive.6 In addition, if misdiagnosed,

it may lead to unnecessary investigations and treatments.

Hydroxychloroquine (HCQ) is a drug frequently used in women with systemic autoimmune rheu-matic diseases (SARDs).7 HCQ has been shown to prevent disease flares in pregnant women with systemic lupus erythematosus (SLE).8 Transpla-cental passage of HCQ has been demonstrated in pregnant women with detectable levels in cord blood and is considered safe during pregnancy.9 10 HCQ is thought to act via inhibition of toll-like receptors (TLRs), which have been implicated in the pathogenesis of cardiac NL and cutaneous lupus.11–13 Although HCQ has not been specifically studied for prevention of cNL, there are encour-aging data on its association with a reduced risk of cardiac NL.14–17 Accordingly, the primary objective of this study was to assess if prenatal exposure to HCQ lowered the risk of cNL with the secondary objective evaluating for a delay in cNL rash onset in HCQ-exposed infants who develop cutaneous involvement.

MeTHOdsstudy populationAll cases of cNL and controls found in three data sources were collected for this multicentre

Key messages

What is already known about this subject? ► In utero exposure to hydroxychloroquine (HCQ) has been associated with a reduced risk of cardiac neonatal lupus. No studies have yet examined the impact of prenatal HCQ exposure on cutaneous neonatal lupus (cNL).

What does this study add? ► This multicenter study showed that prenatal exposure to HCQ was associated with a reduced risk of cNL. Among cNL cases, those exposed to HCQ tend to have later onset of rash.

How might this impact on clinical practice or future developments?

► Preventing cNL could reduce the occurrence of permanent scarring (telangiectasia, epidermal atrophy and cutaneous pigmentation changes).

1743Barsalou J, et al. Ann Rheum Dis 2018;77:1742–1749. doi:10.1136/annrheumdis-2018-213718

Clinical and epidemiological research

retrospective study: the SickKids Neonatal Lupus Erythematosus Database (SickKids NLE Database), the Research Registry for Neonatal Lupus (RRNL) and the French Registry of Neonatal Lupus (French RNL). The SickKids NLE Database was created in 1984 and contains prospectively collected data on anti-Ro±an-ti-La antibody–exposed infants and their mothers living in the Greater Toronto Area, irrespective of the infants’ NL status.17 The RRNL, established in 1994, and the French RNL, estab-lished in 2000, both enrol women that are anti-Ro±anti-La anti-body positive, living in the USA and France, respectively, and have had at least one child with any manifestation of NL. In addition, these two registries collected data on whether siblings of the affected NL children had NL. These registries comprise data collected both prospectively and retrospectively.18 19

Inclusion and exclusion criteriaInclusion criteria for the case–control analyses were (1) infant born to a woman positive for anti-Ro±anti-La antibodies, docu-mented prior to or at the time of pregnancy, (2) known age of onset of cNL and (3) documentation of medication taken during pregnancy. For the primary analysis, a maternal diagnosis of a SARD in which HCQ could be used was required: SLE, Sjogren’s syndrome (SS), dermatomyositis, rheumatoid arthritis or juvenile idiopathic arthritis. Diagnosis had to be made prior to concep-tion to allow time for maternal HCQ blood level to increase. For the secondary analysis related to onset of rash, all cases of cNL irrespective of the maternal diagnosis were included, which meant that mothers could be clinically asymptomatic. The diag-nosis of cNL in both analyses had to be made by a rheumatolo-gist, dermatologist±paediatrician. Infants with cardiac NL were excluded from all analyses because data from the three registries regarding the effects of HCQ on cardiac NL have already been reported.14 15 17

study design, outcome measure and data collectionThe primary analysis was a case–control study to determine whether exposure to HCQ in mothers with a known SARD reduced the risk of cNL. The primary outcome was the devel-opment of cNL. A secondary analysis used a cohort of all chil-dren whot developed cNL, irrespective of maternal diagnosis, to assess if in utero exposure to HCQ delayed the onset of cNL.

Information was extracted on infants (gender, year of birth, age of onset of cNL and siblings with cNL±cardiac NL) and their mothers (age at delivery, diagnosis, anti-Ro±anti-La anti-body positivity, medication intake including HCQ, fluorinated and non-fluorinated steroids, azathioprine, intravenous immu-noglobulins (IVIGs) and plasmapheresis). Children were consid-ered exposed to HCQ if their mother had documented intake of ≥200 mg per day throughout the entire pregnancy. Maternal intake of fluorinated and non-fluorinated steroids, azathioprine and IVIG was defined as intake of any dose at least once during pregnancy. Anti-Ro and anti-La antibodies were measured locally using the following assays: RRNL, commercial Clinical Labo-ratory Improved Amendments (CLIA)-approved laboratories but in majority of cases confirmed in Buyon Clancy laboratory using recombinant Ro52 and La48 and native Ro60 in ELISA; SickKids NLE Database, commercially available ELISAs using affinity-purified Ro and La antigens; and French RNL, coun-terimmunoelectrophoresis and/or by using INNO-LIAT ANA Update (Innogenetics, provided by InGen, Rungis, France in most of the cases). Research ethics board approval was obtained at each participating institution.

statistical analysisFor the primary analyses, characteristics of cases and controls were compared using χ2, Fisher’s exact and Mann-Whitney U tests as appropriate. To account for within-family data correla-tion, generalised estimating equations (GEEs) were used for data analyses.20 21 The following 10 covariates were tested in univari-able GEE as potential factors associated with cNL: maternal age at delivery, maternal diagnosis of primary or secondary SS (higher levels of anti-Ro±anti-La antibodies±epitope specific-ity±genetic background in women with SS may put them at a higher risk of having an affected child),22–24 maternal anti-La antibody positive, maternal use of HCQ, maternal use of fluo-rinated steroids±IVIG±plasmapheresis (use to treat fetuses), maternal intake of non-fluorinated steroids±azathioprine (use to treat mothers), infant’s gender, infant born ≥year 2000 (a time when the safety of HCQ use during pregnancy had been recognised), sibling with cNL±cardiac NL (identify families that may be at increased risk of NL) and registry source (RRNL and French RNL grouped as have same enrolment criteria). Variables with a p value ≤0.05 were selected for inclusion in the multivari-able GEE model. A p value <0.05 in the final multivariable model was considered statistically significant. Two a priori subgroup analyses were performed. The first included only infants born to SLE mothers, to account for possible confounding by indication, as women with SLE may be more likely to be prescribed HCQ than women with other types of SARD. The second, limiting cases to those who developed cNL within ≤4 weeks of life. The rationale was that neonates are not continued on HCQ after delivery, therefore HCQ levels are expected to decline.

The secondary analysis was limited to cNL cases, irrespec-tive of maternal health. Kaplan-Meier survival analysis was performed to assess if HCQ delayed the onset of cNL. A p value <0.05 by the Gehan-Breslow-Wilcoxon test was considered statistically significant. Data analysis was performed using IBM SPSS Statistics V.21.0 (IBM) and R V.3.1.1 (R Foundation for Statistical Computing, Vienna, Austria).

resulTsIncluded and excluded patientsA total of 737 infants were screened (figure 1). Forty-one were excluded due to missing key variables. The overall study popu-lation consisted of 696 infants: 262 cNL cases and 434 controls. Among the 262 cNL cases, 122 infants were born to women with a SARD diagnosed before pregnancy and were therefore included into the primary analysis. The first subgroup analysis included only infants born to women with SLE (cNL cases=85; controls=300). The second subgroup analysis was restricted to infants who devel-oped cNL within 4 weeks of life (cNL cases=48; controls=434). All 262 cNL cases were included into the secondary analysis.

Primary analysis: exposure to HCQ in mothers with a known sArd reduces the risk of cnlPatient characteristicsThe study population consisted of 556 infants: 122 cNL cases and 434 controls. There were 10 twin pairs: cNL status was concor-dant in 8/10 pairs (both twins unaffected). Thirty-eight per cent (N=209) of the study population were siblings: 89 sibling pairs, 7 groups of 3 siblings and 2 groups of 5 siblings. Thirty-seven per cent (N=208) of the study population had at least one sibling with cutaneous (N=85), cardiac (N=107) or both NL features (N=16).

Demographics of the primary analysis study population are shown in table 1. A diagnosis of SLE (48.4%) and SS (48.4%) was found with equal frequency in mothers of cNL cases as compared

1744 Barsalou J, et al. Ann Rheum Dis 2018;77:1742–1749. doi:10.1136/annrheumdis-2018-213718

Clinical and epidemiological research

Figure 1 Diagram of included and excluded patients. A total of 737 infants were screened. Forty-one infants were excluded for missing key variables. A total of 696 infants were eligible: 262 cNL cases and 434 controls. Among the 262 cNL cases, 122 were born to women with a SARD diagnosed before pregnancy and were therefore included into the primary analysis, with 434 controls. The first subgroup analysis included only infants born to women with SLE (cNL cases=85; controls=300). The second subgroup analysis was restricted to cNL cases with onset of rash in the first 4 weeks after birth (cNL cases=48; controls=434). The secondary analysis included 262 cNL cases, regardless of maternal diagnosis: 122 infants born to women with a SARD diagnosed before pregnancy, 10 infants born to women with a SARD diagnosed during pregnancy, 6 infants born to women with autoimmune diseases other than SARD and 124 infants born to women who were asymptomatic or had an undifferentiated autoimmune syndrome (UAS). cNL, cutaneous neonatal lupus; HCQ, hydroxychloroquine; SARD, systemic autoimmune rheumatic disease; SLE, systemic lupus erythematosus.

with mothers of controls for which SLE was the leading diagnosis (62.0%). Nearly all women were anti-Ro antibody positive (99%), but the proportion of women with positive anti-La antibody was significantly higher in mothers of cNL cases (73%) than mothers of controls (48%); p<0.01. In utero exposure to HCQ and to non-fluorinated steroids±azathioprine was higher in controls (34% HCQ exposed and 44% non-fluorinated steroids±azathi-oprine exposed) than cNL cases (16% HCQ exposed and 25% non-fluorinated steroids±azathioprine exposed); p<0.01. Char-acteristics of the 556 infants were similar. The RRNL and French RNL had more cNL cases (65%) versus the SickKids NLE Data-base (35%). The median age of onset of cNL was 6 (range 0–21) weeks old.

Main analysis: factors associated with development of cnlAssociation between the 10 predefined covariates and cNL was first explored with univariable GEE analyses (table 2). Six vari-ables were found to have a p value ≤0.05 and were selected for the multivariable GEE model. In the final model, the odds of in utero exposure to HCQ were significantly lower in cNL cases than in controls (odds ratio (OR) 0.4 (95% CI 0.2 to 0.7); p<0.01) (table 3). A similar association was found for exposure to non-flu-orinated steroids±azathioprine (OR 0.5 (95% CI 0.3 to 0.8); p=0.01). Cases with cNL were more likely to be female (OR 1.7 (95% CI 1.1 to 2.6); p=0.02), exposed to maternal anti-La

antibody (2.7 (95% CI 1.7 to 4.3); p<0.01) and come from the French or RRNL Registries (OR 2.0 (95% CI 1.1 to 3.6); p=0.02).

First subgroup analysis: Infants born to women with sleThis analysis included the 385 infants (cNL cases=85, controls=300) born to women with SLE. Results of the univari-able and multivariable GEEs were similar to those observed for the main analysis (tables 4 and 5) with significantly lower odds of HCQ exposure in cNL cases than in controls (OR 0.4 (95% CI 0.2 to 0.7); p<0.01). In the subgroup of SLE mothers, secondary SS was positively associated with cNL (OR 2.4 (95% CI 1.1 to 5.3); p=0.03), but female gender was no longer significant in the multivariable analysis.

second subgroup analysis: infants who developed cnl within 4 weeks of lifeA total of 482 infants were included (cNL cases=48, controls=434). In univariable analysis, cNL cases were more likely to be exposed to maternal anti-La antibody (OR 3.7 (95% CI 1.7 to 7.8); p<0.01) and come from the French or RRNL Registries (OR 2.1 (95% CI 1.1 to 4.0); p=0.02) but less likely to be exposed to HCQ (OR 0.2 (95% CI 0.1 to 0.6); p<0.01) (table 6). Maternal anti-La antibody positivity (OR 3.2 (95% CI 1.5 to 6.8); p<0.01) and exposure to HCQ (OR 0.2 (95% CI 0.1 to 0.5); p<0.01) remained significant in the multivariable GEE model (table 7).

1745Barsalou J, et al. Ann Rheum Dis 2018;77:1742–1749. doi:10.1136/annrheumdis-2018-213718

Clinical and epidemiological research

Table 1 Characteristics of the study population, primary aim

Maternal characteristics

Patients Cases Controls P values n=546 n=122 n=424

Age at delivery, years (IQR) 533 31 (29–35) 32 (29–35) 0.33

Diagnosis 0.03

SLE 322 59 (48.4) 263 (62.0)

Primary or secondary SS 204 59 (48.4) 145 (34.2)

RA or JIA 19 4 (3.2) 15 (3.6)

Dermatomyositis 1 0 1 (0.2)

Anti-Ro antibody positive 545 121 (100) 419 (99) 0.59

Anti-La antibody positive 518 88 (73) 192 (48) <0.01

Medication use

Hydroxychloroquine* 546 20 (16) 146 (34) <0.01

Fluorinated steroids±IVIG±plasmapheresis

546 7 (6) 17 (4) 0.41

Non-fluorinated steroids±azathioprine

543 31 (25) 186 (44) <0.01

Children characteristics n=556 n=122 n=434

Female:male (%) 554 58:42 48:52 0.05

Born ≥year 2000 556 75 (62) 277 (64) 0.63

Registry, French/RRNL:SickKids (%) 556 65:35 47:53 <0.01†

French RNL 6 106

RRNL 73 97

SickKids NLE Database 43 231

Data presented as number (percentage) unless otherwise specified.*The proportion of women taking HCQ throughout pregnancy among the three datasources were as follows: RRNL N=30/168 (18%), SickKids NLE Database N=76/267 (28%) and French RNL N=60/111 (54%).†P value, French RNL/RRNL compared with SickKids NLE Database.IVIG, intravenous immunoglobulin; JIA, juvenile idiopathic arthritis; N, number; NLE, neonatal lupus erythematosus; RA, rheumatoid arthritis; RRNL, Research Registry for Neonatal Lupus; SLE, systemic lupus erythematosus; SS, Sjogren’s syndrome.

Table 2 Patient characteristics associated with cNL in univariable GEE analysis

Or (95% CI) P values

Maternal characteristics

Age at delivery 1.0 (0.9 to 1.1) 0.39

Primary or secondary SS 1.9 (1.2 to 2.9) <0.01

Anti-La antibody positive 3.0 (1.8 to 4.7) <0.01

Hydroxychloroquine 0.4 (0.2 to 0.6) <0.01

Fluorinated steroids±IVIG±plasmapheresis 1.4 (0.6 to 3.5) 0.47

Non-fluorinated steroids±azathioprine 0.5 (0.3 to 0.7) <0.01

Children characteristics

Female gender 1.5 (1.0 to 2.3) 0.04

Born ≥year 2000 1.1 (0.7 to 1.8) 0.54

Sibling with cNL and/or cardiac NL 1.5 (0.9 to 2.4) 0.10

Registry* 2.1 (1.4 to 3.3) <0.01

Variables with a p value ≤0.05 were selected for inclusion in the multivariable GEE model (p values in bold).*Comparator is the SickKids NLE Database Registry.cNL, cutaneous neonatal lupus; GEE, generalised estimating equation; IVIG, intravenous immunoglobulin; NL, neonatal lupus; NLE, neonatal lupus erythematosus; SS, Sjogren’s syndrome.

Table 3 Patient characteristics associated with cNL in multivariable GEE analyses

Or (95% CI) P values

Maternal characteristics

Anti-La antibody positive 2.7 (1.7 to 4.3) <0.01

Hydroxychloroquine 0.4 (0.2 to 0.7) <0.01

Non-fluorinated steroids±azathioprine 0.5 (0.3 to 0.8) 0.01

Children characteristics

Female gender* 1.7 (1.1 to 2.6) 0.02

Registry 2.0 (1.1 to 3.6) 0.02

*Comparator is the SickKids NLE Database Registry.cNL, cutaneous neonatal lupus; GEE, generalised estimating equation; NLE, neonatal lupus erythematosus.

Table 4 Univariable GEE analysis for factors associated with cNL in infants born to women with SLE

Or (95% CI) P values

Maternal characteristics

Age at delivery 1.0 (0.9 to 1.1) 0.49

Primary or secondary SS 4.4 (2.4 to 7.9) <0.01

Anti-La antibody positive 3.8 (2.1 to 6.9) <0.01

Hydroxychloroquine 0.3 (0.2 to 0.6) <0.01

Fluorinated steroids±IVIG±plasmapheresis 1.7 (0.6 to 4.5) 0.32

Non-fluorinated steroids±azathioprine 0.5 (0.3 to 0.8) <0.01

Children characteristics

Female gender 1.5 (0.9 to 2.4) 0.13

Born ≥year 2000 1.3 (0.8 to 2.1) 0.31

Sibling with cNL and/or cardiac NL 2.5 (1.4 to 4.5) <0.01

Registry* 2.8 (1.6 to 4.7) <0.01

Variables with a p value ≤0.05 were selected for inclusion in the multivariable GEE model (p values in bold).*Comparator is the SickKids NLE Database Registry.cNL, cutaneous neonatal lupus; GEE, generalised estimating equation; IVIG, intravenous immunoglobulin; NL, neonatal lupus; NLE, neonatal lupus erythematosus; SLE, systemic lupus erythematosus; SS, Sjogren’s syndrome.

secondary analysis: in utero exposure to HCQ and timing of cnlPatient characteristicsThe study population consisted of 262 patients with cNL. Twen-ty-three (8.8%) infants were exposed to HCQ. Demographics of

the study cohort are shown in table 8. Characteristics of infants exposed to HCQ differed from those non-exposed with respect to the following variables: maternal diagnoses of SLE or cutaneous lupus (52.2% vs 24.4%; p<0.01) or SS (43.5% vs 21.8%; p=0.04) were more frequent, but the proportion of asymptomatic/undif-ferentiated autoimmune syndrome was lower (4.3% vs 51.7%; p<0.01), and prenatal exposure to non-fluorinated steroids±aza-thioprine (43.5% vs 9.2%; p<0.01) and proportion of infants born ≥year 2000 (91.3% vs 53.1%; p<0.01) were higher.

effect of in utero exposure to HCQ on the age of onset of cnlKaplan-Meier analysis showed no statistically significant differ-ence in the median number of weeks to cNL onset: HCQ exposed 6.0 (95% CI 5.7 to 6.3) versus non-exposed 4.4 (95% CI 3.9 to 5.0) weeks; p=0.21 (figure 2). Although no statistically significant difference was seen, the survival curves showed a delayed onset of rash during the first 6 weeks in infants exposed to HCQ.

dIsCussIOnCutaneous involvement is one of the most common manifestations of NL. In prospective studies, the reported incidence has ranged from 5% to 16%.1 2 Although cutaneous NL is transient and is considered a benign condition, if misdiagnosed, it may lead to unnecessary investigations and treatments. In addition, telangiec-tasia, epidermal atrophy and/or cutaneous pigmentation changes

1746 Barsalou J, et al. Ann Rheum Dis 2018;77:1742–1749. doi:10.1136/annrheumdis-2018-213718

Clinical and epidemiological research

Table 5 Multivariable GEE analysis for factors associated with cNL in infants born to women with SLE

Or (95% CI) P values

Maternal characteristics

Secondary SS 2.4 (1.1 to 5.3) 0.03

Anti-La antibody positive 3.4 (1.8 to 6.3) <0.01

Hydroxychloroquine 0.4 (0.2 to 0.7) <0.01

Non-fluorinated steroids±azathioprine 0.4 (0.2 to 0.8) <0.01

Children characteristics

Registry* 3.5 (1.6 to 7.6) <0.01

*Comparator is the SickKids NLE Database Registry.cNL, cutaneous neonatal lupus; GEE, generalised estimating equation; NLE, neonatal lupus erythematosus; SLE, systemic lupus erythematosus; SS, Sjogren’s syndrome.

Table 6 Univariable GEE analysis for factors associated with early onset cNL

Or (95% CI) P values

Maternal characteristics

Age at delivery 1.0 (0.9 to 1.1) 0.89

Primary or secondary SS 1.8 (1.0 to 3.3) 0.06

Anti-La antibody positive 3.7 (1.7 to 7.8) <0.01

Hydroxychloroquine 0.2 (0.1 to 0.6) <0.01

Fluorinated steroids±IVIG±plasmapheresis 2.1 (0.7 to 6.4) 0.17

Non-fluorinated steroids±azathioprine 0.7 (0.4 to 1.3) 0.27

Children characteristics

Female gender 1.5 (0.8 to 2.8) 0.18

Born ≥year 2000 1.1 (0.6 to 2.0) 0.83

Sibling with cNL and/or cardiac NL 1.6 (0.8 to 3.0) 0.16

Registry* 2.1 (1.1 to 4.0) 0.02

Variables with a p value ≤0.05 were selected for inclusion in the multivariable GEE model (p values in bold).*Comparator is the SickKids NLE Database Registry.cNL, cutaneous neonatal lupus; GEE, generalised estimating equation; IVIG, intravenous immunoglobulin; NL, neonatal lupus; NLE, neonatal lupus erythematosus; SS, Sjogren’s syndrome.

Table 7 Multivariable GEE analysis for factors associated with early onset cNL

Or (95% CI) P values

Maternal characteristics

Anti-La antibody positive 3.2 (1.5 to 6.8) <0.01

Hydroxychloroquine 0.2 (0.1 to 0.5) <0.01

*Comparator is the SickKids NLE Database Registry.cNL, cutaneous neonatal lupus; GEE, generalised estimating equation; NLE, neonatal lupus erythematosus.

Table 8 Characteristics of the cNL population, secondary aim

Patientsexposed to HCQ

non-exposed to HCQ P

values n=262 n=23 n=239

Maternal characteristics

Diagnosis 261 <0.01

Asymptomatic/UAS 1 (4.3) 123 (51.7)

SLE or cutaneous lupus 12 (52.2) 58 (24.4)

Primary or secondary SS 10 (43.5) 52 (21.8)

Overlap connective tissue disease 0 1 (0.4)

RA or JIA 0 4 (1.7)

Antibody status

Anti-Ro antibody positive 262 23 (100) 239 (100)

Anti-La antibody positive 255 19 (86) 173 (74) 0.21

Medication intake

Fluorinated steroids±IVIG 262 2 (9) 9 (4) 0.25

Non-fluorinated steroids±azathioprine

262 10 (44) 22 (9) <0.01

Children characteristics

Female:male (%) 262 52:48 59:41 0.55

Born ≥year 2000 262 21 (91) 127 (53) <0.01

Registry, French/RRNL:SickKids (%) 262 65:35 63:37 *0.82

French RNL 5 9

RRNL 10 141

SickKids NLE Database 8 89

Data presented as number (percentage) unless otherwise specified.*P value, French RNL/RRNL compared with SickKids NLE Database.IVIG, intravenous immunoglobulin; JIA, juvenile idiopathic arthritis; RA, rheumatoid arthritis; RRNL, Research Registry for Neonatal Lupus; SLE, systemic lupus erythematosus; SS, Sjogren's syndrome; UAS, undifferentiated autoimmune syndrome.

may appear in 10%–38% of affected children.5 25 26 There is currently no therapy known to prevent cNL. In this multicentre study of 556 infants born to women with SARDs and exposed to anti-Ro±anti-La antibodies, the odds of exposure to HCQ was lower in cNL cases than in controls, suggesting HCQ may protect against cNL.

HCQ may exert a protective effect through inhibition of TLR activation and signalling.27 In lupus-prone NZBxNZW F1 mice, TLR-7 and TLR-9 were shown to be involved in the initiation and maintenance of interface dermatitis, which is similar to histopatho-logical changes found in cNL.4 13 Moreover, inhibiting TLR-7 and TLR-9 signalling improved dermatitis in these mice. In addition, HCQ is effective in treating subacute cutaneous lupus, which closely resembles cNL.7 Another encouraging link derives from data on the protective effect of HCQ on cardiac NL. Four reports have shown that prenatal HCQ exposure is associated with a lower risk of cardiac NL.14–17

Anti-La antibody positivity has been associated with higher inci-dence of cNL and maternal high titres of anti-La antibody were found to be more common in infants with non-cardiac NL, who predominantly had cutaneous involvement.1 28 Binding of human anti-La antibodies to apoptotic cells at the dermal–epidermal junc-tion and around the basal keratinocytes of murine BALB/c fetuses has been demonstrated.29 In our study, maternal anti-La antibody

positivity was significantly associated with cNL in all of our multi-variable models, with ORs ranging from 2.7 to 3.4. Infant’s female gender was found to be associated with cNL in our main analysis. This finding may be secondary to the observed increased expres-sion of Ro and La antigen on the membrane surface of keratino-cytes exposed to oestrogen.30 31

The impact of maternal health status on the risk of developing cNL remains to be characterised. To our knowledge, there are no studies that have demonstrated that women with SS have a higher risk of having a child with cNL as compared with women with SLE. Our subgroup analysis performed in 385 infants born to women with SLE showed that secondary SS was associated with cNL. Maternal and/or fetal genetic factors could explain the mechanism by which maternal disease modulates the risk of skin involvement in their offspring.22–24 Both the main case–control analysis and the subgroup analysis restricted to mothers with SLE identified a lower exposure of non-fluorinated steroids±azathioprine in infants with cNL. A direct effect of these drugs in the infants is not anticipated

1747Barsalou J, et al. Ann Rheum Dis 2018;77:1742–1749. doi:10.1136/annrheumdis-2018-213718

Clinical and epidemiological research

Figure 2 Kaplan-Meier analysis of time of onset of cNL in HCQ-exposed vs non-exposed infants. Kaplan-Meier analysis of the 262 cNL infants for time to onset of cutaneous involvement in HCQ-exposed (represented by the green line) vs HCQ non-exposed (represented by the blue line) infants. Although HCQ-exposed infants were older (6.0 (95% CI 5.7 to 6.3) weeks) at cNL onset than HCQ non-exposed infants (4.4 (95% CI 3.9 to 5.0) weeks), the difference was not statistically significant (p=0.21). cNL, cutaneous neonatal lupus; HCQ, hydroxychloroquine.

since the placenta acts as a relative barrier to azathioprine’s main metabolites and inactivates non-fluorinated steroids.32 33 Indirect effects through modulation of maternal disease, lowering of auto-antibody titres and/or residual confounding due to the association with maternal disease subtypes are other possible explanations for these findings. In addition, our study was not designed to assess the effect of non-fluorinated steroids±azathioprine on cNL; therefore, no conclusion on the effect of these two drugs on cNL should be made. Infants from the RRNL and French RNL were more likely to have cNL than those enrolled into the SickKids NLE Database. This is most likely a result of the different enrolment criteria between registries. Inclusion criteria for both the RRNL and French RNL require at least one infant with NL. In contrast, the SickKids NLE Database includes all children exposed in utero to maternal anti-Ro±anti-La antibodies, irrespective of the infant’s NL status. Consequently, the SickKids NLE Database had signifi-cantly more controls.

The secondary analysis of this study explored the effect of HCQ on the time of onset of cNL. The retrospective multicentre analysis of the 262 cNL infants irrespective of maternal diagnosis suggests that in utero exposure to HCQ did not significantly delay the onset of rash. Despite having a relatively large sample size, there were only 23 cNL infants exposed to HCQ. The lack of a statistically significant effect of HCQ may be explained by the latter limitation. Interestingly, the Kaplan-Meier curves showed a delayed onset of rash in HCQ-exposed infants during the first 6 weeks of life. This suggests that HCQ could have a protective effect early after birth during the time neonatal blood levels are still within a detectable range. Transplacental passage of HCQ has been demonstrated in neonates with measurable drug levels in cord blood.9 The pharma-cokinetics and pharmacodynamics of HCQ in fetuses and neonates

has not been studied. Therefore, the potential therapeutic dose range required to prevent cNL and time at which infants’ blood levels fall below this threshold remain unknown. HCQ has been shown to be transfered into breast milk at very low concentrations resulting in estimated HCQ ingested doses by breastfed infants of 0.1–0.2 mg/kg/day.9 34 Although we did not collect data on breast feeding, we believe it is unlikely that the amount of HCQ found into breast milk had an impact on timing of cNL.

Our study is the first to address the effect of HCQ exposure on cNL. Cutaneous involvement in NL may result in unnecessary medical consultations, investigations and treatments and lead to permanent scarring.5 25 26 Moreover, the risk of cardiac NL in a preg-nancy following the birth of a child with cNL has been estimated to be 13%.3 Using an inexpensive and safe medication to prevent cNL and therefore minimise these issues is definitely appealing. In addition, we were able to assemble a relatively large sample size for a NL study as our data came from three of the largest NL database/registries. Despite inconclusive findings regarding HCQ's effect on timing of cNL, results of this work are valuable as the survival curves showed a delayed onset of rash during the first 6 weeks of life in infants exposed to HCQ in utero. All analyses performed suggest that prenatal exposure to HCQ may have an impact on the development of cNL and support the need for further studies addressing the underlying pathophysiology of cNL and potential preventive use of maternal HCQ for cNL.

Potential limitations need to be acknowledged. Medication adherence in mothers was not measured specifically during pregnancy, but women were queried about medication intake during routine prenatal rheumatology follow-up and on enrol-ment into their respective database. It is therefore possible that some children categorised as exposed to HCQ were not exposed

1748 Barsalou J, et al. Ann Rheum Dis 2018;77:1742–1749. doi:10.1136/annrheumdis-2018-213718

Clinical and epidemiological research

throughout gestation due to maternal non-adherence to therapy.35 Because women with SLE are more likely to be prescribed HCQ than women with other SARDs, confounding by indication could have occurred in our study. To account for this potential issue, a subgroup analysis was performed and results were consistent with that of the primary analysis. For the secondary analysis, a larger sample size may have been required to show a significant difference. The age of onset of cNL was obtained from parents during postnatal clinic follow-up. Because early cNL may be very subtle, parents might not have noticed the cutaneous lesions when they first appeared and the age of onset reported could have been incorrect.

In this large multicentre study, exposure to HCQ in utero was associated with a reduced risk of cNL, and this association remained in subgroup analyses limited to mothers with SLE and in infants who developed cNL within 4 weeks of life. In addition, although not statically significant, cNL cases exposed to HCQ tend to have later onset of cNL perhaps due to protection conferred by HCQ when neonatal HCQ blood levels still remain within a detectable range.

Author affiliations1division of rheumatology, the Hospital for Sick Children, toronto, ontario, Canada2department of Internal Medicine, Cochin Hospital, Centre de référence Maladies Auto-Immunes et Systémiques rares, Ap-Hp, Université paris descartes-Sorbonne paris Cité, paris, France3Arthritis & rheumatism Associates, pS Wheaton, Maryland, USA4division of rheumatology, new York University School of Medicine, new York City, new York, USA5division of Allergy, Immunology and rheumatology, University of rochester, School of Medicine and dentistry, rochester, USA6department of Statistical Science, University of toronto, Centre for Global Health research, St Michael’s Hospital, toronto, ontario, Canada7department of Medicine, obstetrics and Gynaecology, University of toronto, trIo Fertility, toronto, ontario, Canada

Acknowledgements the authors thank the following individuals for assitance with data collection and extraction: SickKids nLE database, Michelle Wang and Shazia Ali; rrnL, tishaun Middleton and French rnL, Ada Clarke. We wish to thank all the rheumatologists working in the Greater toronto Area for referring patients to dr. Laskin and all of the obstetricians at Mount Sinai Hospital High risk Unit. Similarly, we want to acknowledge physicians who refered patients to the rrnL and French registry of neonatal Lupus. Lastly, the authors thank families who agreed to be part of all three registries.

Contributors Study conception and design: JB, nC-C, JpB, EdS, pMI. Acquisition of data: JB, nC-C, AB, CF-n, US, CAL, nM, KL, JpB, EdS, pMI. Analysis and interpretation of data: JB, nC-C, pB, JpB, EdS, pMI. All authors were involved in drafting the article and/or revising it critically for important intellectual content, and all authors approved the final version. JB and pMI had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Funding Funding for rrnL was provided by national Institute of Arthritis and Musculoskeletal and Skin disease (nIAMS) contract n01-Ar-4-2220-11-0-1 and grant 5r37Ar042455, Eunice Kennedy Shriver national Institute of Child Health and Human development (nICHd) grants r01Hd079951-01A1 and r03 Hd069986, and a Lupus Foundation of America LFA Lifeline program grant to JpB.

Competing interests none declared.

Patient consent not required.

ethics approval the study was approved by the Hospital for Sick Children Ethics Boards (rEB1000037488), rrnL IrB (protocol no. 7820), French rnL IrB pitié-Salpetrière.

Provenance and peer review not commissioned; externally peer reviewed.

data sharing statement there are no unpublished data provided in the manuscript.

RefeRences 1 Cimaz r, Spence dL, Hornberger L, et al. Incidence and spectrum of neonatal

lupus erythematosus: a prospective study of infants born to mothers with anti-ro autoantibodies. J Pediatr 2003;142:678–83.

2 Friedman dM, Kim MY, prIdE Investigators. Utility of cardiac monitoring in fetuses at risk for congenital heart block: the pr Interval and dexamethasone Evaluation (prIdE) prospective study. Circulation 2008;117:485–93.

3 Izmirly pM, Llanos C, Lee LA, et al. Cutaneous manifestations of neonatal lupus and risk of subsequent congenital heart block. Arthritis Rheum 2010;62:1153–7.

4 peñate Y, Guillermo n, rodríguez J, et al. Histopathologic characteristics of neonatal cutaneous lupus erythematosus: description of five cases and literature review. J Cutan Pathol 2009;36:660–7.

5 neiman Ar, Lee LA, Weston WL, et al. Cutaneous manifestations of neonatal lupus without heart block: characteristics of mothers and children enrolled in a national registry. J Pediatr 2000;137:674–80.

6 High WA, Costner MI, scarring p. persistent scarring, atrophy, and dyspigmentation in a preteen girl with neonatal lupus erythematosus. J Am Acad Dermatol 2003;48:626–8.

7 Francès C, Cosnes A, duhaut p, et al. Low blood concentration of hydroxychloroquine in patients with refractory cutaneous lupus erythematosus: a French multicenter prospective study. Arch Dermatol 2012;148:479–84.

8 Clowse ME, Magder L, Witter F, et al. Hydroxychloroquine in lupus pregnancy. Arthritis Rheum 2006;54:3640–7.

9 Costedoat-Chalumeau n, Amoura Z, Aymard G, et al. Evidence of transplacental passage of hydroxychloroquine in humans. Arthritis Rheum 2002;46:1123–4.

10 Costedoat-Chalumeau n, Amoura Z, Huong dL, et al. Safety of hydroxychloroquine in pregnant patients with connective tissue diseases. review of the literature. Autoimmun Rev 2005;4:111–5.

11 Clancy rM, Alvarez d, Komissarova E, et al. ro60-associated single-stranded rnA links inflammation with fetal cardiac fibrosis via ligation of tLrs: a novel pathway to autoimmune-associated heart block. J Immunol 2010;184:2148–55.

12 Alvarez d, Briassouli p, Clancy rM, et al. A novel role of endothelin-1 in linking toll-like receptor 7-mediated inflammation to fibrosis in congenital heart block. J Biol Chem 2011;286:30444–54.

13 Guiducci C, tripodo C, Gong M, et al. Autoimmune skin inflammation is dependent on plasmacytoid dendritic cell activation by nucleic acids via tLr7 and tLr9. J Exp Med 2010;207:2931–42.

14 Izmirly pM, Kim MY, Llanos C, et al. Evaluation of the risk of anti-SSA/ro-SSB/La antibody-associated cardiac manifestations of neonatal lupus in fetuses of mothers with systemic lupus erythematosus exposed to hydroxychloroquine. Ann Rheum Dis 2010;69:1827–30.

15 Izmirly pM, Costedoat-Chalumeau n, pisoni Cn, et al. Maternal use of hydroxychloroquine is associated with a reduced risk of recurrent anti-SSA/ro-antibody-associated cardiac manifestations of neonatal lupus. Circulation 2012;126:76–82.

16 tunks rd, Clowse ME, Miller SG, et al. Maternal autoantibody levels in congenital heart block and potential prophylaxis with antiinflammatory agents. Am J Obstet Gynecol 2013;208:64.e1–7.

17 Barsalou J, Jaeggi E, Laskin CA, et al. prenatal exposure to antimalarials decreases the risk of cardiac but not non-cardiac neonatal lupus: a single-centre cohort study. Rheumatology 2017;56:1552–9.

18 Buyon Jp, Hiebert r, Copel J, et al. Autoimmune-associated congenital heart block: demographics, mortality, morbidity and recurrence rates obtained from a national neonatal lupus registry. J Am Coll Cardiol 1998;31:1658–66.

19 Levesque K, “Lupus néonatal” group, Group of collaborators. description of 214 cases of autoimmune congenital heart block: results of the French neonatal lupus syndrome. Autoimmun Rev 2015;14:1154–60.

20 Hanley JA, negassa A, Edwardes Md, et al. Statistical analysis of correlated data using generalized estimating equations: an orientation. Am J Epidemiol 2003;157:364–75.

21 Homish GG, Edwards Ep, Eiden rd, et al. Analyzing family data: a GEE approach for substance use researchers. Addict Behav 2010;35:558–63.

22 Hernández-Molina G, Avila-Casado C, Cárdenas-Velázquez F, et al. Similarities and differences between primary and secondary Sjögren’s syndrome. J Rheumatol 2010;37:800–8.

23 Buyon Jp, Slade SG, reveille Jd, et al. Autoantibody responses to the "native" 52-kda SS-A/ro protein in neonatal lupus syndromes, systemic lupus erythematosus, and Sjogren’s syndrome. J Immunol 1994;152:3675–84.

24 Colombo G, Brucato A, Coluccio E, et al. dnA typing of maternal HLA in congenital complete heart block: comparison with systemic lupus erythematosus and primary Sjögren’s syndrome. Arthritis Rheum 1999;42:1757–64.

25 Lin SC, Shyur Sd, Wu JY, et al. Facial telangiectasia—an unusual complication of neonatal lupus erythematosus: report of one case. Acta Paediatr Taiwan 2004;45:246–8.

26 Weston WL, Morelli JG, Lee LA. the clinical spectrum of anti-ro-positive cutaneous neonatal lupus erythematosus. J Am Acad Dermatol 1999;40(5 pt 1):675–81.

27 Wallace dJ, Gudsoorkar VS, Weisman MH, et al. new insights into mechanisms of therapeutic effects of antimalarial agents in SLE. Nat Rev Rheumatol 2012;8:522–33.

28 Jaeggi E, Laskin C, Hamilton r, et al. the importance of the level of maternal anti-ro/SSA antibodies as a prognostic marker of the development of cardiac neonatal lupus erythematosus a prospective study of 186 antibody-exposed fetuses and infants. J Am Coll Cardiol 2010;55:2778–84.

1749Barsalou J, et al. Ann Rheum Dis 2018;77:1742–1749. doi:10.1136/annrheumdis-2018-213718

Clinical and epidemiological research

29 tran HB, Macardle pJ, Hiscock J, et al. Anti-La/SSB antibodies transported across the placenta bind apoptotic cells in fetal organs targeted in neonatal lupus. Arthritis Rheum 2002;46:1572–9.

30 norris dA, Lee LA. pathogenesis of cutaneous lupus erythematosus. Clin Dermatol 1985;3:20–35.

31 Jones SK. the effects of hormonal and other stimuli on cell-surface ro/SSA antigen expression by human keratinocytes in vitro: their possible role in the induction of cutaneous lupus lesions. Br J Dermatol 1992;126:554–60.

32 Hutson Jr, Lubetsky A, Walfisch A, et al. the transfer of 6-mercaptopurine in the dually perfused human placenta. Reprod Toxicol 2011;32:349–53.

33 Kemp MW, newnham Jp, Challis JG, et al. the clinical use of corticosteroids in pregnancy. Hum Reprod Update 2016;22:240–59.

34 nation rL, Hackett Lp, dusci LJ, et al. Excretion of hydroxychloroquine in human milk. Br J Clin Pharmacol 1984;17:368–9.

35 Costedoat-Chalumeau n, Houssiau F, Izmirly p, et al. A prospective international study on adherence to treatment in 305 patients with flaring SLE: assessment by drug levels and self-administered questionnaires. Clin Pharmacol Ther 2018;103:1074-1082.

1750 Partington RJ, et al. Ann Rheum Dis 2018;77:1750–1756. doi:10.1136/annrheumdis-2018-213883

Clinical and epidemiological research

ExtEndEd rEport

Incidence, prevalence and treatment burden of polymyalgia rheumatica in the UK over two decades: a population-based studyrichard James partington, Sara Muller, toby Helliwell, Christian d Mallen, Alyshah Abdul Sultan

To cite: partington rJ, Muller S, Helliwell t, et al. Ann Rheum Dis 2018;77:1750–1756.

Handling editor Josef S Smolen

► Additional material is published online only. to view please visit the journal online (http:// dx. doi. org/ 10. 1136/ annrheumdis- 2018- 213883).

Arthritis UK primary Care Centre, Keele University, Keele, UK

Correspondence todr richard James partington, research Institute for primary Care & Health Science, Keele University, Staffordshire, St5 5BG, UK; r. partington@ keele. ac. uk

received 6 June 2018revised 21 August 2018Accepted 28 August 2018published online First 8 october 2018

© Author(s) (or their employer(s)) 2018. no commercial re-use. See rights and permissions. published by BMJ.

Key messages

What is already known about this subject? ► The most recent large scale population estimates of the incidence and prevalence of polymyalgia rheumatica were published in 2006 and no similar sized estimates exist of the duration of glucocorticoid treatment in patients with this condition.

What does this study add? ► This study confirms that the incidence of polymyalgia rheumatica has stabilised since the turn of the century and finds that a significant proportion of patients remain on glucorticoid treatment for much longer than guidelines suggest.

How might this impact on clinical practice or future developments?

► These findings will prompt further work to identify those at risk of prolonged glucocorticoid treatment and its inherent risks.

AbsTrACTObjective polymyalgia rheumatica (pMr) is the most common inflammatory rheumatic disease in older people. Contemporary estimates of the incidence and prevalence are lacking, and no previous study has assessed treatment patterns at a population level. this study aims to address this.Methods We extracted anonymised electronic medical records of patients over the age of 40 years from the Clinical practice research datalink in the period 1990–2016. the absolute rate of pMr per 100 000 person-years was calculated and stratified by age, gender and calendar year. Incidence rate ratios were calculated using a poisson regression model. Among persons with pMr, continuous and total duration of treatment with glucocorticoids (GC) were assessed.results 5 364 005 patients were included who contributed 44 million person-years of follow-up. 42 125 people had an incident diagnosis of pMr during the period. the overall incidence rate of pMr was 95.9 per 100 000 (95% CI 94.9 to 96.8). the incidence of pMr was highest in women, older age groups and those living in the South of England. Incidence appears stable over time. the prevalence of pMr in 2015 was 0.85 %. the median (IQr) continuous GC treatment duration was 15.8 (7.9–31.2) months. However, around 25% of patients received more than 4 years in total of GC therapy.Conclusions the incidence rates of pMr have stabilised. this is the first population-based study to confirm that a significant number of patients with pMr receive prolonged treatment with GC, which can carry significant risks. the early identification of these patients should be a priority in future research.

InTrOduCTIOnPolymyalgia rheumatica (PMR) is an inflamma-tory rheumatic disease affecting older people. Its impact on patients’ lives can be devastating, causing stiffness, severe pain and significant impairment to daily activities.1 Glucocorticoids (GCs) remain the mainstay of treatment.2

The incidence and prevalence of PMR vary depending on geography; as latitude increases, so do PMR rates.3 Previous studies have estimated the prevalence of PMR to lie between 0.1% and 1%4 5 and the incidence between 12 and 113 per 100 000 person-years.6–8 The majority of cases of PMR are treated in primary care (71%–84%)7 9; however, much of the existing literature is based

on secondary care hospital records. Therefore the burden of disease may have been underestimated. One large study (Smeeth et al10) used primary care data to estimate the incidence of PMR, reporting an overall rate of 84 per 100 000 person-years, which was increasing with time. However, the final year of data published in this study was 2001; therefore, more contemporaneous esti-mates of national data are needed to guide health service provision.

PMR is managed with gradually reducing GC therapy, from moderate to low doses.3 Joint guidance released by the American College of Rheumatology and the European League Against Rheumatism advises GC treatment for most patients with PMR should end by 2 years.2 However, it has been suggested a large propor-tion of patients experience symptom flare on cessation, or even reduction, of GC therapy (a ‘symptom tail’).11

The aims of this study are to quantify the overall incidence and prevalence of PMR in the UK using a large population-based database and to investigate prescribing of GCs in those diag-nosed with PMR.

1751Partington RJ, et al. Ann Rheum Dis 2018;77:1750–1756. doi:10.1136/annrheumdis-2018-213883

Clinical and epidemiological research

MeTHOdsdata source and study populationAlmost all healthcare in the UK is delivered by the National Health Service (NHS), a public system funded by taxation that provides free, or low-cost, healthcare to all residents. Around 90% of patient contacts in the UK with the NHS is via primary care,12 and 98% of people who live in the UK are registered with a general practice. We used data from the Clinical Prac-tice Research Datalink (CPRD; July 2017 version), which contains data for around 17 million contributing patients within 718 (7.5% of the total) UK general practices. This database, containing electronic, coded information collected during the course of routine healthcare, is representative of the UK population in terms of age, sex and ethnicity,13 and has been used extensively for primary care research.

IncidenceWe analysed data collected between 1 January 1990 and 1 January 2016. Patients contributed data after the latest of four events: (1) the study start date, (2) the date at which they became 40 years old, (3) the date they registered at a participating practice plus 6 months, or (4) the date at which the practice was adjudged to reach internal quality standards, known as the ‘up-to-standard’ date.

The date at which each follow-up ended was the earliest of five events: (1) the end of study period (1 January 2016), (2) the date when a patient transferred out of a practice, (3) the date of death, (4) the last date of data collection from the practice or (5) the date when they were diagnosed with PMR.

Patients with a Read-coded diagnosis of PMR (codes: N20.00 Polymyalgia rheumatica, N200.00 Giant cell arteritis with poly-myalgia rheumatica) in their general practice record were included as incident cases. The first 6 months following registration with a practice were excluded from the incidence analysis to avoid inclu-sion of prevalent cases which may have been incorrectly recorded at the point of registration.10 To improve case ascertainment, we only considered PMR diagnosis to be valid if patients received at least two prescriptions for oral GCs: one within 6 months of the diagnosis date and the second within 6 months of the first prescrip-tion.10 Patients could have a diagnosis of both PMR and giant cell arteritis (GCA). We looked only at the first occurrence of PMR; therefore, all subsequent person-time and diagnostic codes were excluded. This process is summarised in online supplementary figure 1.

Treatment of PMrTo ascertain trends in the management of PMR, we assessed patterns of GC prescribing in the incident cases of PMR. All GC prescriptions recorded in CPRD using medications from the British National Formulary (BNF) chapter 6.3.2 ‘Glucocorticoid therapy’ were included.14 CPRD contains information about the quantity of medication prescribed, the number of units of medication to be taken each day and the prescription duration. The algorithm used to define duration and dose of GC therapy (detailed in online supplementary figure 2) has been defined elsewhere.15 Kaplan-Meier survival methods were used to calculate the median duration of time from diagnosis until completion of continuous GC therapy. The end of a treatment course was determined to have occurred when no further GC prescriptions occurred for 90 days after the calculated duration of the previous prescription. Patients were censored if they were lost to follow-up prior to stopping treat-ment. The 90-day period was chosen as it is the same as in previous CPRD-based studies of medication use.16 As part of a sensitivity analysis, we recalculated this duration (1) by increasing the interval

between prescriptions to 6 months; or (2) in patients who received a diagnosis with another rheumatological condition either prior to PMR diagnosis or in the 2 years subsequently; or (3) were referred to secondary care rheumatology services.

statistical analysisCrude incidence rates of PMR were calculated by dividing the total number of new cases by the total person-years of follow-up per 100 000 person-years. Incidence rates were stratified by age, gender, region and calendar year. Patient age was grouped into decades. Lexis expansion17 was used to calculate incidence rates by year following the study start date of 1 January 1990.

To compare the absolute rate of PMR by patient characteristics, we used a Poisson regression model and calculated the incidence rate ratios (IRR) for each covariate, including sex, age, region and calendar year of diagnosis. Age-adjusted incidences for each covariate were calculated with direct standardisation, using the sample population structure over the whole study.

For treatment pattern analysis, we calculated the average daily and total dose of GC prescribed, as well as the cumulative treat-ment time and the total number of prescriptions and separate treatment courses each patient received. Dosage calculations were made by converting the strength of all medications to milligrams of prednisolone equivalent using the BNF conversion tables of equiv-alent anti-inflammatory doses.14 Results were stratified by starting GC dose, age and sex.

Point prevalence of PMR was calculated for each calendar year by dividing the total number of patients who have received a diag-nosis of PMR at any time in the past and were alive and contributing data on 31 December of that year (numerator) by the total number of patients alive and contributing data on that date (denominator), thereby including incident and prevalent cases. As part of sensi-tivity analysis, we recalculated the prevalence in patients aged over 55 years in order to compare with a recent study.9 Statistical anal-yses were conducted using Stata V.15.1.

resulTsOverall incidenceA total of 5 364 005 individuals contributed 43.97 million person-years of follow-up in the period 1990–2016. The total number of new occurrences of PMR that fulfilled the GC prescription criteria was 42 145. This equated to 90.4% of the total number of PMR cases recorded during this time. The overall incidence rate of PMR among patients aged 40 years and over was 95.9 (95% CI 94.9 to 96.8) per 100 000 person-years (table 1). The incidence rates were significantly higher at older ages: those aged >70 years were around 10 times (IRR=9.61 (95% CI 9.25 to 9.98)) more likely to have PMR compared with those between the ages of 50 and 59 years. Women were 67% more likely to develop PMR compared with men (IRR=1.67 (95% CI 1.64 to 1.71)). A marked variation in incidence rates by region was found (figure 1), with rates highest in the south-west region of the UK (124.1 (95% CI 120.6 to 127.6)) and lowest in the north-east (65 (95% CI 59.5 to 70.9)).

Incidence of PMr over timeThe variation in incidence rates of PMR over time is displayed in table 2 and figure 2. The rate of diagnosis of PMR dipped a little after 1990 until 1996 before increasing significantly until just after the end of the last century; after this the rate of diagnosis of PMR remained relatively stable between 2003 and 2014.

GC prescribing in PMrIn total 1 242 841 GC prescriptions were issued to patients after a diagnosis with PMR; of these 99.9% contained

1752 Partington RJ, et al. Ann Rheum Dis 2018;77:1750–1756. doi:10.1136/annrheumdis-2018-213883

Clinical and epidemiological research

Table 1 Incidence rates of polymyalgia rheumatica, with incidence rate ratios, stratified by age, sex and region

events (n)Person-time at risk (100 000 years)

rate per 100 000(95% CI)

Incidence rate ratio(95% CI)*

Age-standardised incidence rate† (per 100 000 person-years)

Overall 42 145 439.70 95.9 (94.9 to 96.8)

Age

40–49 409 129.96 3.2 (2.9 to 3.47) 0.11 (0.10 to 0.13)

50–59 3139 113.75 27.6 (26.7 to 28.6) Reference

60–69 9683 91.62 105.7 (103.6 to 107.8) 3.80 (3.65 to 3.96)

70–79 17 620 64.76 272.1 (268.1 to 276.1) 9.61 (9.25 to 9.98)

80+ 10 405 33.05 314.9 (308.9 to 321) 10.58 (10.17 to 11.13)

Sex

Male 13 651 212.06 64.4 (63.3 to 65.5) Reference 69.22

Female 28 494 227.64 125.2 (123.7 to 126.6) 1.67 (1.64 to 1.71) 114.87

Region

North East 500 7.69 65 (59.5 to 70.9) 0.82 (0.75 to 0.90) 62.54

North West 3843 49.36 77.9 (75.4 to 80.4) Reference 77.54

Yorkshire and the Humber 1286 16.91 76.1 (72.0 to 80.3) 0.97 (0.92 to 1.04) 73.62

East Midlands 1461 16.71 87.4 (83.1 to 92.0) 1.14 (1.07 to 1.21) 86.13

West Midlands 4207 41.45 101.5 (98.5 to 104.6) 1.26 (1.21 to 1.32) 98.44

East of England 4698 38.44 122.2 (118.8 to 125.8) 1.56 (1.49 to 1.62) 120.41

South West 4850 39.10 124.1 (120.6 to 127.6) 1.45 (1.39 to 1.51) 112.96

South Central 4754 46.70 101.8 (98.9 to 104.7) 1.29 (1.24 to 1.35) 101.57

London 2901 40.63 71.4 (68.9 to 74.1) 0.97 (0.93 to 1.02) 75.76

South East Coast 5167 43.89 117.7 (114.6 to 121) 1.42 (1.36 to 1.48) 110.23

Northern Ireland 991 13.76 72 (67.7 to 76.6) 0.93 (0.87 to 1.00) 73.06

Scotland 3154 40.05 78.7 (76.0 to 81.5) 1.03 (0.99 to 1.08) 81.51

Wales 4333 45.01 96.3 (93.5 to 99.2) 1.16 (1.11 to 1.21) 90.05

*Adjusted for age, sex, region and year of diagnosis if not stratified as a covariate.†Incidence rate is adjusted by age using overall proportion of person-time contributed per 10-year age category.

Figure 1 Incidence rates of polymyalgia rheumatica (PMR) by region, 1990–2015.

information about quantity of medication prescribed and 48.3% about numeric daily dose. The median time taken for patients to stop continuous therapy was 1.31 years (IQR 0.65–2.6) (figure 3). When the treatment gap was increased to 6 months, this increased to 1.88 years (0.93–4.00). When the total GC treatment time was reviewed, the median duration increased further to 1.93 years (0.95–4.03), meaning around

25% of patients received more than 4 years of therapy. Among patients with a rheumatology diagnosis, or those referred to rheumatology, the median continual duration of GC therapy was greater at 1.49 (IQR 0.73–3.16) and 1.55 (IQR 0.79–3.06) years, respectively. The median first and average daily doses of GC received (in milligrams of prednisolone equiv-alent) were 15 mg (IQR 8–21) and 6 mg (IQR 4–9), respec-tively. However, 7138 (16.9%) patients received on average greater than 10 mg GC per day. The median total dose of GC received (in grams of prednisolone equivalent) was 4 g (IQR 2–8). Repeating analyses stratified by initial GC dose, age and sex were unremarkable, with only patients aged under 50 receiving significantly fewer prescriptions.

Prevalence of PMrThe point prevalence of PMR in 2015 among patients aged over 40 years was 0.85% (table 2) and was markedly different between men and women (0.6% and 1.16%). The prevalence increased to 1.7% (95% CI 1.69% to 1.71%) in patients aged over 55 years.

dIsCussIOnMain findingsThis study estimates the burden of PMR in the UK to be slightly higher than previously estimated. In 2015 around 1 in 120 adults aged over 40 have received a diagnosis of PMR. Overall, the incidence of PMR during the study period 1990–2016 was 95.9 per 100 000 person-years (95% CI 94.9 to 96.8). However, after increasing until 2002, the incidence rate of PMR has stabilised. Almost 50% of patients with PMR received more than 2 years

1753Partington RJ, et al. Ann Rheum Dis 2018;77:1750–1756. doi:10.1136/annrheumdis-2018-213883

Clinical and epidemiological research

Table 2 Incidence rates of polymyalgia rheumatica by calendar year

Year events (n)Person-years at risk per 100 000

rate per 100 000(95% CI)

Incidence rate ratio(95% CI)*

Age-standardised incidence rate†

Point prevalence (%)

Overall 42 145 439.70 95.9 (94.9 to 96.8) 0.84

1990 261 3.30 79.2 (70.1 to 89.4) Reference 76.3 0.34

1991 336 4.54 74 (66.5 to 82.3) 0.91 (0.77 to 1.07) 69.4 0.38

1992 401 5.27 76.1 (69 to 83.9) 0.94 (0.80 to 1.09) 72.1 0.44

1993 464 6.02 77.1 (70.4 to 84.4) 0.95 (0.81 to 1.10) 71.9 0.49

1994 476 6.55 72.7 (66.5 to 79.6) 0.90 (0.77 to 1.04) 68.4 0.52

1995 548 7.06 77.7 (71.4 to 84.4) 0.96 (0.83 to 1.11) 74 0.57

1996 657 8.06 81.5 (75.5 to 88) 1.01 (0.87 to 1.16) 77.4 0.60

1997 754 9.34 80.7 (75.2 to 86.7) 1.01 (0.88 to 1.17) 77.6 0.62

1998 863 10.69 80.7 (75.5 to 86.3) 1.01 (0.88 to 1.16) 76.5 0.64

1999 1239 13.00 95.3 (90.1 to 100.8) 1.20 (1.05 to 1.38) 91.7 0.66

2000 1537 15.85 96.9 (92.2 to 101.9) 1.23 (1.08 to 1.40) 93.7 0.68

2001 1792 17.75 100.9 (96.4 to 105.7) 1.28 (1.13 to 1.46) 98.1 0.71

2002 2131 20.05 106.3 (101.9 to 110.9) 1.36 (1.20 to 1.55) 103.5 0.74

2003 2211 21.49 102.9 (98.7 to 107.3) 1.33 (1.17 to 1.51) 101.4 0.77

2004 2296 22.96 100 (96 to 104.2) 1.30 (1.15 to 1.48) 98.5 0.79

2005 2348 23.73 99 (95 to 103) 1.30 (1.14 to 1.48) 98 0.80

2006 2389 24.12 99.1 (95.2 to 103.1) 1.30 (1.15 to 1.48) 97.7 0.83

2007 2451 24.45 100.3 (96.4 to 104.3) 1.32 (1.16 to 1.50) 99.7 0.83

2008 2495 24.60 101.4 (97.5 to 105.5) 1.33 (1.17 to 1.51) 100.7 0.85

2009 2447 24.64 99.3 (95.5 to 103.3) 1.30 (1.15 to 1.48) 98.2 0.85

2010 2497 24.34 102.6 (98.6 to 106.7) 1.35 (1.19 to 1.53) 101.6 0.86

2011 2379 23.83 99.8 (95.9 to 103.9) 1.32 (1.16 to 1.50) 99.1 0.87

2012 2268 23.50 96.5 (92.6 to 100.6) 1.28 (1.12 to 1.45) 95.9 0.87

2013 2198 22.51 97.6 (93.6 to 101.8) 1.29 (1.14 to 1.47) 96.8 0.88

2014 2037 20.58 99 (94.8 to 103.4) 1.30 (1.14 to 1.48) 97.2 0.88

2015 1603 17.60 91.1 (86.7 to 95.6) 1.20 (1.05 to 1.36) 89.1 0.85

*Adjusted for region, age and gender.†Incidence rate is adjusted by age using overall proportion of person-time contributed per 10-year age category.

of GC therapy following diagnosis, despite guidelines suggesting treatment should have ended.

strengths and limitationsWe have conducted the largest study yet to calculate a true esti-mate of the current incidence, prevalence and real-world treat-ment patterns of patients with PMR. This study uses robust methodology in a large, established database of patients who are representative of the UK population. It therefore is likely to be an accurate estimate of the true burden of PMR. Most patients with PMR are managed exclusively in primary care7 9; therefore, this is the most appropriate setting to conduct this study.

A potential limitation is the ascertainment of cases. This was based on medical codes recorded by the primary care physicians, rather than research classification criteria,18 as there is no suffi-cient detail in medical records and therefore CPRD to allow this. Patients may therefore subsequently be diagnosed with an alter-native condition. However, using GC prescriptions to confirm PMR diagnosis is well established.10 19 Greater than 90% of patients with a diagnosis of PMR received at least two GC prescriptions, showing the diagnosis is likely to be accurate in the vast majority of patients. Furthermore, in the UK diagnoses made in secondary care are communicated to, and recorded in, primary care. Therefore although this study examined patients in primary care, it will also contain information from secondary care.

Comparison with other studiesThe highest incidence, 113 per 100 000 patients, previously reported was a study from the South West of England.4 Although the overall incidence rate we found is lower than this, our esti-mate for this region was slightly higher (124.1 (120.6, 127.6)). In the USA, the most recent estimate of PMR rate reported by Raheel et al8 was 63.9 per 100 000. This is lower than our figure. However, this study was not conducted in primary care, and stricter diagnostic criteria rather than codes were used.20 We included patients from a much larger sample, and while our PMR definition is not ideal our estimates are broadly in line with other studies that have used clinical classification criteria. There-fore we believe that the risk of misclassification is minimal.

Women were more likely to develop PMR, with a female to male ratio of approximately 2:1, reflecting previous studies.10 The strong association between older age and risk of devel-oping PMR has been demonstrated before, with other studies reporting a median age at diagnosis of 709 or 75 years.7 As rates of frailty, aches, pains21 and erythrocyte sedimentation rate (ESR) measurements22 increase with age, it is possible that primary care physicians may overdiagnose PMR in at least some of these patients.

The prevalence of PMR has been found to vary between 0.1% and 1% in North Europe and North America.4 23 The prevalence of 0.85% in 2015 calculated in our study is consistent with this. In a recent study in a single large GP practice in the south of the UK, Yates et al9 reported a prevalence of 2.27% in those aged

1754 Partington RJ, et al. Ann Rheum Dis 2018;77:1750–1756. doi:10.1136/annrheumdis-2018-213883

Clinical and epidemiological research

Figure 2 Overall, male and female incidence of polymyalgia rheumatica, with 95% CI, 1990–2015.

Figure 3 Kaplan-Meier plot showing time to final glucocorticoid (GC) prescription, defined as a gap of greater than 90 days following end of previous prescription.

55 years and over. In our data, the prevalence in this group was 1.7%. This discrepancy could be explained by the higher inci-dence of PMR in the south and east of the UK.

Given PMR is known to preferentially affect people of Northern European descent, these results are likely to be gener-alisable to countries with significant number of people from this

ethnic group. However, the incidence and prevalence figures reported in this study are less generalisable to countries at lower latitudes, as incidence and prevalence rates have been found to reduce with decreasing latitude.5 24 25

The incidence of PMR appears higher in the south of the UK compared with the north. This was also demonstrated by Smeeth

1755Partington RJ, et al. Ann Rheum Dis 2018;77:1750–1756. doi:10.1136/annrheumdis-2018-213883

Clinical and epidemiological research

et al.10 Genetic associations between specific human leucocyte antigen molecules and GCA have been found,26 although none yet for PMR.27 However, as no major variation has been found in the genetic make-up of people between different regions around the UK, it is unlikely to be the reason for this difference.28 Other potential reasons include an association between social class and PMR, a viral aetiological agent, or environmental differences such as reduced vitamin D levels in the north of the UK due to less sunlight exposure, which may lead to vitamin D deficiency being diagnosed preferentially.

Smeeth et al10 found that the incidence of PMR in the UK was increasing until 2001, which we replicated. However following this date, the incidence rate plateaued.

With regard to GC therapy, 75% received a first dose between 8 and 21 mg, which corresponds well to the recom-mended starting dose of 12.5–25 mg.2 The median duration of treatment of patients with GC in our sample is, however, less than that found by Shbeeb et al29 in their recent study into GC prescribing in a cohort of 359 patients with PMR in Olmsted County, Minnesota. The median dose prescribed was similar, at around 5 mg, but the length of treatment was greater, with only 19% of patients discontinuing therapy in the first year of treatment, compared with 27% in our data. A number of reasons for this difference could be suggested; for example, their patients may represent more severe variants of the condi-tion; they defined end of treatment as permanent discontinua-tion of GC therapy rather than a gap of 90 days or 6 months; and their inclusion criteria were stricter. Therefore some of the patients included in our study may have gone on to be reclassified with a different condition and have GC therapy curtailed earlier. Our sensitivity analyses of patients who had a record of referral to secondary care rheumatology services confirmed this group had longer continuous and total treat-ment. Both studies agreed though that a significant proportion of patients were subject to prolonged treatment with GCs.

Previous studies have shown that long-term GC treatment increases a person’s risk of a wide range of medical conditions.22 This is the first study of a large population which confirms the existence of a prolonged ‘symptom tail’ in PMR, wherein a significant number of patients receive a higher average daily dose, a larger total dose, more individual prescriptions of GC and receive their treatment over a longer period of time.

The reason behind this symptom tail could be a more severe subtype of PMR or a different underlying diagnosis, for example rheumatoid arthritis, for which referral for secondary care review may be appropriate. Alternatively, it may repre-sent GCs masking the symptoms of other comorbidities which flare on reduction of GC treatment or adrenal insufficiency following prolonged GC use.

COnClusIOn And ClInICAl IMPlICATIOnsIn conclusion, we have established the burden that PMR places on the UK health service. Due to the ageing popula-tion, the prevalence of PMR in the UK is increasing, although the incidence rates appear to have stabilised. Analysis of high-quality, routinely collected primary care data has enabled us to confirm that a significant proportion of patients with PMR receive prolonged treatment with GC, contrary to previ-ously held norms that cure will be achieved within 2 years. Long-term GC therapy is associated with a number of serious adverse effects,22 which is both dose-dependent30 and dura-tion-dependent31. Early identification of patients who are likely to be subject to prolonged GC therapy is a priority area

for future research. These patients could then be prioritised for referral to secondary care for consideration of GC-sparing agents.Contributors Study design: rJp, SM, tH, CdM, AAS. Literature search: rJp. data management: AAS. data analysis, data interpretation: rJp, AAS. First draft and figures: rJp. Critical revision of drafts: SM, tH, CdM, AAS.

Funding rJp is funded by nHS research and Infrastructure funds. CdM is funded by the nIHr Collaboration for Leadership in Applied Health research and Care West Midlands, the nIHr School for primary Care research, and an nIHr research professorship in General practice, which also supports AAS (nIHr-rp-2014-04-026). tH is funded by an nIHr Clinical Lectureship in General practice. the views expressed are those of the authors and not necessarily those of the nHS, the nIHr or the department of Health. the funder was not involved in the study design; in the collection, analysis and interpretation of data; in the writing of the report; or in the decision to submit the article for publication.

disclaimer this study is based in part on data from the Clinical practice research datalink GoLd database obtained under licence from the UK Medicines and Healthcare products regulatory Agency. However, the interpretation and conclusions contained in this report are those of the author(s) alone.

Competing interests none declared.

Patient consent not required.

ethics approval this study was approved by Cprd’s inhouse Independent Scientific Advisory Committee (ISAC) for MHrA database research (protocol number: 17_203rA).

Provenance and peer review not commissioned; externally peer reviewed.

RefeRences 1 Helliwell t, Brouwer E, pease Ct, et al. development of a provisional core domain set

for polymyalgia rheumatica: report from the oMErACt 12 polymyalgia rheumatica working group. J Rheumatol 2016;43:182–6.

2 dejaco C, Singh Yp, perel p. recommendations for the management of polymyalgia rheumatica: a european league against rheumatism/american college of rheumatology collaborative initiative. Ann Rheum Dis 2015;2015:1799–807.

3 Buttgereit F, dejaco C, Matteson EL, et al. polymyalgia rheumatica and giant cell arteritis: a systematic review. JAMA 2016;315:2442–58.

4 Hayward rA, rathod t, Muller S, et al. Association of polymyalgia rheumatica with socioeconomic status in primary care: a cross-sectional observational study. Arthritis Care Res 2014;66:956–60.

5 Salvarani C, Gabriel SE, o’Fallon WM, et al. Epidemiology of polymyalgia rheumatica in olmsted County, Minnesota, 1970-1991. Arthritis Rheum 1995;38:369–73.

6 Salvarani C, Macchioni p, Zizzi F, et al. Epidemiologic and immunogenetic aspects of polymyalgia rheumatica and giant cell arteritis in northern Italy. Arthritis Rheum 1991;34:351–6.

7 Barraclough K, Liddell WG, du toit J, et al. polymyalgia rheumatica in primary care: a cohort study of the diagnostic criteria and outcome. Fam Pract 2008;25:328–33.

8 raheel S, Shbeeb I, Crowson CS, et al. Epidemiology of polymyalgia rheumatica 2000–2014 and examination of incidence and survival trends over 45 years: a population-based study. Arthritis Care Res 2017;69:1282–5.

9 Yates M, Graham K, Watts rA, et al. the prevalence of giant cell arteritis and polymyalgia rheumatica in a UK primary care population. BMC Musculoskelet Disord 2016;17:285.

10 Smeeth L, Cook C, Hall AJ. Incidence of diagnosed polymyalgia rheumatica and temporal arteritis in the United Kingdom, 1990-2001. Ann Rheum Dis 2006;65:1093–8.

11 Mackie SL, Arat S, da Silva J, et al. polymyalgia rheumatica (pMr) special interest group at oMErACt 11: outcomes of importance for patients with pMr. J Rheumatol 2014;41:819–23.

12 Hippisley-Cox J, Vinogradova Y. trends in consultation rates in general practice 1995/1996 to 2008/2009: analysis of the Qresearch database. London QResearch Inf Cent 2009:1–24.

13 Herrett E, Gallagher AM, Bhaskaran K. Is the Gprd GoLd population comparable to the UK population? Int J Epidemiol 2015;44:827–36.

14 Joint Formulary Committee, 2018. British national Formulary online. https://www. medicinescomplete. com/ mc/ bnflegacy/ 64/ [Accessed 1 Feb 2018].

15 paskins Z, Whittle r, Sultan AA, et al. risk of fracture among patients with polymyalgia rheumatica and giant cell arteritis: a population-based study. BMC Med 2018;16:4.

16 Vinogradova Y, Coupland C, Brindle p, et al. discontinuation and restarting in patients on statin treatment: prospective open cohort study using a primary care database. BMJ 2016;353:3305.

17 Hertz S, Lexis W. In: Heyde CC, Seneta E, Crépel p, Fienberg SE, Gani J, eds. Statisticians of the centuries. new York, nY: Springer new York, 2001: 204–7.

18 dasgupta B, Cimmino MA, Kremers HM, et al. 2012 provisional classification criteria for polymyalgia rheumatica: a european league against rheumatism/american college of rheumatology collaborative initiative. Arthritis & Rheumatism 2012;64:943–54.

1756 Partington RJ, et al. Ann Rheum Dis 2018;77:1750–1756. doi:10.1136/annrheumdis-2018-213883

Clinical and epidemiological research

19 pujades-rodriguez M, duyx B, thomas SL, et al. Associations between polymyalgia rheumatica and giant cell arteritis and 12 cardiovascular diseases. Heart 2016;102:383–9.

20 raheel S. Epidemiology of polymyalgia rheumatica 2000-2014. A population based study. Arthritis Rheumatol 2016;68:4265–6.

21 Buckinx F, rolland Y, reginster JY, et al. Burden of frailty in the elderly population: perspectives for a public health challenge. Arch Public Health 2015;73:19.

22 Moghadam-Kia S, Werth Vp. prevention and treatment of systemic glucocorticoid side effects. Int J Dermatol 2010;49:239–48.

23 Eaton WW, pedersen MG, Atladóttir Ho, et al. the prevalence of 30 ICd-10 autoimmune diseases in denmark. Immunol Res 2010;47):228–31.

24 González-Gay MA, García-porrúa C, Vázquez-Caruncho M, et al. the spectrum of polymyalgia rheumatica in northwestern Spain: incidence and analysis of variables associated with relapse in a 10 year study. J Rheumatol 1999;26:1326–32.

25 Gonzalez-Gay MA, Vazquez-rodriguez tr, Lopez-diaz MJ, et al. Epidemiology of giant cell arteritis and polymyalgia rheumatica. Arthritis Rheum 2009;61:1454–61.

26 Carmona Fd, Mackie SL, Martín JE, et al. A large-scale genetic analysis reveals a strong contribution of the HLA class II region to giant cell arteritis susceptibility. Am J Hum Genet 2015;96:565–80.

27 González-Gay MA, Matteson EL, Castañeda S. polymyalgia rheumatica. Lancet 2017;390:1700–12.

28 Leslie S, Winney B, Hellenthal G, et al. the fine-scale genetic structure of the British population. Nature 2015;519:309–14.

29 Shbeeb I, Challah d, raheel S, et al. Comparable rates of glucocorticoid-associated adverse events in patients with polymyalgia rheumatica and comorbidities in the general population. Arthritis Care Res 2018;70:643–7.

30 Huscher d, thiele K, Gromnica-Ihle E, et al. dose-related patterns of glucocorticoid-induced side effects. Ann Rheum Dis 2009; 68:1119–24.

31 Walsh LJ, Wong CA, oborne J, et al. Adverse effects of oral corticosteroids in relation to dose in patients with lung disease. Thorax 2001; 56:279–84.

1757Kloppenburg M, et al. Ann Rheum Dis 2018;77:1757–1764. doi:10.1136/annrheumdis-2018-213202

Clinical and epidemiological research

ExtEndEd rEport

Etanercept in patients with inflammatory hand osteoarthritis (EHOA): a multicentre, randomised, double-blind, placebo-controlled trialMargreet Kloppenburg,1,2 roberta ramonda,3 Klaus Bobacz,4 Wing-Yee Kwok,1 dirk Elewaut,5,6 tom W J Huizinga,1 Féline p B Kroon,1 Leonardo punzi,3 Josef S Smolen,4 Bert Vander Cruyssen,5 ron Wolterbeek,7 Gust Verbruggen,5 ruth Wittoek5

To cite: Kloppenburg M, ramonda r, Bobacz K, et al. Ann Rheum Dis 2018;77:1757–1764.

Handling editor Francis Berenbaum

► Additional material is published online only. to view please visit the journal online (http:// dx. doi. org/ 10. 1136/ annrheumdis- 2018- 213202).

For numbered affiliations see end of article.

Correspondence toprofessor Margreet Kloppenburg, department of rheumatology, Leiden University Medical Center, Leiden 2300 rC, the netherlands; g. kloppenburg@ lumc. nl

GV and rW are last authors.

received 7 February 2018revised 15 August 2018Accepted 31 August 2018published online First 3 october 2018

© Author(s) (or their employer(s)) 2018. no commercial re-use. See rights and permissions. published by BMJ.

AbsTrACTObjective Hand osteoarthritis is a prevalent disease with limited treatment options. Since joint inflammation is often present, we investigated tumour necrosis factor (tnF) as treatment target in patients with proven joint inflammation in a proof-of-concept study.Methods this 1-year, double-blind, randomised, multicentre trial (ntr1192) enrolled patients with symptomatic erosive inflammatory hand osteoarthritis. patients flaring after non-steroidal anti-inflammatory drug washout were randomised to etanercept (24 weeks 50 mg/week, thereafter 25 mg/week) or placebo. the primary outcome was Visual Analogue Scale (VAS) pain at 24 weeks. Secondary outcomes included clinical and imaging outcomes (radiographs scored using Ghent University Scoring System (GUSS, n=54) and MrIs (n=20)).results of 90 patients randomised to etanercept (n=45) or placebo (n=45), respectively, 12 and 10 discontinued prematurely. More patients on placebo discontinued due to inefficacy (6 vs 3), but fewer due to adverse effects (1 vs 6). the mean between-group difference (Md) in VAS pain was not statistically significantly different (−5.7 (95% CI −15.9 to 4.5), p=0.27 at 24 weeks; − 8.5 (95% CI −18.6 to 1.6), p=0.10 at 1 year; favouring etanercept). In prespecified per-protocol analyses of completers with pain and inflammation at baseline (n=61), Md was −11.8 (95% CI −23.0 to −0.5) (p=0.04) at 1 year. Etanercept-treated joints showed more radiographic remodelling (delta GUSS: Md 2.9 (95% CI 0.5 to 5.4), p=0.02) and less MrI bone marrow lesions (Md −0.22 (95% CI −0.35 to −0.09), p = 0.001); this was more pronounced in joints with baseline inflammation.Conclusion Anti-tnF did not relieve pain effectively after 24 weeks in erosive osteoarthritis. Small subgroup analyses showed a signal for effects on subchondral bone in actively inflamed joints, but future studies to confirm this are warranted.

InTrOduCTIOnOsteoarthritis, commonly occurring in knees, hands and hips, is a prevalent cause of disability.1 Currently, therapeutic options are limited, only moder-ately addressing symptoms and not preventing or retarding disease progression.2 3 Despite consid-erable efforts to find relevant treatment targets, a

lack of suitable candidates has hampered the devel-opment of disease-modifying therapies in osteoar-thritis. Traditionally, osteoarthritis is known as a degenerative disease resulting in bone deformations and cartilage loss, although in the last decades it has become clear that local inflammation is important in its pathophysiology.4 Especially in erosive hand osteoarthritis, characterised by subchondral erosions and cortical destruction in the interpha-langeal joints (IPJs), synovitis is frequently demon-strated.5–8 Recent MRI and ultrasonography studies show that synovitis is associated with pain and future radiographic damage.9–11 Proinflammatory cytokines, including tumour necrosis factor (TNF), are produced in the osteoarthritic joints and are

Key messages

What is already known about this subject? ► Hand osteoarthritis (OA) is a prevalent disease with limited treatment options.

► Evidence for an important role of inflammation in hand OA is accumulating.

► Previous studies of different anti-inflammatory trials led to ambiguous results.

What does this study add? ► In this one-year, double-blind, randomised trial, etanercept, a tumor necrosis factor (TNF) inhibitor, did not relieve pain effectively after 24 weeks in patients with erosive hand OA.

► Small subgroup analyses showed a signal for effects on subchondral bone in actively inflamed joints, but future studies to confirm this are warranted.

How might this impact on clinical practice or future developments?

► This trial does not provide evidence for the use of TNF inhibitors in daily treatment of patients with hand osteoarthritis patients.

► Studies investigating treatment strategies specifically targeting inflammation, for example, short-term anti-inflammatory treatment with glucocorticoids or TNF inhibitors during ‘flares’ of the disease, are warranted.

1758 Kloppenburg M, et al. Ann Rheum Dis 2018;77:1757–1764. doi:10.1136/annrheumdis-2018-213202

Clinical and epidemiological research

Figure 1 Trial profile. ITT, intention-to-treat; PP, per-protocol; TB, tuberculosis.

thought to play a role in the induction of structural damage. Thus, TNF could serve as a potential treatment target.12

Earlier clinical trials in hand osteoarthritis have investigated the efficacy of anti-inflammatory medication, but with ambig-uous results.13–20 Although some studies with (systemic) gluco-corticoids and TNF inhibitors showed symptom relief,14–16 20 most randomised placebo-controlled trials were negative.13 19 Furthermore, systemic glucocorticoids did not seem to influence osteoarthritic synovitis, as assessed by ultrasound or MRI.15 19 The lack of efficacy of anti-inflammatory medication in previous studies might partly be explained by inclusion of patients without proven joint inflammation at baseline. Because of the latter, patients most likely to benefit from these drugs may have been missed in previous trials, leading to negative results. Results of a 1-year randomised trial implying that compared with placebo the

TNF inhibitor adalimumab may lead to less erosive radiographic progression in finger joints with palpable soft tissue swelling (as a measure of inflammation) support this hypothesis.18

The aim of this proof-of-concept study was thus to investigate the efficacy and safety of anti-TNF in patients with erosive hand osteoarthritis with proven joint inflammation.

MeTHOdsstudy design and participantsThe EHOA study, an investigator-initiated, 1-year, double-blind, randomised, placebo-controlled, multicentre trial in patients with symptomatic erosive inflammatory hand osteoarthritis was conducted in four European rheumatology outpatient clinics (Austria, Belgium, Italy and the Netherlands), which serve as

1759Kloppenburg M, et al. Ann Rheum Dis 2018;77:1757–1764. doi:10.1136/annrheumdis-2018-213202

Clinical and epidemiological research

Table 1 Baseline characteristics of the intention-to-treat population

etanercept(n=45)

Placebo(n=45)

All patients(n=90)

Age (years) 59.4 (6.5) 60.1 (8.7) 59.7 (7.6)

Sex

Men 8 (18%) 9 (20%) 17 (19%)

Women 37 (82%) 36 (80%) 73 (81%)

BMI (kg/m2)* 26.3 (3.8) 25.5 (3.8) 25.9 (3.8)

Symptom duration (years)* 8.8 (6.0) 10.7 (8.0) 9.7 (7.1)

Duration after diagnosis (years)* 6.2 (6.2) 7.3 (8.3) 6.8 (7.3)

Inclusion criteria

Use of NSAIDs 42 (93%) 44 (98%) 86 (96%)

Fulfilling ACR hand OA criteria* 42 (96%) 43 (96%) 85 (96%)

≥4 IPJs with osteoarthritic nodes* 41 (91%) 41 (93%) 82 (92%)

≥1 IPJ with soft tissue swelling 42 (93%) 40 (89%) 82 (91%)

≥1 IPJ with positive power Doppler* 40 (93%) 40 (91%) 80 (92%)

J-phase or E-phase* 38 (93%) 37 (95%) 75 (94%)

VAS pain at screening >30 mm 44 (98%) 45 (100%) 89 (99%)

Flare after NSAID washout ≥20 mm 37 (82%) 39 (87%) 76 (84%)

Data are mean (SD) or n (%).*Data not available for all randomised patients: BMI n=89, symptom duration n=86, duration after diagnosis n=87, ACR criteria n=85, paid work n=89, IPJs with osteoarthritic nodes n=89, IPJs with power Doppler signal n=87 and J-phase or E-phase n=80.ACR, American College of Rheumatology; BMI, body mass index; IPJ, interphalangeal joints; NSAID, non-steroidal anti-inflammatory drug; OA, osteoarthritis; VAS, Visual Analogue Scale.

secondary and tertiary care clinics for this condition. Patients with ≥4 IPJs with osteoarthritic nodes, ≥1 IPJ with soft tissue swelling or erythema, ≥1 IPJ with positive power Doppler signal on ultrasound, and ≥1 IPJ with radiographic pre-ero-sive or erosive disease (represented respectively by loss of joint space (J-phase) or by erosions or collapse of the subchondral bone (E-phase) according to the Verbruggen-Veys system) were eligible.5 Clinical inflammation, ultrasound abnormalities and radiographic erosions did not have to be present in the same joint. Patients had to have finger pain >30 on a 100 mm Visual Analogue Scale (VAS) while using a stable dose of non-ste-roidal anti-inflammatory drugs (NSAIDs) for ≥5 days/week. Moreover, an insufficient response to at least two NSAIDs was required. Eligible patients discontinued their NSAIDs, and those with a flare, defined as an increase in VAS pain >20 mm (as indication of modifiable pain), were included. Exclusion criteria were contraindications for TNF inhibitors, such as uncontrolled serious comorbidities, active or recurrent infections, malignancy, and drug or substance abuse. Patients with another autoimmune or inflammatory rheumatic disease, or psoriasis, or patients who were using immunomodulating drugs within 90 days before baseline were also excluded. The complete list of exclusion criteria is provided in the online supplementary appendix.

randomisation and maskingPatients were randomly assigned (1:1) to etanercept or placebo. Randomisation was performed using a random-number list in blocks of four. Randomisation and study medication coding list were supplied to an independent person in each centre, respon-sible for treatment allocation. The study drug was provided as a phial with powder, containing 25 or 50 mg etanercept or placebo, and a prefilled syringe containing 1 mL of sterile water for subcutaneous injection once weekly. Masking was achieved

by the similar appearance of placebo and active drug; patients, healthcare providers and outcome assessors remained masked throughout the study.

ProceduresEtanercept was administered subcutaneously at a dose of 50 mg weekly for the first 24 weeks, followed by 25 mg weekly for the remainder of the study, or placebo. Study visits were after 4, 8, 12, 24, 36 weeks and 1 year. Patients who took analgesics during screening were allowed to continue at the same dose or discontinue if desired. No new analgesics were permitted during the first 12 weeks of the study, except paracetamol (up to 2000 mg/day) as rescue medication, which had to be discontinued 1 day prior to the study visits. A stable dosage for at least 3 months with chondroitin sulfate, glucosamine, bisphosphonate, tetracy-cline, glucocorticoid and oestrogen was allowed to be continued during the study. No intra-articular steroids or hyaluronic acid type drugs were allowed during the study.

At every study visit tender and soft swollen IPJ counts (0–18), grip strength (in kg; using a hand dynamometer (My-Gripper (Yamasa), SH5001 (Saehan), DYNA Test) or vigorimeter), VAS pain in IPJs, and patient and physician global assessment on VAS were assessed by trained research nurses. Every 4 weeks, patients completed the Functional Index for Hand OsteoAr-thritis (FIHOA). The Short-Form 36 (SF-36) was completed at baseline and 1 year.

Hand radiographs were taken at baseline, 24 weeks and 1 year, and read paired in chronological order blinded for patient characteristics and treatment applying the Verbruggen-Veys score (GV, RW) and the Ghent University Scoring System (GUSS, GV).5 21 Progression was defined as (1) transition towards an erosive or remodelling Verbruggen-Veys phase (from normal (N-phase), stationary (S-phase) or J-phase, to E-phase (erosive progression), or from N-phase, S-phase, J-phase or E-phase, to R-phase (remodelling)); or (2) decrease in GUSS score (range 0–300 per joint). Ultrasound of the IPJs was performed at screening, 24 weeks and 1 year by experienced ultrasonogra-phers according to a standardised scoring system.22 Power Doppler signal and synovial thickening were assessed semiquan-titatively (0–3). Contrast-enhanced MRI of the IPJs of one hand was performed in a subset of 20 patients (n=10 in each treatment group) at baseline and 1 year. Images were scored according to the OMERACT hand osteoarthritis MRI score for synovitis and bone marrow lesions (BMLs) (0–3 per joint). Imaging protocols and information on reliability are provided in the online supple-mentary appendix.

Patients were monitored for clinical and laboratory evidence of adverse events on a routine basis throughout the study.

OutcomesThe primary endpoint was VAS pain at 24 weeks. Secondary clin-ical endpoints included VAS pain at 1 year, patient and physician global assessment, health-related quality of life (physical compo-nent summary (PCS) of SF-36), FIHOA, number of tender or swollen joints, and grip strength (right and left hand aver-aged), all at 24 weeks and 1 year. Secondary imaging endpoints were the number of joints with synovial thickening and power Doppler signal on ultrasound at 24 weeks and 1 year, and radio-graphic progression assessed with (1) Verbruggen-Veys scores at 24 weeks and 1 year, and (2) change in GUSS scores over 1 year in the per-protocol population. Change in MRI features over 1 year was defined as post-hoc endpoint.

1760 Kloppenburg M, et al. Ann Rheum Dis 2018;77:1757–1764. doi:10.1136/annrheumdis-2018-213202

Clinical and epidemiological research

Table 2 Joint pain at baseline, 24 weeks and 1 year, and between-group differences at 24 weeks and 1 year in intention-to-treat and per-protocol population

etanercept Placebo between-group difference

baseline 24 weeks 1 year baseline 24 weeks 1 year 24 weeks P values 1 year P values

Intention-to-treat

n=45 n=45 n=90 n=90

VAS pain 71.1(15.7)

39.2(24.7)

35.7(25.1)

68.4(12.8)

46.5(23.4)

45.4(25.7)

−5.7(−15.9 to 4.5)

0.27 −8.5(−18.6 to 1.6)

0.10

Per-protocol

n=32 n=32 n=28 n=38 n=38 n=33 n=70 n=61

VAS pain 69.2(13.9)

44.8(22.8)

35.1(24.4)

67.0(13.0)

40.6(24.7)

45.2(26.1)

−5.6(−16.9 to 5.6)

0.32 −11.8(−23.0 to −0.5)

0.04

Data are mean (SD) and mean difference (95% CI).VAS, Visual Analogue Scale.

Table 3 Secondary clinical and imaging outcome measurements at baseline, and between-group differences at 24 weeks and 1 year

etanercept Placebo between-group difference

baseline baseline 24 weeks P values 1 year P values

Intention-to-treat

Clinical outcomes n=45 n=45 n=89 n=89*

VAS patient global 63.7 (18.8) 66.3 (16.2) −2.5 (−12.6 to 7.6) 0.62 −2.3 (−13.0 to 8.4) 0.67

VAS physician global 58.1 (16.9) 56.1 (13.6) −4.0 (−14.9 to 6.8) 0.53 −2.8 (−17.4 to 11.6) 0.70

FIHOA 9.9 (5.9) 10.9 (9.9) 0.0 (−1.7 to 1.8) 0.97 0.0 (−2.4 to 2.3) 0.98

Grip strength 15.9 (10.8) 17.3 (11.9) 0.4 (−1.7 to 2.4) 0.74 0.0 (−2.2 to 2.1) 0.97

SF-36 PCS† 42.9 (8.4) 42.9 (9.3) 0.7 (−3.6 to 5.1) 0.11

Tender joint count 8.0 (9.0) 6.5 (4.6) −0.4 (−1.8 to 1.0) 0.58 0.4 (−1.4 to 2.1) 0.66

Soft swollen joint count 3.0 (2.2) 2.1 (1.8) −0.03 (−0.8 to 0.8) 0.94 −0.01 (−0.9 to 0.9) 0.99

Ultrasound n=43 n=44 n=89 n=89

Joints with power Doppler, median (IQR) 2.0 (1.0–3.0) 1.0 (1.0–3.0) −0.3 (−1.04 to 0.4) 0.39 −0.01 (−0.7 to 0.7) 0.98

Joints with synovial thickening, median (IQR) 3.0 (1.0–6.0) 2.0 (1.0–5.0) 0.2 (−1.02 to 1.3) 0.07 −0.3 (−1.9 to 1.3) 0.73

Selected patient groups

Radiographsठn=23 n=31 n=54

GUSS 287 (36) 288 (34) 2.9 (0.5 to 5.4) 0.02

MRI‡ n=10 n=10 n=20

Synovitis 1.0 (0.5) 1.4 (0.3) 0.03 (−0.2 to 0.3) 0.81

Bone marrow lesions 0.6 (0.3) 0.7 (0.3) −0.2 (−0.4 to −0.1) 0.001

Data are mean (SD) and mean difference (95% CI) unless otherwise specified. All analyses are performed in the intention-to-treat population, unless otherwise indicated.*One patient on etanercept had no baseline or any follow-up assessment.†Data available from n=32 and n=33 patients in etanercept and placebo groups at baseline.‡In patients from per-protocol population.§Data are presented on joint level, between-group difference represents the between-group difference in change over 1 year on joint level.FIHOA, Functional Index for Hand Osteo Arthritis; GUSS, Ghent University Scoring System; SF-36 PCS, Short-Form 36 physical component scale; VAS, Visual Analogue Scale.

statistical analysisA sample size of 45 patients per group was calculated to provide 80% power to detect a between-group difference in mean 10 cm VAS pain of 1.1 per week based on an estimated group SD of 1.9, based on a previous placebo-controlled trial, using mixed-effect models with a two-sided 0.05 significance level.23

All outcomes were analysed in two populations as defined in the protocol: the intention-to-treat population including all randomised patients receiving at least one dose of study medica-tion, who underwent at least one outcome assessment, and the per-protocol population including only completers with symp-tomatic, inflammatory and radiographic erosive disease.

VAS pain was analysed using linear mixed models for repeated measures (at 4, 8, 12, 24, 36, 48 weeks and 1 year), including treatment, study centre, time (categorical), baseline VAS pain and the interaction between treatment and time as independent variables, using a heterogeneous Toeplitz correlation matrix.

VAS pain at 24 weeks was the primary outcome. VAS pain at 1 year and patient global assessment were analysed similarly.

The secondary outcomes, that is, VAS physician global assess-ment, FIHOA, grip strength, SF-36 PCS, tender and swollen joint counts, and number of joints with ultrasound features, were analysed with generalised estimating equations (GEE), specifying the working correlation as first-order autoregres-sive. Fixed treatment effects were specified in such a way that it excluded a treatment effect on baseline. The SF-36 PCS was calculated using the Medical Outcomes Study (MOS) SF-36 scoring algorithm, applying norm-based scores using age-spe-cific and sex-specific Dutch population-based norm scores (since no such norm values are available for the other countries).24 Scores were standardised to a scale of 0–100, mean of 50 and SD of 10; lower scores represent worse health. Changes in GUSS scores and MRI features over 1 year were both analysed on joint level using GEE, accounting for within-patient clustering

1761Kloppenburg M, et al. Ann Rheum Dis 2018;77:1757–1764. doi:10.1136/annrheumdis-2018-213202

Clinical and epidemiological research

Figure 2 Interaction between etanercept treatment and presence of baseline inflammation in a joint (based on soft swelling (A), power Doppler signal on ultrasound (B) or MRI-detected synovitis (C)) on change in GUSS (A,B) and bone marrow lesions on MRI (C) over 1 year. Error bars represent 95% CIs. GUSS, Ghent University Scoring System.

Table 4 List of reported adverse events

etanercept (n=45)

Placebo (n=45)

Adverse events (n)

Total* 28 24

Infectious problems 11 10

Cardiovascular 1 4

Gastrointestinal 2 2

Skin/mucosal 3 1

Exacerbation osteoarthritis 2 0

Pulmonary 0 1

Neurological 1 0

Malignancy 1 0

Other† 7 6

Serious adverse events (n)

Total 1 1

Development of breast cancer 1 0

Infectious diarrhoea 0 1

Dropouts due to adverse events

Total‡ 6 1

Skin eruptions 2 0

Development of breast cancer 1 0

Infectious diarrhoea 0 1

Progression of polyneuropathy 1 0

Disturbed liver function tests 1 0

Musculoskeletal pain and swelling with submandibular gland swelling

1 0

*P=0.65.†Including back pain, bursitis, deep venous thrombosis, disturbed kidney function, local reaction after trauma, perineum cyst, pharyngodynia, progression of polyneuropathy, radicular pain (n=2), submandibular gland swelling, tendinitis and traumatic fracture.‡P=0.11.

effects, specifying the working correlation exchangeable (pair-wise correlations of repeated measures considered the same). In addition, statistical interaction between the presence of baseline inflammation, reflected by soft swelling, power Doppler signal

or MRI-detected synovitis, and treatment effect on change in GUSS scores or BMLs over 1 year was tested. Analyses were performed with SPSS V.23. No data monitoring committee was appointed. This trial was registered at the Netherlands Trial Register, trial registration number NTR1192.

resulTsPatient populationPatients were recruited between 11 March 2008 and 9 January 2012. There were 284 patients assessed for eligibility, and 91 were randomly assigned to a group, of whom 1 withdrew consent before receiving a study drug, leaving 90 patients in the intention-to-treat population (figure 1). Ten (11%) in the placebo and 12 (13%) in the etanercept group prematurely discontinued the study medication. Baseline characteristics were well balanced between the groups (table 1). The study popula-tion was representative of patients with hand osteoarthritis, with 81% women and a mean age of 59.7 years. As shown in table 1, not all patients fulfilled the extensive inclusion criteria. At 24 weeks and 1 year, n=70 and n=61 were eligible for the per-pro-tocol population, respectively.

Primary and secondary clinical outcomesBetween-group differences after 24 weeks and 1 year in the intention-to-treat analysis were −5.7 (95% CI −15.9 to 4.5) and −8.5 (95% CI −18.6 to 1.6), respectively, favouring etanercept, although not being statistically significant (table 2). At 24 weeks, there were no between-group differences in reported analgesic

1762 Kloppenburg M, et al. Ann Rheum Dis 2018;77:1757–1764. doi:10.1136/annrheumdis-2018-213202

Clinical and epidemiological research

use. In the per-protocol analysis, between-group differences were −5.6 (95% CI −16.9 to 5.6) and −11.8 (95% CI −23.0 to −0.5, p=0.04) at 24 weeks and 1 year, respectively, in favour of etanercept (table 2). Between-group differences in the other clinical outcome parameters were in favour of etanercept, but not statistically significant (table 3).

secondary imaging outcomesOf 54 patients (31 on placebo, 23 on etanercept) in the per-pro-tocol population, radiographs were available for scoring at baseline and 1 year. Their Verbruggen-Veys phases at baseline and the changes during follow-up are provided in online supple-mentary table S2. Over 1 year, 9 (3.3% of eligible joints) and 14 (3.5%) joints showed erosive progression (p=0.86), and 24 (8.0%) and 22 (5.2%) joints showed remodelling in the etaner-cept and placebo groups (p=0.13), respectively. The GUSS score was performed in 862 joints. At baseline the mean (SD; range) GUSS score was 287.6 (34.8; 90–300) per joint and similar in both groups (table 3). The between-group difference in change over 1 year was 2.9 (95% CI 0.5 to 5.4, p=0.02) on joint level, indicating more remodelling in the etanercept group. Moreover, a statistical interaction was observed between the presence of soft swelling at baseline in a joint and the effect of etanercept on the GUSS, resulting in a larger improvement in GUSS in those joints (p=0.008; figure 2). We found similar results for joints with positive power Doppler signal on ultrasound at baseline (p=0.005; figure 2).

In the subgroup with MRI available (n=20), no change and no between-group difference could be observed in MRI-detected synovitis over 1 year, whereas BML scores decreased more in the etanercept group (table 3). A similar statistical interaction was found as with GUSS, namely between etanercept treatment and the presence of MRI-detected synovitis (grade 2 or 3) at baseline (p=0.003; figure 2).

On ultrasound, no between-group differences in the number of joints with power Doppler signal or synovial thickening were observed (table 3).

Adverse eventsFifteen of 45 (33%) patients on etanercept and 12 of 45 (27%) on placebo reported at least one adverse event (p=0.65) (table 4). Infectious problems were most often reported: 11 times in the etanercept group and 10 times in the placebo group. Six (13%) patients in the etanercept compared with one (2%) in the placebo group dropped out due to an adverse event (p=0.11). Two serious adverse events occurred (development of breast cancer in the etanercept group and hospital admission for infectious diarrhoea in the placebo group), both not deemed to be related to the study drug.

dIsCussIOnIn this proof-of-concept study on patients with erosive hand osteoarthritis, no statistically significant effect of treatment with anti-TNF on the primary prespecified clinical endpoint, VAS pain at 24 weeks, was seen. However, in subgroup analyses of patients with signs of active inflammation, anti-TNF treatment led to limited symptomatic relief, as well as structural effects. Patients with signs of inflammation at baseline showed a statisti-cally significant and clinically relevant improvement in pain after 1 year, based on a minimally clinically important difference in pain of 0.37 SD units as used before by others (corresponding to 9.1 mm on a 100 mm VAS in this study).25 Moreover, we demon-strated that the possible beneficial effects of TNF inhibition on

subchondral bone were especially present at baseline in actively inflamed joints, marking a possible interplay between synovitis and subchondral bone.

Although TNF inhibitors have consistently been shown to be effective in other rheumatic diseases like rheumatoid arthritis,26 trials in hand osteoarthritis are scarce and have led to conflicting results, some showing symptom relief and others not.13 14 17 18 The two randomised trials of anti-TNF failed to show benefi-cial effects on symptom relief after 6 months and 1 year.13 18 The difference with previous controlled trials may have been the result of selecting different study populations, since none of these studies specifically included patients with proven joint inflammation as in this trial. Also in the current trial, despite our best intentions, some patients were included who had no inflam-mation at baseline and who indeed appeared to have no benefit of etanercept treatment. In addition, the treatment period in previous studies may have been too short. We found that the reduction in pain was only after 1 year of treatment, at which time we also found an effect on structural damage. Therefore we cannot rule out that the pain reduction could be associated with the observed drug’s structural effects.

This is the first-hand osteoarthritis study investigating the effects of anti-TNF on inflammatory signs on ultrasound and MRI. Remarkably, no treatment response on ultrasonographic or MRI-detected synovitis was seen, although this is in line with trials investigating systemic glucocorticoids in hand osteoar-thritis.15 19 However, the severity of BMLs in a joint did decrease on etanercept treatment. Various studies in knee osteoarthritis investigating treatments primarily targeting subchondral bone have also shown that BMLs, but not synovitis, were modified by these interventions.27–29

Surprisingly, synovitis in itself did not improve even though we observed that the effect of etanercept on radiographic erosive damage and BMLs was more pronounced in actively inflamed joints. This is indicative that the interrelation between synovitis, subchondral bone marrow inflammation and structural damage is not straightforward and clearly distinct in erosive hand osteoar-thritis compared with rheumatoid arthritis. However, ultrasonog-raphy was performed in several centres with different equipment, focusing on the presence or absence of synovitis, which might have hampered sensitivity to change. Moreover, MRI was only performed in a small (random) subset of participants. In any case, it suggests that there may not be a direct role for synovitis in the pathophysiology of osteoarthritis, but rather that synovitis may act as an intermediate factor, negatively influencing the subchondral bone or cartilage through production of cytokines like TNF. Alter-natively, the observed synovitis is occurring secondary to structural changes, with the primary events occurring in the subchondral bone or cartilaginous tissues. This interplay between synovitis and subchondral bone or cartilage confirms the suspected importance of the entire joint structure that has been emphasised before.30

Several pathways have been described by which TNF-alpha, which is known to be produced in the inflamed synovium in osteoarthritis, could induce structural damage, including induc-tion of other proinflammatory cytokines (eg, interleukin (IL)-6 and IL-8), synthesis of matrix metalloproteinases by chondro-cytes and synovial cells, an increased synthesis of cyclo-oxy-genase-pathway products, and an upregulation of nitric oxide production.12 Moreover, TNF-alpha has been shown to induce osteoclastic bone resorption in vitro, and in an animal osteo-arthritis model treatment with anti-TNF appeared to improve subchondral bone structure.31 32

While this study has a number of strengths, such as its randomised, double-blind design, the inclusion of a specific

1763Kloppenburg M, et al. Ann Rheum Dis 2018;77:1757–1764. doi:10.1136/annrheumdis-2018-213202

Clinical and epidemiological research

subset of patients with hand osteoarthritis likely to benefit from treatment, and the inclusion of ultrasound and MRI evaluation, there are also several limitations. The etanercept dose of 50 mg weekly was based on experience in inflammatory rheumatic diseases such as rheumatoid arthritis. With the eye on safety and the uncertainty of clinical efficacy, it was decided to halve the dose after 24 weeks to 25 mg weekly, which may have been too low for the drug to be efficacious. Not all patients fulfilled all the stringent inclusion criteria, but the vast majority fulfilled most criteria. Since this trial was set up as a proof-of-concept study, a prespecified per-protocol analysis was performed including only those who did fulfil all the criteria. Although postrandomisa-tion exclusion may introduce bias, it is known to lead to more informative analyses when applied appropriately.33 Although no important imbalance between the groups was noted in the number of dropouts, numerically more patients on etanercept dropped out due to adverse events and less due to inefficacy. Also, for the analysis of the radiographic data, a number of data were missing, due to absence of or bad-quality radiographs at one or more time points. The potential influence of missing data on the results is unclear; however, specific (preplanned) statis-tical methods were chosen to minimise bias. The study sample size, as calculated, resulted in 90 patients, but the sample size of the per-protocol analysis was smaller. Nevertheless a difference between the groups was seen.

Future research is warranted to elucidate the proposed inter-action between synovium and subchondral bone. This trial does not provide evidence for the use of TNF inhibitors in daily treat-ment of patients with hand osteoarthritis in general. Rather it highlights the possible role of targeted treatment of patients or even joints in which inflammation is present. Studies investi-gating treatment strategies specifically targeting inflammation, for example, short-term anti-inflammatory treatment with glucocorticoids or TNF inhibitors during ‘flares’ of the disease, are warranted.

Author affiliations1department of rheumatology, Leiden University Medical Center, Leiden, the netherlands2department of Clinical Epidemiology, Leiden University Medical Center, Leiden, the netherlands3rheumatology Unit, department of Medicine dIMEd, University of padua, padua, Italy4division of rheumatology, department of Internal Medicine III, Medical University of Vienna, Vienna, Austria5department of rheumatology, Ghent University Hospital, Ghent, Belgium6VIB inflammation research Center, Ghent University, Ghent, Belgium7department of Medical Statistics and Bioinformatics, Leiden University Medical Center, Leiden, the netherlands

Acknowledgements We are indebted to Marion Kortekaas and Carla Campana for performing the ultrasound, to theo Stijnen for his involvement in the study protocol and statistical analysis plan, and to Ida Haugen for her involvement in training FpBK to score hand MrIs.

Contributors MK was the principal investigator and contributed to study design, data collection, data analysis, data interpretation and writing of the report. rr, W-YK, KB, GV contributed to study design, data collection, data interpretation and writing of the report. dE, tWJH, Lp, JSS contributed to study design, data interpretation and writing of the report. FpBK, BVC contributed to data analysis, data interpretation and writing of the report. rWo contributed to data analysis and writing of the report. rWi contributed to study design, data collection, data analysis, data interpretation and writing of the report.

Funding this study was funded by pfizer.

Competing interests MK reports an unrestricted grant paid to the institution and study medication free of charge from pfizer (Wyeth) during the conduct of the study, outside the submitted work. MK reports advisory board fees from Levicept, GlaxoSmithKline, AbbVie, Merck and pfizer, grants from IMI-EFpIA ApproACH, grants from pfizer, and other from AbbVie. W-YK reports an unrestricted grant from pfizer (Wyeth) during the conduct of the study. tWJH/department of rheumatology LUMC

has received grants/lecture fees/consultancy fees from Merck, UCB, Bristol-Myers Squibb, pfizer, roche, Sanofi-Aventis, Crescendo Bioscience and Eli Lilly. JSS reports grants from AbbVie, Lilly, pfizer and roche, personal fees from AbbVie, Amgen, AstraZeneca, Astro, BMS, Celgene, Celltrion, Chugai, Gilead, Glaxo, ILtoo, Janssen, Lilly, MedImmune, MSd, novartis-Sandoz, pfizer, roche, Samsung, Sanofi and UCB, outside the submitted work. rWi reports having a patent GUSS issued.

Patient consent obtained.

ethics approval the trial was approved by the medical ethical committee in each participating centre.

Provenance and peer review not commissioned; externally peer reviewed.

data sharing statement Additional documents (study protocol and statistical analysis plan) are available upon request (contact the corresponding author).

RefeRences 1 GBd 2013 dALYs and HALE Collaborators, Murray CJ, Barber rM, et al. Global,

regional, and national disability-adjusted life years (dALYs) for 306 diseases and injuries and healthy life expectancy (HALE) for 188 countries, 1990-2013: quantifying the epidemiological transition. Lancet 2015;386:2145–91.

2 Kloppenburg M. Hand osteoarthritis-nonpharmacological and pharmacological treatments. Nat Rev Rheumatol 2014;10:242–51.

3 Hochberg MC, Altman rd, April Kt, et al. American College of rheumatology 2012 recommendations for the use of nonpharmacologic and pharmacologic therapies in osteoarthritis of the hand, hip, and knee. Arthritis Care Res 2012;64:465–74.

4 pelletier Jp, Martel-pelletier J, Abramson SB. osteoarthritis, an inflammatory disease: potential implication for the selection of new therapeutic targets. Arthritis Rheum 2001;44:1237–47.

5 Verbruggen G, Veys EM. numerical scoring systems for the anatomic evolution of osteoarthritis of the finger joints. Arthritis Rheum 1996;39:308–20.

6 Kortekaas MC, Kwok WY, reijnierse M, et al. In erosive hand osteoarthritis more inflammatory signs on ultrasound are found than in the rest of hand osteoarthritis. Ann Rheum Dis 2013;72:930–4.

7 Wittoek r, Carron p, Verbruggen G. Structural and inflammatory sonographic findings in erosive and non-erosive osteoarthritis of the interphalangeal finger joints. Ann Rheum Dis 2010;69:2173–6.

8 punzi L, Frigato M, Frallonardo p, et al. Inflammatory osteoarthritis of the hand. Best Pract Res Clin Rheumatol 2010;24:301–12.

9 Yusuf E, Kortekaas MC, Watt I, et al. do knee abnormalities visualised on MrI explain knee pain in knee osteoarthritis? A systematic review. Ann Rheum Dis 2011;70:60–7.

10 Haugen IK, Bøyesen p, Slatkowsky-Christensen B, et al. Associations between MrI-defined synovitis, bone marrow lesions and structural features and measures of pain and physical function in hand osteoarthritis. Ann Rheum Dis 2012;71:899–904.

11 Kortekaas MC, Kwok WY, reijnierse M, et al. Brief report: association of inflammation with development of erosions in patients with hand osteoarthritis: a prospective ultrasonography study. Arthritis Rheumatol 2016;68:392–7.

12 Kapoor M, Martel-pelletier J, Lajeunesse d, et al. role of proinflammatory cytokines in the pathophysiology of osteoarthritis. Nat Rev Rheumatol 2011;7:33–42.

13 Chevalier x, ravaud p, Maheu E, et al. Adalimumab in patients with hand osteoarthritis refractory to analgesics and nSAIds: a randomised, multicentre, double-blind, placebo-controlled trial. Ann Rheum Dis 2015;74:1697–705.

14 Fioravanti A, Fabbroni M, Cerase A, et al. treatment of erosive osteoarthritis of the hands by intra-articular infliximab injections: a pilot study. Rheumatol Int 2009;29:961–5.

15 Keen HI, Wakefield rJ, Hensor EM, et al. response of symptoms and synovitis to intra-muscular methylprednisolone in osteoarthritis of the hand: an ultrasonographic study. Rheumatology 2010;49:1093–100.

16 Kvien tK, Fjeld E, Slatkowsky-Christensen B, et al. Efficacy and safety of a novel synergistic drug candidate, Crx-102, in hand osteoarthritis. Ann Rheum Dis 2008;67:942–8.

17 Magnano Md, Chakravarty EF, Broudy C, et al. A pilot study of tumor necrosis factor inhibition in erosive/inflammatory osteoarthritis of the hands. J Rheumatol 2007;34:1323–7.

18 Verbruggen G, Wittoek r, Vander Cruyssen B, et al. tumour necrosis factor blockade for the treatment of erosive osteoarthritis of the interphalangeal finger joints: a double blind, randomised trial on structure modification. Ann Rheum Dis 2012;71:891–8.

19 Wenham CY, Hensor EM, Grainger AJ, et al. A randomized, double-blind, placebo-controlled trial of low-dose oral prednisolone for treating painful hand osteoarthritis. Rheumatology 2012;51:2286–94.

20 Spolidoro paschoal no, natour J, Machado FS, et al. Effectiveness of triamcinolone hexacetonide intraarticular injection in interphalangeal joints: a 12-week randomized controlled trial in patients with hand osteoarthritis. J Rheumatol 2015;42:1869–77.

21 Verbruggen G, Wittoek r, Vander Cruyssen B, et al. Morbid anatomy of ’erosive osteoarthritis’ of the interphalangeal finger joints: an optimised scoring system to monitor disease progression in affected joints. Ann Rheum Dis 2010;69:862–7.

1764 Kloppenburg M, et al. Ann Rheum Dis 2018;77:1757–1764. doi:10.1136/annrheumdis-2018-213202

Clinical and epidemiological research

22 Keen HI, Lavie F, Wakefield rJ, et al. the development of a preliminary ultrasonographic scoring system for features of hand osteoarthritis. Ann Rheum Dis 2008;67:651–5.

23 Vander Cruyssen B, Moqadam Q, Groeneboer S. xylosan polysulphate as a symptom modifying osteoarthritis drug in patients with finger joint arthritis. Scand J Rheumatol 2008;37.

24 Aaronson nK, Muller M, Cohen pd, et al. translation, validation, and norming of the dutch language version of the SF-36 Health Survey in community and chronic disease populations. J Clin Epidemiol 1998;51:1055–68.

25 Wandel S, Jüni p, tendal B, et al. Effects of glucosamine, chondroitin, or placebo in patients with osteoarthritis of hip or knee: network meta-analysis. BMJ 2010;341:c4675.

26 nam JL, ramiro S, Gaujoux-Viala C, et al. Efficacy of biological disease-modifying antirheumatic drugs: a systematic literature review informing the 2013 update of the EULAr recommendations for the management of rheumatoid arthritis. Ann Rheum Dis 2014;73:516–28.

27 Callaghan MJ, parkes MJ, Hutchinson CE, et al. A randomised trial of a brace for patellofemoral osteoarthritis targeting knee pain and bone marrow lesions. Ann Rheum Dis 2015;74:1164–70.

28 Laslett LL, doré dA, Quinn SJ, et al. Zoledronic acid reduces knee pain and bone marrow lesions over 1 year: a randomised controlled trial. Ann Rheum Dis 2012;71:1322–8.

29 pelletier Jp, roubille C, raynauld Jp, et al. disease-modifying effect of strontium ranelate in a subset of patients from the phase III knee osteoarthritis study SEKoIA using quantitative MrI: reduction in bone marrow lesions protects against cartilage loss. Ann Rheum Dis 2015;74:422–9.

30 Bijlsma JW, Berenbaum F, Lafeber Fp. osteoarthritis: an update with relevance for clinical practice. Lancet 2011;377:2115–26.

31 Bertolini dr, nedwin GE, Bringman tS, et al. Stimulation of bone resorption and inhibition of bone formation in vitro by human tumour necrosis factors. Nature 1986;319:516–8.

32 Ma CH, Lv Q, Yu Yx, et al. protective effects of tumor necrosis factor-α blockade by adalimumab on articular cartilage and subchondral bone in a rat model of osteoarthritis. Braz J Med Biol Res 2015;48:863–70.

33 Fergusson d, Aaron Sd, Guyatt G, et al. post-randomisation exclusions: the intention to treat principle and excluding patients from analysis. BMJ 2002;325:652–4.

1765Nakamura H, et al. Ann Rheum Dis 2018;77:1765–1772. doi:10.1136/annrheumdis-2018-213588

Basic and translational research

ExtEndEd rEport

Ectopic RASGRP2 (CalDAG-GEFI) expression in rheumatoid synovium contributes to the development of destructive arthritisHiroyuki nakamura,1 Sanae Shimamura,1 Shinsuke Yasuda,1 Michihito Kono,1 Michihiro Kono,1 Yuichiro Fujieda,1 Masaru Kato,1 Kenji oku,1 toshiyuki Bohgaki,1 tomohiro Shimizu,2 norimasa Iwasaki,2 tatsuya Atsumi1

To cite: nakamura H, Shimamura S, Yasuda S, et al. Ann Rheum Dis 2018;77:1765–1772.

Handling editor Josef S Smolen

► Additional material is published online only. to view please visit the journal online (http:// dx. doi. org/ 10. 1136/ annrheumdis- 2018- 213588).

1department of rheumatology, Endocrinology and nephrology, Faculty of Medicine and Graduate School of Medicine, Hokkaido University, Sapporo, Hokkaido, Japan2department of orthopedic Surgery, Faculty of Medicine, Hokkaido University, Sapporo, Hokkaido, Japan

Correspondence toShinsuke Yasuda, department of rheumatology, Endocrinology and nephrology, Faculty of Medicine and Graduate School of Medicine, Hokkaido University, Sapporo 060-8638, Japan; syasuda@ med. hokudai. ac. jp

Hn and SS contributed equally.

received 16 April 2018revised 27 June 2018Accepted 14 July 2018published online First 3 August 2018

© Author(s) (or their employer(s)) 2018. no commercial re-use. See rights and permissions. published by BMJ.

ABsTrACTObjectives rheumatoid arthritis (rA) is an autoimmune polyarthritis, in which fibroblast-like synoviocytes (FLS) play a key role in cartilage and bone destruction through tumour-like proliferation and invasiveness. Considering still unsatisfactory remission rate in rA even under treatment with biological disease-modifying antirheumatic drugs, novel therapeutic strategy for treatment-resistant rA is still awaited. In this study, we analysed the expression and function of ras guanine nucleotide-releasing proteins (rASGrps), guanine exchange factors for small Gtpase ras, in FLS as a potential therapeutic target for rA.Methods the expression of RASGRPs mrnA was quantified by a real-time pCr assay in FLS isolated from synovial tissue samples. rASGrp2 protein was also evaluated immunohistochemically. then, we transiently transfected FLS with rASGrp2 expression vector and assessed their proliferation, adhesion, migration and invasion by cellular functional assays and downstream signalling activation using immunoblot. Finally, the therapeutic effect of rASGrp2 silencing was evaluated in type-II collagen-induced arthritis rats.results rASGrp2 was abundantly expressed in FLS from rA synovium, whereas scarcely found in those from osteoarthritis. Expression of rASGrp2 in rA-FLS was enhanced by transforming growth factor-beta. rASGrp2 activated rAp-1, subsequently affecting nuclear factor kappa-light-chain-enhancer of activated B cells pathway and actin dynamics in FLS. rASGrp2-overexpressed FLS had increased abilities of adhesion, migration and interleukin (IL)-6 production. Silencing of rASGrp2 using the intra-articular injection of Rasgrp2-specific sirnAs dampened experimental arthritis in rats by inhibiting pannus formation.Conclusions rASGrp2 was identified to be involved in the pathogenesis of rA by promoting adhesion, migration and IL-6 production from FLS, proposed as a potential novel non-immunosuppressive therapeutic target for rA.

InTrOduCTIOnRheumatoid arthritis (RA) is an autoimmune poly-arthritis characterised by progressive cartilage and bone destruction, leading to poor joint and vital prognosis. Inflammatory cytokines produced by activated macrophages, such as interleukin (IL)-6 and tumour necrosis factor-alpha (TNF-α), cause

a tumour-like proliferation and drive an inva-sive phenotype of fibroblast-like synoviocytes (FLS), resulting in pannus formation.1 FLS are directly involved in cartilage and bone destruc-tion by production of matrix metalloproteinases (MMPs) and activation of osteoclasts through receptor activator of nuclear factor kappa-B ligand (RANKL) and also by its potent invasive nature.2 FLS also contribute to inflammatory amplifier via IL-6 production. When in the presence of IL-17A produced by Th17 cells, excessive IL-6 produc-tion is induced by a positive feedback loop of IL-6 signalling.3 Thus, biological disease-modifying anti-rheumatic drugs targeting these inflammatory cyto-kines have dramatically improved the prognosis of RA. However, the response to biologic therapy depends on patients’ background, in particular disease duration and treatment history. In general, the more drugs the patients experience, the lower the response rates become.4 In addition, opportu-nistic infections related to cytokine blockade are also inevitable clinical disadvantages. Therefore, further exploration for novel therapeutic strategy in treatment-resistant/refractory RA is still awaited. Although cell proliferation/inflammatory cytokines/co-stimulatory molecules have been feasible targets for RA treatment, there has never been a treatment strategy targeting specifically on the motility of FLS.

Ras-mitogen-activated protein kinase (MAPK) pathway is a cascade consisting of signalling mole-cules that control various cellular functions from proliferation, apoptosis, differentiation to stress responses and immune function. Ras-MAPK pathway is extensively involved in tumour-like proliferation of FLS in RA as cellular responses to inflammatory cytokines.5 Ras guanine nucle-otide-releasing proteins (RASGRPs, also called CalDAG-GEFs) that comprise RASGRP1–4 are a calcium and diacylglycerol-regulated guanine exchange factor (GEF) for Ras, which mainly contribute to activation of Ras-MAPK pathway and subsequent cell proliferation.6 We previously reported that RASGRP4 was abundantly expressed in the FLS from a subset of patients with RA and promote proliferation of FLS and MMP-1 produc-tion from FLS. Moreover, intra-articular injection of Rasgrp4-specific small interfering RNA (siRNA) dampened experimental arthritis in rats, suggesting RASGRP4 as a candidate target for treatment of proliferative synovitis.7 In addition, RASGRP1,

1766 Nakamura H, et al. Ann Rheum Dis 2018;77:1765–1772. doi:10.1136/annrheumdis-2018-213588

Basic and translational research

which is preferentially expressed in T cells, but smaller amounts in other leucocytes such as B cell, neutrophils, mast cells and natural killer cells, has recently implicated as one of the disease susceptible gene for RA.8 However, the role of RASGRPs other than RASGRP4 in FLS has remained unclear.

To explore the possibility of suppressing RASGRPs in FLS as a novel treatment for RA, we evaluated the expression of RASGRP1–4 in RA-FLS. During this procedure, we unexpect-edly identified that RASGRP2, which is generally accepted as a GEF for RAP-1 in platelets,9 was highly expressed in RA-FLS.

MATerIAls And MeTHOdsDetailed explanation of each procedure was described in the online Supplementary file 1.

Synovial tissues were acquired during synovectomy or joint replacement surgery from patients with RA or osteoarthritis (OA) after written informed consent. Every patient with RA fulfilled the American College of Rheumatology criteria for the classification of RA.10 In quantitative reverse transcrip-tion (qRT)-PCR experiments for RASGRPs transcripts, FLS from RA with various treatments or from OA at their passages 4–5 were utilised, whereas in RASGRPs overexpression or cytokines stimulation experiment, biologics-pretreated RA or OA-FLS at passages 6–8 were utilised.

FLS were isolated from the synovial tissue samples using type-I collagenase. Total RNA was obtained from FLS using TRIzol RNA Reagent, and the expression of RASGRPs mRNA in FLS was quantified by a real-time qRT-PCR analysis. The syno-vial tissue samples were evaluated immunohistochemically for the presence of RASGRP2. The expression of RASGRP2 mRNA and RASGRP2 protein in FLS cultured under cytokine stimula-tion was evaluated by real-time qRT-PCR and immunoblotting, respectively.

Open reading frame encoding RASGRP2 was cloned from Jurkat cell and was inserted into pcDNA3.1/V5-His-TOPO expression vector. RASGRP2-expression vector was transfected to FLS using Neon Transfection System. RASGRP2-trans-fected FLS were assessed in vitro for proliferation ability using 5-bromo-2′-deoxyuridine (BrdU)-based assay, migration using wound-healing assay by scratching, adhesion using extracellular matrices (ECM) coated or non-coated plastic plate, migration and invasion using two-chamber chemotaxis assay in the pres-ence or absence of ECM. The production of RA-related proteins from FLS was evaluated by real-time qRT-PCR or ELISA. Signal-ling pathway in FLS was assessed by immunoblotting and actin dynamics by immunofluorescence.

Collagen-induced arthritis (CIA) was installed in a 7-week old female Lewis rats by immunisation at the base of the tail with bovine type-II collagen emulsified in Incomplete Freund’s Adju-vant, as previously described.11 On day 14, the CIA rats received an intra-articular injection of 50 µL solution that contained 10 µM siRNA-atelocollagen complex into both the right and left ankle joints. Arthritis score,12 diameter, radiographic severity on micro-CT13 and pannus formation14 in the specimens of the ankle joints were evaluated.

resulTsrAsGrP2 is abundantly expressed in rheumatoid synoviumTo evaluate the expression of RASGRPs in FLS, we obtained syno-vial tissue samples from patients with RA or OA who underwent knee joint replacement surgery. Levels of RasGRP1–4 transcripts were quantified in early passages of cultured FLS from patients with RA or OA using real-time qRT-PCR analyses. RASGRP2

as well as RASGRP4 mRNA were highly expressed in FLS from some of our patients with RA compared with those from patients with OA (figure 1A). Synovial tissue samples were evaluated immunohistochemically for the presence of RASGRP2/4. FLS in synovial tissue were identified by H&E stain and by immuno-histochemistry using antibody against cadherin-11, a marker of FLS as well as a possible treatment target in RA.15 RASGRP2 protein was abundant in FLS and vascular endothelium from RA synovial tissues, whereas scarcely found in OA synovial tissues (figure 1B,C). RASGRP4 positive areas were increased in FLS from RA synovial tissues compared with OA synovial tissues (online Supplementary figure S1), as we previously reported.7 The expression of RASGRP2 was also confirmed in the lining layer, in particular pannus formation site, of the synovial tissues from CIA mice (online Supplementary figure S2). Thus, here we identified for the first time that RASGRP2 was abundantly expressed in FLS from patients with RA and CIA mice, indicating the involvement of this GEF in the pathogenesis of destructive arthritis.

We next determined whether inflammatory cytokines have any effects on the expression of RASGRP2 since the expres-sion levels of RASGRP2 mRNA tended to be lower in FLS from patients with RA treated with biologics than those without biologics (figure 1A). We evaluated the expression of RASGRP2 in FLS cultured under stimulation of cytokines using qRT-PCR analyses and immunoblotting. Stimulation with transforming growth factor-beta (TGF-β) significantly increased the expres-sion of RASGRP2 mRNA transcription as well as RASGRP2 protein in RA-FLS. Vascular endothelial growth factor (VEGF) and platelet-derived growth factor (PDGF) seemed to increase RASGRP2 protein in RA-FLS, whereas increment of mRNA was not statistically significant (figure 1D). These enhancement of RASGRP2 expression by the growth factors was not found in FLS from patients with OA (online Supplementary figure S3). TNF-α, IL-6, IL-1β, interferon-gamma or IL-17A stimulation did not significantly increase the expression of RASGRP2 at transcription/protein levels (figure 1D).

rAsGrP2 promotes adhesion/migration and Il-6 production of FlsTo determine the function of RASGRP2 in FLS, we established RASGRP2-overexpressed FLS (RASGRP2+FLS) by transfecting RASGRP2 expression vector. This approach was taken because FLS tend to lose their expression of RASGRP2 during passages when cultures in the absence of growth factors/inflammatory cytokines (online Supplementary figure S4). First, we evaluated cell proliferation of RASGRP2+FLS by BrdU assay. As a result, cell proliferation of RASGRP2+FLS did not differ from that of control FLS transfected with empty vector (figure 2A). In contrast, overexpression of RASGRP4 increased proliferation of FLS (online Supplementary figure S5), consistent with our previous report.7 Second, we assessed cell migration of RASGRP2+FLS by wound healing assay. RASGRP2+FLS increased the ability of migration and promoted wound healing after 24 and 48 hours compared with control FLS (figure 2B,C). Third, to evaluate cell adhesion of RASGRP2+FLS, we applied FLS into poly-carbonate membrane inserts in a 24-well plate and observed the membrane after 24 hours. Adhesion of RASGRP2+FLS increased onto the membrane with ECM coating, but not onto the membrane without ECM coating (figure 2D,E). This finding suggests that RASGRP2 might increase the adhesive ability of FLS through activation of integrin as found in platelets.9 Because activated integrin binds ECM and mediates cell adhesion,16 no

1767Nakamura H, et al. Ann Rheum Dis 2018;77:1765–1772. doi:10.1136/annrheumdis-2018-213588

Basic and translational research

Figure 1 Expression of Ras guanine nucleotide-releasing proteins (RASGRPs) in fibroblast-like synoviocytes (FLS). (A) Levels of RASGRPs transcript in FLS from patients with rheumatoid arthritis (RA) or osteoarthritis (OA). Patients (3–8) with RA were treated with biologics. (B) Representative synovial tissue specimens of patients with RA or OA stained with H&E, anticadherin-11 antibody or anti-RASGRP2 antibody. Arrows show FLS and arrowheads endothelium. Original magnification ×200. (C) RASGRP2 positive area was quantified in the lining layer of the synovial tissues from patients with RA or OA. Values are the mean±SD. (D) Change of RASGRP2 transcript and protein level in RA-FLS after cytokine stimulation (10 ng/mL, 24 hours). In this experiment, biologics-pretreated RA-FLS at passages 6–8 were utilised. Upper: RASGRP2 mRNA expression levels shown as the mean±SEM (n=3). Lower: immunoblotting using anti-RASGRP2 or anti-β-actin antibody. Results are representative of independent three experiments. Lanes were run on the same gel but were non-contiguous. **P<0.01, t-test.

increment of adhesion onto non-ECM-coating membrane in RASGRP2+FLS also may support this molecular mechanism. We also evaluated the number of FLS passing through the membrane into a lower well-plate using serum as chemoattractants. Passing through the polycarbonate membrane represents cell migration, meanwhile passing through the polycarbonate membrane with ECM coating represents cell invasion requiring ECM proteol-ysis step in addition to migration. RASGRP2+FLS increased the ability of migration consistent with the results of wound healing assay, but not the ability of invasion (figure 2F). These findings indicate that RASGRP2 is involved in adhesive and migrating function of FLS.

We also asked if overexpression of RASGRP2 affects production of RA-related cytokines, growth factors and proteinases from FLS. The expression of IL-6 mRNA was increased in RASGRP2+FLS, whereas none of the expression of RANKL, MMP-1, MMP-3 or VEGF mRNA differ between

RASGRP2+ and control FLS (figure 2G). We confirmed that IL-6 concentrations were increased in the culture supernatant of RASGRP2+FLS by ELISA (figure 2H). These data shows that RASGRP2 is associated with IL-6 production of FLS, but not with RANKL, MMPs and VEGF. Thus, we can understand that RASGRP2 does not increase the invasive ability since RASGRP2 is not associated with MMPs production, main ECM proteolytic enzymes. Those results provided the evidence that overexpres-sion of RASGRP2 is involved in RA phenotype of FLS through increased ability of adhesion, migration and IL-6 production.

rAsGrP2 affects nuclear factor kappa-light-chain-enhancer of activated B cells (nF-κB) pathway and actin dynamics in Fls via rAP-1 activationTo clarify the molecular mechanisms in which RASGRP2 promotes adhesion, migration and IL-6 production of FLS, we

1768 Nakamura H, et al. Ann Rheum Dis 2018;77:1765–1772. doi:10.1136/annrheumdis-2018-213588

Basic and translational research

Figure 2 Function of Ras guanine nucleotide-releasing protein 2 (RASGRP2)-overexpressed fibroblast-like synoviocytes (FLS). (A) Bromodeoxyuridine taken in by proliferative FLS was quantified using ELISA. (B) Wound was made in 12-well plate with confluent FLS using pipette tips. Wound healing through FLS migration was observed sequentially (0 hour, 24 hours and 48 hours) by light microscopy. Original magnification ×40. (C) Wound area was quantified. (D) Light microscope images of polycarbonate membrane with or without extracellular matrix (ECM) coating. The membrane was stained with adhesive FLS. Original magnification ×20. (E) FLS adhesive area was quantified. (F) The number of FLS passing through the polycarbonate membrane with ECM (invasive FLS) or without ECM (migrating FLS) was quantified using fluorimetric assay. (G) Levels of IL-6, RANKL, MMP-1, MMP-3 and VEGF transcript in FLS 48 hours after transfection. (H) Interleukin (IL)-6 concentration in culture supernatant of FLS was measured by ELISA. In these experiments, biologics non-pretreated RA-FLS at passages 6–8 were utilised. All values are the mean±SEM of independent four experiments. Panels B and D are representative data. *P<0.05, **P<0.01, t-test.

assessed evoked changes in the signalling pathway by overexpres-sion of RASGRP2 in FLS. First, we evaluated three well-defined Ras-MAPK pathways: the extracellular signal-regulated kinase (ERK), the p38 and the c-Jun N-terminal kinase. Our immu-noblotting assay revealed that signalling proteins belonging to these three Ras-MAPK pathways were not activated by overex-pression of RASGRP2, at least at their steady state (figure 3A). The results are consistent with the fact that overexpression of RASGRP2 did not affect cell proliferation as shown in figure 2A. Similarly, the transcription of MMPs, RANKL and VEGF is regu-lated by Ras-MAPK pathways,17 thus, unaffected by RASGRP2 (figure 2G). On the other hand, overexpression of RASGRP4

activated the ERK and p38 pathway as expected, because RASGRP4 is one of Ras GEFs (online Supplementary figure S5B).

We next evaluated RAP-1 activation by active RAP-1 pulldown assay. GTP-bound RAP-1, the activated form, was pulled down from cell lysates using the glutathione resin through the specif-ically GTP-binding protein and was then evaluated by immuno-blotting using anti-RAP-1 antibody. We confirmed that RAP-1 was activated by RASGRP2 in FLS (figure 3B). RAP-1 forms a complex with inhibitor κB kinases (IκB) and mediates the activa-tion of NF-κB, a major transcription factor of IL-6.18 We demon-strated that overexpression of RASGRP2 caused degradation

1769Nakamura H, et al. Ann Rheum Dis 2018;77:1765–1772. doi:10.1136/annrheumdis-2018-213588

Basic and translational research

Figure 3 Signalling pathway of RASGRP2-overexpressed FLS. (A and B) FLS lysates 72 hours after transfection were analysed by immunoblotting with the indicated antibodies. GTP-bound RAP-1, the activated form of RAP-1, was pulled down from cell lysates using the glutathione resin through the specifically GTP-binding protein. Results are representative of independent three experiments. (C) Double immunofluorescence detection of β-actin (green) and RASGRP2 (red) on FLS 72 hours after transfection. Original magnification ×400. In these experiment, biologics non-pretreated RA-FLS at passages 6–8 were utilised. Representative data from three independent experiments. ERK, extracellular signal-regulated kinase; FLS, fibroblast like synviocytes; JNK, c-Jun N-terminal kinase; MAPK, mitogen-activated protein kinase; NF-κB, nuclear factor kappa-light-chain-enhancer of activated B cells; RASGRP, Ras guanine nucleotide-releasing proteins.

of the IκB proteins and phosphorylation of the p65 subunit of NF-κB though RAP-1 activation (figure 3B), leading to increased IL-6 production.

RAP-1 contributes to integrin-mediated cellular adhesion and migration,19 and RASGRP2 is an essential activator of RAP-1. The mutation of RASGRP2 gene causes severe defects in the β-subunits of integrins.9 Several members of the β1 integrins, in particular α5/β1 integrin, are highly expressed by FLS and facilitate anchoring of FLS to components in the cartilage ECM.20 These integrins also modulate cell adhesion and migration through regulation of actin dynamics. Although flow cytometry analysis showed that the expression level of β1 integrin subunit did not differ between RASGRP2+ and control FLS (online Supplementary figure S6), it is reasonable because RAP-1 activates integrin by inducing conformational change.21 We evaluated the change of actin dynamics as a result of inte-grin activation by immunofluorescence. Gathered and polymer-ised actin filaments form spindle shape of the cells in control FLS. In contrast, actin remodelling by accelerated polymerisa-tion/depolymerisation cycles of actin filaments was observed in RASGRP2+FLS, including foot process formation participating in the cell adhesion and movement (figure 3C). These findings indicate that RASGRP2 promotes adhesion and migration of FLS through changing actin dynamics.

Intra-articular injection of Rasgrp2-specific sirnA hinders the development of CIA in ratsTo clarify whether RASGRP2 could be a therapeutic target for RA in vivo, we evaluated the potential clinical efficacy of RASGRP2 knockdown using an animal model of RA, CIA in rats. The rats received an intra-articular injection of negative control siRNA or Rasgrp2-specific siRNA into both of the ankle joints. Atelocollagen was concomitantly injected into the ankle joints to be resistant to nucleases and be efficiently transduced into cells, thereby allowing long-term gene silencing.22 Intra-articular injection of Rasgrp2-specific siRNAs significantly improved both

the clinical arthritis score and the diameter of the ankle joints in the arthritic rats compared with the rats injected with control siRNA (figure 4A). Micro-CT revealed that arthritic rats injected with the Rasgrp2-specific siRNAs had significantly lower erosion scores compared with the rats injected with control siRNA (figure 4B,C). The RASGRP2+ area in the synovial lining was significantly decreased in relevant tissues of the rats that received intra-articular injections of Rasgrp2-specific siRNAs compared with those received control siRNA (figure 4D,E). Arthritic rats with Rasgrp2-knockdown had significantly lower pannus scores in the ankle joints relative to arthritic rats with control siRNA (figure 4D,F).

Intra-articular injection of Rasgrp2-specific and Rasgrp4-specific sirnAs in CIA ratsNext, we compared the effect of Rasgrp2 knockdown, Rasgrp4 knockdown and Rasgrp2/4 double knockdown in CIA rats. Silencing of RASGRP4 seemed to have a bigger effect size in improving clinical arthritis score and the diameter of the ankle joints than silencing of RASGRP2 (online Supplemen-tary figure S7A). In contrast, silencing of RASGRP2 and that of RASGRP4 showed a comparable inhibitory effect on bone erosion and pannus formation (online Supplementary figure S7B,C,D), although silencing of either RASGRP2 or RASGRP4 did not completely inhibit bone erosion/pannus formation (online Supplementary figure S7B). Rasgrp2/4 double knock-down showed a synergistic effect in decreasing bone erosion and pannus scores, even half—each dose of siRNA was given in combination (online Supplementary figure S7C,D).

dIsCussIOnIn this study, we identified for the first time that RASGRP2 was ectopically and highly expressed in FLS from a subset of patients with RA. Although RASGRP2 is known as a factor essential for platelet activation, its contribution in inflammatory arthritis has

1770 Nakamura H, et al. Ann Rheum Dis 2018;77:1765–1772. doi:10.1136/annrheumdis-2018-213588

Basic and translational research

Figure 4 Intra-articular injection of Rasgrp2-specific small interfering RNA (siRNA) to rats with collagen-induced arthritis (CIA). (A) Arthritis scores and ankle diameters of ankle joints were monitored for 35 days in the CIA rats. The rats received intra-articular injection of negative control siRNA (n=8), Rasgrp2-specific siRNA (1) s167867 (n=4) or (2) s167865 (n=4), respectively. (B) Representative micro-CT images of the rats’ ankles. Arrows show ankle joints. (C) Erosion scores were determined in control and Rasgrp2-specific siRNAs group (n=8 per group). (D) Representative ankle joints specimens stained with H&E or immunohistochemistry using anti-RASGRP2 antibody in the CIA rats. Arrows show pannus formation sites. Original magnification ×20 (H&E) and ×40 (RASGRP2). (E) RASGRP2+ area in the synovial lining and (F) pannus scores were determined in control and Rasgrp2-specific siRNAs group (n=8 per group). Values are the mean±SEM. *P<0.05, **P<0.01, t-test.

not been evaluated. Therefore, ectopic RASGRP2 expression in inflammatory synovium was an unexpected result. RASGRP2 was induced in RA-FLS, at least in response to PDGF, VEGF and TGF-β stimulation. Supporting our data, PDGF, VEGF and TGF-β are increased in synovial tissue and fluid of patients with RA.23–25 Moreover, PDGF is reported as a growth factor for FLS,23 whereas TGF-β stimulates proliferation of FLS and can induce synovitis in rat joints.24 VEGF is known to induce angio-genesis in synovial tissue and interacts with inflammatory signal-ling to promote and maintain synovitis in RA.25 We propose that these growth factors are involved in the pathogenesis of RA, at least in part, mediated by increasing the expression of RASGRP2 in FLS. Interestingly, enhancement of RASGRP2 expression by the growth factors was not found in OA-FLS. Although it has been unclear how RASGRP2 expression is regulated in FLS, epigenetic status might affect the expression of RASGRP2.

RASGRP2 was originally cloned and named as CalDAG-GEFI by Kawasaki and coworkers from cortex cDNA library as an activator for RAP-1.26 27 More wide expression of RASGRP2 was identified

in platelets, lymphocytes, neutrophils and endothelial cells, as dual guanine exchange factor of Rap and Ras, but preferentially of the former small GTPase.28–31 RASGRP2 activates RAP-1 and subsequently integrin to perform proper adhesive function and inside-out signalling.9 Patients with mutation of RASGRP2 gene and Rasgrp2-−/− mice lack adhesive function of platelets and cause severe bleeding.32 33 RASGRP2 is also essential for lymphocytes and neutrophils to adhere onto stimulated vascular walls and to migrate into inflammatory sites33 34 and for blood vessel formation of endothelial cells in response to VEGF stimulation.35

The present study showed that overexpression of RASGRP2 in FLS activated RAP-1, which resulted in increased ability of adhe-sion and migration through changing actin dynamics. RASGRP2 simultaneously activated NF-κB pathway, resulting in IL-6 production. In contrast, we previously reported that RASGRP4 promotes cell proliferation and MMP-1 production.7 Here we propose that RASGRP2/4 in concert promotes vicious cycle of inflammatory cytokines, tumour-like proliferation and invasive-ness of FLS in patients with RA (figure 5).

1771Nakamura H, et al. Ann Rheum Dis 2018;77:1765–1772. doi:10.1136/annrheumdis-2018-213588

Basic and translational research

Figure 5 Graphical abstract. Expression of Ras guanine nucleotide-releasing proteins (RASGRP2) in fibroblast-like synoviocytes (FLS) is enhanced by transforming growth factor-beta (TGF-β). RASGRP2 in FLS activates RAP-1 and subsequently increases abilities of adhesion onto extracellular matrices and migration by changing actin dynamics. In addition, RASGRP2 affects nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) pathway and increases interleukin (IL)-6 production from FLS. On the other hand, RASGRP4 is increased by tumour necrosis factor-alpha (TNF-α) stimulation and promotes cell proliferation and matrix metalloproteinase -1 (MMP-1) production through Ras, extracellular signal-regulated kinase (ERK) pathway. RASGRP2/4 in concert contributes to the pathophysiology of rheumatoid arthritis.

Silencing of RASGRP2 by intra-articular injection of Rasgrp2-specific siRNAs dampened experimental arthritis in rats by inhibiting pannus formation. The fact that RASGRP2 in FLS is relevant to development of destructive synovitis proposes this RAP-1 GEF as a potential therapeutic target for RA. However, platelets, endothelial cells and other inflammatory cells as well as FLS could also express RASGRP2 in rheumatoid synovium. Intra-articular injection of Rasgrp2-specific siRNAs might affect these cells in addition to FLS. Because contribution of RASGRP2 to the immune system has not been known, dampening of RASGRP2 is expected as a novel strategy in the treatment for RA with minimum immunosuppressive side effects. Of course we have to take into account the possibility of platelet dysfunction when RASGRP2 is systemically abrogated, complete blockade of this signalling protein is rather unlikely by administration of a specific chemical inhibitor. In conclusion, we identified RASGRP2 to be involved in the pathogenesis of RA by promoting adhesion, migration and IL-6 production from FLS, and thus we propose RASGRP2 as a potential novel non-immunosuppressive therapeutic target for RA, in concert with RASGRP4.Acknowledgements the authors acknowledge dr tomohiro onodera and members the lower limb group of orthopedic surgery for providing synovial tissue. We also thank dr Yusuke ohba for technical support and valuable discussion.

Contributors Hn, SS and SY conceptualised the study. Hn, SS, SY and MK developed the methodology. Hn, SS, MK and tS conducted the investigation. MK and nI provided resources. Hn and SY wrote the original draft of the manuscript. YF, MK, Ko, tB, nI and tA reviewed and edited the manuscript. SY and tA acquired funding. MK, YF, tB, nI and tA supervised the study.

Funding this work was supported by grants from the Japan Society for the promotion of Science, Grant-in-Aid for scientific research (KAKEnHI, project number: 15K09514 and 18K0838008), Japan Agency for Medical research and development

(project number: pH42180003) and from Bristol Myers Squibb (Bristol-Myers Squibb rA Clinical Investigation Grant).

Competing interests SY has accepted honoraria from Mitsubishi tanabe pharma Co., Chugai pharmaceutical Co., Astellas pharma and GlaxoSmithKline and grant from Bristol-Myers Squibb. tA has accepted honoraria from Mitsubishi tanabe pharma Co., Chugai pharmaceutical Co., Astellas pharma and pfizer. other authors have declared that no conflict of interest exists.

Patient consent not required.

ethics approval Human and Animal Ethics Committee of the Hokkaido University Hospital (approval number: 008–0103 and 17–0031).

Provenance and peer review not commissioned; externally peer reviewed.

data sharing statement data are available from the corresponding author upon request.

RefeRences 1 Smolen JS, Aletaha d, redlich K. the pathogenesis of rheumatoid arthritis: new

insights from old clinical data? Nat Rev Rheumatol 2012;8:235–43. 2 Bartok B, Firestein GS. Fibroblast-like synoviocytes: key effector cells in rheumatoid

arthritis. Immunol Rev 2010;233:233–55. 3 ogura H, Murakami M, okuyama Y, et al. Interleukin-17 promotes autoimmunity

by triggering a positive-feedback loop via interleukin-6 induction. Immunity 2008;29:628–36.

4 Smolen JS, Aletaha d. rheumatoid arthritis therapy reappraisal: strategies, opportunities and challenges. Nat Rev Rheumatol 2015;11:276–89.

5 thalhamer t, McGrath MA, Harnett MM. MApKs and their relevance to arthritis and inflammation. Rheumatology 2008;47:409–14.

6 Ebinu Jo, Bottorff dA, Chan EY, et al. rasGrp, a ras guanyl nucleotide-releasing protein with calcium- and diacylglycerol-binding motifs. Science 1998;280:1082–6.

7 Kono M, Yasuda S, Stevens rL, et al. ras guanine nucleotide-releasing protein 4 is aberrantly expressed in the fibroblast-like synoviocytes of patients with rheumatoid arthritis and controls their proliferation. Arthritis Rheumatol 2015;67:396–407.

8 Eyre S, Bowes J, diogo d, et al. High-density genetic mapping identifies new susceptibility loci for rheumatoid arthritis. Nat Genet 2012;44:1336–40.

1772 Nakamura H, et al. Ann Rheum Dis 2018;77:1765–1772. doi:10.1136/annrheumdis-2018-213588

Basic and translational research

9 pasvolsky r, Feigelson SW, Kilic SS, et al. A LAd-III syndrome is associated with defective expression of the rap-1 activator CaldAG-GEFI in lymphocytes, neutrophils, and platelets. J Exp Med 2007;204:1571–82.

10 Aletaha d, neogi t, Silman AJ, et al. rheumatoid arthritis classification criteria: an American College of rheumatology/European League Against rheumatism collaborative initiative. Arthritis Rheum 2010;2010:2569–81.

11 Earp JC, dubois dC, Molano dS, et al. Modeling corticosteroid effects in a rat model of rheumatoid arthritis I: mechanistic disease progression model for the time course of collagen-induced arthritis in Lewis rats. J Pharmacol Exp Ther 2008;326:532–45.

12 Brahn E, Banquerigo ML, Firestein GS, et al. Collagen induced arthritis: reversal by mercaptoethylguanidine, a novel antiinflammatory agent with a combined mechanism of action. J Rheumatol 1998;25:1785–93.

13 pine pr, Chang B, Schoettler n, et al. Inflammation and bone erosion are suppressed in models of rheumatoid arthritis following treatment with a novel Syk inhibitor. Clin Immunol 2007;124:244–57.

14 Wruck CJ, Fragoulis A, Gurzynski A, et al. role of oxidative stress in rheumatoid arthritis: insights from the nrf2-knockout mice. Ann Rheum Dis 2011;70:844–50.

15 Lee dM, Kiener Hp, Agarwal SK, et al. Cadherin-11 in synovial lining formation and pathology in arthritis. Science 2007;315:1006–10.

16 rinaldi n, Schwarz-Eywill M, Weis d, et al. Increased expression of integrins on fibroblast-like synoviocytes from rheumatoid arthritis in vitro correlates with enhanced binding to extracellular matrix proteins. Ann Rheum Dis 1997;56:45–51.

17 paunovic V, Harnett MM. Mitogen-activated protein kinases as therapeutic targets for rheumatoid arthritis. Drugs 2013;73:101–15.

18 teo H, Ghosh S, Luesch H, et al. telomere-independent rap1 is an IKK adaptor and regulates nF-kappaB-dependent gene expression. Nat Cell Biol 2010;12:758–67.

19 priego n, Arechederra M, Sequera C, et al. C3G knock-down enhances migration and invasion by increasing rap1-mediated p38α activation, while it impairs tumor growth through p38α-independent mechanisms. Oncotarget 2016;7:45060–78.

20 Kitagawa A, Miura Y, Saura r, et al. Anchorage on fibronectin via VLA-5 (alpha5beta1 integrin) protects rheumatoid synovial cells from Fas-induced apoptosis. Ann Rheum Dis 2006;65:721–7.

21 Shattil SJ, Kim C, Ginsberg MH. the final steps of integrin activation: the end game. Nat Rev Mol Cell Biol 2010;11:288–300.

22 Minakuchi Y, takeshita F, Kosaka n, et al. Atelocollagen-mediated synthetic small interfering rnA delivery for effective gene silencing in vitro and in vivo. Nucleic Acids Res 2004;32:e109.

23 Charbonneau M, Lavoie rr, Lauzier A, et al. platelet-derived growth factor receptor activation promotes the prodestructive invadosome-forming phenotype of synoviocytes from patients with rheumatoid arthritis. J Immunol 2016;196:3264–75.

24 Higgs r. rheumatoid arthritis: Synergistic effects of growth factors drive an rA phenotype in fibroblast-like synoviocytes. Nat Rev Rheumatol 2010;6:383.

25 Konisti S, Kiriakidis S, paleolog EM. Hypoxia–a key regulator of angiogenesis and inflammation in rheumatoid arthritis. Nat Rev Rheumatol 2012;8:153–62.

26 Kawasaki H, Springett GM, toki S, et al. A rap guanine nucleotide exchange factor enriched highly in the basal ganglia. Proc Natl Acad Sci U S A 1998;95:13278–83.

27 toki S, Kawasaki H, tashiro n, et al. Guanine nucleotide exchange factors CaldAG-GEFI and CaldAG-GEFII are colocalized in striatal projection neurons. J Comp Neurol 2001;437:398–407.

28 Clyde-Smith J, Silins G, Gartside M, et al. Characterization of rasGrp2, a plasma membrane-targeted, dual specificity ras/rap exchange factor. J Biol Chem 2000;275:32260–7.

29 ohba Y, Mochizuki n, Matsuo K, et al. rap2 as a slowly responding molecular switch in the rap1 signaling cascade. Mol Cell Biol 2000;20:6074–83.

30 ohba Y, Ikuta K, ogura A, et al. requirement for C3G-dependent rap1 activation for cell adhesion and embryogenesis. Embo J 2001;20:3333–41.

31 ohba Y, Mochizuki n, Yamashita S, et al. regulatory proteins of r-ras, tC21/r-ras2, and M-ras/r-ras3. J Biol Chem 2000;275:20020–6.

32 Canault M, Ghalloussi d, Grosdidier C, et al. Human CaldAG-GEFI gene (rASGrp2) mutation affects platelet function and causes severe bleeding. J Exp Med 2014;211:1349–62.

33 Bergmeier W, Goerge t, Wang HW, et al. Mice lacking the signaling molecule CaldAG-GEFI represent a model for leukocyte adhesion deficiency type III. J Clin Invest 2007;117:1699–707.

34 Ghandour H, Cullere x, Alvarez A, et al. Essential role for rap1 Gtpase and its guanine exchange factor CaldAG-GEFI in LFA-1 but not VLA-4 integrin mediated human t-cell adhesion. Blood 2007;110:3682–90.

35 Suzuki K, takahashi S, Haramoto Y, et al. xrASGrp2 is essential for blood vessel formation during xenopus development. Int J Dev Biol 2010;54:609–15.

1773Rivellese F, et al. Ann Rheum Dis 2018;77:1773–1781. doi:10.1136/annrheumdis-2018-213418

Basic and translational research

ExtEndEd rEport

Mast cells in early rheumatoid arthritis associate with disease severity and support B cell autoantibody productionFelice rivellese,1 daniele Mauro,1 Alessandra nerviani,1 Sara pagani,1 Liliane Fossati-Jimack,1 tobias Messemaker,2 Fina A S Kurreeman,2 rené E M toes,2 Andreas ramming,3 Simon rauber,3 Georg Schett,3 Gareth W Jones,4 Simon A Jones,4 Francesca Wanda rossi,5 Amato de paulis,5 Gianni Marone,5,6 Mohey Eldin M El Shikh,1 Frances Humby,1 Costantino pitzalis1

To cite: rivellese F, Mauro d, nerviani A, et al. Ann Rheum Dis 2018;77:1773–1781.

Handling editor Josef S Smolen

► Additional material is published online only. to view please visit the journal online (http:// dx. doi. org/ 10. 1136/ annrheumdis- 2018- 213418).

For numbered affiliations see end of article.

Correspondence toprofessor Costantino pitzalis, Centre for Experimental Medicine & rheumatology, William Harvey research Institute, Barts and the London School of Medicine & dentistry, Queen Mary University of London, London EC1M 6BQ, UK; c. pitzalis@ qmul. ac. uk

received 14 March 2018revised 26 July 2018Accepted 27 July 2018published online First 20 August 2018

© Author(s) (or their employer(s)) 2018. no commercial re-use. See rights and permissions. published by BMJ.

ABsTrACTObjectives Mast cells (MCs) are involved in the pathogenesis of rheumatoid arthritis (rA). However, their contribution remains controversial. to establish their role in rA, we analysed their presence in the synovium of treatment-naïve patients with early rA and their association and functional relationship with histological features of synovitis.Methods Synovial tissue was obtained by ultrasound-guided biopsy from treatment-naïve patients with early rA (n=99). Immune cells (Cd3/Cd20/Cd138/Cd68) and their relationship with Cd117+MCs in synovial tissue were analysed by immunohistochemistry (IHC) and immunofluorescence (IF). the functional involvement of MCs in ectopic lymphoid structures (ELS) was investigated in vitro, by coculturing MCs with naïve B cells and anticitrullinated protein antibodies (ACpA)-producing B cell clones, and in vivo in interleukin-27 receptor alpha (IL27ra)-deficient and control mice during antigen-induced arthritis (AIA).results High synovial MC counts are associated with local and systemic inflammation, autoantibody positivity and high disease activity. IHC/IF showed that MCs reside at the outer border of lymphoid aggregates. Furthermore, human MCs promote the activation and differentiation of naïve B cells and induce the production of ACpA, mainly via contact-dependent interactions. In AIA, synovial MC numbers increase in IL27ra deficient mice, in association with ELS and worse disease activity.Conclusions Synovial MCs identify early rA patients with a severe clinical form of synovitis characterised by the presence of ELS.

InTrOduCTIOnMast cells (MCs) are tissue-resident cells of the innate immunity, involved in several physiological and pathological processes, including infections, cancer and chronic inflammatory diseases.1 2 They are present in the synovial membrane (SM) and have been implicated in contributing to the inflammatory response in several rheumatic diseases,3 including rheumatoid arthritis (RA).4 Notably, MCs are present in healthy synovia,5 but their number significantly increases accom-panying the cellular hyperplasia characteristic of RA synovitis.6–8 Many MC mediators have direct

proarthritogenic effects,9–12 and MCs can be acti-vated by several stimuli present in the synovium/synovial fluid, such as anticitrullinated protein antibodies (ACPA) IgG immune complexes.13 However, recent evidence suggests that MC contri-bution to autoimmune diseases can be complex and multifaceted.14 In the context of RA, for example, human MCs have been shown to exert immunomodulatory functions in vitro.15In vivo, initial findings were contrasting,16–18 most likely because of the use of animal models in which MC depletion was accompanied by anomalies of other immune cells.19 In recent years, thanks to the devel-opment of new specific models of MC depletion,20 their contribution has been confirmed to be essen-tial in collagen-induced arthritis (CIA) but redun-dant in serum-transfer arthritis.21 Additionally, their depletion in the preclinical phases of CIA, rather than in the established phases, was shown to influence the disease outcome.22 These results in vivo suggest that the contribution of MCs to RA may be different at various disease stages, that is, essential during the early phases (assessed by CIA) but somehow dispensable during the late effector phases (serum transfer). However, while these models are self-resolving, in RA, there is chronic inflammation with a perpetuation of the aber-rant autoimmune response; therefore, the results cannot be easily translated to the clinical setting. Overall, despite the substantial amount of data, the role of MCs in RA remains to be clarified.4 As RA is well-recognised as a heterogeneous syndrome in terms of genetic predisposition, pathogenesis, clinical23 24 and histological25 features, it could be hypothesised that MC presence and functions in the synovium may be different in various disease subsets. To explore this hypothesis, we systemati-cally analysed the presence of MCs in the synovia of a large cohort of disease-modifying antirheu-matic drugs (DMARD)-naïve patients with early RA. Furthermore, we assessed their interactions with immune cells at synovial level and analysed in vitro the crosstalk of MCs with B cells. Finally, we evaluated the relationship between MC synovial infiltration and ectopic lymphoid structures (ELS) in an experimental model of synovial ectopic lymphoid neogenesis.

1774 Rivellese F, et al. Ann Rheum Dis 2018;77:1773–1781. doi:10.1136/annrheumdis-2018-213418

Basic and translational research

Table 1 Summary of patient characteristics (n=99)

Age (years), mean (SD), range 52 (16) 19–89

Sex (female), % 70.7%

Disease duration (Months), mean (SD), range 6 (3) 1–12

ACPA+, % 75.8%

RF+, % 73.7%

ESR (mm/hour), mean (SD), range 38 (30) 2–120

CRP (mg/L), mean (SD), range 17 (25) 0–162

DAS-28, mean (SD), range 5.62 (1.41) 1.88–8.92

TJC, mean (SD), range 11.33 (7.14) 1–28

SJC, mean (SD), range 7.33 (5.88) 1–26

VAS (PGA), mean (SD), range 66.25 (24.57) 0–100

HAQ, mean (SD), range 1.51 (0.79) 0–3

ACPA, anticitrullinated protein antibodies; CRP, C reactive protein; DAS-28, disease activity score 28 joints; ESR, erythrocyte sedimentation rate; HAQ, health assessment questionnaire; PGA, Patient Global disease Activity; RF, rheumatoid factor; SJC, swollen joint count; TJC, tender joint count; VAS visual analogue scale.

MeTHOdsPatient samples and ultrasound-guided synovial biopsySynovial tissue was obtained by ultrasound-guided synovial biopsy from DMARD-naïve patients with early (<12 months) RA (n=99), enrolled in the Pathobiology of Early Arthritis Cohort of the Centre for Experimental Medicine and Rheumatology of Queen Mary University (London), as previously described.26 All patients fulfilled the 2010 European League Against Rheuma-tism criteria for RA.27

Histological analyses of synovial samplesSynovial sections underwent standard H&E staining and semiquantitative (SQ) assessment of synovitis according to a previously validated score (Krenn).28 Sequentially cut sections underwent immunohistochemical (IHC) staining and SQ assess-ment (0–4) for immune cells, as previously reported29 and auto-mated image analysis and counting for CD117-positive MCs. Patients were classified into high, medium and low MC groups (>66th, 33rd–66th, <33rd percentiles, respectively) (online supplementary methods).

Peripheral blood-derived MCsCD34+ haematopoietic stem cells (StemPro CD34+ kit, ThermoFisher) were differentiated into MCs as previously described30 (online supplementary methods and online supple-mentary figure 1 for MC purity).

naïve B cell isolation and coculture with MCsIgD+ B cells isolated by immunomagnetic sorting (Miltenyi) from tonsil mononuclear cells were cultured for 7 days alone or together with MCs, in the presence of TLR-9 ligand CpG oligodeoxynucleotides (ODN-2006, Invitrogen), in contact or separated by a Transwell membrane. In parallel experiments, B cells were marked with CFSE (Biolegend) to measure prolifer-ation. Where indicated, mouse antihuman CD154 (CD40L) or isotype control (Biolegend) were added at a concentration of 0.1–10 µg/mL (online supplementary methods).

Visualisation of MC–B cell interactionPlease see online supplementary methods.

ACPA B cell clone and coculture with MCsACPA-producing immortalised B cells (2×105) were obtained as described (Germar et al. Manuscript submitted and ref 31) and cultured as indicated in online supplementary methods.

Flow cytometryFlow cytometry staining was performed as previously described32 (online supplementary methods).

Measurement of immunoglobulinsPlease check online supplementary methods.

Antigen-induced arthritis (AIA) in wild-type and interleukin-27 receptor (IL27ra) knock outAIA was induced in adult (8–12 weeks) IL27ra−/− mice and age-matched/sex-matched WT as previously described.33 Five-micrometre sections of synovia from animals culled at different intervals (days 3, 10 and 35) were processed and anal-ysed as described.33 To visualise MCs, sequentially cut sections were stained with acidic Toluidine Blue (Sigma) 0.1% solution (pH 2.0~2.5) (online supplementary methods).

statistical analysesMeasures of central tendency and dispersions and statistical analyses are indicated in each figure legend and in online supple-mentary methods. P values of <0.05 were considered statisti-cally significant.

resuLTsMCs strongly associate with defined histological features of synovitis and markers of disease activity in patients with early rATo evaluate the association of MCs with different clinical and histological phenotypes of RA in an unbiased setting, we studied their presence in the SMs of patients with early RA, naïve to treatment with DMARDs. Table 1 summarises the demographic features of the patient cohort, which are as expected for a popu-lation with early untreated RA, that is, active disease, with a mean disease activity score (DAS)-28 of 5.62, high inflamma-tory markers, with a mean erythrocyte sedimentation rate (ESR) of 38 mm/hour and approximately 70% autoantibody positive – rheumatoid factor (RF) or ACPA.

First, we assessed the correlation of MC density with markers of both local (synovial) and systemic inflammation and disease activity. Figure 1A demonstrates that MC counts significantly correlate with inflammatory markers (ESR and C reactive protein (CRP)), disease activity (DAS-28) and synovial inflam-mation (Krenn Score). Interestingly, MCs correlated with other immune cells in SMs, with particularly high correlation indexes for B cell and T cells (Spearman r 0.617 and 0.519, respectively, p<0.001). Since MC infiltration in synovial specimens was heterogenous, we stratified patients according to the number of MCs into three groups (low, medium and high MC counts), as shown in figure 1B. Consistent with the strong correlation shown in figure 1A, high T cell and B cell scores were predominant in patients with high MC counts (figure 1C,D). As these data indicate that MCs correlate with synovial inflammation and the degree of lymphocyte infiltration, we evaluated the presence of MCs in three classified forms of synovitis (pathotypes): lymphoid, myeloid and pauci-immune/fibroid.25 Interestingly, more than 80% of patients with high MC synovial counts displayed a lymphoid-rich patho-type, characterised by synovial T cell and B cell aggregates; vice versa, as expected, the number of MCs was significantly higher in patients with a lymphoid pathotype (figure 1E,F). These data indi-cate that MCs are strongly associated with lymphoid aggregates in the synovia of patients with early RA. Next, we investigated

1775Rivellese F, et al. Ann Rheum Dis 2018;77:1773–1781. doi:10.1136/annrheumdis-2018-213418

Basic and translational research

Figure 1 Association of high MC counts with synovial inflammation, disease severity and lymphoid aggregates. (A) Correlations of MC numbers with inflammatory markers, disease activity and histological scores. Line at p=0.05. (B) Patients stratified according to MC numbers into low (<33rd percentile) medium (33rd–66th percentile) and high (>66th percentile) groups. (C and D) Distribution of CD3 (C) and CD20 (D) scores in patients stratified according to MC numbers as in panel B. (E) Distribution of pathotypes in patients stratified according to MC numbers. (F) MC density in patients stratified according to pathotypes. n=99. *p=0.05, Spearman correlation in panel A, χ2 in panels C–E, one-way analysis of variance with Bonferroni post hoc test in panel F. CRP, C reactive protein; DAS-28, disease activity score 28 joints; ESR, erythrocyte sedimentation rate; HAQ, health assessment questionnaire; MCs, mast cells; SJ, swollen joint count; TJ, tender joint count.

the heterogeneity of synovial MCs. Two types of MCs have been described in humans, expressing tryptase alone (MC_T) or tryptase and chymase (MC_TC), with the following distribution in synovium: predominance of MC_TC in normal synovium,5 expan-sion of both in RA,7 with relative increase of MC_T described in early8 and late RA.34 By performing double immunofluorescence in a subgroup of patients from our early RA cohort (n=15), we

found both types of MCs expressed in the synovia, with signifi-cantly higher levels of MC_TC (online supplementary figure S2A) and an average ratio MC_T:MC_TC of 1:3. When patients were stratified according to pathotypes, we observed a signif-icant increase of both types of MCs in the lymphoid pathotype (online supplementary figure S2B), with the ratio MC_T:MC_TC shifting from 1:6 (fibroid) to 1:2 (lymphoid). Additionally, MC_T,

1776 Rivellese F, et al. Ann Rheum Dis 2018;77:1773–1781. doi:10.1136/annrheumdis-2018-213418

Basic and translational research

Table 2 Clinical phenotype of patients stratified according to MC numbers.

MCs

Low(n=32)

Medium(n=33)

High (n=34)

Age, mean (SD)

52 (16) 50 (15) 54 (17) 52 (15) 0.300

Female, % 70.7 68.8 71.9 72.7 0.704

ESR, mean (SD)

38 (30) 27 (28) 42 (30) 45 (30) 0.033

CRP, mean (SD)

17 (25) 10 (29) 21 (23) 18 (24) 0.174

RF+, % 73.7 68.8 60.6 91.2 0.031

ACPA+, % 75.8 78.1 60.6 88.2 0.009

DAS-28, mean (SD)

5.65 (1.41) 4.97 (1.54) 6.05 (1.22) 5.91 (1.24) 0.003

Fisher’s exact test or one - way  analysis of variance (ANOVA) as appropriate.ACPA, anticitrullinated protein antibodies; CRP, C reactive protein; DAS-28, disease activity score 28 joints; ESR, erythrocyte sedimentation rate; MCs, mast cells; RF, rheumatoid factor.

and not MC_TC, showed a significant correlation with synovial inflammation (online supplementary figure S2C and D). Online supplementary figure S2E shows a representative image with an abundance of MC_T in a patient with lymphoid pathotype. These data suggest an enrichment of tryptase expressing synovial MCs (MC_T) in the lymphoid pathotype in association with the degree of inflammation.

Finally, we assessed the clinical phenotype of patients strat-ified according to MCs. As shown in table 2, patients with medium and high MC counts have significantly raised ESR and disease activity (DAS-28) compared with low MC counts, and patients with high MCs have a significantly higher prev-alence of autoantibody positivity (ACPA and RF) compared with low and medium MCs. To exclude that the association of MCs with disease severity was exclusively driven by their association with lymphoid cells, we performed additional analyses excluding lymphoid patients and found that MCs were significantly correlated with ESR (Spearman r=0.272, p=0.007), CRP (r=0.217, p=0.033) and DAS-28 (r=0.308, p=0.002). Overall, this suggests that the stratification of patients according to synovial MCs identifies patients with a severe clinical phenotype.

MCs interact with T and B cells in follicular structures in rA synovium and tonsil tissueHaving established the presence of MCs in the synovial biop-sies from early RA patients with lymphoid-rich sysnovitis, we next investigated the distribution of MCs in the synovia of patients with a lymphoid pathotype. A representative example of the synovial histology of these patients is shown in figure 2A. IHC staining of sequential sections confirmed the presence of CD117+ synovial MCs (figure 2B). By immuno-fluorescence, we identified MCs bordering lymphoid aggre-gates, in close contact with B and T cells (figure 2C). A similar distribution was observed in the highly organised secondary lymphoid organs (SLOs) from tonsil tissue, used as controls (figure 2D).

Thus, MCs reside on the outer boundary of B and T cell aggre-gates and are a histological feature of both synovial ELS and SLOs.

MCs enhance B cell survival, proliferation and differentiation and production of class-switched Ig and ACPAAs MCs were found in the proximity of B and T cell aggregates in SMs, and because the activation of B cells towards the production of autoantibodies locally contributes to the pathogenic process in RA,29 we hypothesised that human MCs could influence the activation of B cells. To test this hypothesis, we cultured naïve B cells isolated from tonsils with in vitro differentiated human MCs, using the TLR9 ligand CpG to boost B cell activation.35 MCs enhanced the survival of naïve B cells (figure 3A) with a significant increase in IgG secretion but only minor changes in IgM production (figure 3B). Since CpG per se lacks the ability to induce a full differentiation of naïve B cells,36 37 the production of IgG on coculture of MCs with naïve B cells suggest that MCs can provide additional signals allowing B cell differentiation and the isotype switch towards IgG (figure 3B). Interestingly, cell contact was not necessary to induce the MC-mediated effect on B cell survival, indicating that soluble factors were sufficient (figure 3C). Additionally, MCs were enhancing the prolifera-tion of CFSE-labelled naïve B cells, and this effect was again not dependent on cell contact (figure 3D). On the contrary, the production of IgG was significantly dependent on cell contact, suggesting that membrane-bound factors were responsible (figure 3E). Similarly, MCs were able to enhance the production of RA-specific autoantibodies (ACPA) by B cells (figure 3F), an effect again dependent on cell contact. Collectively, these data indicate that MCs can induce the survival, proliferation and differentiation of naïve B cells towards IgG-secreting B cells via indirect and direct cell–cell contact.

Next, we investigated the mechanisms by which human MCs promote IgG production by B cells and demonstrated that this is CD40L dependent (figure 3G), confirming previous reports indicating that murine MCs mediate B cell activation through this cell surface costimulatory molecule.38 39

To further confirm the ability of MCs to induce the differ-entiation of naïve B cells towards antibody-producing memory B cells, we analysed B cells by flow cytometry after 7 days of coculture with MCs. Figure 3H shows that coculture of naïve B cell with MCs increased the number of antibody-producing memory B cells (CD27+/CD38+). Interestingly, this effect could be inhibited by treatment with anti-CD40L in a dose-de-pendent manner, further confirming the ability of MCs to induce of B cell differentiation via CD40L-CD40 interaction (figure 3H).

As cell contact was crucial for the MC-induced differentiation of B cells, we performed phalloidin staining on MCs and B cells after 24 hours of coculture, which showed actin reorganisation in the region of contact between MCs and B cells (figure 3I). This suggests an active cellular interaction between MCs and B cells.

synovial MC infiltration occurs early and is associated with eLs and disease severity in AIA in IL27ra-deficient miceHaving demonstrated that human MCs enhance B cell activation and differentiation in vitro and are associated with synovial ELS in patients with early RA, we wished to investigate in vivo the relevance of the interaction between MCs and B cells within ELS in the pathogenesis of arthritis. To this end, we examined MCs synovial infiltration in AIA, a model in which acute inflamma-tory arthritis is induced by intra-articular injection of methylated bovine serum albumin following systemic immunisation with the same antigen. We used IL27ra-deficient mice, which develop exacerbated synovitis comprising ELS.33

1777Rivellese F, et al. Ann Rheum Dis 2018;77:1773–1781. doi:10.1136/annrheumdis-2018-213418

Basic and translational research

Figure 2 MCs border synovial lymphoid aggregates (A) IHC staining for CD3, CD20, CD138 and CD68 in an ultrasound-guided biopsy of a patient with early RA classified as lymphoid pathotype. Semiquantitative (SQ) scores for each marker are indicated. (B) IHC staining for CD117 (c-kit) showing synovial MCs. High magnification, on the right, shows synovial MCs (arrows) close to cellular aggregates. (C) Immunofluorescence staining of the RA synovia showing the interactions between CD20+ B cells (in green) and CD3+ T cells (in red), forming an ectopic lymphoid aggregate, and CD117+ MCs (in clear blue). The high magnification shows MCs surrounded by B and T cells at the edge of one aggregate. (D) Immunofluorescence of a human tonsil showing MCs (clear blue) at the edge of a germinal centre formed by aggregates of B (green) and T cells (red). DAPI, nuclei in dark blue.

Figure 4A,B shows that MCs were present already in the early phases of AIA (3 days after intra-articular injection), with their numbers further increasing at day 10 (d10) and day 35 (d35) postarthritis induction. Importantly, in IL27ra-deficient mice, synovial MC infiltration was significantly higher compared with wild-type littermates at d10 and d35 (figure 4C,D). Moreover,

at the peak of inflammation (d10) synovial MCs showed a posi-tive correlation with arthritis index, synovial infiltrate and the presence of erosions (figure 4E). Finally, in IL27ra-deficient animals at d10, there was a trend towards a correlation between MCs and the area of lymphoid aggregates (figure 4F), and MCs were found at the borders of lymphoid aggregates (figure 4G),

1778 Rivellese F, et al. Ann Rheum Dis 2018;77:1773–1781. doi:10.1136/annrheumdis-2018-213418

Basic and translational research

Figure 3 MCs support survival, proliferation and differentiation of naïve B cells into antibody secreting cells and induce the production of ACPA autoantibodies. (A) Proportion of live B cells, measured by flow cytometry, after 7 days of culture without (o) or with (■) MCs, at a ratio of 1:6, together with CpG ODN at 1 µg/mL, n=4 (B) IgG and IgM measured by ELISA in the supernatants of naïve B cells harvested after 7 days of culture, n=13 for IgG and n=7 for IgM. (C) Proportion of live cells, when naïve B cells were cultured alone or cocultured with MCs in contact or in transwell, n=3. (D) Proliferation measured by CFSE staining and flow cytometry after 7 days of culture. Representative histograms on the left and cumulative data with division index on the right, n=3. (E) IgG production, naïve B cells cultured alone or cocultured with MCs in contact or in transwell, n=3. (F) ACPA measured by CCP2 ELISA on coculture ACPA producing B cell clone with MCs in contact or transwell, n=3. (G) IgG production on inhibition of CD40L in the coculture of MCs and B cells. n=3 (H) Proportion of antibody-producing cells (CD27+ and CD38+), after gating on live/CD117−/CD19+. Representative histograms from three independent experiments. (I) IF of MCs (light blue) and B cells (red) after 24 hours of coculture and CpG triggering. Actin reorganisation is shown in green (phalloidin). Nuclei in blue (DAPI). Measure bar 5 µm. Representative image of three independent experiments. *P<0.05, Mann-Whitney (A and B), one-way analysis of variance with Bonferroni post hoc (C,D, E, F and G). n=number of independent experiments with n MC donors and n B cell donors. ACPA, anticitrullinated protein antibodies; CFSE, carboxyfluorescein succinimidyl ester; CpG ODN, CpG oligodeoxynucleotides; DAPI, 4′,6-diamidino-2-phenylindole;  IF, immunofluorescence; MCs, mast cells.

1779Rivellese F, et al. Ann Rheum Dis 2018;77:1773–1781. doi:10.1136/annrheumdis-2018-213418

Basic and translational research

Figure 4 Mast cells (MCs) associate with ELS and disease severity in antigen-induced arthritis in IL27ra deficient mice. (A and B) Representative images (A) of toluidine blue staining showing metachromatic MCs (arrows) at different time-points during antigen-induced arthritis, with results summarised in panel B; n=>6 for each time-point. (C) Comparison of MC infiltration in WT and IL27ra KO mice and (D) representative images of toluidine blue staining with metachromatic MCs (red arrows) at day 35 time-point. N=>6/ time-point/group. (E) Correlation of synovial MC numbers with arthritis index, synovial infiltrate and erosions at day 10. (F) Correlation of MC number with ELS area in IL27ra KO mice at day 10 and (G) representative image. Size bars 100 µm, unless specified. *P<0.05, one-way analysis of variance with Bonferroni in B, Mann-Whitney comparing WT and IL27ra KO  at each timepoint in panel C, Spearman correlation in panels E and F. ELS, ectopic lymphoid structures; IL27ra, interleukin-27 receptor alpha;KO, knock out; MCs, mast cells; WT, wild type.

reflecting the observations in RA patients with a lymphoid pathotype.

Overall, these observations confirm the association of MCs with ectopic lymphoid neogenesis and disease severity, during arthritis induction in vivo.

dIsCussIOnIn this manuscript, we present the first systematic anal-ysis of MCs in the SM of patients with early RA obtained by ultrasound-guided synovial biopsy prior to therapeutic intervention. Our data demonstrate that synovial MCs are strongly

1780 Rivellese F, et al. Ann Rheum Dis 2018;77:1773–1781. doi:10.1136/annrheumdis-2018-213418

Basic and translational research

associated with specific pathobiological and clinical phenotypes potentially linked to their ability to induce the activation/differ-entiation of B cells and the production of ACPA autoantibodies.

Although MCs have long known to be part of the inflamma-tory infiltrate in RA, their presence in the SM has only been anal-ysed in a few studies, which described increased MC numbers and mediators in the SM and synovial fluid of patients with RA, possibly in correlation with disease activity.6–8 More recently, in a study describing the immunopathological characteristics of ultra-sound-defined synovitis in patients with RA in remission, the pres-ence of synovial MCs and B cells at baseline was associated with disease reactivation at follow-up.40 At the same time, studies in vitro and in vivo have yielded contradictory results, so that their exact contribution is still unclear.4

To obtain further insight into the relevance of MCs in RA, we analysed the SM of a large cohort of patients with early (<12 months) RA (n=99), unbiased by treatment and disease duration.

Our data demonstrate high synovial MC counts in patients with a severe clinical phenotype at baseline. Furthermore, MCs correlate with synovial inflammation and, in particular, with ELS, which are found in approximately 40% of patients with RA and have been associated with disease severity, T cell priming and auto-antibody production, including the local ongoing production of class-switched autoantibodies, such as ACPA.29 41 Interestingly, we found a higher prevalence of tryptase-expressing MCs (MC_T) in the synovia of patients with ELS. Although the concept of MC heterogeneity is still ill-defined,42 the abundance of MC_T in asso-ciation with the lymphoid pathotype would be in line with the proinflammatory role of this subpopulation described in asthma.43 The strong association of MCs with ELS led us to hypothesise that MCs could modulate the local adaptive immune response. The ability of human MCs to influence T cells interaction has already been shown.44 45 Murine MCs have been shown to activate B cells and promote their differentiation toward effector cells.38 46 Interestingly, murine MCs were also shown to control, via CD40/CD40L interaction, the expansion and differentiation of inter-leukin-10 competent B cells, which is in line with their immuno-modulatory functions.39 The interaction of human MCs and B cells has been poorly studied: in the context of allergic responses, MCs have been shown to express CD40L and induce IgE production by B cells,47 48 while in the context of cancer, they have been shown to activate lymphoplasmacytic cells via CD40L.49 Nonetheless, the interaction between human MCs and B cells has never been studied in the context of autoimmune diseases characterised by local B cell responses.

Here, we show that MCs are at the border of B and T cell aggre-gates in the synovia of patients with RA, similarly to the distribu-tion observed in SLOs. Furthermore, we show that human MCs support the survival, activation, proliferation and differentiation of naive B cells into IgG-producing B cells, in line with recent results in mice.46 This effect is dependent on cell contact and, specifically, on CD40L–CD40 interaction. Importantly, we demonstrate that MCs induce ACPA production by B cells, also in a contact-de-pendent manner, although the relevance of CD40L expression by MCs in the induction of ACPA remains to be confirmed. Furthermore, we provide the first evidence of an active interac-tion between human MCs and B cells, as previously shown for MC–T cells50 and MC–Dendritic Cells.51In vivo, we used AIA to assess the time course and magnitude of synovial MC infiltration, and IL27ra-deficient mice as a model of exacerbated synovitis accompanied by ELS formation,33 thus resembling RA patients with a lymphoid pathotype. In both wild-type and IL27ra deficient animals, we observed an early synovial MC infiltration, before the formation of ELS, further increasing at later stages. Whether the

early synovial infiltration of MCs means that they are essential to lymphoid neogenesis remains to be established. In fact, in addition to the direct modulation of lymphocyte activation, MCs are also known to produce several mediators,52 including angiogenic and lymphangiogenic factors,53 54 that could support the organisation of ELS.55 Interestingly, MC numbers were significantly higher in IL27ra-deficient animals, in association with synovial ELS and a worse disease outcome. As IL27ra deficiency has been shown to enhance the activation of MCs in the context of Th2 responses,56 57 it will be of interest to further assess the specific role of IL27-medi-ated MC inhibition in autoimmune diseases.

In conclusion, our study points to the relevance of MCs in RA and their role as novel markers of synovial inflammation. In fact, their presence contributes to the definition of an MC-rich highly inflamed synovial pathotype and helps identifying patients with a severe clinical phenotype.

Author affiliations1Centre for Experimental Medicine and rheumatology, William Harvey research Institute, Barts and the London School of Medicine and dentistry, Queen Mary University of London, London, UK2department of rheumatology, Leiden University Medical Center, Leiden, the netherlands3department of Internal Medicine 3–rheumatology and Immunology, Friedrich-Alexander-University Erlangen-nürnberg (FAU) and Universitätsklinikum Erlangen, Erlangen, Germany4division of Infection and Immunity, School of Medicine, Cardiff University, Cardiff, UK5department of translational Medical Sciences and Center for Basic and Clinical Immunology research (CISI), World Allergy organization (WAo) Center of Excellence, University of naples Federico II, naples, Italy6Institute of Experimental Endocrinology and oncology ’Gateano Salvatore’ (IEoS), national research Council (Cnr), naples, Italy

Acknowledgements We would like to thank drs d Baeten, H Spits and H U Scherer for providing the immortalised ACpA-producing B cell lines.

Contributors Fr: study design, experiments, data acquisition, data analysis, manuscript preparation and revision; dM: study design, experiments, data acquisition and data analysis; LF-J and GWJ: experiments, data acquisition and data analysis (animal data); Sp and tM: experiments and data acquisition; An and FH: data acquisition and data analysis (clinical data); Ar, Sr, FASK, rEMt, GS, SAJ , FWr, Adp, GM and MEMES: interpretation of experimental results and manuscript revision; Cp: study design, interpretation of experimental results, manuscript preparation and revision. Fr wrote the manuscript, and all authors critically revised its final preparation and approved its submission.

Funding the research leading to these results has received funding from the people programme (Marie Curie Actions) of the European Union’s Seventh Framework programme (Fp7/2007 2013) under rEA grant agreement n° 608765. this work was supported in part by grants from the Ministero dell’Istruzione, Università e ricerca (MIUr) and regione Campania CISILab project, CrÈME project, and tIMInG project. the pathobiology of Early Arthritis Cohort (pEAC) was funded by the MrC grant 36661. Additional funding from MrC funded Maximising therapeutic Utility for rheumatoid Arthritis using genetic and genomic tissue responses to stratify medicines (MAtUrA) Grant ref: Mr/K015346/1 and ArUK funded Experimental Arthritis treatment Centre (EAtC) Grant ref: 20022. the animal work was supported by Arthritis research UK Grants reference 20305 and 20770.

Competing interests none declared.

Patient consent not required.

ethics approval All procedures were performed following written informed consent and were approved by the research Ethics Committee – Queen Mary University – rEC 05/Q0703/198.

Provenance and peer review not commissioned; externally peer reviewed.

RefeRences 1 Voehringer d. protective and pathological roles of mast cells and basophils. Nat Rev

Immunol 2013;13:362–75. 2 Varricchi G, raap U, rivellese F, et al. Human mast cells and basophils-How are they

similar how are they different? Immunol Rev 2018;282:8–34. 3 Suurmond J, van der Velden d, Kuiper J, et al. Mast cells in rheumatic disease. Eur J

Pharmacol 2016;778:116–24.

1781Rivellese F, et al. Ann Rheum Dis 2018;77:1773–1781. doi:10.1136/annrheumdis-2018-213418

Basic and translational research

4 rivellese F, nerviani A, rossi FW, et al. Mast cells in rheumatoid arthritis: friends or foes? Autoimmun Rev 2017;16:557–63.

5 de paulis A, Marinò I, Ciccarelli A, et al. Human synovial mast cells. I. Ultrastructural in situ and in vitro immunologic characterization. Arthritis Rheum 1996;39:1222–33.

6 Crisp AJ, Chapman CM, Kirkham SE, et al. Articular mastocytosis in rheumatoid arthritis. Arthritis Rheum 1984;27:845–51.

7 Gotis-Graham I, Mcneil Hp. Mast cell responses in rheumatoid synovium. Association of the MCtC subset with matrix turnover and clinical progression. Arthritis Rheum 1997;40:479–89.

8 Gotis-Graham I, Smith Md, parker A, et al. Synovial mast cell responses during clinical improvement in early rheumatoid arthritis. Ann Rheum Dis 1998;57:664–71.

9 Hueber AJ, Asquith dL, Miller AM, J Ha, M Ma, et al. Mast cells express IL-17A in rheumatoid arthritis synovium. J Immunol 2010;184:3336–40.

10 Mcneil Hp, Shin K, Campbell IK, et al. the mouse mast cell-restricted tetramer-forming tryptases mouse mast cell protease 6 and mouse mast cell protease 7 are critical mediators in inflammatory arthritis. Arthritis Rheum 2008;58:2338–46.

11 Sawamukai n, Yukawa S, Saito K, et al. Mast cell-derived tryptase inhibits apoptosis of human rheumatoid synovial fibroblasts via rho-mediated signaling. Arthritis Rheum 2010;62:952–9.

12 noordenbos t, Blijdorp I, Chen S, et al. Human mast cells capture, store, and release bioactive, exogenous IL-17A. J Leukoc Biol 2016;100:453–62.

13 Suurmond J, rivellese F, dorjée AL, et al. toll-like receptor triggering augments activation of human mast cells by anti-citrullinated protein antibodies. Ann Rheum Dis 2015;74:1915–23.

14 Galli SJ, Kalesnikoff J, Grimbaldeston MA, a GM, et al. Mast cells as "tunable" effector and immunoregulatory cells: recent advances. Annu Rev Immunol 2005;23:749–86.

15 rivellese F, Suurmond J, Habets K, et al. Ability of interleukin-33- and immune complex-triggered activation of human mast cells to down-regulate monocyte-mediated immune responses. Arthritis Rheumatol 2015;67:2343–53.

16 Lee dM, Friend dS, Gurish MF, et al. Mast cells: a cellular link between autoantibodies and inflammatory arthritis. Science 2002;297:1689–92.

17 pitman n, Asquith dL, Murphy G, et al. Collagen-induced arthritis is not impaired in mast cell-deficient mice. Ann Rheum Dis 2011;70:1170–1.

18 Zhou JS, xing W, Friend dS, et al. Mast cell deficiency in Kit(W-sh) mice does not impair antibody-mediated arthritis. J Exp Med 2007;204:2797–802.

19 Brown MA, Hatfield JK. Mast cells are important modifiers of autoimmune disease: with so much evidence, why is there still controversy? Front Immunol 2012;3:147.

20 reber LL, Marichal t, Galli SJ. new models for analyzing mast cell functions in vivo. Trends Immunol 2012;33:613–25.

21 Schubert n, dudeck J, Liu p, et al. Mast cell promotion of t cell-driven antigen-induced arthritis despite being dispensable for antibody-induced arthritis in which t cells are bypassed. Arthritis Rheumatol 2015;67:903–13.

22 van der Velden d, Lagraauw HM, Wezel A, et al. Mast cell depletion in the preclinical phase of collagen-induced arthritis reduces clinical outcome by lowering the inflammatory cytokine profile. Arthritis Res Ther 2016;18:138.

23 Smolen JS, Aletaha d, McInnes IB. rheumatoid arthritis. Lancet 2016;6736:1–16. 24 Firestein GS, McInnes IB. Immunopathogenesis of rheumatoid Arthritis. Immunity

2017;46:183–96. 25 pitzalis C, Kelly S, Humby F. new learnings on the pathophysiology of rA from synovial

biopsies. Curr Opin Rheumatol 2013;25:334–44. 26 Kelly S, Humby F, Filer A, et al. Ultrasound-guided synovial biopsy: a safe,

well-tolerated and reliable technique for obtaining high-quality synovial tissue from both large and small joints in early arthritis patients. Ann Rheum Dis 2015;74:611–7.

27 Aletaha d, neogi t, Silman AJ, et al. 2010 rheumatoid arthritis classification criteria: an American College of rheumatology/European League Against rheumatism collaborative initiative. Ann Rheum Dis 2010;69:1580–8.

28 Krenn V, Morawietz L, Burmester Gr, et al. Synovitis score: discrimination between chronic low-grade and high-grade synovitis. Histopathology 2006;49:358–64.

29 Humby F, Bombardieri M, Manzo A, et al. Ectopic lymphoid structures support ongoing production of class-switched autoantibodies in rheumatoid synovium. PLoS Med 2009;6:e1.

30 rådinger M, Jensen BM, Kuehn HS, et al. Generation, isolation, and maintenance of human mast cells and mast cell lines derived from peripheral blood or cord blood. Curr Protoc Immunol 2010;37.

31 Kwakkenbos MJ, diehl SA, Yasuda E, et al. Generation of stable monoclonal antibody-producing B cell receptor-positive human memory B cells by genetic programming. Nat Med 2010;16:123–8.

32 rivellese F, Suurmond J, de paulis A, et al. IgE and IL-33-mediated triggering of human basophils inhibits tLr4-induced monocyte activation. Eur J Immunol 2014;44:3045–55.

33 Jones GW, Bombardieri M, Greenhill CJ, et al. Interleukin-27 inhibits ectopic lymphoid-like structure development in early inflammatory arthritis. J Exp Med 2015;212:1793–802.

34 tetlow LC, Woolley dE. distribution, activation and tryptase/chymase phenotype of mast cells in the rheumatoid lesion. Ann Rheum Dis 1995;54:549–55.

35 Hua Z, Hou B. tLr signaling in B-cell development and activation. Cell Mol Immunol 2013;10:103–6.

36 Bernasconi nL, onai n, Lanzavecchia A. A role for toll-like receptors in acquired immunity: up-regulation of tLr9 by BCr triggering in naive B cells and constitutive expression in memory B cells. Blood 2003;101:4500–4.

37 Jiang W, Lederman MM, Harding CV, et al. tLr9 stimulation drives naïve B cells to proliferate and to attain enhanced antigen presenting function. Eur J Immunol 2007;37:2205–13.

38 Merluzzi S, Frossi B, Gri G, et al. Mast cells enhance proliferation of B lymphocytes and drive their differentiation toward IgA-secreting plasma cells. Blood 2010;115:2810–7.

39 Mion F, d’Incà F, danelli L, et al. Mast cells control the expansion and differentiation of IL-10-competent B cells. J Immunol 2014;193:4568–79.

40 ramírez J, Celis r, Usategui A, et al. Immunopathologic characterization of ultrasound-defined synovitis in rheumatoid arthritis patients in clinical remission. Arthritis Res Ther 2016;18:74.

41 Bombardieri M, Lewis M, pitzalis C. Ectopic lymphoid neogenesis in rheumatic autoimmune diseases. Nat Rev Rheumatol 2017;13:141–54.

42 Frossi B, Mion F, Sibilano r, et al. Is it time for a new classification of mast cells? What do we know about mast cell heterogeneity? Immunol Rev 2018;282:35–46.

43 Bradding p, Arthur G. Mast cells in asthma--state of the art. Clin Exp Allergy 2016;46:194–263.

44 Suurmond J, van Heemst J, van Heiningen J, et al. Communication between human mast cells and Cd4(+) t cells through antigen-dependent interactions. Eur J Immunol 2013;43:1758–68.

45 Suurmond J, Habets KL, dorjée AL, et al. Expansion of th17 cells by human mast cells is driven by inflammasome-independent IL-1β. J Immunol 2016;197:4473–81.

46 palm AK, Garcia-Faroldi G, Lundberg M, et al. Activated mast cells promote differentiation of B cells into effector cells. Sci Rep 2016;6:20531.

47 Gauchat JF, Henchoz S, Mazzei G, et al. Induction of human IgE synthesis in B cells by mast cells and basophils. Nature 1993;365:340–3.

48 pawankar r, okuda M, Yssel H, et al. nasal mast cells in perennial allergic rhinitics exhibit increased expression of the Fc epsilonrI, Cd40L, IL-4, and IL-13, and can induce IgE synthesis in B cells. J Clin Invest 1997;99:1492–9.

49 tournilhac o, Santos dd, xu L, et al. Mast cells in Waldenstrom’s macroglobulinemia support lymphoplasmacytic cell growth through Cd154/Cd40 signaling. Ann Oncol 2006;17:1275–82.

50 Gaudenzio n, Espagnolle n, Mars Lt, et al. Cell-cell cooperation at the t helper cell/mast cell immunological synapse. Blood 2009;114:4979–88.

51 Carroll-portillo A, Cannon JL, te riet J, et al. Mast cells and dendritic cells form synapses that facilitate antigen transfer for t cell activation. J Cell Biol 2015;210:851–64.

52 Mukai K, tsai M, Saito H, et al. Mast cells as sources of cytokines, chemokines, and growth factors. Immunol Rev 2018;282:121–50.

53. detoraki A, Staiano rI, Granata F, et al. Vascular endothelial growth factors synthesized by human lung mast cells exert angiogenic effects. J Allergy Clin Immunol 2009;123:1142–9.

54 Varricchi G, Loffredo S, Galdiero Mr, et al. Innate effector cells in angiogenesis and lymphangiogenesis. Curr Opin Immunol 2018;53:152–60.

55 ruddle nH. Lymphatic vessels and tertiary lymphoid organs. J Clin Invest 2014;124:953–9.

56 Artis d, Villarino A, Silverman M, et al. the IL-27 receptor (WSx-1) is an inhibitor of innate and adaptive elements of type 2 immunity. J Immunol 2004;173:5626–34.

57 Miyazaki Y, Inoue H, Matsumura M, et al. Exacerbation of experimental allergic asthma by augmented th2 responses in WSx-1-deficient mice. J Immunol 2005;175:2401–7.

1782 Pamfil C, et al. Ann Rheum Dis 2018;77:1782–1789. doi:10.1136/annrheumdis-2018-213485

Basic and translational research

ExtEndEd rEport

Intrarenal activation of adaptive immune effectors is associated with tubular damage and impaired renal function in lupus nephritisCristina pamfil,1,2 Zuzanna Makowska,3 Aurélie de Groof,2 Gaëlle tilman,1,2 Sepideh Babaei,3 Christine Galant,2,4 pauline Montigny,1,2 nathalie demoulin,5,6 Michel Jadoul,5,6 Selda Aydin,4 ralf Lesche,3 Fiona Mcdonald,3 Frédéric A Houssiau,1,2 Bernard r Lauwerys1,2

To cite: pamfil C, Makowska Z, de Groof A, et al. Ann Rheum Dis 2018;77:1782–1789.

Handling editor Josef S Smolen

► Additional material is published online only. to view please visit the journal online (http:// dx. doi. org/ 10. 1136/ annrheumdis- 2018- 213485).

1department of rheumatology, Cliniques Universitaires Saint-Luc, Brussels, Belgium2pôle de pathologies rhumatismales systémiques et inflammatoires, Institut de recherche Expérimentale et Clinique, Université catholique de Louvain, Brussels, Belgium3pharmaceuticals, research and development, Bayer AG, Berlin, Germany4department of pathology, Cliniques Universitaires Saint-Luc, Brussels, Belgium5department of nephrology, Cliniques Universitaires Saint-Luc, Brussels, Belgium6pôle de néphrologie, Institut de recherche Expérimentale et Clinique, Université catholique de Louvain, Brussels, Belgium

Correspondence todr Bernard r Lauwerys, department of rheumatology, Cliniques Universitaires Saint-Luc, Brussels 1200, Belgium; bernard. lauwerys@ uclouvain. be

Cp and ZM contributed equally.

received 27 March 2018revised 12 July 2018Accepted 16 July 2018published online First 31 July 2018

© Author(s) (or their employer(s)) 2018. re-use permitted under CC BY. published by BMJ.

ABsTrACTObjectives Chronic renal impairment remains a feared complication of lupus nephritis (Ln). the present work aimed at identifying mechanisms and markers of disease severity in renal tissue samples from patients with Ln.Methods We performed high-throughput transcriptomic studies (Illumina HumanHt-12 v4 Expression BeadChip) on archived kidney biopsies from 32 patients with Ln and eight controls (pretransplant donors). Histological staging (glomerular and tubular scores) and immunohistochemistry experiments were performed on the same and on a replication set of 37 Ln kidney biopsy samples.results A group of Ln samples was identified by unsupervised clustering studies based on their gene expression features, that is, the overexpression of transcripts involved in antigen presentation, t and B cell activation. these samples were characterised by a significantly lower estimated glomerular filtration rate (eGFr) at the time of biopsy (t0) compared with the other systemic lupus erythematosus samples. Yet, apparent disease duration at t0, double-stranded dnA antibody titres at t0 and other relevant characteristics (serum C3, proteinuria, histological scores, numbers of previous flares) were not different between groups.Immunohistochemistry studies confirmed the association between interstitial infiltration by adaptive immune effectors and decreased renal function in the same and in a replication group of Ln kidney biopsies. this was associated with transcriptomic, histological and immunohistochemical evidence of renal tubular cell involvement.Conclusion Interstitial infiltration of Ln kidney biopsies by adaptive immune effectors is associated with impaired renal tubular cell function and decreased eGFr. these results open new perspectives in evaluating and treating patients with Ln, focusing on intrarenal mechanisms of immune cell activation.

InTrOduCTIOnLupus nephritis (LN) is a severe complication of systemic lupus erythematosus (SLE), initiated by the deposition of anti-double stranded DNA (anti-dsDNA) autoantibodies in glomerular basement membranes.1–3 Treatment of LN requires the use of immunosuppressants, in order to avoid evolu-tion to end-stage renal disease.4–6 Yet, despite major

improvements in treatment strategies over the last decades, a significant proportion of patients display renal damage accrual, and 10% develop renal failure after 10 years.7 8 Because the disease mainly affects young women, these numbers represent a heavy toll in terms of quality of life and health expenditures.

In recent years, it appeared that the lupus kidney is not just a passive target of systemic autoimmunity features, but also hosts mechanisms that contribute to the pathogenesis of the disease, independently of the systemic activation of immune effectors.9 10 The identification of anti-dsDNA-producing long-lived plasma cells in kidneys from SLE mice is a good illustration of how second wave immune effectors take advantage of the local inflammatory environ-ment to perpetuate the production of pathogenic autoantibodies.11 12 In human SLE kidneys, the presence of T and B cell aggregates and germinal centres is another indication of the development of an organised immune response in the target organ.10

In this study, we intended to establish a compre-hensive map of intrarenal molecular profiles associated with disease severity, by performing high-throughput transcriptomic studies on kidney biopsies from patients with LN, followed for a long period of time in a single LN reference centre, and controls (cadaveric donors). An independent set of LN kidney biopsies was used for histolog-ical and immunohistochemistry confirmation of our findings. In all patients, longitudinal clinical and biological data were collected systematically, which enabled us to correlate transcriptomic and/or immunohistochemistry patterns with longitu-dinal markers of renal and systemic disease activity. Although we used a ‘bottom-up’ approach to our analyses (ie, predefined clinical variables were not used to categorise the samples; instead, the samples were categorised based on their transcriptomic characteristics), our hypothesis was that intrarenal molecular patterns in our samples were associated with impaired renal function.

During the course of our study, our observa-tions made us focus on tubulointerstitial involve-ment as a variable associated with renal outcomes in LN. Evaluation of tubulointerstitial structures is not part of the present International Society of Nephrology/Renal Pathology Society (ISN/RPS) classification of LN. Yet, several groups developed

1783Pamfil C, et al. Ann Rheum Dis 2018;77:1782–1789. doi:10.1136/annrheumdis-2018-213485

Basic and translational research

Table 1 Characteristics of the patients with SLE included in this study

discovery set (transcriptomic and immunohistochemistry studies)

Confirmation set (immunohistochemistry studies)

Number of patients (female/male)

32 (23/9) 37 (34/3)

Age (years, median, IQR) 30 (35.7–35.7) 27 (22–40)

Apparent renal disease duration (months; median, range)

4 (0–144) 0 (0–156)

Duration of follow-up after the biopsy (months; median, range)

82 (5–258) 109 (37–256)

Ethnicity

Caucasian 26 31

African 3 3

Asiatic 3 1

Other 0 1

SLEDAI (median; range) 16 (4–25) 16 (6–24)

Serum dsDNA antibody titres (U/mL; median, IQR)

87.3 (32.2–291.3) 300.5 (66.7–686.7)

Serum C3 (mg/dL, median, IQR)

64 (48.7–80.2) 58.0 (33–68.5)

eGFR (mL/min/1.73 m2, median, IQR)

73.0 (48.7–93.0) 82.0 (60.0–104.0)

Proteinuria (g/24 hours, median, IQR)

2.6 (1.9–4.4) 2.0 (1.2–5.2)

Immunosuppressive therapy at biopsy

None (patients, n) 16 19

Methylprednisolone (patients, n; dose range)

7 (4–16) 16 (4–6)

Azathioprine (patients, n) 5 5

Mycophenolate mofetil (patients, n)

4 0

Glomerular histology

Class III (+V) 4 6

Class IV (+V) 27 31

Class V 1 0

Activity index (median, IQR) 8 (5–11) 10 (5–12)

Chronicity index (median, IQR)

1 (1–2) 1 (0–2)

There are no significant differences between the groups using Mann-Whitney and Χ2 tests.eGFR, estimated glomerular filtration rate; SLE, systemic lupus erythematosus; SLEDAI, SLE Disease Activity Index.

semiquantitative scores of interstitial fibrosis, inflammation and tubular cell damage in order to address the issue in the context of the disease. In this study, we also took advantage of the recent identification of Syndecan-1 (SDC1, CD138) as a marker of renal tubular cell stress. SDC1 is a lectin expressed at the surface of plasma cells, but also epithelial cells. In epithelial cells, cell surface SDC1 is cleaved by inflammation-induced proteases (eg, ADAMTS, MMP), and loss of cell surface SDC1 is associated with higher susceptibility to cell damage, in gastroenterological, but also renal disorders.13–19

PATIenTs And MeTHOdsPatients and kidney biopsiesAll patients with LN included in this study were recruited in a single centre (Université catholique de Louvain, Brussels, Belgium). They met the 1982 American College of Rheuma-tology revised classification criteria for the diagnosis of SLE,20 and had biopsy-proven nephritis. Patients’ demographics are displayed in table 1. See online supplementary file 1 for addi-tional specifications and ethical approval.

High-throughput transcriptomic studiesRNA was extracted from 60 SLE and 27 control biopsies using NucleoSpin technology (Macherey-Nagel). The samples were tested for RNA integrity by Bioanalyzer (Agilent) measurements. Samples with an RNA integrity number >6 and RNA quantity >50 ng were used for target synthesis (ie, from 32 SLE and eight control biopsies). There was no significant bias in clinical, biolog-ical or histological characteristics between the samples with good quality RNA and the samples dropped from the study. See online supplementary file 1 for the description of HumanHT-12 v4 Expression BeadChip (Illumina) hybridisations.

Histological and immunohistochemistry experimentsISN/RPS classification of the SLE renal biopsies and glomerular activity/chronicity indices (Morel-Maroger semiquantitative scores) were retrieved from the medical files of the patients. In addition, a semiquantitative score for renal tubular cell atrophy and interstitial fibrosis (0–3) was generated using H&E stained slides.

See online supplementary file 1 for the description of CD3, CD20, CD8 and SDC1 immunostaining experiments.

statistical analysesSee online supplementary file 1.21 22

resulTsHigh-density transcriptomic studies (Illumina BeadChip arrays) were performed using renal biopsies from 32 patients with LN and eight controls. All patients had proliferative (class III or class IV) nephritis with or without membranous (class V) involve-ment, except for one who had an isolated class V (table 1). After normalisation of the transcriptomic data, we found that LN samples were characterised by the overexpression of 478 tran-scripts compared with controls. Further data mining indicated that these transcripts were significantly enriched in the following pathways: inflammatory response, antigen presentation, leuco-cyte activation, interferon signature and chemokine signalling (figure 1A, online supplementary table 1). Conversely, 329 tran-scripts were downregulated in LN compared with control renal biopsies (online supplementary table 2). In the latter, pathway analyses showed a significant enrichment in the following gene ontologies: mitochondrion, oxidation reduction, organic acid

transport, hydrogen ion transmembrane transporter activity and glutathione transferase activity, many of which point at renal proximal tubular cell functions.

LN samples were not homogeneous in the differential expres-sion of these transcripts, and unsupervised hierarchical clustering studies distributed the biopsies in three groups (figures 1B and 2). Out of these, one group of samples overexpressed transcripts belonging to pathways pointing at local activation of adaptive immune effectors: antigen presentation, T and B cell activation. From a clinical and biological point of view, these patients had a significantly lower estimated glomerular filtration rate (eGFR) at baseline. Yet, there were no clinical or biological factors explaining this difference: renal disease duration, number of renal flares prior to the renal biopsy, number of renal flares after

1784 Pamfil C, et al. Ann Rheum Dis 2018;77:1782–1789. doi:10.1136/annrheumdis-2018-213485

Basic and translational research

Figure 1 Transcripts overexpressed in systemic lupus erythematosus (SLE) compared with control kidney biopsies. (A) Gene expression was determined in kidney biopsies by microarray analyses. Moderated t-tests with Benjamini-Hochberg corrections for multiple comparisons identified 478 transcripts upregulated in SLE compared with control biopsies with a p value <0.05. The figure displays log2-transformed mean-centred cyclic loess-normalised intensities across all samples. Pathway analyses indicated that a significant percentage of transcripts clustered into the indicated gene ontology terms. The complete lists of transcripts and pathways are available in online supplementary table 1. (B) An unsupervised hierarchical clustering algorithm (Pearson-centred algorithm, Ward’s linkage rule) using the same transcripts distributed the samples in three clusters, identified by a colour (same colours as in figure 2).

biopsy, glomerular activity and chronicity indices; even anti-dsDNA antibody titres were not different in these compared with the other LN samples (figure 2, table 2). Intriguingly, eGFR was still significantly lower in these patients at 1 year, and at last follow-up visit (figure 2, table 2). Univariate analyses indicated a link between eGFR at follow-up and eGFR at baseline, renal molecular cluster, glomerular chronicity index in the biopsy and induction therapy with cyclophosphamide, but only eGFR at baseline remained as an independent variable predictive of eGFR at follow-up using multivariate analyses.

Thus, our transcriptomic data indicated a link between the presence of transcripts associated with the activation of an adaptive immune response in the lupus kidney, and decreased renal function at baseline. We therefore performed immuno-histochemistry experiments aiming at the detection of CD3+ T, CD8+ T and CD20+ B cells in renal biopsies from patients with LN. In view of previous reports indicating the presence of tertiary lymphoid structures in LN biopsies, we also tried to detect CD21+ follicular dendritic cells. At first, we stained the samples used in the high-throughput transcriptomic studies. Out of the 30 stained biopsy specimens available, we found very little CD21 positive samples (n=3). In these few samples, CD21+ cells colocalised with CD3+ and CD20+ cells, indicating the presence of tertiary lymphoid structures. Otherwise, CD3+ and CD20+ cell aggregates were found in the majority of the samples with either a periglomerular or an intertubular distribu-tion (figure 3A,B). After digital quantification of the stains, we confirmed that the group of samples enriched in antigen presen-tation, T and B cell activation transcripts were significantly more positive for CD3, CD8 and CD20 than the other LN samples

(figure 3C). Accordingly, digital quantification of the CD3 and CD8 stains was significantly higher in samples from patients with a baseline eGFR≤60 mL/min/1.73 m2 compared with >60 mL/min/1.73 m2. Although CD20 correlated significantly with the CD3 stain (Spearman r=0.4985, p=0.0112), and negatively with eGFR (Spearman r=−0.4947, r=0.0119), there was no significant difference in CD20 quantification between eGFR≤60 and eGFR>60 mL/min/1.73 m2 (online supplementary figure 1).

These immunohistochemistry experiments were repeated in an independent set of 37 LN biopsies. Again, CD21 positive cells and tertiary lymphoid structures were found in three samples only. Aggregates of CD3+, CD8+ and CD20+ cells were found in a majority of samples in the kidney interstitium. After digital quantification, we also found that CD3 and CD20 correlated significantly with each other (Spearman r=0.4950, p=0.0034). CD3+, CD8+ and CD20+ cells were significantly higher in the biopsy samples from patients with an eGFR≤60 mL/min compared with >60 mL/min in this second cohort of samples (figure 3D).

In order to understand the link between the presence of adap-tive immune effectors in lupus kidney biopsies and decreased kidney function, we went back to the high-throughput transcrip-tomic data and identified transcripts and pathways that correlated (positively or negatively) with CD3, CD4, CD8 and CD19 tran-scripts (there was no transcript corresponding to CD20 on the Illumina platform) in lupus kidney biopsy samples. Not surpris-ingly, a positive correlation was found between the expression of these transcripts and transcripts involved in inflammatory responses, adaptive immune responses, antigen presentation, T cell activation and chemotaxis. By contrast, a negative correlation

1785Pamfil C, et al. Ann Rheum Dis 2018;77:1782–1789. doi:10.1136/annrheumdis-2018-213485

Basic and translational research

Figure 2 Transcriptomic profiling singles out a subgroup of systemic lupus erythematosus (SLE) kidney biopsy samples with decreased estimated glomerular filtration rate (eGFR). (A) A cluster of SLE kidney biopsy samples, identified by hierarchical clustering algorithm (Pearson-centred algorithm, Ward’s linkage rule), is characterised by the overexpression of 271 transcripts (by moderated t-test, with Bonferroni correction for multiple comparisons) compared with the other lupus samples, significantly enriched in the indicated pathways. Samples designated by a black box are SLE samples. The complete lists of transcripts and pathways are available in online supplementary table 3. (B) Distribution of biological data in sera of patients with SLE at thetime of the biopsy and at 1 year, according to the transcriptomic clusters determined in (A). Horizontal lines represent the median values and IQRs. P values were calculated by Mann-Whitney test. APC, antigen-presenting cells.

was found with transcripts associated with renal tubular cell functions: oxidation reduction, mitochondrion, peroxisome and proton transport (figure 4A). Accordingly, digital quantification of the CD3+ T and CD20+ B cell stains in the kidney intersti-tium of these samples was significantly associated with increased histological scores of renal tubular cell damage (CD3: Spearman r=0.6229, p=0.0015; CD20: Spearman r=0.4771, p=0.0213). Renal tubular cell damage itself displayed a significant negative correlation with eGFR (Spearman r=−0.4732, p=0.0226).

Finally, we evaluated the loss of SDC1 expression at the surface of renal tubular cells by immunohistochemistry. SDC1 is cleaved at the surface of renal tubular cells by MMP7 (online supple-mentary file 1 and supplementary figure 2). In both cohorts of lupus biopsy samples, we found that SDC1 was expressed by a few infiltrating plasma cells, but the vast majority of stained cells were renal tubular cells (figure 4B). Quantitative evaluation of the SDC1 stain correlated positively with the expression of tran-scripts associated with renal tubular cell functions (figure 4A). Hence, it was significantly higher in biopsies where no tubular damage was evidenced at histological evaluation. Accordingly, it was significantly higher in patients with an eGFR higher than 60 mL/min, due to the significant association between tubular damage and eGFR (figure 4C). Of note, quantitative evaluation of the SDC1 stain correlated negatively with transcripts associ-ated with T, B and antigen-presenting cell activation in the same set of samples (figure 4A), providing further support to the link between the presence of adaptive immune effectors in the lupus kidney, and renal tubular cell damage-mediated impairment of renal function.

dIsCussIOnThe development of new therapeutic strategies using immuno-suppressive drugs (intravenous cyclophosphamide, mycophe-nolate mofetil in association with corticosteroids) significantly improved renal outcomes in LN.23 Yet, up to 10% of the patients develop end-stage renal disease at 10 years, which is a major concern in a population of mainly young female patients. In this perspective, several studies were performed in order to identify early markers of disease severity in patients with LN. Thus, evidence indicates that decreased eGFR and high glomer-ular chronicity indices at histological evaluation of baseline kidney biopsies are associated with poor renal outcomes.24–29 More recently, it appeared that reaching a proteinuria <0.7–0.8 g/24 hours at 1 year after initiation of therapy was predic-tive of a good renal outcome at 10 years, an observation that potentially translates into a treat-to-target approach of LN.30 31 Yet, our current approach to the care of patients with LN is still hampered by an incomplete understanding of the pathogenic events characteristic of the disease.

Our present results bring important additional information on markers and mechanisms of renal inflammation in LN, by pointing at the role of activated T and B cells in the kidney interstitium as potential mediators of renal tubular cell damage. In two independent groups of kidney biopsies, we did not find an association between decreased renal function and markers of systemic disease activity or histological indices of glomer-ular involvement (activity and chronicity indices). By contrast, a decreased eGFR was observed in patients with histological evidence of tubular damage, and transcriptomic evidence of T and B cell activation in the kidney, which was further confirmed by immunohistochemistry.

In a recent study, Parikh et al evaluated the expression of 511 immune response genes in longitudinal (before and after

1786 Pamfil C, et al. Ann Rheum Dis 2018;77:1782–1789. doi:10.1136/annrheumdis-2018-213485

Basic and translational research

Table 2 Clinical, biological and histological characteristics of patients with LN categorised according to the transcriptomic clusters determined in figure 2A

APC, T and B cell cluster Other ln samples P values

Number of patients 14 18

Sex (male/female) 5/9 14/4 0.3997*

Ethnicity (Caucasians/non-Caucasians) 3/11 15/3 0.7321*

Age (years, median, IQR) 28.5 (25.5–36.2) 31.0 (24.7–38.5) 0.9545†

Baseline

eGFR (mL/min/1.72 m2, median, IQR) 49.5 (29.2–69.2) 88 (72.0–129.3) 0.0007†

Proteinuria (g/24 hours, median, IQR) 3.39 (2.04–5.88) 2.45 (1.14–3.89) 0.0839†

Serum C3 (mg/dL, median, IQR) 70.5 (42.5–93.2) 64 (51.5–77.5) 0.6213†

Serum dsDNA antibody titres (U/mL; median; IQR) 34.1 (14.9–315.1) 155.5 (77.1–306.8) 0.0654†

SLEDAI-2K (median; IQR) 18 (13–20) 16 (16–20) 0.9229†

Glomerular histology 0.5127*

Class III (+V) 3 2

Class IV (+V) 11 15

Class V 0 1

Activity index (median, IQR) 10 (4–13) 7 (5–10) 0.1759†

Chronicity index (median, IQR) 2 (1–2) 1 (1–2) 0.1544†

Apparent renal disease duration (median, months; IQR) 4.5 (1–39) 3.5 (2.7–16.2) 0.9847†

Number of renal flares at biopsy (median; months; IQR) 1 (1–2) 1 (1–1) 0.1832†

Immunosuppressive therapy at biopsy 0.3086*

None (patients, n) 9 7

Methylprednisolone (patients, n; mg/day; dose range) 2 (4–16) 5 (8–32)

Azathioprine (patients, n) 1 5

Mycophenolate mofetil (patients, n) 2 2

Intravenous cyclophosphamide (Euro-Lupus) (patients, n) 1 0

At year 1

eGFR (mL/min/1.72 m2, median, IQR) 54.5 (45.7–96.2) 101.5 (80.5–121.8) 0.0150†

Proteinuria (g/24 hours, median, IQR) 0.97 (0.34–2.51) 0.87 (0.13–2.0) 0.5590†

Serum C3 (mg/dL, median, IQR) 87.55 (76.7–107.3) 86.5 (70.5–103.0) 0.8196†

Serum dsDNA antibody titres (U/mL; median; IQR) 13.3 (8.5–46.2) 33.5 (17.7–72.1) 0.0984†

Immunosuppressive therapy after renal biopsy 0.1038*

Intravenous steroids alone (patients, n) 2 0

Mycophenolate mofetil (patients, n) 7 5

Intravenous cyclophosphamide (Euro-Lupus) (patients, n) 3 11

Intravenous cyclophosphamide (NIH) (patients, n) 1 1

Rituximab 1 0

At follow-up

Duration of follow-up (months; median; IQR) 91 (20–129) 86 (41–136) 0.8197†

Total number of renal flares at follow-up (n; median; IQR) 1.5 (1.0–3.0) 1.5 (1.0–2.0) 0.4835†

eGFR (mL/min/1.72 m2, median, IQR) 51.0 (11.7–72.0) 88.5 (67.7–114.5) 0.0215†

ESRD (patients, n) 5 3 0.1476‡

Significant P values are in bold.*P values calculated by Χ2 tests.†P values calculated by Mann-Whitney tests.‡P values calculated by Gehan-Breslow-Wilcoxon test.APC, antigen-presenting cells, eGFR, estimated glomerular filtration rate; ESRD, end- stage renal disease; LN, lupus nephritis; NIH, National Institutes of Health; SLEDAI, SLE Disease Activity Index.

administration of induction therapy) kidney biopsies from four patients with LN who did not respond to therapy, compared with five patients who displayed a complete response, and four controls. In line with our own results, they reported that tran-scripts associated with leucocyte infiltration, T cell activation, complement pathway and interferon signature increased between baseline and follow-up in non-responders.32 In the present study, we instead used an unbiased analytical approach, grouping the samples based on their molecular characteristics rather than predefined clinical categories, resulting in the identification of

a link between T and B cell infiltration and activation, tubular damage and decreased renal function in LN.

Tubulointerstitial lesions were previously already identified as a prognostic marker in LN.33–37 Interstitial infiltration by inflammatory cells and tubular atrophy potentially result from or are amplified by several kinds of insults in the context of LN: proteinuria, deposition of immune complexes in the interstitium, rupture of the Bowman’s capsule and presentation of cryptic antigens by juxtaglomerular antigen-presenting cells, induction of proinflammatory molecules at the surface of renal tubular cells

1787Pamfil C, et al. Ann Rheum Dis 2018;77:1782–1789. doi:10.1136/annrheumdis-2018-213485

Basic and translational research

Figure 3 The presence of adaptive immune effectors in the systemic lupus erythematosus (SLE) kidney interstitium is associated with decreased estimated glomerular filtration rate (eGFR). (A) CD21-negative and (B) CD21-positive T and B cell aggregates in the interstitium are shown in serial sections from two SLE kidney biopsies. (C) Digital quantification of the CD3, CD8 and CD20 stains (Log2 scale) confirms higher numbers of T and B cells in SLE kidney biopsy fragments from the cluster of samples overexpressing T and B cell-associated transcripts in the microarray data. (D) Digital quantification of the CD3, CD8 and CD20 stains in an independent group of SLE kidney biopsies indicates higher numbers of T and B cells in patients with decreased baseline eGFR. Horizontal lines represent the median values and IQRs. P values were calculated by Mann-Whitney test. APC, antigen-presenting cells.

(eg, ICAM-1, CD40).38–43 Several lines of evidence are in favour of an antigen-dependent local expansion of adaptive immune effectors. Thus, Chang et al demonstrated the presence of cells

from the adaptive immune system in the interstitium of lupus kidney biopsies, organised either in T and B cell aggregates, or in germinal centres in the presence of CD21+ follicular dendritic cells. By sequencing the Ig repertoire of one germinal centre and four T:B cell aggregates, they found evidence of clonal restriction in all samples tested, and extensive somatic hypermutation in all but one T and B cell aggregate, in line with an antigen-driven clonal selection.10 In a later publication, authors from the same group found that vimentin is overexpressed by mononuclear cells infiltrating the renal interstitium in lupus biopsies, and is a recurrent target of locally produced immunoglobulins.44 Simi-larly, evaluation of the local T cell receptor repertoire is also indicative of a clonal expansion of a restricted population of T cells.45 46 Local antigens, driving the proliferation of these T cells in the lupus kidney, were not identified yet.

Our high-density transcriptomic data provide further support to the hypothesis that infiltrating adaptive immune effectors display a toxic effect on renal tubular cells, thereby contributing to the development of renal failure. Thus, transcripts associated with CD3, CD4, CD8 and B cells displayed strong correlations with each other, which is not a surprise in the context of an organised immune response. Little CD21+ follicular dendritic cells were found, but this could be biased by the small size of the renal biopsy samples. Negative correlations were found with transcripts associated with renal tubular cell functions. The link between the presence of T and B cells in the renal interstitium and tubular cell damage was confirmed by immunohistochem-istry and histological scoring of biopsy specimens. The stron-gest and most consistent correlations between cell infiltration and renal tubular cell involvement were observed for CD3 and CD8 positive cells. These data are in line with observations by Couzi et al, who reported that CD8 T cells are the most preva-lent T cells infiltrating lupus kidneys, and also found a significant correlation between CD8 T cells in the kidney interstitium and decreased renal function in 25 patients with LN.47 The possi-bility that in particular CD8 T cells play an amplificatory role in LN-related renal lesions is also emphasised by the observation that the presence of a T effector-memory signature in CD8 T cells isolated from the peripheral blood of patients with LN is associated with a higher probability of renal relapses and a higher number of disease flares per month.48 Based on data from the same group, that is, presence of a similar CD8 effector-memory T cell signature in the blood of patients with ANCA-associated vasculitis (AAV) and a higher rate of renal relapses, it is tempting to speculate that secondary activation of immune effectors against autoantigens locally modified by an initial renal injury is not restricted to LN. Thus, evidence indicates that intersti-tial infiltration by CD3 T cells and tubular damage is associated with poor response to rituximab therapy and lower eGFR at 12 months in AAV.49 Similarly, previous studies had identified the presence of interstitial inflammation as a significant prognostic marker in patients with AAV treated with cyclophosphamide and corticosteroids.50

That renal tubular cells in the lupus kidney are particularly targeted by infiltrating adaptive immune effectors is also illus-trated by our SDC1 data. Although SDC1 is expressed by a few plasma cells infiltrating the lupus kidney, this lectin is mainly expressed by renal tubular cells.13–15 SDC1 is cleaved from the surface of epithelial cells by inflammation-induced proteases.16 17 In the lupus kidney, MMP7 is a good candidate in view of the strong correlation between the loss of SDC1 and MMP7 expres-sion, and our in vitro data demonstrating the loss of SDC1 expression at the surface of a renal tubular cell line after expo-sure to MMP7. In inflammatory bowel diseases, loss of SDC1

1788 Pamfil C, et al. Ann Rheum Dis 2018;77:1782–1789. doi:10.1136/annrheumdis-2018-213485

Basic and translational research

Figure 4 The presence of adaptive immune effectors in the systemic lupus erythematosus (SLE) kidney interstitium is associated with impaired renal tubular cell phenotype and functions. (A) The microarray data set was analysed to identify genes that show >0.4 or <−0.4 value of Pearson correlation to CD3, CD8, CD4, CD19 transcripts from the same microarray data set, or to the digital quantification of SDC1 in the histological specimens from the same biopsies. The resulting gene sets were then annotated using DAVID software and the selected top pathway enrichment scores (−log10 p value) are shown as different hues of green (for the sets of genes showing positive correlation with the respective variable) and red (negative correlation). Grey squares indicate that the pathway is not enriched in the list of transcripts. Transcripts and pathways are listed in online supplementary tables 4–8. (B) H&E and SDC1 stains on consecutive slides from patients without (upper line) or with (lower line) tubular infiltration by mononuclear cells. The arrow indicates CD138-positive infiltrating plasmablasts. (C) Quantification of the SDC1 stain in SLE kidney biopsies used in the microarray experiments indicates lower SDC1 in patients with higher renal tubular cell damage and lower estimated glomerular filtration rate (eGFR). P values were calculated by Mann-Whitney tests.

at the surface of intestinal epithelial cells is associated with a higher susceptibility to cell damage,18 19 and the same holds true in renal tubular cells in view of our transcriptomic and histolog-ical data linking SDC1 expression and renal tubular cell viability and function. Using the SDC1 stain as a marker of renal tubular cell integrity, we confirmed the negative correlation with the expression of genes associated with the activation of adaptive immune effectors.

Elevated SDC1 concentrations are found in the sera of patients with SLE as compared with controls.15 It will be of interest to evaluate whether serum SDC1 in these patients is associated with renal disease and poorer renal outcomes. Overall, our data are a strong trigger to evaluate the link between markers of renal tubular cell damage and poorer renal outcomes in LN individuals, together with an assessment of the distribution and activation patterns of immune cells in urine specimens of these patients. In addition, mechanistic data are needed, in order to provide definite evidence about the pathogenic amplificatory role of CD8 T cells in LN. Also, while our observations were made in a mainly Caucasian population of patients with LN, it will be of importance to extend our studies to ethnically diverse populations, in particular due to the known influence of ethnicity on renal outcomes in LN. Would such observations confirm our data, they would open new perspectives in evaluating and treating patients with LN, focusing on intrarenal mechanisms of immune cell activation.

Contributors ZM, AdG, Gt, SB, CG, rL, FMd, BrL: acquisition of experimental data, data analysis, drafting of the manuscript. nd, MJ, SA, FAH: acquisition of clinical data, data analysis, drafting of the manuscript.

Funding the research leading to these results has received support from the Innovative Medicines Initiative Joint Undertaking under grant agreement number 115565, resources of which are composed of financial contribution from the European Union’s Seventh Framework programme (Fp7/2007-2013) and EFpIA companies’ in-kind contribution. this work was also supported by ForEUM (Foundation for research in rheumatology) in the context of the rEFrACt project. Cp was funded in part by a EULAr scientific training bursary, and in part by a mobility grant from the department of rheumatology, ’Iuliu Hatieganu’ University of Medicine and pharmacy, Cluj-napoca, romania. AdG is funded by the ’Chaire UCL/UCB sur les rhumatismes Systémiques et Inflammatoires’. pM is funded by the Fondation Saint-Luc, Cliniques Universitaires Saint-Luc, Brussels, Belgium. BrL is funded in part by the Fonds national de la recherche Scientifique (FnrS, communauté française de Belgique).

Competing interests none declared.

Patient consent obtained.

ethics approval the Ethical Committee of the Université catholique de Louvain (UCL) approved the study and initially ruled that patient consent was not required for the use of residual corporal material, in agreement with Belgian regulations on human studies. However, in order to ensure formal compliance with German regulations on the use of residual corporal material, we did ask all patients with SLE included in the study to sign a consent form, after re-approval by the UCL Ethical Committee.

Provenance and peer review not commissioned; externally peer reviewed.

data sharing statement raw data are available upon request from the corresponding author.

1789Pamfil C, et al. Ann Rheum Dis 2018;77:1782–1789. doi:10.1136/annrheumdis-2018-213485

Basic and translational research

Open access this is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https:// creativecommons. org/ licenses/ by/ 4. 0/.

RefeRences 1 Hahn BH. Antibodies to dnA. N Engl J Med 1998;338:1359–68. 2 Mjelle JE, rekvig op, Van der Vlag J, et al. nephritogenic antibodies bind in glomeruli

through interaction with exposed chromatin fragments and not with renal cross-reactive antigens. Autoimmunity 2011;44:373–83.

3 Krishnan Mr, Wang C, Marion tn. Anti-dnA autoantibodies initiate experimental lupus nephritis by binding directly to the glomerular basement membrane in mice. Kidney Int 2012;82:184–92.

4 Bertsias GK, tektonidou M, Amoura Z, et al. Joint European League Against rheumatism and European renal Association-European dialysis and transplant Association (EULAr/ErA-EdtA) recommendations for the management of adult and paediatric lupus nephritis. Ann Rheum Dis 2012;71:1771–82.

5 Hahn BH, McMahon MA, Wilkinson A, et al. American College of rheumatology guidelines for screening, treatment, and management of lupus nephritis. Arthritis Care Res 2012;64:797–808.

6 Houssiau FA, Lauwerys Br. Current management of lupus nephritis. Best Pract Res Clin Rheumatol 2013;27:319–28.

7 Vandepapelière J, Aydin S, Cosyns Jp, et al. prognosis of proliferative lupus nephritis subsets in the Louvain Lupus nephritis inception Cohort. Lupus 2014;23:159–65.

8 Zhang L, Lee G, Liu x, et al. Long-term outcomes of end-stage kidney disease for patients with lupus nephritis. Kidney Int 2016;89:1337–45.

9 Starke C, Frey S, Wellmann U, et al. High frequency of autoantibody-secreting cells and long-lived plasma cells within inflamed kidneys of nZB/W F1 lupus mice. Eur J Immunol 2011;41:2107–12.

10 Chang A, Henderson SG, Brandt d, et al. In situ B cell-mediated immune responses and tubulointerstitial inflammation in human lupus nephritis. J Immunol 2011;186:1849–60.

11 taddeo A, Khodadadi L, Voigt C, et al. Long-lived plasma cells are early and constantly generated in new Zealand Black/new Zealand White F1 mice and their therapeutic depletion requires a combined targeting of autoreactive plasma cells and their precursors. Arthritis Res Ther 2015;17:39.

12 Espeli M, Bökers S, Giannico G, et al. Local renal autoantibody production in lupus nephritis. J Am Soc Nephrol 2011;22:296–305.

13 Celie JW, reijmers rM, Slot EM, et al. tubulointerstitial heparan sulfate proteoglycan changes in human renal diseases correlate with leukocyte influx and proteinuria. Am J Physiol Renal Physiol 2008;294:F253–F263.

14 rops AL, Götte M, Baselmans MH, et al. Syndecan-1 deficiency aggravates anti-glomerular basement membrane nephritis. Kidney Int 2007;72:1204–15.

15 Kim KJ, Kim JY, Baek IW, et al. Elevated serum levels of syndecan-1 are associated with renal involvement in patients with systemic lupus erythematosus. J Rheumatol 2015;42:202–9.

16 teng YH, Aquino rS, park pW. Molecular functions of syndecan-1 in disease. Matrix Biol 2012;31:3–16.

17 Zhang S, Qing Q, Wang Q, et al. Syndecan-1 and Heparanase. Inflamm Bowel Dis 2013;19:1025–33.

18 Zhang Y, Wang Z, Liu J, et al. Suppressing Syndecan-1 Shedding Ameliorates Intestinal Epithelial Inflammation through Inhibiting nF-κB pathway and tnF-α. Gastroenterol Res Pract 2016;2016:1–8.

19 Zhang Y, Wang Z, Liu J, et al. Cell surface-anchored syndecan-1 ameliorates intestinal inflammation and neutrophil transmigration in ulcerative colitis. J Cell Mol Med 2017;21:13–25.

20 tan EM, Cohen AS, Fries JF, et al. the 1982 revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum 1982;25:1271–7.

21 Huang daW, Sherman Bt, Lempicki rA. Systematic and integrative analysis of large gene lists using dAVId bioinformatics resources. Nat Protoc 2009;4:44–57.

22 Huang daW, Sherman Bt, Lempicki rA. Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res 2009;37:1–13.

23 Fiehn C, Hajjar Y, Mueller K, et al. Improved clinical outcome of lupus nephritis during the past decade: importance of early diagnosis and treatment. Ann Rheum Dis 2003;62:435–9.

24 dall’Era M, Stone d, Levesque V, et al. Identification of biomarkers that predict response to treatment of lupus nephritis with mycophenolate mofetil or pulse cyclophosphamide. Arthritis Care Res 2011;63:351–7.

25 Ayodele oE, okpechi IG, Swanepoel Cr. predictors of poor renal outcome in patients with biopsy-proven lupus nephritis. Nephrology 2010;15:482–90.

26 Hiramatsu n, Kuroiwa t, Ikeuchi H, et al. revised classification of lupus nephritis is valuable in predicting renal outcome with an indication of the proportion of glomeruli affected by chronic lesions. Rheumatology 2008;47:702–7.

27 Contreras G, pardo V, Cely C, et al. Factors associated with poor outcomes in patients with lupus nephritis. Lupus 2005;14:890–5.

28 Mok CC, Ying KY, tang S, et al. predictors and outcome of renal flares after successful cyclophosphamide treatment for diffuse proliferative lupus glomerulonephritis. Arthritis Rheum 2004;50:2559–68.

29 Korbet SM, Lewis EJ, Schwartz MM, et al. Factors predictive of outcome in severe lupus nephritis. Lupus nephritis Collaborative Study Group. Am J Kidney Dis 2000;35:904–14.

30 dall’Era M, Cisternas MG, Smilek dE, et al. predictors of long-term renal outcome in lupus nephritis trials: lessons learned from the Euro-Lupus nephritis cohort. Arthritis Rheumatol 2015;67:1305–13.

31 tamirou F, Lauwerys Br, dall’Era M, et al. A proteinuria cut-off level of 0.7 g/day after 12 months of treatment best predicts long-term renal outcome in lupus nephritis: data from the MAIntAIn nephritis trial. Lupus Sci Med 2015;2:e000123.

32 parikh SV, Malvar A, Song H, et al. Molecular imaging of the kidney in lupus nephritis to characterize response to treatment. Transl Res 2017;182:1–13.

33 Hsieh C, Chang A, Brandt d, et al. predicting outcomes of lupus nephritis with tubulointerstitial inflammation and scarring. Arthritis Care Res 2011;63:865–74.

34 Ferraccioli G, romano G. renal interstitial cells, proteinuria and progression of lupus nephritis: new frontiers for old factors. Lupus 2008;17:533–40.

35 Broder A, Mowrey WB, Khan Hn, et al. tubulointerstitial damage predicts end stage renal disease in lupus nephritis with preserved to moderately impaired renal function: A retrospective cohort study. Semin Arthritis Rheum 2018;47:545–51.

36 daniel L, Sichez H, Giorgi r, et al. tubular lesions and tubular cell adhesion molecules for the prognosis of lupus nephritis. Kidney Int 2001;60:2215–21.

37 Yu F, Wu LH, tan Y, et al. tubulointerstitial lesions of patients with lupus nephritis classified by the 2003 International Society of nephrology and renal pathology Society system. Kidney Int 2010;77:820–9.

38 Yap dY, Yung S, Zhang Q, et al. Serum level of proximal renal tubular epithelial cell-binding immunoglobulin G in patients with lupus nephritis. Lupus 2016;25:46–53.

39 Bonanni A, Vaglio A, Bruschi M, et al. Multi-antibody composition in lupus nephritis: isotype and antigen specificity make the difference. Autoimmun Rev 2015;14:692–702.

40 ronda n, Cravedi p, Benozzi L, et al. Early proinflammatory activation of renal tubular cells by normal and pathologic IgG. Nephron Exp Nephrol 2005;100:e77–e84.

41 Eddy AA. proteinuria and interstitial injury. Nephrol Dial Transplant 2004;19:277–81. 42 theilig F. Spread of glomerular to tubulointerstitial disease with a focus on

proteinuria. Ann Anat 2010;192:125–32. 43 Zheng L, Sinniah r, Hsu SI, Si-h H. pathogenic role of nF-kappaB activation in

tubulointerstitial inflammatory lesions in human lupus nephritis. J Histochem Cytochem 2008;56:517–29.

44 Kinloch AJ, Chang A, Ko K, et al. Vimentin is a dominant target of in situ humoral immunity in human lupus tubulointerstitial nephritis. Arthritis Rheumatol 2014;66:3359–70.

45 Murata H, Matsumura r, Koyama A, et al. t cell receptor repertoire of t cells in the kidneys of patients with lupus nephritis. Arthritis Rheum 2002;46:2141–7.

46 Winchester r, Wiesendanger M, Zhang HZ, et al. Immunologic characteristics of intrarenal t cells: trafficking of expanded Cd8+ t cell β-chain clonotypes in progressive lupus nephritis. Arthritis Rheum 2012;64:1589–600.

47 Couzi L, Merville p, deminière C, et al. predominance of Cd8+ t lymphocytes among periglomerular infiltrating cells and link to the prognosis of class III and class IV lupus nephritis. Arthritis Rheum 2007;56:2362–70.

48 McKinney EF, Lyons pA, Carr EJ, et al. A Cd8+ t cell transcription signature predicts prognosis in autoimmune disease. Nat Med 2010;16:586–91.

49 Berden AE, Jones rB, Erasmus dd, et al. tubular lesions predict renal outcome in antineutrophil cytoplasmic antibody-associated glomerulonephritis after rituximab therapy. J Am Soc Nephrol 2012;23:313–21.

50 de Lind van Wijngaarden rA, Hauer HA, Wolterbeek r, et al. Clinical and histologic determinants of renal outcome in AnCA-associated vasculitis: A prospective analysis of 100 patients with severe renal involvement. J Am Soc Nephrol 2006;17:2264–74.

1790 Pieterse E, et al. Ann Rheum Dis 2018;77:1790–1798. doi:10.1136/annrheumdis-2018-213223

Basic and translational research

ExtEndEd rEport

Cleaved N-terminal histone tails distinguish between NADPH oxidase (NOX)-dependent and NOX-independent pathways of neutrophil extracellular trap formationElmar pieterse,1 nils rother,1 Cansu Yanginlar,1 Jelle Gerretsen,2 Sebastian Boeltz,3 Luis Enrique Munoz,3 Martin Herrmann,3 peter pickkers,2 Luuk B Hilbrands,1 Johan van der Vlag1

To cite: pieterse E, rother n, Yanginlar C, et al. Ann Rheum Dis 2018;77:1790–1798.

Handling editor Josef S Smolen

► Additional material is published online only. to view please visit the journal online (http:// dx. doi. org/ 10. 1136/ annrheumdis- 2018- 213223).

For numbered affiliations see end of article.

Correspondence todr Johan van der Vlag, department of nephrology (480), radboud Institute for Molecular Life Sciences, radboud University Medical Center, nijmegen 6525 GA, the netherlands; johan. vandervlag@ radboudumc. nl

Ep and nr contributed equally.

received 11 February 2018revised 21 July 2018Accepted 23 July 2018published online First 17 August 2018

© Author(s) (or their employer(s)) 2018. no commercial re-use. See rights and permissions. published by BMJ.

ABsTrACTObjectives neutrophil extracellular traps (nEts) act in various rheumatic diseases. Although nEt formation was originally described as a nicotinamide adenine dinucleotide phosphate (nAdpH) oxidase (nox)-dependent pathway, it appears that there are also nox-independent pathways of nEt release. Currently, no tools are available that can discriminate between both nEt-forming pathways. We aimed to develop a serological method allowing the discrimination between nEts generated through nox-dependent or nox-independent pathways.Methods Histones from in vitro generated nox-dependent and nox-independent nEts were characterised with a panel of lupus-derived antibodies against n-terminal histone tails using immunofluorescence microscopy, western blot and ELISA. nEts in patients with nEt-associated diseases, that is, rheumatoid arthritis (rA), systemic lupus erythematosus (SLE), psoriatic arthritis (psA) and sepsis, were characterised in sandwich ELISAs employing antibodies against myeloperoxidase (Mpo) and n-terminal histone tails as detecting and capturing antibodies, respectively. Functional responses of endothelial cells to nox-dependent and nox-independent nEts were assessed as well.results neutrophil elastase cleaves the n-terminal tails of core histones during nox-dependent, but not during nox-independent nEt formation. Consequently, the detection of Mpo–histone complexes with antibodies against n-terminal histone tails allows discrimination between nEts formed through a nox-dependent or nox-independent manner. Characterisation of in vivo circulating nEts revealed the presence of nox-independent nEts in rA, SLE and sepsis, but nox-dependent nEts in psA. nox-independent nEts displayed an increased capacity to activate endothelial cells when compared with nox-dependent nEts.Conclusions these results indicate heterogeneity in nEt-forming pathways in vivo and highlight the need for disease-specific strategies to prevent nEt-mediated pathology.

InTrOduCTIOnIn addition to phagocytosis and degranulation, neutrophils can release the so-called neutrophil

extracellular traps (NETs) into the extracellular space.1 NETs are web-like structures of chromatin and granule-derived antimicrobial proteins that serve to entrap and eliminate pathogens that are too large for phagocytic uptake or have evolved ways to circumvent phagocytosis.2 Despite their pivotal role in host defense, NETs have also been linked to a wide variety of infectious, haematological and autoimmune diseases.3 4

To date, two distinct pathways involved in NET release have been identified.5 In the canonical pathway of NET release, often referred to as nico-tinamide adenine dinucleotide phosphate (NADPH) oxidase (NOX)-dependent or lytic NET formation, the assembly of NOX at cytoplasmic or phago-somal membranes and the subsequent generation of reactive oxygen species (ROS) initiates a series of events which culminate in explosive cell death accompanied by NET release.2 4 6 Among others, the exposure of neutrophils to phorbol myristate acetate (PMA) or lipopolysaccharides (LPS) lead to NOX-dependent (lytic) NET formation. The signal-ling events underlying the non-canonical pathway of NET release, often referred to as NOX-inde-pendent or non-lytic NET formation, have been less studied. However, it has become clear that the NOX-mediated production of ROS is not involved in non-lytic NET formation.7 During NOX-inde-pendent NET formation, neutrophils are not lysed, but instead they release NETs through blebbing of the nuclear envelope and the subsequent vesicular inside-out transportation of NETs.7 8 Activated platelets appear to be potent inducers of NOX-in-dependent NET formation.9 10

The ELISA-based detection of complexes consisting of myeloperoxidase (MPO) and DNA is currently often employed when attempting to quan-tify NET formation in vivo in biological fluids.11–14 However, the detection of MPO–DNA complexes fails to distinguish between NETs that have been generated through NOX-dependent or NOX-in-dependent pathways, as both contain MPO and DNA.10 Discrimination between NOX-dependent and NOX-independent NETs is important to deci-pher differences in physiological function, as well as differences in pathophysiological relevance.5 Since it has previously been demonstrated that histones are subjected to post-transcriptional modifications

1791Pieterse E, et al. Ann Rheum Dis 2018;77:1790–1798. doi:10.1136/annrheumdis-2018-213223

Basic and translational research

at their N-terminal regions during NOX-dependent NET forma-tion,15–17 we explored whether the use of antihistone antibodies directed against N-terminal histone tails could be utilised in MPO-based ELISAs in order to discriminate between NOX-de-pendent or NOX-independent NETs.

MATerIAls And MeTHOdsA more elaborate description of the methods is described in the online Supplementary methods.

Isolation of neutrophilsNeutrophils were isolated by Ficoll density gradient centrifuga-tion as described previously.10

Characterisation of in vitro generated neTsNeutrophils were stimulated with 50 nM PMA or 10 µg/mL of LPS from Pseudomonas aeruginosa (Sigma-Aldrich) to generate NOX-dependent NETs. NOX-independent NETs were generated as described previously.10 NETs were either directly visualised and characterised by immunofluorescence microscopy or isolated with 5 U/mL micrococcal nuclease for further charac-terisation with ELISAs.

Preparation of cell extracts, sodium dodecyl sulfate polyacrylamide gel electrophoresis (sds PAGe) and Western blottingPreparation of cell extracts, SDS PAGE and Western blotting was performed as described previously.18

Anti-histone antibodiesA panel of lupus-derived monoclonal antibodies consisting of #34, KM-2, LG11-2 and BT164 was used to detect N-terminal histone tails in NOX-dependent and NOX-independent NETs.

endothelial cell experimentsHuman umbilical vein endothelial cells were exposed to NOX-dependent or NOX-independent NETs up to 24 hours. Functional responses were determined by immunofluores-cence microscopy or quantitative real time-PCR (qRT-PCR) analyses.

Human materialAll analyses of patient and healthy volunteer material were performed in accordance with institutional guidelines. Demo-graphical and clinical characteristics of study cohorts are in the online Supplementary information.

statistical analysesValues are expressed as mean±SEM. Significance was either determined by Student’s t-test or one-way analysis of variance followed by Bonferroni correction using GraphPad Prism V.5.0 (La Jolla, USA). P values less than 0.05 were considered as statis-tically significant.

resulTsn-terminal histone tails are lacking in nOX-dependent neTsIt was recently described that neutrophil elastase exerts proteo-lytic activity on core histones during NOX-dependent NET formation.19 As the N-terminal regions of histones appear to be particularly sensitive to degradation by proteases,20 we hypoth-esised that histones present within NOX-dependent NETs may lack N-terminal tails as the result of their proteolysis. To test this

hypothesis, we stimulated neutrophils with LPS or PMA to form NOX-dependent NETs and subsequently analysed these NETs by immunofluorescence microscopy for their reactivity with anti-bodies against N-terminal histone tails. After 2 hours of stimula-tion, reactivity of antihistone antibodies against N-terminal tails was maintained or enhanced in LPS-exposed or PMA-exposed neutrophils, respectively (online Supplementary figure 1A). Although the chromatin was clearly decondensated at this time point, NET-like structures were not observed yet. At a later time point (4 hours), NET-like extracellular stretches of DNA could be clearly identified in response to both stimuli. However, at this time point, our panel of monoclonal antibodies against N-ter-minal histone tails had lost reactivity with NETs (figure 1A,B, left panel; online Supplementary figure 1A). In contrast, a poly-clonal antibody reacting with the C-terminal region of all core histones showed clear reactivity with NETs at this time point (figure 1A,B, right panel). Additionally, we used necrostatin-1, an inhibitor of mixed lineage kinase domain-like protein pore formation that has been suggested to play a role in NET forma-tion, although conflicting results have been reported in the liter-ature.21 22 Nevertheless, we observed that necrostatin-1 prevents lysis of neutrophils exposed to LPS and could prevent the loss of N-terminal histone tails in neutrophils (online Supplemen-tary figure 1B). Necrostatin-1 clearly arrested neutrophils at the stage of maximum intracellular chromatin decondensation, where reactivity of antibodies against N-terminal histone tails was preserved. Altogether, these results show that histones are present within NOX-dependent NETs, but that N-terminal histone tails disappear when neutrophils lyse and release their NETs extracellularly.

The loss of n-terminal histone tails in nOX-dependent neTs is mediated by extracellular activity of neutrophil elastaseTo assess whether neutrophil elastase is responsible for the loss of N-terminal histone tails in NOX-dependent NETs, we co-incubated neutrophils with sivelestat or alpha1-anti-trypsin while stimulating them with LPS or PMA for 4 hours. Sivelestat and alpha1-antitrypsin are cell impermeable and therefore inhibit the proteolytic activity of neutrophil elas-tase extracellularly.23 24 Indeed, in the presence of both elas-tase inhibitors, reactivity of NOX-dependent NETs with antihistone antibodies against N-terminal tails was preserved (figure 1C,D; online Supplementary figure 2A, second and third panel). GW311616A, a potent cell-permeable inhibitor of elastase, fully blocked NOX-dependent NET formation (online Supplementary figure 2A, fourth panel), confirming earlier work that intracellular elastase activity is required for NOX-dependent NET formation.19 Indeed, elastase translo-cates to the nucleus during early stages of NOX-dependent NET formation; yet, the reactivity against N-terminal histone antibodies is preserved and only disappears after cell lysis (online Supplementary figure 3). Inhibition of matrix metal-loproteinases or cathepsins with batimastat or chymostatin, respectively, did not prevent loss of N-terminal histone tails (online Supplementary figure 2A, fifth and sixth panel). The proteolytic cleavage of N-terminal histone tails after NOX-de-pendent NET formation was also confirmed by Western blot analyses. It appeared that cleaved histones are predominantly present in neutrophils 3 and 4 hours after PMA stimulation, that is, after cells lyse and release chromatin extracellularly, whereas platelet-induced NET formation did not result in histone cleavage (online Supplementary figure 2B). Alto-gether, these data indicate that neutrophil elastase mixes with

1792 Pieterse E, et al. Ann Rheum Dis 2018;77:1790–1798. doi:10.1136/annrheumdis-2018-213223

Basic and translational research

Figure 1 Histones in NOX-dependent NETs lack N-terminal histone tails. (A) Representative images (n=3 donors) showing that reactivity of an antibody against histone H2B-K12Ac is lost during NOX-dependent NET formation by LPS (left panel), while a pan antibody against C-terminal regions of all core histones remains reactive (right panel). Additional stainings are included in the online Supplementary figure 1. (B) Quantification of reactivity of antihistone antibodies during NOX-dependent NET formation induced by PMA (blue triangles; n=2) or LPS (green squares; n=3). Reactivity was quantified after 120 and 240 min of stimulation. (C) Representative images (n=3 donors) showing that the presence of sivelestat, an inhibitor of elastase, preserves reactivity of the anti-H2B-K12Ac antibody with NOX-dependent (LPS induced) NETs. Additional stainings with other protease inhibitors are included in the online Supplementary figure 2. (D) Effects of elastase inhibitors (sivelestat and hAAT) on reactivity of antihistone antibodies with NOX-dependent NETs induced by PMA (blue triangles; n=2) or LPS (green squares; n=2). ***P<0.001, **P<0.01, *P<0.05 PMA compared with control, ###P<0.001, ##P<0.01, #P<0.05 LPS compared with control. Scale bar=20 µm. hAAT, human alpha1-antitrypsin; LPS, lipopolysaccharide; NET, neutrophil extracellular trap; PMA, phorbol myristate acetate.

chromatin to facilitate early stages of NOX-dependent NET formation,19 but that the proteolysis of N-terminal histone tails by neutrophil elastase is mainly occurring extracellularly, thus after NET release.

n-terminal histone tails remain present in nOX-independent neTsNext, we explored whether histones present within NOX-in-dependent NETs also lack their N-terminal tails. To evaluate this, we exposed neutrophils to activated platelets and anal-ysed these NETs by immunofluorescence microscopy for their reactivity with antibodies against the N-terminal histone tails.

In contrast to NOX-dependent NETs, NOX-independent NETs showed a strong reactivity with our panel of antihistone antibodies against N-terminal tails (figure 2A). NOX-inde-pendent NETs appeared within 1 hour after stimulation with activated platelets (online Supplementary figure 4A). At this time point, stimulation of neutrophils with PMA or LPS had not yet resulted in chromatin decondensation or NET release, thereby confirming an earlier report that NOX-independent NET formation is a much faster process than NOX-depen-dent NET formation.25 Platelets activated by LPS, ADP or thrombin appeared equally potent in inducing fast NOX-in-dependent NET formation (online Supplementary figure

1793Pieterse E, et al. Ann Rheum Dis 2018;77:1790–1798. doi:10.1136/annrheumdis-2018-213223

Basic and translational research

Figure 2 N-terminal histone tails are present in NOX-independent NETs. (A) Activated platelets induced rapid NET formation after 1 hour of stimulation, whereas exposure to LPS does not cause NET release at this timepoint. Antibodies recognising N-terminal histone tails show strong reactivity against platelet-induced NETs (representative of n=3 donors). (B) The presence of N-terminal histone tails in isolated NETs, induced by PMA (4 hour), LPS (4 hour) or activated platelets (1 hour) (n=5 donors), was quantified by ELISA in which NETs were coated as antigen. The reactivity of our panel of antihistone antibodies was plotted as a ratio to the reactivity of a monoclonal antibody recognising NET-associated DNA. Green lines represent reactivity of our antibody panel with 1 µg/mL of apoptotic chromatin, which was obtained as described in Papayannopoulos et al.19. (C) Stimulation of neutrophils with LPS or activated platelets yields NETs containing MPO. Inhibition of NOX with DPI prevents NET release by LPS, but not by activated platelets (representative of n=3 donors). (D) The presence of MPO in isolated NETs, induced by either PMA, LPS or activated platelets (n=4 donors), was quantified in the same way as outlined for panel B. (E) NOX-independent NETs (platelets), but not NOX-dependent NETs (PMA, LPS) can be measured in sandwich ELISAs employing antihistone antibodies against N-terminal tails and anti-MPO as detecting antibody (n=4 donors). Obtained reactivity of these ELISAs was plotted as a ratio compared with the reactivity of an ELISA with anti-DNA as capturing antibody. (F) Supernatants from neutrophils undergoing NET formation (LPS, PMA) but not those from neutrophils undergoing apoptotic (4NQO 4 hour), secondary necrotic (4NQO 30 hours) or necrotic cell death (S. aureus) contain soluble MPO–DNA complexes (n=2 donors). ***P<0.001, **P<0.01, *P<0.05 compared with PMA and LPS in (B) and (E) and compared with v in (F). Scale bar=20 µm. DPI, diphenyleneiodonium; hAAT, human alpha1-antitrypsin; LPS, lipopolysaccharide; MPO, myeloperoxidase; 4NQO, 4-nitroquinoline 1-oxide; NET, neutrophil extracellular trap; PMA, phorbol myristate acetate.

4B). Reactivity of NOX-independent NETs with antihistone antibodies was preserved up to 4 hours after stimulation with activated platelets (online Supplementary figure 5A). In addition to platelets, exposure of neutrophils to the calcium ionophore A23187, a reported stimulus of NOX-indepen-dent NET formation,21 26 yielded NETs recognised by anti-histone antibodies directed against N-terminal tails (online Supplementary figure 5B). To mimic the presence of protease inhibitors in the extracellular environment, we also induced NET formation with PMA-activated and LPS-activated plate-lets in the presence of 10% autologous serum. In line with the previous reports,27 we observed a general reduction in the number of NETs in the presence of serum; however, the presence of serum did not essentially alter the difference in staining of N-terminal histone tails between the NOX-depen-dent and NOX-independent pathway (online Supplementary figure 6). To more precisely quantify the differences in reac-tivity of antihistone antibodies against N-terminal tails within NOX-independent and NOX-dependent NETs, we applied an ELISA in which NETs were coated as antigens after which the reactivity of our antibody panel was evaluated. This ELISA revealed the low and high reactivity of our antihistone anti-body panel with, respectively, NOX-dependent and NOX-in-dependent NETs (figure 2B). In summary, NOX-independent NETs contain N-terminal histone tails in contrast to NOX-de-pendent NETs.

nOX-independent neTs are detected in sandwich elIsAs employing anti-MPO and antihistone antibodies directed against n-terminal tailsIn order to allow discrimination between NOX-independent and NOX-dependent NETs in biological fluids, we aimed to establish a sandwich ELISA employing anti-MPO and anti-histone antibodies directed against N-terminal tails. For this purpose, we first validated the presence of MPO within both NOX-dependent and NOX-independent NETs. Immuno-fluorescence microscopy confirmed that MPO was present within both NOX-dependent and NOX-independent NETs (figure 2C, online Supplementary figure 7), with NET-as-sociated MPO being equally present in NOX-dependent and NOX-independent NETs when corrected for the total amount of NET-associated DNA (figure 2D). To confirm that the observed NETs were derived from either NOX-de-pendent or NOX-independent pathways, we also stimulated neutrophils in the presence of the NOX inhibitor diphenyl-eneiodonium (DPI), which confirmed that activated platelets induced NOX-independent NET formation whereas both PMA and LPS triggered NOX-dependent NET formation (figure 2C). Next, we employed sandwich ELISAs using our panel of antihistone antibodies against N-terminal tails or a commercially available pan anti-histone antibody reacting with the C-terminal regions of all core histones as capturing antibodies and anti-MPO as detecting antibody. Complexes

1794 Pieterse E, et al. Ann Rheum Dis 2018;77:1790–1798. doi:10.1136/annrheumdis-2018-213223

Basic and translational research

Figure 3 Different forms of NETs in sepsis and autoimmune disease. (A) Circulating NETs in patients with sepsis (n=10) versus HC (n=10). Cut-off values for positivity in either the MPO–DNA, MPO–H2BAc or MPO–H4Ac ELISA was calculated by the mean value+3× SD of the HC group. Positivity in all three ELISAs was classified as NI-NETs, whereas sole positivity in the MPO–DNA ELISA was classified as ND-NETs. (B) Kinetics of in vivo NET formation in two healthy volunteers undergoing experimental endotoxemia. (C) The 10% pooled plasma of the two healthy volunteers from graph B was used for ex vivo NET induction in healthy donor neutrophils. Plasma sampled at time point 90 min after LPS challenge induced NETs, in contrast to plasma sampled 60 min prior to LPS challenge. Of note, these NETs stained positive for H2B-K12Ac (yellow arrows), indicating NOX-independent NET formation. (D) Circulating NETs in patients with PsA (n=20), RA (n=20) and SLE (n=20) were characterised in MPO–DNA ELISAs (X-axis) and MPO–H2BAc ELISAs (Y-axis) and based on cut-off values for positivity (red lines), NETs were classified as being derived from NI or ND pathways; some samples were unclassifiable (MPO–DNA negative, MPO–H2BAc positive). Scale bar=20 µm. HC, healthy controls; LPS, lipopolysaccharide; MPO, myeloperoxidase; ND-NET, NOX-dependent NETs; NET, neutrophil extracellular trap; NI-NETs, NOX-independent NETs; PsA, psoriatic arthritis; RA, rheumatoid arthritis; SLE, systemic lupus erythematosus; Uncl, unclassifiable.

consisting of MPO and histones with intact N-terminal tails could only be detected in case of NOX-independent NETs (figure 2E). In contrast, the pan antibody against C-terminal histone regions showed similar reactivity for both NOX-de-pendent and NOX-independent NETs. Importantly, other neutrophil cell death pathways such as apoptosis, secondary necrosis and Stahylococcus aureus toxin-induced necrosis, did not result in the detection of MPO–DNA complexes, indi-cating that the detection of soluble MPO–DNA complexes most likely is specific for NET formation (figure 2F). In conclusion, the detection of MPO–histone complexes in a sandwich ELISA with antihistone antibodies against N-ter-minal tails as capturing antibodies allows measurement of NOX-independent, but not NOX-dependent NETs.

Circulating NETs in patients with septic shock, rheumatoid arthritis (RA) and systemic lupus erythematosus (SLE) are derived from NOX-independent pathwaysBeing able to discriminate between NETs derived through NOX-dependent or NOX-independent pathways, we next assessed whether we could detect NOX-dependent or NOX-independent NETs in blood plasma obtained from patients with prototype NET-associated diseases; that is, septic shock,28 RA,13 psoriatic arthritis (PsA)29 and SLE.30 We first measured circulating MPO–DNA complexes and then evaluated whether positivity in the MPO–DNA ELISA

coincided with positivity in additional ELISAs employing anti-histone antibodies directed against N-terminal tails (H2B-K12Ac and/or H4-K8,12,16Ac). Positivity in all ELISAs was classified as NET formation through NOX-inde-pendent pathways, whereas positivity in only the MPO–DNA ELISA was classified as NOX-dependent NET formation. We first validated this approach in patients with septic shock, as sepsis can be perceived as a positive control for NOX-in-dependent NET formation.31 Positivity in the MPO–DNA ELISA was found in 7/10 patients with sepsis. All patients except one were also positive in the MPO–H2B-K12Ac and MPO–H4-K8,12,16Ac ELISAs, thus indeed revealing NETs derived from NOX-independent pathways (figure 3A). One patient was borderline positive in the MPO–DNA ELISA, but did not reach positivity in the MPO–H2B-K12Ac or MPO–H4-K812,16Ac ELISAs, and these NETs were thus classified as being derived through a NOX-dependent pathway. As NOX-dependent and NOX-independent NET formation also differ in their kinetics of NET release, we evaluated the kinetics of NET formation in two healthy volunteers enrolled in a human endotoxemia trial, a systemic inflammatory model partially mimicking sepsis. A rapid extrusion of NETs (ie, within 1 hour after intravenous LPS challenge) hereby supports a NOX-independent pathway to be involved. Indeed, NETs were detectable in plasma sampled at 30 and 60 min after LPS administration, reaching

1795Pieterse E, et al. Ann Rheum Dis 2018;77:1790–1798. doi:10.1136/annrheumdis-2018-213223

Basic and translational research

Figure 4 Stimulatory potential of NOX-independent NETs on endothelial cells. (A) Assessment of nuclear translocation of NFκB p65 after 24 hours of exposure to LPS (positive control), NI-NETs and ND-NETs (n=3 donors). The percentage of endothelial cells with nuclear positivity for NFκB p65 (illustrated by red arrows) was scored in three independent fields and plotted in the right panel. (B) Quantitative real time-PCR analyses reveals the upregulation of adhesion markers (ICAM-1, VCAM-1) and cytokines (IL-8, IL-6) particularly in response to NI-NETs (n=3 donors). (C) Representative images (n=3 donors) confirming upregulation of ICAM-1 at the protein level after 24 hours exposure to NI-NETs. ***P<0.001, **P<0.01, *P<0.05 compared with control if not specified otherwise. Scale bar=20 µm. ICAM-1, intercellular adhesion molecule-1; IL, interleukin; LPS, lipopolysaccharide; NFκB, nuclear factor kappa-light-chain-enhancer of activated B cells; ND-NETs, NOX-dependent NETs; NET, neutrophil extracellular trap; NI-NETs, NOX-independent NETs; VCAM-1, vascular cell adhesion molecule-1.

its peak at 90 min (figure 3B). After 90 min, the concentra-tion of NETs declined again. Pooled plasma from these two LPS-challenged volunteers was also capable of inducing NET formation in healthy donor neutrophils ex vivo. Of note, the extruded NETs stained positive with our antibody against the N-terminal histone tail of H2B, thereby confirming NOX-in-dependent NET release (figure 3C). Having validated that our ELISAs are suitable for the characterisation of in vivo circulating NETs, we subsequently analysed whether circu-lating NETs in RA, PsA and SLE patients are derived from NOX-dependent or NOX-independent pathways. Positivity in the MPO–DNA ELISA was found in 8/20 patients with PsA (40%), 14/20 patients with RA (70%) and 6/20 patients with SLE (30%). Whereas all patients with PsA except one were negative in the MPO–H2B-K12Ac ELISA, patients with RA and SLE were mainly positive in the MPO–H2B-K12Ac ELISA (figure 3D). The detection of NOX-independent NETs in SLE plasma is in line with earlier reports showing that low-density granulocytes (LDGs) present in SLE patients

undergo spontaneous NOX-independent NET formation.32 Indeed, spontaneously formed NETs from LDGs of patients with SLE had a strong reactivity with our antibody panel against N-terminal histone tails (online Supplementary figure 8). Taken together, NETs in PsA seem derived from a NOX-dependent pathway, whereas NETs in RA and SLE are generated through a NOX-independent pathway. Thus, these data show that NETs are detectable in the circulation of patients with sepsis or autoimmune disease, but that the molecular pathway through which they originate (NOX-de-pendent vs NOX-independent) differs between diseases.

nOX-dependent and nOX-independent neTs differ in their endothelial cell stimulatory potentialTo demonstrate that the ability to discriminate between NOX-de-pendent and NOX-independent NETs is relevant, we finally performed functional assays in which the potential of NOX-de-pendent and NOX-independent NETs to activate endothelial

1796 Pieterse E, et al. Ann Rheum Dis 2018;77:1790–1798. doi:10.1136/annrheumdis-2018-213223

Basic and translational research

Figure 5 Schematic representation of distinction between NETs that have been generated through NOX-dependent or NOX-independent pathways of NET formation using antibodies recognising N-terminal histone tails. The formation of ROS during NOX-dependent NET formation leads to the activation of neutrophil elastase that subsequently degrades N-terminal histone tails within NETs (upper panel). During NOX-independent NET formation, neutrophil elastase remains proteolytically inactive, thereby yielding NETs that are recognised by antibodies directed against N-terminal histone tails (bottom panel). MPO, myeloperoxidase; NET, neutrophil extracellular trap; ROS, reactive oxygen species.

cells in vitro was evaluated. To this end, endothelial cells were exposed to NOX-dependent and NOX-independent NETs at a concentration of 200 ng/mL. NOX-independent NETs induced the nuclear translocation of cytosolic NFκB p65 (figure 4A) and increased mRNA levels of adhesion markers (intercellular adhesion molecule 1 (ICAM-1), vascular cell adhesion mole-cule-1 (VCAM-1)), as well as proinflammatory cytokines (inter-leukin (IL)-8, IL-6) (figure 4B). Of note, the increased ICAM-1 mRNA expression in endothelial cells exposed to NOX-indepen-dent NETs was also confirmed at the protein level (figure 4C). NOX-dependent NETs had significantly less immunostimulatory effects on endothelial cells compared with NOX-independent NETs. Collectively, these data further highlight the relevance of the ability to discriminate between NOX-dependent and NOX-independent NETs, since they show different immuno-stimulatory potential at disease-relevant concentrations.

dIsCussIOnThis study shows that the predominant protease within NETs, neutrophil elastase, cleaves N-terminal histone tails from NETs formed via a NOX-dependent pathway, a phenomenon that does not occur during NOX-independent NET formation. This observation was exploited to design ELISAs for the detection of MPO–histone complexes that allowed the discrimination between NETs derived from NOX-dependent or NOX-inde-pendent pathways. Whereas NOX-independent NETs could be captured in such ELISAs, NOX-dependent NETs were not, as these were negative for N-terminal histone tails.

We have previously shown that elastase has proteolytic activity in NOX-dependent NETs, but not in NOX-independent NETs.10 Why the proteolytic activity of elastase differs between NOX-de-pendent and NOX-independent NETs is not completely clear, but is likely explained by the differential requirement of NOX-de-rived ROS during both pathways of NET formation. Hence, it was previously demonstrated that antioxidants prevent elastase activity on exposure of neutrophils to PMA.33 The degrada-tion of core histones during NOX-dependent NET formation was previously described and was shown to be required for NOX-dependent NETs to be released.19 In addition to intra-cellular histone cleavage during the process of NET formation, our data indicate that histone cleavage continues after NETs have been expelled. Hence, the loss of N-terminal histone tails only occurred after neutrophils had been lysed to release their NET. Experiments with sivelestat and alpha1-antitrypsin, which both inhibit extracellular elastase activity only,34 35 support this conclusion as they both prevented the loss of N-terminal tails of NET-associated histones. We speculate that the extracellular elas-tase-mediated degradation of N-terminal histone tails in NETs may function to potentiate the antimicrobial activity of NET-as-sociated histones,36 reduce their cytotoxicity to host cells37 or facilitate the timely removal of NETs by nucleases.38 The loss of N-terminal histone tails in NOX-dependent NETs may also explain their reduced immunostimulatory effects on endothelial cells compared with NOX-independent NETs; however, other explanations including differences in (total or modified) protein content15 or NET conformation10 could also account for this.

To our knowledge, evidence supporting the in vivo existence of NOX-independent NET formation has been exclusively described in the pathogenic context of sepsis.31 Intriguing in vivo imaging during sepsis has shown that neutrophils become lodged to the sinusoids of the liver, where activated platelets interact with neutrophils to induce NET release independent from NOX-mediated neutrophil lysis.39 Direct platelet–neutrophil

interactions do not per se seem to be required for NOX-in-dependent NET formation as is illustrated by the finding that plasma from patients with sepsis induced NOX-independent NET formation in the absence of platelets in our ex vivo exper-iments. The presence of soluble P-selectin or platelet-derived microparticles, generated during platelet activation in vivo, may explain NOX-independent NET formation in response to septic plasma ex vivo.40 41 Similar to sepsis, our data suggest that in vivo formed NETs in patients with RA and SLE are generated through a similar NOX-independent pathway. Indeed, platelets have an activated phenotype in both RA and SLE.42–44 Further-more, besides platelets, other circulating factors in patients with RA and SLE (eg, immune complexes or apoptotic cell-derived microparticles) have previously been shown to induce NOX-in-dependent NET formation ex vivo.26 45 Notably, depending of the nature of immune complexes, also NOX-dependent NET formation may occur.46 The fact that circulating NETs in SLE appear to have been generated through a NOX-independent pathway may explain why NOX deficiency, potentially reducing NOX-dependent NET formation, does not ameliorate disease activity in MRL/lpr mice,47 nor in a pristane-induced lupus model.48 Similarly, the finding that individuals with chronic granulomatous disease, that is, patients who fail to produce ROS through NOX, have a propensity of developing SLE has ques-tioned the involvement of NOX-dependent NET formation in SLE.49–51

It is worth noting that not all patients within the screened cohorts were positive for circulating NETs. For instance, in case of SLE, only 6 out of 20 patients were positive for circulating NETs. Since all SLE patients were treated with immunosup-pressive agents, had low disease activity (SLE Disease Activity Index <4) and did not have signs of lupus nephritis, these data imply that a direct causal relationship between circulating NETs and disease activity cannot be made. Similarly, 3 out of 10 patients with septic shock were negative for circulating NETs, while all patients were severely ill and required vasopressor treatment with norepinephrine. Altogether, these data suggest that NETs do

1797Pieterse E, et al. Ann Rheum Dis 2018;77:1790–1798. doi:10.1136/annrheumdis-2018-213223

Basic and translational research

not per se contribute to disease activity or that additional factors are required to turn NETs into pathogenic structures. As this was not the aim of our study, these observations warrant further investigation with larger cohorts containing more heterogeneity in terms of disease activity.

In summary, we conclude that N-terminal histone tails are lacking in NOX-dependent, but not in NOX-independent NETs due to a different activity of neutrophil elastase in both conditions. Therefore, we propose that the use of antibodies recognising N-terminal histone tails in MPO-based ELISAs can discriminate between NETs that have been generated through either NOX-dependent or NOX-independent NET formation pathways (figure 5). Identifying the involved pathways through which NETs emerge in different disease contexts will facilitate the identification of appropriate molecular targets for future therapeutics to inhibit NET formation.

Author affiliations1department of nephrology, radboud Institute for Molecular Life Sciences, radboud University Medical Center, nijmegen, the netherlands2department of Intensive Care Medicine, radboud Institute for Molecular Life Sciences, radboud University Medical Center, nijmegen, the netherlands3department of Internal Medicine 3, rheumatology and Immunology, Friedrich-Alexander-University Erlangen-nürnberg (FAU), Universitätsklinikum Erlangen, Erlangen, Germany

Contributors Ep, nr and CY performed the experiments, interpreted the data and wrote the manuscript. JG, SB, pp, LM and MH provided patient material and wrote the manuscript. LH and JV supervised the research, interpreted the data and wrote the manuscript.

Funding this research was supported by the radboud Institute of Molecular Life Sciences (rIMLS) phd program and the department of nephrology of the radboud University Medical Center.

Competing interests none declared.

Patient consent not required.

ethics approval Ethical committees of the radboud University Medical Center and the University Hospital Erlangen.

Provenance and peer review not commissioned; externally peer reviewed.

RefeRences 1 Brinkmann V, reichard U, Goosmann C, et al. neutrophil extracellular traps kill

bacteria. Science 2004;303:1532–5. 2 Brinkmann V, Zychlinsky A. neutrophil extracellular traps: is immunity the second

function of chromatin? J Cell Biol 2012;198:773–83. 3 Kaplan MJ, radic M. neutrophil extracellular traps: double-edged swords of innate

immunity. J Immunol 2012;189:2689–95. 4 Yang H, Biermann MH, Brauner JM, et al. new insights into neutrophil extracellular

traps: mechanisms of formation and role in inflammation. Front Immunol 2016;7:302. 5 Jorch SK, Kubes p. An emerging role for neutrophil extracellular traps in noninfectious

disease. Nat Med 2017;23:279–87. 6 Stoiber W, obermayer A, Steinbacher p, et al. the role of reactive oxygen species

(roS) in the formation of extracellular traps (Ets) in humans. Biomolecules 2015;5:702–23.

7 Yipp BG, Kubes p. nEtosis: how vital is it? Blood 2013;122:2784–94. 8 Yipp BG, petri B, Salina d, et al. Infection-induced nEtosis is a dynamic process

involving neutrophil multitasking in vivo. Nat Med 2012;18:1386–93. 9 Clark Sr, Ma AC, tavener SA, et al. platelet tLr4 activates neutrophil extracellular

traps to ensnare bacteria in septic blood. Nat Med 2007;13:463–9. 10 pieterse E, rother n, Yanginlar C, et al. neutrophils discriminate between

lipopolysaccharides of different bacterial sources and selectively release neutrophil extracellular traps. Front Immunol 2016;7:484.

11 Caudrillier A, Kessenbrock K, Gilliss BM, et al. platelets induce neutrophil extracellular traps in transfusion-related acute lung injury. J Clin Invest 2012;122:2661–71.

12 nakazawa d, Shida H, tomaru U, et al. Enhanced formation and disordered regulation of nEts in myeloperoxidase-AnCA-associated microscopic polyangiitis. J Am Soc Nephrol 2014;25:990–7.

13 Sur Chowdhury C, Giaglis S, Walker UA, et al. Enhanced neutrophil extracellular trap generation in rheumatoid arthritis: analysis of underlying signal transduction pathways and potential diagnostic utility. Arthritis Res Ther 2014;16:r122.

14 Wang H, Sha LL, Ma tt, et al. Circulating level of neutrophil extracellular traps is not a useful biomarker for assessing disease activity in antineutrophil cytoplasmic antibody-associated vasculitis. PLoS One 2016;11:e0148197.

15 Knight JS, Carmona-rivera C, Kaplan MJ. proteins derived from neutrophil extracellular traps may serve as self-antigens and mediate organ damage in autoimmune diseases. Front Immunol 2012;3:380.

16 Liu CL, tangsombatvisit S, rosenberg JM, et al. Specific post-translational histone modifications of neutrophil extracellular traps as immunogens and potential targets of lupus autoantibodies. Arthritis Res Ther 2012;14:r25.

17 pieterse E, Hofstra J, Berden J, et al. Acetylated histones contribute to the immunostimulatory potential of neutrophil extracellular traps in systemic lupus erythematosus. Clin Exp Immunol 2015;179:68–74.

18 dieker JW, Fransen JH, van Bavel CC, et al. Apoptosis-induced acetylation of histones is pathogenic in systemic lupus erythematosus. Arthritis Rheum 2007;56:1921–33.

19 papayannopoulos V, Metzler Kd, Hakkim A, et al. neutrophil elastase and myeloperoxidase regulate the formation of neutrophil extracellular traps. J Cell Biol 2010;191:677–91.

20 Zhou p, Wu E, Alam HB, et al. Histone cleavage as a mechanism for epigenetic regulation: current insights and perspectives. Curr Mol Med 2014;14:1164–72.

21 Kenny EF, Herzig A, Krüger r, et al. diverse stimuli engage different neutrophil extracellular trap pathways. Elife 2017;6.

22 desai J, Kumar SV, Mulay Sr, et al. pMA and crystal-induced neutrophil extracellular trap formation involves rIpK1-rIpK3-MLKL signaling. Eur J Immunol 2016;46:223–9.

23 Sun Z, Yang p. role of imbalance between neutrophil elastase and α1-antitrypsin in cancer development and progression. Lancet Oncol 2004;5:182–90.

24 Zeiher BG, Matsuoka S, Kawabata K, et al. neutrophil elastase and acute lung injury: prospects for sivelestat and other neutrophil elastase inhibitors as therapeutics. Crit Care Med 2002;30:S281–S287.

25 pilsczek FH, Salina d, poon KK, et al. A novel mechanism of rapid nuclear neutrophil extracellular trap formation in response to Staphylococcus aureus. J Immunol 2010;185:7413–25.

26 Kraaij t, tengström FC, Kamerling SW, et al. A novel method for high-throughput detection and quantification of neutrophil extracellular traps reveals roS-independent nEt release with immune complexes. Autoimmun Rev 2016;15:577–84.

27 Kamoshida G, Kikuchi-Ueda t, nishida S, et al. Spontaneous formation of neutrophil extracellular traps in serum-free culture conditions. FEBS Open Bio 2017;7:877–86.

28 Camicia G, pozner r, de Larrañaga G. neutrophil extracellular traps in sepsis. Shock 2014;42:286–94.

29 Hu SC, Yu HS, Yen FL, et al. neutrophil extracellular trap formation is increased in psoriasis and induces human β-defensin-2 production in epidermal keratinocytes. Sci Rep 2016;6:31119.

30 Yu Y, Su K. neutrophil extracellular traps and systemic lupus erythematosus. J Clin Cell Immunol 2013;4.

31 Ma AC, Kubes p. platelets, neutrophils, and neutrophil extracellular traps (nEts) in sepsis. J Thromb Haemost 2008;6:415–20.

32 Lood C, Blanco Lp, purmalek MM, et al. neutrophil extracellular traps enriched in oxidized mitochondrial dnA are interferogenic and contribute to lupus-like disease. Nat Med 2016;22:146–53.

33 Iwamura H, Moore Ar, Willoughby dA. Interaction between neutrophil-derived elastase and reactive oxygen species in cartilage degradation. Biochim Biophys Acta 1993;1156:295–301.

34 Hagio t, Matsumoto S, nakao S, et al. Sivelestat, a specific neutrophil elastase inhibitor, prevented phorbol myristate acetate-induced acute lung injury in conscious rabbits. Pulm Pharmacol Ther 2005;18:285–90.

35 Sun Z, Yang p. role of imbalance between neutrophil elastase and alpha 1-antitrypsin in cancer development and progression. Lancet Oncol 2004;5:182–90.

36 Kawasaki H, Iwamuro S. potential roles of histones in host defense as antimicrobial agents. Infect Disord Drug Targets 2008;8:195–205.

37 xu J, Zhang x, pelayo r, et al. Extracellular histones are major mediators of death in sepsis. Nat Med 2009;15:1318–21.

38 dhaenens M, Glibert p, Meert p, et al. Histone proteolysis: a proposal for categorization into ’clipping’ and ’degradation’. Bioessays 2015;37:70–9.

39 Mcdonald B, Urrutia r, Yipp BG, et al. Intravascular neutrophil extracellular traps capture bacteria from the bloodstream during sepsis. Cell Host Microbe 2012;12:324–33.

40 dieker J, tel J, pieterse E, et al. Circulating apoptotic microparticles in systemic lupus erythematosus patients drive the activation of dendritic cell subsets and prime neutrophils for nEtosis. Arthritis Rheumatol 2016;68:462–72.

41 Etulain J, Martinod K, Wong SL, et al. p-selectin promotes neutrophil extracellular trap formation in mice. Blood 2015;126:242–6.

42 Joseph JE, Harrison p, Mackie IJ, et al. Increased circulating platelet-leucocyte complexes and platelet activation in patients with antiphospholipid syndrome, systemic lupus erythematosus and rheumatoid arthritis. Br J Haematol 2001;115:451–9.

43 Lood C, tydén H, Gullstrand B, et al. platelet activation and anti-phospholipid antibodies collaborate in the activation of the complement system on platelets in systemic lupus erythematosus. PLoS One 2014;9:e99386.

1798 Pieterse E, et al. Ann Rheum Dis 2018;77:1790–1798. doi:10.1136/annrheumdis-2018-213223

Basic and translational research

44 pamuk GE, Vural o, turgut B, et al. Increased platelet activation markers in rheumatoid arthritis: are they related with subclinical atherosclerosis? Platelets 2008;19:146–54.

45 rother n, pieterse E, Lubbers J, et al. Acetylated histones in apoptotic microparticles drive the formation of neutrophil extracellular traps in active lupus nephritis. Front Immunol 2017;8:1136.

46 Garcia-romo GS, Caielli S, Vega B, et al. netting neutrophils are major inducers of type I IFn production in pediatric systemic lupus erythematosus. Sci Transl Med 2011 3:73ra20.

47 Campbell AM, Kashgarian M, Shlomchik MJ. nAdpH oxidase inhibits the pathogenesis of systemic lupus erythematosus. Sci Transl Med 2012;4:157ra141.

48 Kienhöfer d, Hahn J, Stoof J, et al. Experimental lupus is aggravated in mouse strains with impaired induction of neutrophil extracellular traps. JCI Insight 2017;2.

49 Albarrán-planelles C, Jiménez-Gallo d, Linares-Barrios M, et al. [Lupus-like associated with chronic granulomatous disease]. Rev Clin Esp 2013;213:465–6.

50 Ben Abdallah Chabchoub r, turki H, Mahfoudh A. [Systemic lupus erythematosus in a boy with chronic granulomatous disease: case report and review of the literature]. Arch Pediatr 2014;21:1364–6.

51 xie C, Cole t, McLean C, et al. Association between discoid lupus erythematosus and chronic granulomatous disease – report of two cases and review of the literature. Pediatr Dermatol 2016;33:e114–e120.

1799Qi Y, et al. Ann Rheum Dis 2018;77:1799–1809. doi:10.1136/annrheumdis-2018-213028

Basic and translational research

ExtEndEd rEport

Increased autophagy is cytoprotective against podocyte injury induced by antibody and interferon-α in lupus nephritisYuan-yuan Qi ,1 xu-jie Zhou ,1 Fa-juan Cheng ,1 ping Hou ,1 Ya-li ren ,2 Su-xia Wang ,2 Ming-hui Zhao ,1 Li Yang ,1 Jennifer Martinez ,3 Hong Zhang 1

To cite: Qi Y, Zhou x, Cheng F, et al. Ann Rheum Dis 2018;77:1799–1809.

Handling editor Josef S Smolen

► Additional material is published online only. to view please visit the journal online (http:// dx. doi. org/ 10. 1136/ annrheumdis- 2018- 213028).

1renal division, peking University First Hospital; peking University Institute of nephrology; Key Laboratory of renal disease, Ministry of Health of China; Key Laboratory of Chronic Kidney disease prevention and treatment (peking University), Ministry of Education, Beijing, China2Laboratory of Electron Microscopy, Ultrastructural pathology Center, peking University First Hospital, Beijing, China3Immunity, Inflammation, and disease Laboratory, nIEHS, national Institutes of Health, research triangle park, Carolina, USA

Correspondence todr xu-jie Zhou, renal division, peking University First Hospital; peking University Institute of nephrology; Key Laboratory of renal disease, Ministry of Health of China; Key Laboratory of Chronic Kidney disease prevention and treatment (peking University), Ministry of Education, Beijing 100034, China; zhouxujie@ bjmu. edu. cn

received 16 January 2018revised 25 July 2018Accepted 19 August 2018published online First 12 September 2018

© Author(s) (or their employer(s)) 2018. no commercial re-use. See rights and permissions. published by BMJ.

Key messages

What is already known about this subject? ► More recent studies suggested that defects in autophagy contribute to the pathogenesis of SLE, especially in adaptive immunity.

► Occurrence and progression of lupus nephritis (LN) is the end result of complex interactions between regulation of immune responses and pathological process by renal resident cells, but there is still a lot of missing information for an establishment on the role of autophagy in pathogenesis of LN and as a therapy target.

What does this study add? ► This study demonstrates that autophagy is activated in LN, especially in podocyte.

► In vitro assays, many of the most important mediators of the disease—patients’ sera, patients’IgG and IFN-α —could induce autophagy in both murine and human podocytes.

► Autophagy activation negatively associated with podocyte injury.

► In intervention, autophagy activator could protect podocyte injury whereas its inhibitor aggravated it.

How might this impact on clinical practice or future developments?

► As autoantibodies and IFN play an important role in a variety of cell populations, autophagy may also be relevant for biomarker, translational and therapeutic studies in several immune related diseases.

► New insights into the molecular mechanisms of podocyte autophagy in LN may shed promising therapeutic strategies.

ABsTrACTObjective More recent studies suggested that defects in autophagy contribute to the pathogenesis of SLE, especially in adaptive immunity. occurrence and progression of lupus nephritis (Ln) is the end result of complex interactions between regulation of immune responses and pathological process by renal resident cells, but there is still a lot of missing information for an establishment on the role of autophagy in pathogenesis of Ln and as a therapy target.Methods Systemic and organ-specific aetiologies of autophagy were first evaluated by autophagy protein quantification in tissue homogenates in MrLlpr/

lpr lupus prone and female C57BL mice. Analysis of gene expression was also adopted in human blood and urine sediments. then, some key mediators of the disease, including complement inactivated serum, IgG from patients with Ln (IgG-Ln) and interferon (IFn)-α were chosen to induce podocyte autophagy. podocyte injuries including apoptosis, podocin derangement, albumin filtration and wound healing were monitored simultaneously with autophagy steady-state and flux.results Elevated LC3B in kidney homogenates and increased autophagosomes in podocyte from MrLlpr/

lpr were observed. In humans, mrnA levels of some key autophagy genes were increased in blood and urinary sediments, and podocyte autophagosomes were observed in renal biopsies from patients with Ln. Complement inactivated serum, IgG-Ln and IFn-α could induce podocyte autophagy in a time-dependent and dosage-dependent manner, and by reactive oxygen species production and mtorC1 inhibition, respectively. Autophagy inhibition aggravated podocyte damage whereas its inducer relieved the injury.Conclusion podocyte autophagy is activated in lupus-prone mice and patients with lupus nephritis. Increased autophagy is cytoprotective against antibody and interferon-α induced podocyte injury.

Systemic lupus erythematosus (SLE) is a spectrum of autoimmune disease in which defects can occur at multiple points of the immune cascade, resulting in a striking heterogeneity of clinical presenta-tions.1–4 The central pathology is the recognition of, and reaction to, self-antigens by the immune system, leading to type I interferon signalling and the production of pathogenic autoantibodies, which mediate much of the disease.1–4

More recent genetic and cellular studies suggested that defects in the evolutionarily

conserved catabolic pathway of autophagy contrib-uted to the pathogenesis of SLE, especially in the adaptive immune response.5–12 Hypothesis-free genome-wide association studies and further follow-up studies revealed that variants locating in autophagy-related genes associated with suscepti-bility to SLE.13–22 Increased ATG5, MAP1LC3B and BECN1 expressions were observed in blood of patients with lupus.13 18 23 And exposure to serum from patients with active SLE in SH-SY5Y cells was associated with increased formation of autophagic vacuoles.24 It was observed that

1800 Qi Y, et al. Ann Rheum Dis 2018;77:1799–1809. doi:10.1136/annrheumdis-2018-213028

Basic and translational research

autophagy was upregulated in B cells during plasma cell differentiation in SLE,25 and macrophages from patients with SLE also exhibited increased levels of autophagy.26 In a toll-like receptor 7 transgenic mouse model of lupus, ATG5-de-pendent B-cell autophagy was necessary for the induction of lupus-like disease.27 In addition, autophagy in T cells from both lupus-prone mice and patients with SLE was deregulated and not distributed homogeneously.28–30 More recently, it was observed that, in transgenic mouse models, disruption of LC3-associated phagocytosis (LAP) resulted in defective clear-ance of exogenous dead cells, leading to increased serum levels of anti-dsDNA antibodies, anti-nuclear antigens and inflam-matory cytokines—all of which were hallmarks of SLE.10 31 However, to what extent autophagy contributes to the organ damage in human SLE remains obscure.

Lupus nephritis (LN) is one of the most serious mani-festations of SLE and is an important predictor of poor outcome. Occurrence and progression of LN is the end result of complex interactions between regulation of immune responses and pathological process involved by renal resi-dent cells. Similar to other diseases, autophagy is also impli-cated in the pathological process of renal diseases. Studies from renal cells in culture, human kidney tissues and exper-imental animal models implicate that autophagy regulates many critical aspects of normal and disease conditions in the kidney.32–35 By specific gene knockout mice, it is indicated that constitutive and induced autophagy is a major protec-tive mechanism against podocyte ageing and glomerular injury.36 Also, LAP-deficient animals displayed indications of kidney damage and exhibited increased functional markers of kidney injury,10 such as increased serum creatinine, blood urea nitrogen and proteinuria. And variants of APOL1 and MTMR3, key drivers in autophagy, have been associated with LNs.37–39

Rapamycin/sirolimus, as an inhibitor of TOR, would acti-vate autophagy and was observed to prolong survival, main-tain normal renal function, normalise proteinuria, restore protein levels of podocyte-specific markers including nephrin and podocin, reduce anti-dsDNA titres, ameliorate histolog-ical lesions, and reduce Akt and mammalian target of rapa-mycin (mTOR) activation in LN.11 40–45 But there is still a lot of missing information necessary for a full establishment on the role of autophagy in pathogenesis of LN and as a therapy target. Thus, the aim for the current study is to reveal the role and function of autophagy in kidney pathology in LN, checking whether modulating autophagic activity in certain cell types may be therapeutically beneficial.

MeTHOdsPatients and healthy donorsAll the 44 patients, recruited from Peking University First Hospital, fulfilled the American College of Rheumatology revised criteria for the classification of SLE and were diag-nosed with LN by renal biopsy. A total of 37 volunteer controls were reported to be healthy by medical examination.

Quantification of autophagy-related gene expressionThe quantifications of autophagy-related gene expressions were conducted using both peripheral blood mononuclear cells (PBMCs, 13 patients with LN and 12 healthy donors) and urinary sediments (16 patients with LN and 15 healthy donors) (online supplementary methods and table 1).

Purification of IgG from patient seraIgG were purified from sera of five patients with LN and five healthy donors by protein G affinity chromatography (GE Healthcare).

MiceTen female MRLlpr/lpr lupus-prone mice and 10 control female C57BL mice were acquired from Nanjing Medical Univer-sity. Thymus, spleen, inguinal lymph nodes and kidneys were harvested at 6 months.46 Lupus nephritis was proven by pathology (online supplementary figure 1). All the proce-dures were approved by the Institutional Animal Care and Use Committee of Peking University First Hospital.

eleCTrOn MICrOsCOPyKidney tissue for electron microscopy was fixed in 3% glutar-aldehyde and 1% osmium tetroxide (OsO4), followed by dehy-dration in graded ethanol series and washed with acetone, and finally embedded in Epon 812. Ultrathin sections were stained with uranyl acetate and lead citrate and examined with a trans-mission electron microscope (JEM-1230; JEOL, Japan).

A total of 20 human renal biopsy specimens were examined by light microscopy, immunofluorescence (IF) microscopy and transmission electron microscopy (TEM). The average number of autophagosomes per podocyte was counted.

Cell culture and treatmentsThe immortalised human podocyte cell line was a gift from Moin A Saleem47 and the mouse podocyte was purchased from Cell Resource Centre of Peking Union Medical College (3111C0001CCC000230). Both human and mouse podocytes were cultured and differentiated as described in online supple-mentary methods.

Podocytes were maintained and exposed to complement inac-tivated sera from patients with LN,24 purified IgG from patients with LN, purified IgG from healthy donors, and commercial IgG (Sigma, I4506) and interferon (IFN)-α 2 (PBL Interferon Source, 11 105-1) for human podocyte or IFN-α (PBL Interferon Source, 12 100-1) for mouse podocyte. For autophagy inhibition, 5 mM 3-methyladenine (3-MA, M9281; Enzo) and 20 µM bafilomycin A1 (BafA1, BML-CM110; Enzo) were used. And for autophagy induction, 400 µM rapamycin (BML-A275; Enzo) was added.

lentiviral vector infectionThe immortalised human podocyte cell line was seeded at 1×105 cells/well in six-well plates and infected with lentivirus vector ATG5 shRNA (Hanbio, Shanghai, China) for 12 hours after 30 min incubation with polybrene. Then, the media was changed regularly and experiments were performed after 10 days in differentiation condition. An empty lentiviral vector (Hanbio) was used as the negative control (online supplementary figure 2).

Western blottingThe primary antibodies against SQSTM1/p62 protein (5114; CST), LC3B (L7543; Sigma) and ACTB/β-actin (sc-47778; Santa Cruz), podocin (ab50339; Abcam), RPS6KB (2708P; CST), phospho-RPS6KB (Thr389, 9205S; CST), RPS6 (2317S; CST), phospho-RPS6 (Ser240/244, 2215S; CST), EIF4B (3592P), phospho-EIF4B (Ser422, 3591; CST) and the horseradish perox-idase–conjugated secondary antibodies (sc-2004 and sc-2005; Santa Cruz) were used.

1801Qi Y, et al. Ann Rheum Dis 2018;77:1799–1809. doi:10.1136/annrheumdis-2018-213028

Basic and translational research

Immunocytochemical stainingIndirect immunofluorescence staining was performed as described previously.48 49 The primary antibodies included 1:100 anti-LC3B (L7543; Sigma) or 1:100 podocin (ab50339; Abcam). Images were captured by IF microscopy (Zeiss, Oberkochen, Germany).

mrFP-GFP-lC3To analyse autophagy flux, cells were infected with dual-tagged LC3 (mRFP-GFP-LC3) adenovirus (Hanbio Biotechnology) according to the manufacturer’s instructions. GFP and mRFP dots/cell and the ratio of mRFP/GFP were used for quantitation of autophagy flux, and 20 cells were counted in each group.

Flow cytometry assayPodocyte apoptosis was determined by flow cytometry using an Annexin V-PE apoptosis detection kit (559763; BD Biosciences, USA) following the manufacturer’s protocol. Then 10 000 cells were analysed by FACScan flow cytometer (BD Biosciences) to analyse cellular apoptosis. The levels of intracellular reactive oxygen species (ROS) were determined by 2′, 7′-dichlorofluo-rescin diacetate (Sigma) assay.

In vitro ‘wound healing’ assayIt is assumed that foot process retraction is a migration event caused by podocyte injury.50 Podocytes were thus seeded in six-well tissue culture dishes 10 days in 37°C for differentia-tion. At the time point 0 hour, the cell monolayer was wounded. Afterward, the cell culture medium was replaced by test medium containing either standard medium or medium supple-mented with IFN-α with 3-MA or rapamycin or shATG5, or IFN-α, 3-MA, rapamycin, shATG5 alone. The wounds closure were observed at 12 hours by phase contrast microscopy and documented by photography. Each condition was examined in triplicates.

Albumin influx assayThe filtration barrier function of podocyte monolayer was eval-uated by a simple albumin influx assay as described previously.51

sTATIsTICAl AnAlysIsAll the analysis was done using SPSS V.14.0. All the results are presented as mean±SD. Two group comparisons were assessed by two-tailed t-test. Values of p<0.05 were considered statisti-cally significant.

(Please refer to online supplementary methods for detailed information).

resulTselevated lC3 quantification in kidney homogenates of Mrllpr/

lpr miceWe first examined the LC3B protein quantification in the thymus, spleen, inguinal lymph node and kidney tissue homogenate from 6-month-old lupus-prone MRLlpr/lpr mice and C57BL6/J mice. It was observed that LC3B-II was decreased in MRLlpr/lpr spleen homogenate but increased in fraction of lymph nodes (online supplementary figure 3). No significant difference was observed in the thymus. Five (50%) MRLlpr/lpr murine lupus mice were histologically proven with glomerulonephritis at 6 months, and the level of LC3B-II significantly increased in kidney tissue homogenate from MRLlpr/lpr mice with nephritis, compared with MRLlpr/lpr mice without nephritis or C57BL6/J mice (figure 1A).

Increased frequency of observed autophagosome in Mrllpr/lpr podocyteAutophagy level detected from homogenates of the whole organ included organelles from infiltrated cells as well as from renal resident cells. We thus further directly checked the presence of autophagosomes in renal resident cells using TEM (online supplementary figure 4). It was observed that autophagic vacu-oles were present in podocytes instead of endothelial cells, mesangial cells or tubular epithelial cells. And the observation was present in four out of five MRLlpr/lpr mice with nephritis, but absent in MRLlpr/lpr mice without renal impairment or C57BL6/J mice (figure 1B).

Increased autophagy mrnA levels in PBMC and urinary sediment from human patients with lnCompared with healthy controls, 31 out of 117 Kyoto Encyclo-paedia of Genes and Genomes/KEGG autophagy genes (26%) were differentially expressed in PBMC from patients with LN. Twenty-three genes were significantly upregulated and eight genes were significantly downregulated in patients with LN (figure 1C, p<0.05; details in online supplementary table 2). MTMR3 (myotubularin-related protein 3) (case vs controls, 7.55±0.12 vs 7.12±0.26; p=1.75×10−5) and RRAGD (Ras-re-lated GTP binding D) (8.58±0.30 vs 7.56±0.36; p=7.28×10−8) showed the most significance and can survive from multiple testing (p threshold 0.05/117=4.27×10−4) (figure 1D).

Urinary sediment analysis indicated significantly increased mRNA levels of ULK1 (p=0.035), BECN1 (p=0.021) and MAP1LC3B (p=0.04) (figure 1E) in patients with LN compared with healthy controls. But ATG5, ATG12 and ATG16 showed non-significantly differed expression (data not shown).

Observed podocyte autophagosome in renal biopsy from human patients with lnAutophagosomes in podocyte were further analysed by TEM in 20 patients with LN (each five patients were selected in every subgroup of types II, III, IV and V according to 2003 ISN/RPS classification of LN). Podocyte autophagosomes could be observed in 17 out of 20 patients (85%) with LN (figure 1F). And only three patients showed no podocyte autophagosome and were all belonging to subclass II. However, the number of autophagosomes per podocyte (ranged from 0 to 9) correlated with no clinical data (data not shown).

sera from patients with ln could induce podocyte autophagy in vitroWe hypothesised that a serum factor present with LN may induce podocyte autophagy. To test for the presence of such a factor, we incubated human podocytes in the presence of complement inac-tivated sera from healthy donors or LN. It indicated that auto-phagy was stimulated by sera from patients with LN compared with sera from healthy donors (1:10 dilution) (figure 1G). As sera contained multiple mediators in disease pathogenesis, we then focused on IFN-α and IgG from patients with LN (IgG-LN).

IFn-α could induce podocyte autophagy in vitroIFN-α, a central cytokine in the pathogenesis of SLE, can induce and accelerate SLE tissue damage in patients and mice. It remained unknown whether IFN-α can induce autophagy in human and murine podocytes. We observed that treatment with IFN-α induced autophagy in both human and murine podo-cytes, as indicated by an increase of LC3B-II and a decrease of p62 expression. And this kind of induction was time and dose

1802 Qi Y, et al. Ann Rheum Dis 2018;77:1799–1809. doi:10.1136/annrheumdis-2018-213028

Basic and translational research

Figure 1 Upregulated autophagy in MRLlpr/lpr mice, patients with lupus nephritis (LN) and in podocytes treated by sera from patients with LN. (A) Western blot analysis of autophagy marker LC3B was shown with long and short exposure in kidney tissue homogenate from MRLlpr/lpr mice. Quantification of LC3B-II levels relative to ACTB or LC3B-I was present with mean and range. (B) A representative human podocyte autophagosome by transmission electron micrographs (TEMs) is shown (scale bar 1 µm, 200 nm and enlarged image). (C) Heatmap of genes in autophagy pathway (pathway: hsa04140) expression in 13 patients with LN and 12 healthy donors. A total of 128 genes were included in autophagy pathway according to the Kyoto Encyclopaedia of Genes and Genomes in which 11 genes were not included in the Illumina HT-12 v4 Expression BeadChip. Among them, MTMR3 and RRAGD showed the most significance (D). (E) Scatter plot from selected autophagy gene expressions in peripheral blood mononuclear cells and urinary sediments (data were expressed relative to their respective 18S expression with means±SEM). (F) A representative human podocyte autophagosome by TEM is shown (scale bar 2 µm and 0.5 µm). (G) Sera from patients with LN could induce podocyte autophagy in vitro. Densitometry analysis of LC3B-II levels relative to GAPDH is also shown. Values are expressed as mean. P value is indicated in the figure.

dependent (figure 2A–D). Our results also showed enhanced LC3-II and p62 expression levels in IFN (−) podocytes from 6 h to 48 h, indicating that autophagic vacuole accumulation was partially attributed to impaired autophagic degradation.

By indirect immunofluorescence, a large number of GFP-LC3B puncta formed in the cytoplasm and the average number of GFP-LC3B puncta per cell was significantly increased after

exposure to IFN-α (figure 2E–H) compared with untreated cells. And by checking autophagic flux by GFP-RFP-LC3 system, there were more GFP+RFP+LC3 puncta (autophagosomes) as well as GFP-RFP+LC3 puncta (autophagolysosomes) in podocytes under IFN-α treatment than in untreated controls, indicative of accumulation of LC3. When co-treated with the lysosomal inhibitor (bafilomycin A) to block autophagic flux, more

1803Qi Y, et al. Ann Rheum Dis 2018;77:1799–1809. doi:10.1136/annrheumdis-2018-213028

Basic and translational research

Figure 2 Autophagy was induced by interferon (IFN)-α in podocyte. (A–D) Western blot analysis of autophagy markers LC3B and p62 in human and mouse podocytes after exposure to 1000 U/mL IFN-α at various time points and after exposure to different concentrations of IFN-α for 48 hours. (E–H) Immunofluorescence staining and quantative changes of LC3B (green) in human and mouse podocytes after exposure to 1000 U/mL IFN-α for 48 hours. Quantitation of autophagy vacuoles (green dots) were shown in histogram with mean±SEM from three experiments. GFP-LC3B displayed a diffuse cytoplasmic distribution in the absence of autophagy. In contrast, after the induction of autophagy, LC3B forms autophagic membranes and can be observed as visible puncta representing autophagic vacuoles. (I and K) Fluorescence imaging of mRFP-GFP-LC3 acquired by confocal microscopy and (J and L) quantitation of dots per cell and mRFP:GFP ratio in podocytes with or without bafilomycin A1 treatment. The GFP signal (green) can be quenched in a lysosomal environment; in contrast, the RFP signal (red) is more stable in an acidic environment. Therefore, autophagosomes (GFP+RFP+LC3 puncta) were labelled with yellow (merge from green and red) and autolysosomes (GFP-RFP+LC3 puncta) red colour. By detecting and analysing the two different fluorescent signals, the autophagic flux can be monitored. Scale bar 20 000 nm. *p<0.05. **p<0.01. (M–P) Western blotting of LC3B-II with co-treatment of bafilomycin A and IFN-α, compared with single or no treatment (untreated vs BafA1; IFN-a vs IFN-a +BafA1). **p<0.01. HPC, human podocyte; MPC, murine podocyte.

autophagosomes (GFP+RFP+LC3) were observed (figure 2I–L). By Western blotting, it was observed LC3B-II protein quantifica-tion was significantly increased with co-treatment of bafilomycin A and IFN-α, compared with single (IFN-α treatment alone or bafilomycin A alone) or no treatment. It suggested that the treat-ment increased the synthesis of autophagy-related membranes instead of autophagic flux block (figure 2M–P).

IgG isolated from patients with ln could induce podocyte autophagy in vitroBecause most of the pathogenic autoantibodies were of IgG isotype, to examine the likely link between autoantibodies and the induction of autophagy in podocytes, IgGs were isolated. IgG from the sera of human patients with LN (IgG-LN), enrolled healthy donors (IgG-Control) and commercial immunoglob-ulin from healthy donors were adopted to stimulate podocytes in vitro. Podocytes exposed to IgG-LN displayed significantly elevated LC3B-II levels compared with exposure to IgG-Con-trol and commercial IgG. This increase occurred in both a time-dependent and dose-dependent manner, peaking at 6 hours post-treatment (figure 3A–F). Simultaneously, the levels of p62 decreased in a time-dependent and dose-dependent manner in IgG-LN-treated podocytes. It was thus indicative of potent induction of autophagy.

Indirect immunofluorescence experiments also demonstrated an increased number of LC3-positive dots in podocytes with increased autophagic activity in response to IgG-LN. In untreated or IgG-Control-treated cells, only a few LC3 puncta were detected, but increased number of LC3-positive puncta could be observed after exposure to IgG-LN addition (figure 3G–H). Assays using GFP-RFP-LC3 further confirmed this. Podocytes treated with IgG-LN displayed increased mature GFP-RFP+autophagolyso-somes. An additive effect in increased levels of GFP+RFP+LC3 puncta (autophagosomes) in the presence of both IgG-LN and lysosomal inhibition (with bafilomycin A1) indicated that IgG-LN enhanced autophagic flux (figure 3I–J). Increased levels of LC3-II in the presence of lysosomal inhibition were also confirmed by Western blotting, which increased the confidence that an induc-tion of autophagic process was observed (figure 3K).

IgG-ln regulated generation of rOs and podocyte apoptosisROS production was observed to be considerably increased in podocytes treated with IgG-LN in a dose-dependent manner (figure 3L). To study the effect of IgG-LN on podocyte apop-tosis in vitro, podocytes were further treated with IgG-LN. Flow cytometry revealed that apoptosis rates significantly increased to threefold when treated with IgG-LN (13.9%) compared with no treatment (4.65%) (figure 3M–N).

1804 Qi Y, et al. Ann Rheum Dis 2018;77:1799–1809. doi:10.1136/annrheumdis-2018-213028

Basic and translational research

Figure 3 Autophagy induced by IgG extracted from patients with lupus nephritis played a protective role in podocyte apoptosis. (A–F) Western blot analysis of autophagy markers LC3B and p62 at various time points (100 µg/mL IgG) and at different IgG concentrations (exposure to IgG for 6 hours). (G–H) Immunofluorescence staining and quantative changes of LC3B (green) of untreated podocytes and after exposure to 100 μg/mL IgG from patients with lupus nephritis and healthy donors for 6 hours. Quantitation of autophagy vacuoles (green dots in G) were shown in histogram with mean±SEM from three experiments. (I–J) Fluorescence imaging of mRFP-GFP-LC3 acquired by confocal microscopy and quantitation of dots per cell and mRFP:GFP ratio in podocytes with or without bafilomycin A1 treatment. (K) Western blotting of LC3B-II with co-treatment of bafilomycin A and IgG-LN, compared with single or no treatment (untreated vs BafA1; IFN-a vs IFN-a+BafA1). **p<0.01>. (L) Reactive oxygen species levels in podocytes after treatment with various concentrations of IgG-LN for 6 hours by flow cytometric analysis and quantitation of geometric means were shown in histogram with mean±SEM from three experiments. (M–N) Representatives of Annexin V-PE/7-AAD double staining in different groups and quantative of apoptosis was shown in histogram (N) with mean±SEM from three experiments. Scale bar 20 000 nm. *p<0.05, **p<0.01, ***p<0.001.

IFn-α treatment linked mTOrC1 inhibition and podocyte damageIt was reported that type I interferon induced autophagy in certain human cancer cell lines involving mTORC1 signalling. The activity of mTORC1 can be monitored by phosphoryla-tion changes in downstream target proteins like RPS6KB and EIF4EBP1.

To examine the effect of IFN-α on mTORC1, human and mouse podocytes were treated with 1000 U for 6, 24 and 48 hours. We observed decreased phosphorylation of RPS6KB (Thr389) and decreased phosphorylation changes of two

RPS6KB-dependent proteins, RPS6 (Ser240/Ser242) and EIF4B (Ser422) (figure 4A–B), indicating that IFN-α attenuated mTORC1 signalling. Direct inhibition of mTORC1 activity can lead to induction of autophagy.

IFN-α treatment was not observed to affect cell apoptosis (online supplementary figure 5). Thus, maintenance of integrity of the structure and function of podocytes were further checked. By immunofluorescence staining, it was observed that podocin was in derangement (figure 4C–D). Expression of podocin, a podocyte-specific marker, was detected to be decreased in IFN-α using western blotting analysis (figure 4E). The scratch gap

1805Qi Y, et al. Ann Rheum Dis 2018;77:1799–1809. doi:10.1136/annrheumdis-2018-213028

Basic and translational research

Figure 4 Detection of mTORC1 activity and autophagy protected podocytes from interferon (IFN)-α-induced injury. (A–B) Western blot analysis of mTORC1 downstream phosphorylation of RPS6 (Ser240/Ser242), EIF4B (Ser422) and RPS6KB (Thr389) in human and mouse podocytes after exposure to 1000 U/mL IFN-α for 48 hours. (C–D) Immunofluorescence staining showed the localisation of podocin (green) in human and mouse podocytes after exposure to 1000 U/mL IFN-α for 48 hours. Scale bar 20 µm. (E) Western blotting of podocin under different treatments. (F) Podocyte monolayer on collagen-coated transwell filters and albumin permeability across podocyte monolayer was determined in human and mouse podocytes after exposure to 1000 U/mL IFN-α for 48 hours and was shown in histogram (G–H) with mean±SEM from three experiments. Scale bar 20 000 nm. *p<0.05, **p<0.01.

1806 Qi Y, et al. Ann Rheum Dis 2018;77:1799–1809. doi:10.1136/annrheumdis-2018-213028

Basic and translational research

dimension with IFN-α treatment for 12 hours was larger than with no treatment (online supplementary figure 6), implying some compromise of the migration capability by IFN-α treat-ment. To assess the functional consequence, we examined the filtration barrier function of podocyte by using a paracellular permeability influx assay, which measured the albumin flux rate across the differentiated podocyte monolayer. Differen-tiated podocytes were incubated with IFN-α for 12 hours and then subjected to albumin influx assay (figure 4F). As shown in figure 4G–H, addition of IFN-α increased albumin influx across the podocyte monolayer from 4.84±1.00 µg/mL to 8.14±0.32 µg/mL in the human podocyte and from 3.52±0.45 µg/mL to 6.43±0.96 µg/mL in the mouse podocyte. These results indi-cated that the filtration barrier function of podocytes was severely impaired by IFN-α.

Autophagy inhibition aggravated podocyte damage whereas its inducer relieved the injuryIt was important to demonstrate an effect resulting from inhi-bition or stimulation of autophagy. We thus applied chemical sequestration inhibitors of autophagy, including 3-MA and wortmannin, which inhibited class I phosphoinositide 3-kinases (PI3Ks) as well as class III PtdIns3Ks. As they were not entirely specific, some data were confirmed using RNA silencing of ATG5. On the contrary, the most commonly used inducer of autophagy was rapamycin, an allosteric inhibitor of mTORC1, and was adopted as autophagy activator.

In IgG-LN-induced podocyte apoptosis, addition of autophagy inhibitor further increased frequencies of podocyte apoptosis, from 13.9% with IgG-LN alone to 19.8% and 24.6% with addi-tion of wortmannin and 3-MA, respectively, whereas it seemed that addition of rapamycin could moderately decrease apop-tosis induced by IgG-LN, from 13.9% to 11.3% (figure 3M). It suggested that induction of autophagy partly protected podocyte from IgG-LN-induced apoptosis and the inhibition of autophagy may aggravate .

It was observed that both IFN-α treatment and shATG5 could decrease podocin quantification, whereas rapamycin could partially restore its expression in IFN-α-treated cells. Inhibition of autophagy by 3-MA made podocin derangement more pronounced. In contrast, upregulation of autophagy with rapamycin could ameliorate podocin derangement to some extent (figure 4C–D). Compared with untreated cells, all IFN-a, 3-MA and shATG5 12-hour treatments could weaken the scratch healing potential to some extent. Combi-nation of IFN-α and shATG5 and 3-MA made worse impaired podocyte migration. However, rapamycin can restore podo-cyte migration to some extent (online supplementary figure 6). In paracellular permeability influx assay, it was observed that inhibition of autophagy by 3-MA or shATG5 alone can significantly increase podocyte layer permeability, even more significant than addition of IFN-α. However, when combina-tional treatment was applied, a higher loss of podocyte layer permeability was observed. Inhibition of autophagy by adding 3-MA (16.00±0.80 µg/mL in human podocyte, 13.30±0.99 µg/mL in mouse podocyte) and shAtg5 (14.50±0.68 µg/mL in human podocyte) in IFN-α worsened podocyte barrier func-tion with increased albumin influx. Induction of autophagy by rapamycin (6.64±0.48 µg/mL in human podocyte, 4.34±0.60 µg/mL in mouse podocyte) ameliorated this kind of podocyte injury (figure 4G–H).

dIsCussIOnAutophagy, an intracellular degrading system, is highly depen-dent on lysosomes and has existed universally in eukaryotic cells.52 As a ‘nonselective’ degradation system for long-lived cytoplasmic proteins and dysfunctional organelles, it has been repeatedly confirmed to be actively involved in pathogenesis in SLE, especially in systemic immunity.10 25–28 30 36 38 53 54 The patho-genesis of LN involves pathomechanisms outside and inside the kidney. But there is a lack of investigation on autophagy inside the kidney in LN. Although some observational studies indicated autophagy involved in some other immune-related nephritis in recent years,35 36 inconsistent conclusions have been made. There have been studies that provide evidence to support a cyto-protective role of autophagy and others that support deleterious effects of autophagy. It is plausible that it is context dependent.

In the current study, we first checked whether autophagy is involved in pathogenesis of LN. The data indicated that auto-phagy is activated in LN, especially in podocyte. In lupus-prone MRLlpr/lpr mice, we observed that the levels of autophagy varied in different immune organs of MRLlpr/lpr comparing with C57BL mice. In kidney tissues, the levels of autophagy were significantly elevated in MRLlpr/lpr mice with renal injury comparing with MRLlpr/lpr mice without renal injury and C57BL control mice. In addition, in patients with LN, we observed that the expression of autophagy-related genes were increased in PBMC and urinary sediments. It reinforced the notion that both aberrant systemic immunity and intrarenal immunopathology contribute to lupus nephritis, in which autophagy may exert an important effect. However, as PBMC and urinary sediments were all mixtures of different cell types, they may be not optimal specimens, but they both shed light on autophagy upregulation, which was also consistent with all the previous reports in SLE.28–30 Isolation of blood lineage–specific lymphocytes and urine podocytes surely will be more illuminating in future non-invasive biomarker studies in the field of autophagy.

To understand the precise role of autophagy activation in renal injury response and the pathogenesis of LN, we further checked relations between autophagy induction and podocyte injury.

We observed that many of the most important mediators of the disease—patients’ sera, patients’ IgG and IFN-α—could induce autophagy in both murine and human podocytes. In general, baseline autophagy in mammalian systems occurs under normal conditions, but can be stimulated by various patholo-gies including by ischaemic, toxic, immunological and oxida-tive insults. Other autophagy stimulators are pharmacological agents like the macrolide antibiotic rapamycin, the best-known trigger for autophagy, which acts by inhibiting mTOR. Although previous studies showed that deregulated autophagy plays an essential role for the induction of SLE, the precise nature of the stimulus leading to increased autophagic activity in lupus is not defined. We checked the effect of key known mediators of lupus, including sera from patients, IgG-LN and IFN-α. Although complement was inactivated, sera contain multiple layers of mediators in disease pathogenesis, so we further focused on IgG from patients with LN (IgG-LN) and IFN-α. The produc-tion of pathogenic autoantibodies and type I interferon signal-ling are key pathogenic mechanisms in the development of SLE and LN.2–4 55 It was reported that autoimmune mice with systemic inflammation but impaired autoantibody production were protected from LN.55 It has been estimated that anti-dsDNA deposition in LN accounts for no more than 10%–20% of overall eluted IgG overall, implying that IgG not recognising DNA represents the vast majority of antibodies in glomeruli.

1807Qi Y, et al. Ann Rheum Dis 2018;77:1799–1809. doi:10.1136/annrheumdis-2018-213028

Basic and translational research

And it was reported that anti-podocyte antibodies in lupus nephritis were multiple and secondary events after autoantibody deposition included complement deposition, and activation of podocyte intracellular mediators such as phosphorylated protein kinase C and free oxygen radicals. And ROS played a key role in mediating autophagy formation in podocytes and determined podocyte depletion. We thus speculated that increased genera-tion of ROS was associated with the IgG-induced autophagy in podocytes. Exposure of NZB/W mice to IFN-α can accelerate pathogenic autoantibody production, proteinuria development and glomerular IgG deposition.56 And type I IFNs synthesised by resident renal cells (mesangial and endothelial cells) induced renal dysfunction, glomerulonephritis, crescent formation and tubulointerstitial nephritis.57

We observed that autophagy activation negatively associated with podocyte injury. Terminally differentiated podocytes are vulnerable to various kinds of injury, and the loss of podocytes is considered a key feature of progressive glomerular disease. Podo-cyte injury is responsible for proteinuria, and loss of podocytes by cell death or detachment is a critical step for the progression of glomerular diseases. In LN, podocytes are injured at the struc-tural and molecular level.58 In the study, autophagosomes were observed in the podocyte but not in other glomerular resident cells (such as mesangial cells or endothelial cells), suggesting the regulation and function of autophagy is likely cell type and context specific. In clinical situations, it is difficult to demonstrate autophagy clearly in tissues of formalin-fixed and paraffin-em-bedded biopsy samples retrospectively because tissues fixed in formalin have low or no LC3 detectable. More recently, LC3B or GFP-LC3 has been used to monitor autophagy through indirect IF or direct fluorescence microscopy. However, it was reported that IF assay pointing to LC3 could not differentiate cell types in kidney tissues.35 36 Thus, confirmatory data on involved cell types may warrant future transgenic mice study. We observed that autophagy levels negatively associated with podocyte apop-tosis, derangement of podocin, paracellular permeability influx and albumin influx, which were all indicators of podocyte injury. And we observed autophagosomes in all proliferative classes except from some patients with class II. As type class II was char-acterised by fewer immune deposits and less podocyte injury compared with classes III–V, which may be the underlying cause for the difference in which autophagy was cytoprotective against damage caused by IgG-LN. Some observational studies also suggested increased autophagosomes were detected in podocytes in FSGS, membranous nephropathy and IgA nephropathy.35 36 It thus highlighted the fundamental role of autophagy for the long-time maintenance of glomerular podocytes in LN and other immune-related glomerulonephritis. But we observed no clinical correlation. The reasons may be several. First, the occurrence of LN is the interaction outcome between systemic/local immunity and renal intrinsic mechanism by resident cells. Every patient has his own key pathophysiological pathway. Such is also the fact that no single histological lesion or score could precisely predict clinical findings. Second, to correlate every single histo-logical lesion with clinical parameters, it may need hundreds of patients. Lastly but not the least, the formation of autophago-somes is to protect podocyte against damage, which is a feed-back or compensatory mechanism instead of direct damage. The production of autoantibodies to self-antigens followed by formation of immune complexes in glomeruli was the core step in leading to nephritis in SLE. Serum IgG from patients with SLE could induce the apoptosis of endothelial cell59 and induce auto-phagy of T lymphocytes.29 And anti-dsDNA antibodies could bind to podocytes in vitro and in vivo, resulting in segmental

effacement of podocyte foot processes and proteinuria.60 61 Our data provided evidence that serum IgG from patients with SLE could induce apoptosis and also autophagy in podocyte. With the increase of the concentration of serum IgG from patients with SLE, more ROS produced and levels of LC3B increased in podocytes, suggesting that ROS mediated podocyte autophagy induction by IgG from patients with SLE. In contrast, consistent to previous report,62 we observed that IFN-α was not involved in podocyte loss, but impairment of cell structure and function. We observed that the induction of autophagy correlated with mTORC1 inhibition in human and mouse podocytes and that autophagy ameliorated IFN-α-induced podocyte injury. Taken together, our data indicated that activation of podocyte auto-phagy protected from cell injury at least under conditions of IFN and IgG-LN, which was in accordance with the data that LAP-deficient mice showed nephritis and loss of autophagy in podocytes resulted in a markedly increased susceptibility to glomerulosclerosis. Similarly, based on reported in vivo models of kidney injury, autophagy is often regarded as a renoprotec-tive factor in the development and progression of renal diseases. But a defect in autophagy regulation itself is not (yet) proven to be able to cause human kidney diseases. In this study, we also observed an association of autophagy with LN and then iden-tified some factors that could lead to its occurrence, in which autophagy played a protective role in feedback. This explains some mechanism of autophagy and its role in LN. Future inves-tigations, for instance by using cell-type-specific targeted auto-phagic gene knockout mice in LN, are necessary to elucidate and clarify the precise functional role of autophagy in renal injuries. As autoantibodies and IFN play an important role in a variety of cell populations, autophagy may also be relevant for biomarker, translational and therapeutic studies in several other diseases.

Last but not the least, in interventional study, we observed autophagy activator could protect podocyte injury whereas its inhibitor aggravated it. Rapamycin, also known as sirolimus, is considered to be non-nephrotoxic in the normal kidney. In most pathophysiological cases, rapamycin showed renal protec-tion, including decreasing proteinuria in lupus-prone mice.63 However, it may also produce renal injury, including increasing proteinuria in long-term therapy. The possible reason is, inhi-bition of mTOR activates autophagy, but prolonged starvation reactivates mTOR with degradation of autolysosomal products, and mTOR reactivation is required to regenerate functional lysosomes and completion of autophagic process.64 In Mtor pod-KO mice or prolonged rapamycin treatment, mTOR activity is blocked at both points.65 Thus, inhibition of mTOR activates autophagy, but mTOR reactivation is also blocked, resulting in disruption of autophagic flux. Development of de novo or worsening proteinuria is well recognised in patients with long-term use of rapamycin. Given that disruption of the autophagic pathway may play a role in the pathogenesis of proteinuria, therapy with mTOR inhibitors can be a double-edged sword with both favourable and unfavourable consequences. In our current assay, rapamycin was added in quite a short time period, and it supported its effects on autophagy activation and renal protection. Future more specific and selective regulators of the autophagic machinery (without off-target effects) are urgently needed. Intermittent upregulation of autophagy may be more effective with fewer side effects than chronic use of such strategy. And studies using targeted autophagic gene knockout mice or transgenic mice are necessary to elucidate and clarify the precise therapeutic role of autophagy in lupus nephritis.

We systemically checked autophagy in both lupus-prone mice and human patients, confirming the involvement of autophagy

1808 Qi Y, et al. Ann Rheum Dis 2018;77:1799–1809. doi:10.1136/annrheumdis-2018-213028

Basic and translational research

Figure 5 Schematic of the key observations of the current study.

in LN, as the exact effect that autophagy might exert in renal disease is multifaceted and complex, depending on different stimulations, types of disease or cells. We have thus further taken multiple mediators of the disease in checking their effect in two different podocyte cell lines and by combining different auto-phagy detection systems, different types of injuries and different drug interventions. Combining assays with multiple confirma-tion, we concluded that increased autophagy is cytoprotective against podocyte injury induced by antibody and interferon-α in LN. However, there were also some limitations of the current study. First, we checked different autophagy levels in lupus-prone mice; similar to previous reports,28 53 it was observed that different tissues showed different autophagy states. But different mice models, metabolic state and time of assay may all be confounding factors. Future gene knockout or transgenic mice study will be more informative in explaining cause effect and intervention value. Second, we observed only IgG from patients with LN could induce autophagy, but how differences in the type of autoantibodies or glycosylation affect autophagy is not clear. Third, an optimal condition and therapeutic window in which induction of autophagy would yield protective effects is still not clear. Fourth, whether autophagy levels could be monitored as disease biomarker is still an open question.

In the current study, we evaluated the levels of autophagy in kidney in MRLlpr/lpr mice and patients with LN and further explored the role of autophagy in podocyte injury of LN. Notably, we found that induction of autophagy (rapamycin) protected podocyte from IgG-LN-induced apoptosis and IFN-α induced derangement of podocin and increased albumin influx. Likewise, drugs acting as autophagy inducers such as rapamycin were shown to be beneficial in LN treatment. Our findings proved the protective role of autophagy in LN, especially in the podocyte (figure 5). New insights into the molecular mecha-nisms of podocyte autophagy in LN may shed promising ther-apeutic strategies.Acknowledgements We thank all the members of the laboratory for technical assistance. We also thank the patients, their families and healthy donors for their cooperation and for giving consent to participate in this study.

Contributors x-JZ ang HZ conceived the idea. Y-YQ, x-JZ, F-JC, pH, Y-Lr and S-xW conducted the assays and were involved in the acquisition and analysis of

data. Y-YQ, x-JZ, F-JC, LY, JM and HZ wrote the manuscript. Y-YQ, x-JZ, M-HZ, LY, JM and HZ revised the manuscript critically for important intellectual content. x-JZ supervised the research group and has given the final approval of the version to be published.

Funding this work was supported by grants from the national Science Foundation of China (no. 81570629), national Key research and development program of China (2016YFC0904102), the training program of the Major research plan of the national natural Science Foundation of China (91642120), Beijing nova program (Z171100001117023), Beijing Youth top-notch talent Support program (2017000021223ZK31), natural Science Foundation for Innovation research Group of China (81621092), Beijing natural Science Foundation (7152148), the University of Michigan Health System–peking University Health Science Center Joint Institute for translational and Clinical research (BMU2017JI007), and Chinese Society of nephrology (15020030591).

disclaimer the funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.

Competing interests none declared.

Patient consent obtained.

ethics approval Institutional review Board of the peking University First Hospital.

Provenance and peer review not commissioned; externally peer reviewed.

data sharing statement no additional unpublished data are available.

RefeRences 1 deng Y, tsao Bp. Updates in lupus genetics. Curr Rheumatol Rep 2017;19:68. 2 tsokos GC. Systemic lupus erythematosus. N Engl J Med 2011;365:2110–21. 3 Kaul A, Gordon C, Crow MK, et al. Systemic lupus erythematosus. Nat Rev Dis Primers

2016;2:16039. 4 tsokos GC, Lo MS, Costa reis p, et al. new insights into the immunopathogenesis of

systemic lupus erythematosus. Nat Rev Rheumatol 2016;12:716–30. 5 Wu dJ, Adamopoulos IE. Autophagy and autoimmunity. Clin Immunol

2017;176:55–62. 6 Zhou xJ, Zhang H. Autophagy in immunity: implications in etiology of autoimmune/

autoinflammatory diseases. Autophagy 2012;8:1286–99. 7 rockel JS, Kapoor M. Autophagy: controlling cell fate in rheumatic diseases. Nat Rev

Rheumatol 2016;12:517–31. 8 Mistry p, Kaplan MJ. Cell death in the pathogenesis of systemic lupus erythematosus

and lupus nephritis. Clin Immunol 2017;185:59–73. 9 Kemp MG, Apoptosis CB. Crosstalk between apoptosis and autophagy: environmental

genotoxins, infection, and innate immunity. J Cell Death 2017;9:1179670716685085. 10 Martinez J, Cunha Ld, park S, et al. noncanonical autophagy inhibits the

autoinflammatory, lupus-like response to dying cells. Nature 2016;533:115–9. 11 Gros F, Muller S. pharmacological regulators of autophagy and their link with

modulators of lupus disease. Br J Pharmacol 2014;171:4337–59. 12 pan Q, Gao C, Chen Y, et al. Update on the role of autophagy in systemic lupus

erythematosus: A novel therapeutic target. Biomed Pharmacother 2015;71:190–3.

1809Qi Y, et al. Ann Rheum Dis 2018;77:1799–1809. doi:10.1136/annrheumdis-2018-213028

Basic and translational research

13 Qi YY, Zhou xJ, nath SK, et al. A rare variant (rs933717) at FBxo31-MAp1LC3B in Chinese is associated with systemic lupus erythematosus. Arthritis Rheumatol 2018;70:287–97.

14 Zhang YM, Zhou xJ, Cheng FJ, et al. Autophagy-related gene LrrK2 is likely a susceptibility gene for systemic lupus erythematosus in northern Han Chinese. Oncotarget 2017;8:13754–61.

15 Zhou xJ, nath SK, Qi YY, et al. Brief report: identification of MtMr3 as a novel susceptibility gene for lupus nephritis in northern Han Chinese by shared-gene analysis with IgA nephropathy. Arthritis Rheumatol 2014;66:2842–8.

16 Molineros JE, Yang W, Zhou xJ, et al. Confirmation of five novel susceptibility loci for systemic lupus erythematosus (SLE) and integrated network analysis of 82 SLE susceptibility loci. Hum Mol Genet 2017;26:ddx026–1216.

17 Zhang YM, Cheng FJ, Zhou xJ, et al. detecting genetic associations between AtG5 and lupus nephritis by trans-eQtL. J Immunol Res 2015;2015:153132:1–7.

18 Zhou xJ, Lu xL, Lv JC, et al. Genetic association of prdM1-AtG5 intergenic region and autophagy with systemic lupus erythematosus in a Chinese population. Ann Rheum Dis 2011;70:1330–7.

19 Sun C, Molineros JE, Looger LL, et al. High-density genotyping of immune-related loci identifies new SLE risk variants in individuals with Asian ancestry. Nat Genet 2016;48:323–30.

20 Zhang YM, Cheng FJ, Zhou xJ, et al. rare variants of AtG5 Are likely to be associated with chinese patients with systemic lupus erythematosus. Medicine 2015;94:e939.

21 Schuster C, Gerold Kd, Schober K, et al. the autoimmunity-associated gene CLEC16A modulates thymic epithelial cell autophagy and alters t cell selection. Immunity 2015;42:942–52.

22 Yang W, tang H, Zhang Y, et al. Meta-analysis followed by replication identifies loci in or near CdKn1B, tEt3, Cd80, drAM1, and ArId5B as associated with systemic lupus erythematosus in Asians. Am J Hum Genet 2013;92:41–51.

23 Wu ZZ, Zhang JJ, Gao CC, et al. Expression of autophagy related genes mtor, Becline-1, LC3 and p62 in the peripheral blood mononuclear cells of systemic lupus erythematosus. Am J Clin Exp Immunol 2017;6:1–8.

24 towns r, Kabeya Y, Yoshimori t, et al. Sera from patients with type 2 diabetes and neuropathy induce autophagy and colocalization with mitochondria in SY5Y cells. Autophagy 2005;1:163–70.

25 Clarke AJ, Ellinghaus U, Cortini A, et al. Autophagy is activated in systemic lupus erythematosus and required for plasmablast development. Ann Rheum Dis 2015;74:912–20.

26 Li B, Yue Y, dong C, et al. Blockade of macrophage autophagy ameliorates activated lymphocytes-derived dnA induced murine lupus possibly via inhibition of proinflammatory cytokine production. Clin Exp Rheumatol 2014;32:705–14.

27 Weindel CG, richey LJ, Bolland S, et al. B cell autophagy mediates tLr7-dependent autoimmunity and inflammation. Autophagy 2015;11:1010–24.

28 Gros F, Arnold J, page n, et al. Macroautophagy is deregulated in murine and human lupus t lymphocytes. Autophagy 2012;8:1113–23.

29 Alessandri C, Barbati C, Vacirca d, et al. t lymphocytes from patients with systemic lupus erythematosus are resistant to induction of autophagy. Faseb J 2012;26:4722–32.

30 pierdominici M, Vomero M, Barbati C, et al. role of autophagy in immunity and autoimmunity, with a special focus on systemic lupus erythematosus. Faseb J 2012;26:1400–12.

31 Heckmann BL, Boada-romero E, Cunha Ld, et al. LC3-Associated phagocytosis and inflammation. J Mol Biol 2017;429:3561–76.

32 Wang Z, Choi ME. Autophagy in kidney health and disease. Antioxid Redox Signal 2014;20:519–37.

33 Huber tB, Edelstein CL, Hartleben B, et al. Emerging role of autophagy in kidney function, diseases and aging. Autophagy 2012;8:1009–31.

34 Anders HJ, Schlondorff do. Innate immune receptors and autophagy: implications for autoimmune kidney injury. Kidney Int 2010;78:29–37.

35 Hartleben B, Wanner n, Huber tB. Autophagy in glomerular health and disease. Semin Nephrol 2014;34:42–52.

36 Hartleben B, Gödel M, Meyer-Schwesinger C, et al. Autophagy influences glomerular disease susceptibility and maintains podocyte homeostasis in aging mice. J Clin Invest 2010;120:1084–96.

37 Freedman BI, Langefeld Cd, Andringa KK, et al. End-stage renal disease in African Americans with lupus nephritis is associated with ApoL1. Arthritis Rheumatol 2014;66:390–6.

38 Wang L, Law HK. the role of autophagy in lupus nephritis. Int J Mol Sci 2015;16:25154–67.

39 Iwamoto t, niewold tB. Genetics of human lupus nephritis. Clin Immunol 2017;185:32–9.

40 Stylianou K, petrakis I, Mavroeidi V, et al. the pI3K/Akt/mtor pathway is activated in murine lupus nephritis and downregulated by rapamycin. Nephrol Dial Transplant 2011;26:498–508.

41 reddy pS, Legault HM, Sypek Jp, et al. Mapping similarities in mtor pathway perturbations in mouse lupus nephritis models and human lupus nephritis. Arthritis Res Ther 2008;10:r127.

42 nagai K, tominaga t, Ueda S, et al. Mesangial cell mammalian target of rapamycin complex 1 activation results in mesangial expansion. J Am Soc Nephrol 2017;28:2879–85.

43 oaks Z, Winans t, Huang n, et al. Activation of the mechanistic target of rapamycin in SLE: explosion of evidence in the last five years. Curr Rheumatol Rep 2016;18:73.

44 Bride KL, Vincent t, Smith-Whitley K, et al. Sirolimus is effective in relapsed/refractory autoimmune cytopenias: results of a prospective multi-institutional trial. Blood 2016;127:17–28.

45 Zhou xJ, Cheng FJ, Zhang H. Emerging view of autophagy in systemic lupus erythematosus. Int Rev Immunol 2015;34:280–92.

46 Andrews BS, Eisenberg rA, theofilopoulos An, et al. Spontaneous murine lupus-like syndromes. Clinical and immunopathological manifestations in several strains. J Exp Med 1978;148:1198–215.

47 Saleem MA, o’Hare MJ, reiser J, et al. A conditionally immortalized human podocyte cell line demonstrating nephrin and podocin expression. J Am Soc Nephrol 2002;13:630–8.

48 Liu WJ, Luo Mn, tan J, et al. Autophagy activation reduces renal tubular injury induced by urinary proteins. Autophagy 2014;10:243–56.

49 Fang L, Zhou Y, Cao H, et al. Autophagy attenuates diabetic glomerular damage through protection of hyperglycemia-induced podocyte injury. PLoS One 2013;8:e60546.

50 reiser J, oh J, Shirato I, et al. podocyte migration during nephrotic syndrome requires a coordinated interplay between cathepsin L and alpha3 integrin. J Biol Chem 2004;279:34827–32.

51 rico M, Mukherjee A, Konieczkowski M, et al. Wt1-interacting protein and Zo-1 translocate into podocyte nuclei after puromycin aminonucleoside treatment. Am J Physiol Renal Physiol 2005;289:F431–F441.

52 Choi AM, ryter SW, Levine B. Autophagy in human health and disease. N Engl J Med 2013;368:651–62.

53 Wang F, Li B, Schall n, et al. Assessing autophagy in mouse models and patients with systemic autoimmune diseases. Cells 2017;6:E16:16.

54 Weindel CG, richey LJ, Mehta AJ, et al. Autophagy in dendritic cells and B cells is critical for the inflammatory state of tLr7-mediated autoimmunity. J Immunol 2017;198:1081–92.

55 Lorenz G, desai J, Anders HJ. Lupus nephritis: update on mechanisms of systemic autoimmunity and kidney immunopathology. Curr Opin Nephrol Hypertens 2014;23:211–7.

56 Mathian A, Weinberg A, Gallegos M, et al. IFn-alpha induces early lethal lupus in preautoimmune (new Zealand Black x new Zealand White) F1 but not in BALB/c mice. J Immunol 2005;174:2499–506.

57 Fairhurst AM, xie C, Fu Y, et al. type I interferons produced by resident renal cells may promote end-organ disease in autoantibody-mediated glomerulonephritis. J Immunol 2009;183:6831–8.

58 dos Santos M, poletti pt, Milhoransa p, et al. Unraveling the podocyte injury in lupus nephritis: Clinical and experimental approaches. Semin Arthritis Rheum 2017;46:632–41.

59 van paassen p, duijvestijn A, debrus-palmans L, et al. Induction of endothelial cell apoptosis by IgG antibodies from SLE patients with nephropathy: a potential role for anti-endothelial cell antibodies. Ann N Y Acad Sci 2007;1108:147–56.

60 Chan tM, Leung JK, Ho SK, et al. Mesangial cell-binding anti-dnA antibodies in patients with systemic lupus erythematosus. J Am Soc Nephrol 2002;13:1219–29.

61 Zhao Z, Weinstein E, tuzova M, et al. Cross-reactivity of human lupus anti-dnA antibodies with alpha-actinin and nephritogenic potential. Arthritis Rheum 2005;52:522–30.

62 Migliorini A, Angelotti ML, Mulay Sr, et al. the antiviral cytokines IFn-α and IFn-β modulate parietal epithelial cells and promote podocyte loss: implications for IFn toxicity, viral glomerulonephritis, and glomerular regeneration. Am J Pathol 2013;183:431–40.

63 Warner LM, Adams LM, Sehgal Sn. rapamycin prolongs survival and arrests pathophysiologic changes in murine systemic lupus erythematosus. Arthritis Rheum 1994;37:289–97.

64 Yu L, Mcphee CK, Zheng L, et al. termination of autophagy and reformation of lysosomes regulated by mtor. Nature 2010;465:942–6.

65 Cinà dp, onay t, paltoo A, et al. Mtor regulates autophagic flux in the glomerulus. Autophagy 2012;8:696–8.

1810 van den Hoogen LL, et al. Ann Rheum Dis 2018;77:1810–1814. doi:10.1136/annrheumdis-2018-213497

Basic and translational research

ConCise report

Galectin-9 is an easy to measure biomarker for the interferon signature in systemic lupus erythematosus and antiphospholipid syndromeLucas L van den Hoogen,1,2 Joël A G van roon,1,2 Jorre s Mertens,1,2,3 Judith Wienke,1 Ana pinheiro Lopes,1,2 Wilco de Jager,1 Marzia rossato,1,2,4 Aridaman pandit,1,2 Catharina G K Wichers,1,2 Femke van Wijk,1 ruth D e Fritsch-stork,2,5,6 timothy r D J radstake1,2

To cite: van den Hoogen LL, van roon JAG, Mertens Js, et al. Ann Rheum Dis 2018;77:1810–1814.

Handling editor Josef s smolen

► Additional material is published online only. to view please visit the journal online (http:// dx. doi. org/ 10. 1136/ annrheumdis- 2018- 213497).

For numbered affiliations see end of article.

Correspondence toDr Lucas L van den Hoogen, rheumatology and Clinical immunology, Universitair Medisch Centrum Utrecht, Utrecht 3508 GA, the netherlands; l. l. vandenhoogen@ umcutrecht. nl

rDeF-s and trDJr are joint senior authors.

received 29 March 2018revised 12 July 2018Accepted 2 August 2018published online First 5 september 2018

© Author(s) (or their employer(s)) 2018. no commercial re-use. see rights and permissions. published by BMJ.

Key messages

What is already known about this subject? ► The interferon signature is an imporant biomarker for disease activity and vascular disease in systemic lupus erythematosus and antiphospholipid syndrome.

► The interferon signature is assessed by gene-expression analysis which hampers its implication in clinical practice.

What does this study add? ► Galectin-9 correlates with disease activity and is a robust and easy to measure serum biomarker to detect the interferon signature in systemic lupus erythematosus and antiphospholipid syndrome.

► Galectin-9 outperforms CXCL-10 or traditional markers of disease activity to detect the interferon signature in systemic lupus erythematosus and antiphospholipid syndrome.

How might this impact on clinical practice or futue developments?

► Serum levels of galectin-9 may aid clinical decision making in steering anti-interferon targeted therapies in patients with systemic lupus erythematosus and antiphospholipid syndrome.

ABsTrACTObjective the interferon (iFn) signature is related to disease activity and vascular disease in systemic lupus erythematosus (sLe) and antiphospholipid syndrome (Aps) and represents a promising therapeutic target. Quantification of the iFn signature is currently performed by gene expression analysis, limiting its current applicability in clinical practice. therefore, the objective of this study was to establish an easy to measure biomarker for the iFn signature.Methods serum levels of galectin-9, CXCL-10 (ip-10) and tumour necrosis factor receptor type ii (tnF-rii) were measured in patients with sLe, sLe+Aps and primary Aps (pAps) and healthy controls (n=148) after an initial screening of serum analytes in a smaller cohort (n=43). Analytes were correlated to measures of disease activity and the iFn signature. the performance of galectin-9, CXCL-10 and tnF-rii as biomarkers to detect the iFn signature was assessed by receiver operating characteristic curves.results Galectin-9, CXCL-10 and tnF-rii were elevated in patients with sLe, sLe+Aps and pAps (p<0.05) and correlated with disease activity and tissue factor expression. Galectin-9 correlated stronger than CXCL-10 or tnF-rii with the iFn score (r=0.70, p<0.001) and was superior to CXCL-10 or tnF-rii in detecting the iFn signature (area under the curve (AUC) 0.86). importantly, in patients with sLe(±Aps), galectin-9 was also superior to anti-dsDnA antibody (AUC 0.70), or complement C3 (AUC 0.70) and C4 (AUC 0.78) levels in detecting the iFn signature.Conclusion Galectin-9 is a novel, easy to measure hence clinically applicable biomarker to detect the iFn signature in patients with systemic autoimmune diseases such as sLe and Aps.

Systemic lupus erythematosus (SLE) is a chronic relapsing autoimmune disease in which immune complexes of autoantibodies are deposited in tissues inducing tissue damage. Approximately 20% of patients with SLE have antiphospholipid syndrome (APS), defined as the persistent presence of anti-phospholipid (aPL) antibodies in patients who experienced at least one thrombotic or predefined obstetric complication. APS also affects the patients without an underlying disease and is then termed primary APS (PAPS). The pathogenesis of SLE and

APS is partly overlapping including shared genetic risk loci and perturbations in both the innate and adaptive immune system, although both conditions are only rarely studied together.1

Transcriptomic studies in SLE and APS have revealed a markedly increased expression of inter-feron (IFN) inducible genes, known as the IFN signature which is present in ~75% of patients with SLE (±APS) and ~50% of patients with PAPS.2–5 In SLE, the IFN signature is associated with elevated autoantibody levels, disease activity, future flares and congenital heart block.3 4 6 7 In (P)APS, the IFN signature has only recently been reported and has been linked to endothelial progenitor cell dysfunction and increases in proin-flammatory monocytes.2 5

1811van den Hoogen LL, et al. Ann Rheum Dis 2018;77:1810–1814. doi:10.1136/annrheumdis-2018-213497

Basic and translational research

Table 1 Clinical characteristics

HC (n=27) sLE (n=50) sLE+APs (n=40) PAPs (n=29)

Female (%) 93 96 95 97

Age 43 (34–50) 40 (28–48) 45 (37–53) 40 (33–50)

Disease manifestations

SLEDAI 4 (2–6) 4 (1–5) –

Malar rash (%) 66 55 0

Discoid rash (%) 22 18 0

Photosensitivity (%) 42 53 0

Oral ulcers (%) 36 35 0

Arthritis (%) 70 65 0

Serositis (%) 28 20 0

Lupus nephritis (%) 60 45 0

Neurologic disorder (%) 4 18 10

Haematological disorder (%) 60 85 35

Arterial thrombosis (%) 10 43 59

Venous thrombosis (%) 4 60 38

Obstetrical morbidity (%) 6 23 31

Current drug use

Hydroxychloroquine (%) 80 55 21

Prednisone (%) 64 45 0

Azathioprine (%) 34 33 0

Oral anticoagulant (%) 2 75 62

Aspirin (%) 22 28 48

Serology

C3 (g/L) 0.88 (0.69–1.03) 0.79 (0.68–0.92) –

C4 (g/L) 0.16 (0.13–0.20) 0.15 (0.11–0.22) –

a-dsDNA (IU/mL) 27 (6–114) 14 (5–58) –

Lupus anticoagulant (%) 13 65 82

IgG aCL (%) 15 78 86

IgM aCL (%) 11 15 38

IgG aβ2GPI (%) 7 26 35

IgM aβ2GPI (%) 7 5 10

Medians with IQR or percentage of total.aβ2GPI, anti-β2 glycoprotein I antibodies; aCL, anticardiolipin antibodies; a-dsDNA, anti-double stranded DNA antibodies; APS, antiphospholipid syndrome; HC, healthy control; PAPS, primary APS; SLE, systemic lupus erythematosus; SLEDAI, SLE Disease Activity Index.

The IFN signature is a promising therapeutic target. Anifrolumab, a monoclonal antibody against IFNAR, the receptor for type I IFN, showed clinical efficacy in a phase IIb clinical trial in SLE, particularly in those patients with an IFN signature.8 The detection of the IFN signature may therefore soon be used to guide treatment decisions in SLE and possibly other autoimmune diseases.

The IFN signature is measured by gene expression analysis. Therefore, easier to measure biomarkers for IFN signature detection are urgently needed. Measuring IFNα in periph-eral blood seems the most straightforward method. However, IFNα is difficult to detect by standard techniques9 and the 12 different IFNα proteins cannot be measured by a single assay. In addition, other type I IFNs (including IFNβ) as well as type II and III IFN (IFNγ and IFNλ) contribute to the IFN signa-ture.10 Therefore, surrogate markers for the IFN signature were searched for Siglec-1 and CD64 expression on mono-cytes, and serum levels of CXCL-10 correlate with the IFN signature in SLE and serve as alternatives to detect the IFN signature in SLE.6 11–13

Here, using an identification and replication approach we identified that galectin-9, CXCL-10 and tumour necrosis factor receptor type II (TNF-RII) correlate with the IFN signature and disease activity in patients with SLE and APS. Importantly,

galectin-9 outperformed CXCL-10 or traditional markers of disease activity as an easy measurable biomarker to detect the IFN signature both in SLE and APS.

METHOdsBlood was drawn from patients with SLE, SLE+APS, PAPS and healthy controls (HC, table 1). Patients with PAPS did not meet the American College of Rheumatology criteria for SLE.14 The medical ethical committee of the UMC Utrecht approved the study and written informed consent was obtained.

Clinical and laboratory assessmentsSerum analytes were measured as described in the online supple-mentary methods15 in an initial identification cohort (n=43), followed by a replication step (n=148). For serum analytes, elevated levels were defined as 2 SDs above the mean of HC. Anti-dsDNA antibodies and complement levels were determined by EliA or nephelometry. Details on the assessment of the IFN signature, tissue factor (TF) expression and mRNA expression of galectin-9 in dendritic cells (DC) are provided in the online supplementary methods.

statisticsAll tests were conducted two sided at an α level of 0.05. Statis-tical tests used are provided in the figure legends.

1812 van den Hoogen LL, et al. Ann Rheum Dis 2018;77:1810–1814. doi:10.1136/annrheumdis-2018-213497

Basic and translational research

Figure 1 Galectin-9 is a biomarker for disease activity and tissue factor expression in SLE and APS. (A) Galectin-9, CXCL-10 and TNF-RII were measured by Luminex in sera of patients with SLE, SLE+APS and PAPS (Statistics: Kruskal-Wallis with post hoc Dunn’s). Dotted line represents the mean of HC+2 SD. (B) Correlation of galectin-9, CXCL-10 and TNF-RII with disease activity in patients with SLE and SLE+APS. (C) Correlation of galectin-9, CXCL-10 and TNF-RII tissue factor (TF) expression in monocytes in patients with SLE+APS and PAPS as assessed by qPCR (-dCq, delta quantification cycle). Statistics: (A) Kruskal-Wallis followed by post hoc Dunn’s. (B, C) Spearman correlation coefficients. APS, antiphospholipid syndrome; HC, healthy control; PAPS, primary APS; SLE, systemic lupus erythematosus; SLEDAI, SLE Disease Activity Index; TNF-RII, tumour necrosis factor receptor type II.

rEsuLTsGalectin-9, CXCL-10 and TNF-rII are biomarkers of disease activity in sLE and APsIn an initial screening of 22 serum analytes, galectin-9 (r=0.81, p<0.001), CXCL-10 (r=0.72, p<0.001) and TNF-RII (r=0.42, p=0.007) correlated significantly with the IFN score in patients with SLE, SLE+APS and PAPS (n=43, online supplementary table and figure 1). We next assessed these analytes in a larger replication step (n=148). Galectin-9, CXCL-10 and TNF-RII levels were significantly elevated in patients with SLE, SLE+APS and PAPS as compared with HC (figure 1A). Elevated levels of galectin-9 were present in 74%, 68% and 41% of patients with SLE, SLE+APS and PAPS, respectively, compared with 46%, 55% and 45% for CXCL-10% and 46%, 50% and 41% for TNF-RII. All three analytes correlated with anti-dsDNA antibodies and complement levels in patients with SLE (online supplementary figure 2). Furthermore, galectin-9 and TNF-RII, not CXCL-10, correlated with disease activity as assessed by SLE Disease Activity Index (figure 1B). In patients with APS, the increased expression of TF by monocytes is induced by aPL antibodies and reflects a prothrombotic phenotype,1 and we found significant

correlations between galectin-9, CXCL-10 and TNF-RII and the expression of TF by monocytes in patients with APS (figure 1C).

Galectin-9 is a biomarker to detect the IFN signature in sLE and APsA superior correlation of galectin-9 with the IFN score (r=0.70, p<0.001) as compared with CXCL-10 (r=0.52, p<0.001) and TNF-RII (r=0.46, p<0.001) was found (figure 2A). Receiver operating characteristic curve analysis revealed the highest area under the curve (AUC, 0.86) for galectin-9, followed by CXCL-10 (0.78) and TNF-RII (0.75) for the detection of the IFN signature (figure 2B) among patients with SLE or APS. When focusing only on patients with SLE(±APS), galectin-9 (AUC 0.84) remained a supe-rior biomarker to detect the IFN signature compared with CXCL-10 (AUC 0.75) or TNF-RII (AUC 0.70) or traditional markers of SLE disease activity such as anti-dsDNA anti-bodies (AUC 0.70) or complement levels (C3: AUC 0.70, C4: AUC 0.78) (figure 2C). Elevated levels of galectin-9 had a sensitivity of 84% and a positive predictive value of 91% of detecting the IFN signature among patients with SLE(±APS)

1813van den Hoogen LL, et al. Ann Rheum Dis 2018;77:1810–1814. doi:10.1136/annrheumdis-2018-213497

Basic and translational research

Figure 2 In SLE and APS, galectin-9 outperforms CXCL-10 and traditional markers for disease activity as biomarker for the IFN signature and is upregulated in dendritic cells. (A) Correlation of galectin-9, CXCL-10 and TNF-RII with the IFN score. (B) ROC curve analysis of galectin-9, CXCL-10 and TNF-RII in the detection of the IFN signature in patients with SLE, SLE+APS and PAPS. (C) ROC curve analysis of galactin-9, CXCL-10 and TNF-RII as compared with traditional markers of SLE disease activity in detecting the IFN signature among patients with SLE±APS. (D) Test characteristics of elevated galectin-9, CXCL-10, TNF-RII, anti-dsDNA antibodies or reduced complement levels to detect the IFN signature in patients with SLE(±APS). (E) Expression of galectin-9 coding gene LGALS9 in myeloid and plasmacytoid DCs as assessed by RNAseq in patients with SLE and APS stratified by IFN status. (F) Expression of LGALS9 assessed by qPCR after in vitro stimulation of DCs with IFNα for 3 hours (n=3/group). Statistics: (A) Spearman correlation coefficients. (E) ANOVA with post hoc Tukey’s or Student’s t-test. (F) Student’s t-test. ANOVA, analysis of variance; APS, antiphospholipid syndrome; AUC, area under the curve; HC, healthy control; IFN, interferon; mDC, myeloid dendritic cells; NPV, negative predictive value; PAPS, primary APS; pDC, plasmacytoid dendritic cells; PPV, positive predictive value; ROC, receiver operating characteristic; SLE, systemic lupus erythematosus; TNF-RII, tumour necrosis factor receptor type II.

(figure 2D). In multivariate linear regression analysis galectin-9 was an independent predictor for the IFN score, and remained significant after correction for anti-dsDNA antibodies, complement levels and disease activity.

Increased expression of galectin-9 in dCs of IFN-positive patients with sLE and APsGalectin-9 is a β-galactoside-binding lectin, encoded by the gene LGALS9 which is highly expressed in DCs.16 Therefore, we explored the expression of LGALS9 in myeloid and plasmacy-toid DCs (mDC/pDC) in SLE and APS. LGALS9 expression was markedly higher in circulating DCs of patients with SLE and APS with an IFN signature and IFNα increased LGALS9 expression in vitro in mDCs and pDCs (figure 2E,F).

dIsCussIONHere we report that galectin-9, CXCL-10 and TNF-RII are elevated and associated with disease activity in SLE and APS. Notably, we identified galectin-9 as a robust and easy to measure biomarker to detect the IFN signature, outperforming tradi-tional biomarkers for the IFN signature such as CXCL-10.

Measuring the IFN signature in patients with SLE might in the future be used in clinical decision-making.8 The necessity to quantify gene expression and the lack of a uniformly accepted gene set or scoring system hamper its implementation in clin-ical practice. Furthermore, serum biomarkers are more easily standardised and are less laborious as compared with previously proposed biomarkers for the IFN signature that are assessed by flow cytometry.12 13 In this respect, galectin-9 is a promising

1814 van den Hoogen LL, et al. Ann Rheum Dis 2018;77:1810–1814. doi:10.1136/annrheumdis-2018-213497

Basic and translational research

stable biomarker, which is superior to CXCL-10 in the detection of the IFN signature, both in SLE and APS, and can furthermore be reliably detected in serum samples stored for long periods.17

Besides serving as a biomarker, galectin-9 may play a role in the pathogenesis of SLE and APS. Galectin-9 induces the matura-tion of DCs by increasing the expression of HLA-DR and costim-ulatory molecules and galectin-9 produced by DCs activates T cells.16 Galectin-9 expression was markedly increased in DCs of IFN-positive patients with SLE and APS and may therefore drive the unabated activation of DCs seen in these patients. Further-more, in the pristane-induced SLE mouse model, knockout of LGALS9 results in reduced nephritis and arthritis,18 supporting a pathogenic role for galectin-9 in the pathogenesis of SLE. Besides IFNα, galectin-9 may also be induced by other proinflammatory mediators implicated in the pathophysiology of autoimmune diseases,19 hence the applicability of galectin-9 as a biomarker for immunopathology in other diseases warrants further study.

The treatment of SLE and APS has remained largely unchanged for decades. The revolution in the treatment of other rheumatic conditions, most notably rheumatoid arthritis, in the last decades has not been paralleled in SLE and APS in which few clinical trials have met their primary outcome. A major reason for the failure of these trials is the molecular heterogeneity of these diseases. Consequently, for SLE and APS there is an urgent need for a more personalised treatment approach, guided by the molecular phenotype rather than clinical diagnosis. Treatment of patients on the basis of the presence of an IFN signature could represent such a molecular phenotype, supported by the results from the anifrolumab trial.8 Galectin-9 may serve as a robust and easy to measure biomarker to detect the IFN signature in patients with SLE and APS, as well as other (systemic) autoimmune diseases.

Author affiliations1Laboratory of translational immunology, University Medical Centre Utrecht, Utrecht, the netherlands2Department of rheumatology and Clinical immunology, University Medical Centre Utrecht, Utrecht, the netherlands3Department of Dermatology, radboud University Medical Centre, nijmegen, the netherlands4Department of Biotechnology, University of Verona, Verona, italy51st Medical Department, Hanusch Hospital, Ludwig Boltzmann institute of osteology at the Hanusch Hospital of WGKK and AUVA trauma Centre Meidling, Vienna, Austria6Faculty of Medicine, sigmund Freud private University, Vienna, Austria

Acknowledgements We thank the Multiplex Core Facility of the Laboratory of translational immunology (UMC Utrecht) for performing the in-house developed and validated multiplex immunoassays.

Contributors All authors were involved in study design, interpretation of data, drafting of the work and gave final approval of the version published. LLvdH was involved in data collection and performed the analysis.

Funding this project has received funding from the european Union’s Horizon 2020 research and innovation programme under grant agreement no 731944, Harmonicss.

Competing interests none declared.

Patient consent obtained.

Ethics approval Medisch ethische toetsingscommissie Utrecht.

Provenance and peer review not commissioned; externally peer reviewed.

data sharing statement original data are available upon reasonable request.

RefeRences 1 van den Hoogen LL, van roon JA, radstake tr, et al. Delineating the deranged

immune system in the antiphospholipid syndrome. Autoimmun Rev 2016;15:50–60. 2 Grenn rC, Yalavarthi s, Gandhi AA, et al. endothelial progenitor dysfunction

associates with a type i interferon signature in primary antiphospholipid syndrome. Ann Rheum Dis 2017;76:450–7.

3 Kirou KA, Lee C, George s, et al. Activation of the interferon-alpha pathway identifies a subgroup of systemic lupus erythematosus patients with distinct serologic features and active disease. Arthritis Rheum 2005;52:1491–503.

4 Hoffman rW, Merrill Jt, Alarcón-riquelme MM, et al. Gene expression and pharmacodynamic changes in 1,760 systemic lupus erythematosus patients from two phase iii trials of baff blockade with tabalumab. Arthritis Rheumatol 2017;69:643–54.

5 van den Hoogen LL, Fritsch-stork rD, Versnel MA, et al. Monocyte type i interferon signature in antiphospholipid syndrome is related to proinflammatory monocyte subsets, hydroxychloroquine and statin use. Ann Rheum Dis 2016;75:e81.

6 rose t, Grützkau A, Klotsche J, et al. Are interferon-related biomarkers advantageous for monitoring disease activity in systemic lupus erythematosus? A longitudinal benchmark study. Rheumatology 2017;56:1618–26.

7 Lisney Ar, szelinski F, reiter K, et al. High maternal expression of siGLeC1 on monocytes as a surrogate marker of a type i interferon signature is a risk factor for the development of autoimmune congenital heart block. Ann Rheum Dis 2017;76:1476–80.

8 Furie r, Khamashta M, Merrill Jt, et al. Anifrolumab, an anti-interferon-α receptor monoclonal antibody, in moderate-to-severe systemic lupus erythematosus. Arthritis Rheumatol 2017;69:376–86.

9 rodero Mp, Decalf J, Bondet V, et al. Detection of interferon alpha protein reveals differential levels and cellular sources in disease. J Exp Med 2017;214:1547–55.

10 Chiche L, Jourde-Chiche n, Whalen e, et al. Modular transcriptional repertoire analyses of adults with systemic lupus erythematosus reveal distinct type i and type ii interferon signatures. Arthritis Rheumatol 2014;66:1583–95.

11 rose t, Grützkau A, Hirseland H, et al. iFnα and its response proteins, ip-10 and siGLeC-1, are biomarkers of disease activity in systemic lupus erythematosus. Ann Rheum Dis 2013;72:1639–45.

12 York Mr, nagai t, Mangini AJ, et al. A macrophage marker, siglec-1, is increased on circulating monocytes in patients with systemic sclerosis and induced by type i interferons and toll-like receptor agonists. Arthritis Rheum 2007;56:1010–20.

13 Li Y, Lee pY, Kellner es, et al. Monocyte surface expression of Fcgamma receptor ri (CD64), a biomarker reflecting type-i interferon levels in systemic lupus erythematosus. Arthritis Res Ther 2010;12:r90–12.

14 Hochberg MC. Updating the American College of rheumatology revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum 1997;40:1725.

15 Bellutti enders F, van Wijk F, scholman r, et al. Correlation of CXCL10, tumor necrosis factor receptor type ii, and galectin 9 with disease activity in juvenile dermatomyositis. Arthritis Rheumatol 2014;66:2281–9.

16 John s, Mishra r. Galectin-9: from cell biology to complex disease dynamics. J Biosci 2016;41:507–34.

17 scholman rC, Giovannone B, Hiddingh s, et al. effect of anticoagulants on 162 circulating immune related proteins in healthy subjects. Cytokine 2018;106:114–24.

18 Zeggar s, Watanabe Ks, teshigawara s. Lgals9 deficiency attenuates nephritis and arthritis in pristane-induced lupus model of BALB/c mice. Arthritis Rheumatol 2018.

19 Gieseke F, Kruchen A, tzaribachev n, et al. proinflammatory stimuli induce galectin-9 in human mesenchymal stromal cells to suppress t-cell proliferation. Eur J Immunol 2013;43:2741–9.

1815Conway R, et al. Ann Rheum Dis 2018;77:1815–1824. doi:10.1136/annrheumdis-2018-213488

Basic and translational research

ExtEndEd rEport

Interleukin 12 and interleukin 23 play key pathogenic roles in inflammatory and proliferative pathways in giant cell arteritisrichard Conway,1,2 Lorraine o’neill,1 Geraldine M McCarthy,3 Conor C Murphy,4 Aurelie Fabre,5 Susan Kennedy,5 douglas J Veale,1 Sarah M Wade,6 Ursula Fearon,6 Eamonn S Molloy1

To cite: Conway r, o’neill L, McCarthy GM, et al. Ann Rheum Dis 2018;77:1815–1824.

Handling editor Josef S Smolen

► Additional material is published online only. to view please visit the journal online (http:// dx. doi. org/ 10. 1136/ annrheumdis- 2018- 213488).

1Centre for Arthritis and rheumatic diseases, St Vincent’s University Hospital dublin, Academic Medical Centre, dublin 4, Ireland2CArd newman research Fellow, University College dublin, dublin, Ireland3Mater Misericordiae University Hospital, dublin Academic Medical Centre, dublin, Ireland4rCSI department of ophthalmology, royal College of Surgeons of Ireland, royal Victoria Eye and Ear Hospital, dublin, Ireland5department of pathology, St Vincent’s University Hospital, dublin, Ireland6department of Molecular rheumatology, School of Medicine, trinity Biomedical Sciences Institute, trinity College dublin, dublin, Ireland

Correspondence todr richard Conway, Centre for Arthritis and rheumatic diseases, St Vincent’s University Hospital, dublin 4, d04 Y8V0, Ireland; drrichardconway@ gmail. com

received 28 March 2018revised 18 July 2018Accepted 19 July 2018published online First 10 August 2018

© Author(s) (or their employer(s)) 2018. no commercial re-use. See rights and permissions. published by BMJ.

ABsTrACTObjectives the pathogenesis of giant cell arteritis (GCA) remains unclear. tH1 and tH17 pathways are implicated, but the proximal initiators and effector cytokines are unknown. our aim was to assess the role of interleukin 12 (IL-12) and interleukin 23 (IL-23) in GCA pathogenesis.Methods IL-12 and IL-23 expression were quantified by immunohistochemistry in temporal artery biopsies (tABs). temporal artery (tA) explant, peripheral blood mononuclear cell (pBMC) and myofibroblast outgrowth culture models were established. pBMCs and tA explants were cultured for 24 hours in the presence or absence of IL-12 (50 ng/mL) or IL-23 (10 ng/mL). Gene expression in tA was quantified by real-time pCr and cytokine secretion by ELISA. Myofibroblast outgrowths were quantified following 28-day culture.results Immunohistochemistry demonstrated increased expression of interleukin 12p35 (IL-12p35) and interleukin 23p19 (IL-23p19) in biopsy-positive tAs, localised to inflammatory cells. IL-12p35 tA expression was significantly increased in those with cranial ischaemic complications (p=0.026) and large vessel vasculitis (p=0.006). IL-23p19 tA expression was increased in those with two or more relapses (p=0.007). In pBMC cultures, exogenous IL-12 significantly increased interleukin 6 (IL-6) (p=0.009), interleukin 22 (IL-22) (p=0.003) and interferon γ (IFn-γ) (p=0.0001) and decreased interleukin 8 (IL-8) (p=0.0006) secretion, while exogenous IL-23 significantly increased IL-6 (p=0.029), IL-22 (p=0.001), interleukin 17A (IL-17A) (p=0.0003) and interleukin 17F (IL-17F) (p=0.012) secretion. In ex vivo tA explants, IL-23 significantly increased gene expression of IL-8 (p=0.0001) and CCL-20 (p=0.027) and protein expression of IL-6 (p=0.002) and IL-8 (p=0.004). IL-12 (p=0.0005) and IL-23 (p<0.0001) stimulation increased the quantity of myofibroblast outgrowths from tABs.Conclusion IL-12 and IL-23 play central and distinct roles in stimulating inflammatory and proliferative pathways relevant to GCA pathogenesis.

InTrOduCTIOnGiant cell arteritis (GCA) is the most common form of systemic vasculitis.1 However, its patho-genesis remains incompletely understood.2 While the initiating trigger for GCA is unknown, one of the primary events is the activation of arterial

wall resident dendritic cells (DCs) through toll-like receptors.3 Activated DCs release mediators which initiate the process ultimately resulting in the arterial wall inflammation characteristic of GCA.4 Central to this is the stimulation of two distinct helper T-cell subsets, T helper 1 (TH1) and T helper 17 (TH17) cells. TH1 cells secrete interferon γ (IFN-γ) which is hypothesised to be primarily responsible for the ischaemic manifesta-tions of GCA. TH17 cells secrete several cytokines, including interleukin 17A (IL-17A) and inter-leukin 22 (IL-22), hypothesised to be primarily responsible for the constitutional and polymyalgic symptoms of GCA.2

Interleukin 12 (IL-12) and interleukin 23 (IL-23) are heterodimeric members of the IL-12 super-family of cytokines, secreted by DCs with subse-quent effects on T-lymphocytes.5–7 IL-12 induces T-lymphocytes to differentiate towards the TH1 pathway and regulates TH1 cell function.8 The prin-cipal cellular target of IL-23 is TH17 cells where it promotes release of IL-17A.9 While IL-12 and IL-23 have been identified as important cytokines in the pathogenesis of several diseases, there are limited data in GCA.10 11

Plasma IL-12 is virtually undetectable in healthy controls, increased in untreated patients with GCA and persistently elevated despite gluco-corticoid treatment.12 Monocyte gene expression of interleukin 12p35 (IL-12p35) and interleukin 12p40 (IL-12p40) are increased in patients with GCA compared with healthy controls and persists despite glucocorticoid treatment.12 IL-12p35 and IL-12p40 gene expression are persistently increased in temporal arteries (TAs) following glucocorticoid treatment.12 In one small study, higher serum IL-12p40 predicted a relapsing disease course and relapsing patients appeared to have persistent elevations in IL-12p40 mRNA but paradoxically lower expression of IL-12p35, IL-12p40 and interleukin 23p19 (IL-23p19) by immunohistochemistry.13 TH17 cells, a key targetof IL-23, are increased in the blood of patients with GCA.12 IL-23p19 and IL-17A have previously been demonstrated to be increased at mRNA level in inflamed TAs from patients with GCA.13 14 The aim of the current study was to assess the expres-sion and effects of IL-12 and IL-23 on inflamma-tory and proliferative pathways in GCA.

1816 Conway R, et al. Ann Rheum Dis 2018;77:1815–1824. doi:10.1136/annrheumdis-2018-213488

Basic and translational research

MeTHOdsGCA registryPatients included in the current study were recruited from the participants in an existing GCA Registry, as previously reported (see online supplementary methods).15 All patients were receiving glucocorticoids at the time of peripheral blood mononuclear cell (PBMC) sampling and temporal artery biopsy (TAB). All patients gave written informed consent and the study received ethical approval.

TA immunohistochemistryImmunohistochemistry for IL-12p35 and IL-23p19 was performed, see online supplementary methods. All slides were stained and scored under blinded conditions by an independent senior histopathologist (AF). IL-12p35 and IL-23p19 expres-sion were scored using a semiquantitative method based on that described by Chatelain et al.16 The proportion of inflammatory cells (excluding smooth muscle cells and vessels) stained was scored with 0=no staining, 1+=10–25%, 2+=25–50% and 3+=>50% of inflammatory cells. For each slide, the entire arte-rial section present was examined and scored.

We compared IL-12 and IL-23 expression in patients with and without cranial ischaemic complications (CICs), headache, scalp tenderness, TA abnormality, large vessel vasculitis (LVV) (definitive evidence of vasculitis of aorta or major branches on CT angiogram), jaw claudication, polymyalgia or constitutional symptoms at initial presentation. We compared IL-12 and IL-23 expression in those with and without GCA relapse (recurrence of symptoms of active GCA with or without elevated acute-phase reactants), multiple (≥2) relapses and those subsequently prescribed an immunosuppressive agent for the treatment of GCA. GCA relapse was defined as the occurrence of symp-toms of active GCA (including polymyalgia) with or without elevated acute-phase reactants (erythrocyte sedimentation rate (ESR) >30 mm/hour, C-reactive protein (CRP) >10 mg/L) in a patient previously in remission. Isolated fluctuations in acute-phase reactants or alterations in glucocorticoid dose without the presence of symptoms attributable to GCA were not classified as a relapse. We evaluated correlations between IL-12 and IL-23 expression and CRP, ESR, haemoglobin and platelet count.

Isolation and culture of PBMCsPBMCs were isolated from blood obtained from patients with GCA at their baseline assessment. PBMCs were seeded at a cell density of 1×106 cells/250 µL and cultured in the presence or absence of 10 ng/mL of IL-23 (R&D Systems, Minneapolis, Minnesota, USA), 50 ng/mL of IL-12 (R&D Systems) or the combination of both for 24 hours in humidified air at 37oC in 5% CO2, and cultured supernatants were harvested.

TA explant culture modelTABs from patients with suspected GCA were used to establish an ex vivo TA explant culture model as described previously,15 outlined in online supplementary methods. TA explants were cultured in the presence or absence of 10 ng/mL of IL-23 (R&D Systems), 50 ng/mL of IL-12 (R&D Systems) or the combination of both IL-12 and IL-23 for 24 hours in humidified air at 37oC in 5% CO2. Culture supernatants were harvested and TA explants were processed for RNA isolation and gene expression.

mrnA extraction and PCr analysismRNA was extracted from TA explants, see online supple-mentary methods. Relative quantification of gene expression

was quantified by SYBR Green Real-Time PCR on the Applied Biosystems QuantStudio Real-Time PCR System. Gene quan-tification of signal transducer and activator of transcription 3 (STAT3), interleukin 6 (IL-6), interleukin 8 (IL-8), interleukin 10 (IL-10), IL-12p35, IL-12p40, IL-23p19, IFN-γ, IL-17A, tumour necrosis factor-α (TNF-α), chemokine (c-c motif) ligand 20 (CCL-20), angiopoietin2 (ANG-2), Regulated upon Activa-tion, Normal T cell Expressed and Secreted (RANTES), vascular endothelial growth factor (VEGF) and vascular cell adhesion molecule 1 (VCAM-1) was performed using mRNA-specific primers as shown in online supplementary Table S1. Relative changes in gene expression were determined using the ΔΔCT method with normalisation to the expression of the housekeeper ribosomal protein lateral stalk subunit p0 (RPLPO) as an endog-enous control.

Myofibroblast outgrowth culture modelTo examine effects on migration/invasion in GCA, a myofi-broblast outgrowth culture model was established as described previously.15 Matrigel (50 µL) was plated in 96-well culture plates and allowed to polymerise at 37°C in 5% CO2 for 30 min. Serial TA explant sections were carefully embedded in the Matrigel and cultured in 200 µL Dulbecco’s Modified Eagle’s Medium supplemented with 10% fetal bovine serum, 10 mL of 1 mmol/L 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (Gibco-BRL), penicillin (100 units/mL; Gibco-BRL), strepto-mycin (100 units/mL; Gibco-BRL) and amphotericin B (0.25 µg/mL; Gibco-BRL). TA explants were cultured in the presence or absence 10 ng/mL of IL-23, 50 pg/mL of IL-12 or the combina-tion of both over a time course of 1–28 days. Media and exper-imental agents were replenished every 3 days. Myofibroblast outgrowths were assessed using phase-contrast microscopy and photographed at day 28. The number of outgrowths per high power field (hpf) was counted by two independent observers (RC and UF). The mean number of outgrowths per hpf was calculated for each condition.

Cytokine and chemokine quantificationSee online online supplementary methods.

statistical analysisDescriptive statistics were reported as mean and SD, median and IQR or number (n) and percentages as appropriate. For between-group comparisons, χ2 tests were used for categorical variables and independent samples t-tests for continuous vari-ables. Student’s t-tests were used to analyse parametric data. For non-parametric data, Wilcoxon signed rank test for related samples and Mann Whitney U test for unrelated samples were used. Statistical significance was set at p<0.05 throughout. All analyses were performed using IBM SPSS Statistics (IBM Corp. Released 2011. IBM SPSS Statistics for Windows, V.24.0. Armonk, New York: IBM Corp.) and GraphPad Prism V.7.03 for Windows (GraphPad Software, La Jolla, California, USA, www. graphpad. com).

resulTsIl-12 and Il-23 immunohistochemistryRepresentative images of TA histology H&E are shown in online supplementary figure S1 showing varying combinations of luminal narrowing due to intimal hyperplasia, fragmenta-tion of the internal elastic lamina, mononuclear inflammatory infiltrates and multinucleated giant cells (GCs) in positive TABs. IL-12 and IL-23 expression in positive controls are shown in

1817Conway R, et al. Ann Rheum Dis 2018;77:1815–1824. doi:10.1136/annrheumdis-2018-213488

Basic and translational research

online supplementary figure S2. IL-12 and IL-23 were absent in biopsy-negative sections and were variably expressed in inflam-matory cells in biopsy-positive sections. GCs showed intense expression of IL-12 and IL-23 (figure 1).

IL-12 expression in TAs was significantly greater in patients with CICs (p=0.026), LVV (p=0.006) or without headache (p=0.030). IL-23 expression was not increased in those with CICs (p=0.967), LVV (p=0.264) or without headache (p=0.634). There was no significant difference in IL-12 or IL-23 expression in those with scalp tenderness (p=0.245 and p=0.720, respec-tively) or TA abnormality on clinical examination (p=0.698 and p=0.606, respectively). IL-12 (p=0.051) and IL-23 expression were not increased in those with jaw claudication (p=0.892). IL-12 and IL-23 expression were not different based on polymyalgic

(p=0.664 and p=0.265, respectively) or constitutional (p=0.101 and p=0.798, respectively) symptoms. IL-12 and IL-23 expression were not significantly associated with the occurrence of relapses of GCA (p=0.562 and p=0.470, respectively) or the need for the use of another agent in addition to glucocorticoids (p=0.117 and p=0.140, respectively). However, IL-23 (p=0.007), but not IL-12 (p=0.053) expression was significantly increased in those with two or more relapses (figure 2).

There was a moderate correlation between IL-12 expression and baseline CRP levels, r=0.487 (p=0.004). There was no correlation between IL-12 expression and ESR (r=0.114, p=0.527), haemo-globin (r=0.041, p=0.822) or platelets (r=0.11, p=0.556). There was no correlation between IL-23 expression and CRP (r=0.203, p=0.257), ESR (r=0.166, p=0.356), haemoglobin (r=0.102,

Figure 1 Interleukin 12 (IL-12) and interleukin 23 (IL-23) Immunohistochemistry. Representative images of the microscopic analysis of temporal artery (TA) sections staining for IL-12 and IL-23. Blue arrows indicate representative examples of specific cellular staining. (A) 0, negative staining (100x); (B) 1+IL-12 expression (100x); (C) 2+IL-12 expression (100x); (D) 3+IL-12 expression (100x); (E) IL-12 expression in giant cell (GC) (400x); (F) IL-12 expression in blood vessel (vx) endothelial cells (400x); (G) IL-23 expression in GC (400x); (H) 1+IL-23 expression (100x); (I) 2+IL-23 expression (100x) and (J) 3+IL-23 expression (100x). SM, smooth muscle. IL-12 was expressed in inflammatory cells in biopsy-positive TAs (n=29) but not in biopsy-negative TAs (n=4) (K). IL-23 was expressed in inflammatory cells in biopsy-positive TAs (n=29) but not in biopsy-negative TAs (n=4) (L).

1818 Conway R, et al. Ann Rheum Dis 2018;77:1815–1824. doi:10.1136/annrheumdis-2018-213488

Basic and translational research

Figure 2 Interleukin 12 (IL-12) and interleukin 23 (IL-23) expression and clinical features. IL-12 expression (A) was significantly increased in those with cranial ischaemic complications (CICs), and those with imaging evidence of large vessel vasculitis (LVV), and significantly decreased in those with headache. There was no difference in IL-12 expression in those with jaw claudication, scalp tenderness or an abnormal temporal artery (TA) on clinical examination. IL-12 expression was not increased in those with relapses, multiple relapses and in those requiring a second-line treatment. IL-23 expression (B) was not increased in those with CICs, headache, scalp tenderness, LVV, jaw claudication or in those with an abnormal TA on clinical examination. There was no difference in IL-23 expression in those with and without relapses; however, IL-23 expression was significantly increased in those with multiple relapses. IL-23 expression was not increased in those requiring a second-line treatment. Data were analysed using Mann Whitney U test and are expressed as mean± SE of the mean (SEM). n=33. *p<0.05, **p<0.01.

1819Conway R, et al. Ann Rheum Dis 2018;77:1815–1824. doi:10.1136/annrheumdis-2018-213488

Basic and translational research

p=0.571) or platelets (r=−0.203, p=0.273). IL-12 expression was increased in biopsies with GCs (p=0.034) and decreased in biopsies with intimal hyperplasia (p=0.045). IL-23 was increased in biopsies with GCs (p=0.032) and no different in those with intimal hyperplasia (p=0.681).

Il-12 and Il-23 stimulation in PBMC culture modelIL-12 stimulation significantly increased IL-6 (p=0.009, n=17) (figure 3A), IFN-γ (p=0.0001, n=14) (figure 3E) and IL-22 (p=0.003, n=16) (figure 3F) secretion. IL-12 significantly decreased IL-8 (p=0.0006, n=15) secretion (figure 3B). IL-12 did not increase IL-17A (p=0.669, n=16) (figure 3G), inter-leukin 17F (IL-17F) (p=0.074, n=9) (figure 3H), interleukin 2 (IL-2) (p=0.875, n=4), interleukin-1β (IL-1β) (p=0.875, n=4) or TNF-α (p=0.125, n=3) secretion.

IL-23 stimulation significantly increased IL-6 (p=0.029, n=40) (figure 3C), IL-17A (p=0.0003, n=16) (figure 3G), IL-17F (p=0.012, n=9) (figure 3H) and IL-22 (p=0.001, n=16) (figure 3F) secretion. IL-23 stimulation did not increase IL-8 (p=0.554, n=41) (figure 3D), IFN-γ (p=0.903, n=14) (figure 3E), IL-2 (p=0.125, n=4), IL-1β (p=0.875, n=4) or TNF-α (p=1, n=3) secretion.

effect of Il-12 and Il-23 stimulation on gene expression in TA explantsIL-12 stimulation significantly increased IL-8 (p=0.048, n=6), IL-6 (p=0.048, n=6) and ANG-2 (p=0.048, n=6) mRNA expression (figure 4A-C). IL-12 stimulation did not increase CCL-20 (p=0.683, n=6), STAT3 (p=1, n=6) or RANTES (p=0.364, n=6) gene expression (figure 4D-F). IL-23 stimula-tion significantly increased IL-8 (p=0.0001, n=13) and CCL-20 (p=0.027, n=9) mRNA expression (figure 4G,K). IL-23 stim-ulation did not increase IL-6 (p=0.380, n=11), STAT3 (n=8, p=0.382), ANG-2 (n=3, p=0.5) or RANTES (n=3, p=0.5) gene expression (figure 4H-J,K). IFN-γ, IL-10, IL-17A, TNF-α,

VEGF and VCAM-1 were consistently at or below the limit of detection and further analysis was not possible.

Il-12 and Il-23 stimulation in TA explant culture modelIL-12 stimulation did not increase IL-6 (p=0.855, n=14) (figure 5A), IL-8 (p=0.903 n=14) (figure 5B), IL-1β (p=0.978, n=15) (figure 5E) or TNF-α (p=0.679, n=15) (figure 5F) secretion. IL-23 stimulation increased IL-6 (p=0.002, n=61) (figure 5C) and IL-8 (p=0.004, n=60) (figure 5D) secretion. IL-23 stimulation did not increase IL-1β (p=0.389, n=15) (figure 5E) or TNF-α (p=0.252 , n=15) (figure 5F) secretion. IFN-γ, IL-2, IL-17A, IL-17F and IL-22 were consistently at or below the limit of detection and therefore not evaluable.

Il-12 and Il-23 stimulation in myofibroblast outgrowth culture modelFollowing 28-day culture, myofibroblast outgrowths were evident extending from the TAB sections into the Matrigel matrix as indicated by red arrows (figure 6A-D). IL-12 (p=0.0005, n=20) (figure 6I), IL-23 (p<0.0001, n=33) (figure 6J) and IL-12 and IL-23 co-stimulation (p<0.0001, n=20) (figure 6K) significantly increased the quantity of myofibroblast outgrowths compared with basal conditions with no evident additive effect with combination treatment (figure 6L).

effects in patient subgroupsWe further analysed the results of our experiments in specific disease subgroups, biopsy-positive GCA, biopsy-negative GCA and disease controls; these are illustrated in online supplemen-tary figures S3–S14. For the majority of outcomes, there were no differences between the groups. However, some outcomes were significant only in the biopsy-positive GCA group; IL-12 stimu-lated IL-6 (p=0.03) (online supplementary figure S3) and IL-23 stimulated IL-17A (p=0.031) (online supplementary figure S4)

Figure 3 Effect of interleukin 12 (IL-12) and interleukin 23 (IL-23) stimulation on peripheral blood mononuclear cells. IL-12 stimulation significantly increased interleukin 6 (IL-6) secretion at 24 hours, n=17 (A). IL-12 stimulation significantly decreased interleukin 8 (IL-8) secretion at 24 hours, n=15 (B). IL-23 stimulation significantly increased IL-6 secretion at 24 hours, n=40 (C). IL-23 stimulation did not increase IL-8 secretion at 24 hours, n=40 (D). IL-12 but not IL-23 stimulation significantly increased interferon γ (IFN-γ) secretion at 24 hours, n=14 (E). IL-12 and IL-23 stimulation significantly increased interleukin 22 (IL-22) secretion at 24 hours, n=16 (F). IL-23 but not IL-12 stimulation significantly increased interleukin 17A (IL-17A), n=16 (G) and interleukin 17F (IL-17F), n=9 (H) secretion at 24 hours. Data were analysed using Wilcoxon signed rank test and are expressed as mean±SE of the mean (SEM). *p<0.05, **p<0.01, ***p<0.001.

1820 Conway R, et al. Ann Rheum Dis 2018;77:1815–1824. doi:10.1136/annrheumdis-2018-213488

Basic and translational research

in PBMC cultures, and IL-23 stimulated IL-6 (p=0.040) (supple-mentary figure S5) and IL-8 (p=0.037) (online supplementary figure S6) in TA cultures. In the myofibroblast outgrowth model, significant increases were seen in both GCA groups but not in disease controls (online supplementary figure S7).

dIsCussIOnWe have investigated the potential pathogenic roles of IL-12 and IL-23 in a variety of different models using TA tissue and cells from patients, thus closely reflecting the in vivo

microenvironment. We have demonstrated the increased expres-sion of IL-12p35 and IL-23p19 at protein level in inflammatory cells using immunohistochemistry in TABs from patients with biopsy-positive GCA. IL-12 and IL-23 were both increased in those with GCs, and IL-12 was decreased in those with intimal hyperplasia. The relevance of this differential expression is unclear; however, differential expression of IL-12 and IL-23 has previously been reported.13 Expression of IL-12p35 was significantly increased in patients who had suffered CICs. This suggests that IL-12, and by extension the TH1 pathway, may be

Figure 4 Effect of interleukin 12 (IL-12) and interleukin 23 (IL-23) stimulation on mRNA expression in temporal artery explants. IL-12 significantly increased interleukin 8 (IL-8) n=6 (A), interleukin 6 (IL-6) n=6 (B) and ANG-2 n=6 (C) mRNA expression. IL-12 did not increase STAT3 n=6 (D), CCL-20 n=6 (E) or RANTES n=6 (F) mRNA expression. IL-23 significantly increased IL-8 n=13 (G) and CCL-20 n=9 (K) mRNA expression. IL-23 did not increase IL-6 n=11 (H), STAT3 n=8 (J), ANG-2 n=3 (I) or RANTES n=3 (L) mRNA expression. Data were analysed using Wilcoxon signed rank test and are expressed as mean± SE of the mean (SEM) compared with endogenous controls. *p<0.05, ***p<0.001.

1821Conway R, et al. Ann Rheum Dis 2018;77:1815–1824. doi:10.1136/annrheumdis-2018-213488

Basic and translational research

the key contributor to ischaemic events in GCA. IL-12p35 was also significantly increased in those with LVV, while IL-23p19 expression was significantly greater in those experiencing two or more relapses. IL-12p35 and IL-23p19 protein expression in TA sections from four patients has previously been reported.13 A subsequent study from the same group also reported the identifi-cation of intracellular IL-23p19 but not IL-12p40 in endothelial cells in TA sections, suggesting a potential proinflammatory role of intracellular IL-23p19 monomers.17 IL-12p35 and IL-23p19 expression have not previously been reported to have an associa-tion with disease outcomes; however, in contrast to our findings for IL-23p19, higher expression of IL-17A has been negatively associated with disease relapse.14 We have seen in previous studies, for example, with TNF-α, that immunohistochem-ical evidence of cytokine presence in TAB does not necessarily imply pathophysiological or therapeutic significance; therefore, our extension of our immunohistochemical findings with both functional responses in explant culture models and a previously reported clinical study is important.18 19

We used PBMC and TA explant culture models to further examine the pathways related to IL-12 and IL-23 in GCA patho-genesis. Stimulation with either IL-12 or IL-23 increased IL-6 secretion from PBMCs; in contrast, in the TA explant culture model, IL-23 increased IL-6 at protein level, whereas IL-12 increased IL-6 at gene level. This suggests a differential effect possibly from different lymphocyte populations in the TA or indeed from non-lymphocyte cell types. IL-12 and IL-23 have previously been demonstrated to stimulate IL-6 release from T-cells in murine models, consistent with what we have shown here in ex vivo human models.20 21 In active GCA, IL-6 is increased in both serum and tissue.22 23 Persistent elevations of

IL-6 are associated with relapsing disease and higher glucocorti-coid requirements.24 Paradoxically, increased production of IL-6 is protective from CICs.25 The potential for both IL-12 and IL-23 to increase IL-6 levels is intriguing given the recent evidence of the effect of IL-6 blockade in GCA in two randomised controlled trials.26 27

The functional importance of IL-8 in GCA has not been explored until now.28 High levels of IL-8 secretion from GCA TAs and the suppression of IL-8 mRNA expression with dexa-methasone have been reported.19 Our group has previously shown that acute-phase serum amyloid-A (A-SAA) augments the production of IL-8 in PBMC and TA explant culture models, demonstrating the inducibility of this chemokine.18 The classic function of IL-8 is neutrophil recruitment; however, important ancillary roles in angiogenesis and in the proliferation and migra-tion of fibroblasts and of tumour cells have been identified.29–31 IL-12 decreased the release of IL-8 from PBMCs, whereas in TA explants it increased IL-8 gene expression. IL-23, in contrast, increased the release of IL-8 from TA explants but had no effect in PBMCs. The changes observed may represent a bystander effect but given the high levels of IL-8 found in patients with GCA warrant further investigation.

Our findings that IL-12 stimulated IFN-γ and IL-23 stimulated interleukin 17 (IL-17) release from PBMCs are consistent with the known roles of these cytokine axes; confirming the function-ality of these pathways in GCA.8 12 32 IFN-γ is increased in GCA plasma and TABs and persists despite clinically efficacious treat-ment.12 33 Culturing normal TAs with recombinant IFN-γ results in macrophage infiltration, while blockade of IFN-γ ameliorates macrophage infiltration, GC formation and chemokine secre-tion.33 IL-17A is increased at both gene and protein levels in

Figure 5 Effect of interleukin 12 (IL-12) and interleukin 23 (IL-23) stimulation of temporal artery explants. IL-12 stimulation did not increase interleukin 6 (IL-6) n=17 (A), interleukin 8 (IL-8) n=15 (B), interleukin 1β (IL-1β) n=15 (E) or tumour necrosis factor α (TNF-α) n=15 (F) secretion at 24 hours, n=17 (A). IL-23 stimulation significantly increased IL-6 n=61 (C) and IL-8 n=60 (D) secretion at 24 hours. IL-23 stimulation did not increase IL-1β n=15 (E) or TNF-α n=15 (F) secretion at 24 hours. Data were analysed using Wilcoxon signed rank test and are expressed as mean±SE of the mean (SEM). **p<0.01.

1822 Conway R, et al. Ann Rheum Dis 2018;77:1815–1824. doi:10.1136/annrheumdis-2018-213488

Basic and translational research

TABs from patients with GCA and correlates with the acute-phase response.14 34 35 IL-17A stimulates lymphocyte and macro-phage chemotaxis and induces expression of IL-6, IL-8, matrix metalloproteinase (MMP)-236–38 and MMP-9 (34–36). We have also demonstrated that IL-12 and IL-23 stimulated the secretion of IL-22 from PBMCs. Increased expression of IL-22 in GCA and its stimulatory effects on viability and gene expression in arterial cells have recently been reported.39

We demonstrated the ability of IL-12 and IL-23 to stimulate myofibroblast outgrowths. The magnitude was similar with both cytokines and there was no evident additive effect. The ability of vascular smooth muscle cells (VSMCs) to transform to an invasive phenotype followed by migration to the intimal layer and subsequent proliferation is believed to be a key feature in the pathogenesis of GCA; in particular in the development of ischaemic events.40 Endothelin-1 has recently been shown to stimulate VSMC migration across the arterial wall.41 Previous work from our group has shown the ability of A-SAA (but not

TNF-α) to stimulate myofibroblast outgrowths, suggesting the importance of specific inflammatory pathways in this process.15

In general, similar effects were observed across all three patient groups in our study; significant differential effects were, however, observed in the biopsy-positive GCA group for IL-12 stimulated IL-6 and IL-23 stimulated IL-17A in PBMC cultures, and IL-23stimulated IL-6 and IL-8 in TA cultures. There was a trend towards greater increases in biopsy-positive patients for some other outcomes and it may be that these patients have more marked responses; however, we did see biological responses to IL-12 and IL-23 stimulation across the subgroups. This is in contrast to the majority of other cytokines studied in GCA where responses have been reported in biopsy-positive patients but not biopsy-negative patients.28 33 One explanation for this may be due to activity of IL-12 and IL-23 on stromal cells in addition to their activity on inflammatory cells.

There are several limitations to this study. We have used functional ex vivo models; replication in other models such

Figure 6 Effect of interleukin 12 (IL-12) and interleukin 23 (IL-23) stimulation in the myofibroblast outgrowth culture model. Representative images of the microscopic analysis of myofibroblast outgrowths (red arrows) from temporal artery sections. Experiments representing two patients are shown, first patient (A–D), second patient (E–H). (A), (E) Basal (Dulbecco’s Modified Eagle’s Medium (DMEM)); (B), (F) IL-23 (10 ng/mL in DMEM); (C), (G) IL-12 (50 pg/mL in DMEM); (D),(H) IL-12 (50 pg/mL)+IL-23 (10 ng/mL) (in DMEM). Original magnification 100x. IL-12 stimulation (50 pg/mL) significantly increased myofibroblast outgrowths at day 28, n=20 (I). IL-23 stimulation (10 ng/mL) significantly increased myofibroblast outgrowths at day 28, n=33 (J). IL-12 (50 pg/mL) and IL-23 (10 ng/mL) co-stimulation significantly increased myofibroblast outgrowths at day 28, n=20 (K). No significant additive effect was seen with IL-12 and IL-23 co-stimulation, n=20 (L). Data were analysed using Wilcoxon signed rank test and are expressed as mean±SE of the mean (SEM). *p<0.05, **p<0.01, ****p<0.0001.

1823Conway R, et al. Ann Rheum Dis 2018;77:1815–1824. doi:10.1136/annrheumdis-2018-213488

Basic and translational research

as the TA-severe combined immunodeficiency (SCID) chimera mouse would help confirm our results.42 The included patients were receiving glucocorticoids at the time of blood and tissue sampling, which may have influenced the results; however, GCA requires urgent treatment to prevent ischaemic complications and it would be unethical to withhold treatment to facilitate research sampling. Our results were not corrected for multiple testing; this is consistent with other papers in this field but does increase the risk of a type 1 error.28 43 We detected IL-12 and IL-23 at the protein level by immunohistochemistry but not at the gene level by PCR; these discordant results may possibly be explained by the low level of mRNA for these cytokines and the advantage of immunohistochemistry in terms of the visualisation of single cells. Finally, the disease control TAs used in this study represent a heterogeneous group of other diseases which while representing histologically normal arterial tissue may not be a true healthy control.

In conclusion, our results suggest key roles for IL-12 and IL-23 in the inflammatory and proliferative pathways involved in GCA pathogenesis.Acknowledgements We thank phil Gallagher, Karen Creevey, Michelle trenkmann, Hannah Convery, trudy McGarry, Megan Hanlon, Monika Biniecka and Mary Canavan for their technical support.

Contributors rC planned the study, performed experiments, analysed data and wrote the first draft of the manuscript. Lo’n, SK, SMW and AF performed experiments, analysed data, revised the manuscript for intellectual content and approved the final version. GMMcC, CCM, dJV, UF and ESM planned the study, contributed to the data analysis, revised the manuscript for intellectual content and approved the final version.

Funding the authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

Competing interests none declared.

Patient consent not required.

ethics approval this study was approved by St Vincent’s University Hospital ethics committee.

Provenance and peer review not commissioned; externally peer reviewed.

Author note the results from this study were presented at the European League Against rheumatism Congress 2018 and have been published in abstract form: Conway r et al. Interleukin-12 and Interleukin-23 are Key pathogenic players in Giant Cell Arteritis [abstract SAt0542]. Ann rheumdis.2018;77(Suppl):(A1125).

RefeRences 1 Crowson CS, Matteson EL, Myasoedova E, et al. the lifetime risk of adult-onset

rheumatoid arthritis and other inflammatory autoimmune rheumatic diseases. Arthritis Rheum 2011;63:633–9.

2 Weyand CM, Goronzy JJ. Immune mechanisms in medium and large-vessel vasculitis. Nat Rev Rheumatol 2013;9:731–40.

3 o’neill L, Molloy ES. the role of toll like receptors in giant cell arteritis. Rheumatology 2016;55:1921–31.

4 Ma-Krupa W, Jeon MS, Spoerl S, et al. Activation of arterial wall dendritic cells and breakdown of self-tolerance in giant cell arteritis. J Exp Med 2004;199:173–83.

5 Floss dM, Schröder J, Franke M, et al. Insights into IL-23 biology: From structure to function. Cytokine Growth Factor Rev 2015;26:569–78.

6 Langrish CL, McKenzie BS, Wilson nJ, et al. IL-12 and IL-23: master regulators of innate and adaptive immunity. Immunol Rev 2004;202:96–105.

7 Weyand CM, Younge Br, Goronzy JJ. IFn-γ and IL-17: the two faces of t-cell pathology in giant cell arteritis. Curr Opin Rheumatol 2011;23:43–9.

8 teng MW, Bowman Ep, McElwee JJ, et al. IL-12 and IL-23 cytokines: from discovery to targeted therapies for immune-mediated inflammatory diseases. Nat Med 2015;21:719–29.

9 Murphy CA, Langrish CL, Chen Y, et al. divergent pro- and antiinflammatory roles for IL-23 and IL-12 in joint autoimmune inflammation. J Exp Med 2003;198:1951–7.

10 Sherlock Jp, Joyce-Shaikh B, turner Sp, et al. IL-23 induces spondyloarthropathy by acting on ror-γt+ Cd3+Cd4-Cd8- entheseal resident t cells. Nat Med 2012;18:1069–76.

11 Chan Jr, Blumenschein W, Murphy E, et al. IL-23 stimulates epidermal hyperplasia via tnF and IL-20r2-dependent mechanisms with implications for psoriasis pathogenesis. J Exp Med 2006;203:2577–87.

12 deng J, Younge Br, olshen rA, et al. th17 and th1 t-cell responses in giant cell arteritis. Circulation 2010;121:906–15.

13 Visvanathan S, rahman MU, Hoffman GS, et al. tissue and serum markers of inflammation during the follow-up of patients with giant-cell arteritis--a prospective longitudinal study. Rheumatology 2011;50:2061–70.

14 Espígol-Frigolé G, Corbera-Bellalta M, planas-rigol E, et al. Increased IL-17A expression in temporal artery lesions is a predictor of sustained response to glucocorticoid treatment in patients with giant-cell arteritis. Ann Rheum Dis 2013;72:1481–7.

15 o’neill L, rooney p, Molloy d, et al. regulation of Inflammation and Angiogenesis in Giant Cell Arteritis by Acute-phase Serum Amyloid A. Arthritis Rheumatol 2015;67:2447–56.

16 Chatelain d, duhaut p, Schmidt J, et al. pathological features of temporal arteries in patients with giant cell arteritis presenting with permanent visual loss. Ann Rheum Dis 2009;68:84–8.

17 Espígol-Frigolé G, planas-rigol E, ohnuki H, et al. Identification of IL-23p19 as an endothelial proinflammatory peptide that promotes gp130-StAt3 signaling. Sci Signal 2016;9:ra28.

18 Conway r, o’neill L, o’Flynn E, et al. Ustekinumab for the treatment of refractory giant cell arteritis. Ann Rheum Dis 2016;75:1578–9.

19 Hoffman GS, Cid MC, rendt-Zagar KE, et al. Infliximab for maintenance of glucocorticosteroid-induced remission of giant cell arteritis: a randomized trial. Ann Intern Med 2007;146:621–30.

20 Yen d, Cheung J, Scheerens H, et al. IL-23 is essential for t cell-mediated colitis and promotes inflammation via IL-17 and IL-6. J Clin Invest 2006;116:1310–6.

21 La Flamme AC, Macdonald AS, pearce EJ. role of IL-6 in directing the initial immune response to schistosome eggs. J Immunol 2000;164:2419–26.

22 van der Geest KS, Abdulahad WH, rutgers A, et al. Serum markers associated with disease activity in giant cell arteritis and polymyalgia rheumatica. Rheumatology 2015;54:1397–402.

23 Hernández-rodríguez J, Segarra M, Vilardell C, et al. tissue production of pro-inflammatory cytokines (IL-1beta, tnFalpha and IL-6) correlates with the intensity of the systemic inflammatory response and with corticosteroid requirements in giant-cell arteritis. Rheumatology 2004;43:294–301.

24 García-Martínez A, Hernández-rodríguez J, Espígol-Frigolé G, et al. Clinical relevance of persistently elevated circulating cytokines (tumor necrosis factor alpha and interleukin-6) in the long-term followup of patients with giant cell arteritis. Arthritis Care Res 2010;62:835–41.

25 Hernández-rodríguez J, Segarra M, Vilardell C, et al. Elevated production of interleukin-6 is associated with a lower incidence of disease-related ischemic events in patients with giant-cell arteritis: angiogenic activity of interleukin-6 as a potential protective mechanism. Circulation 2003;107:2428–34.

26 Villiger pM, Adler S, Kuchen S, et al. tocilizumab for induction and maintenance of remission in giant cell arteritis: a phase 2, randomised, double-blind, placebo-controlled trial. The Lancet 2016;387:1921–7.

27 Stone JH, tuckwell K, dimonaco S, et al. trial of tocilizumab in Giant-Cell Arteritis. N Engl J Med 2017;377:317–28.

28 Corbera-Bellalta M, García-Martínez A, Lozano E, et al. Changes in biomarkers after therapeutic intervention in temporal arteries cultured in Matrigel: a new model for preclinical studies in giant-cell arteritis. Ann Rheum Dis 2014;73:616–23.

29 Li A, dubey S, Varney ML, et al. IL-8 directly enhanced endothelial cell survival, proliferation, and matrix metalloproteinases production and regulated angiogenesis. J Immunol 2003;170:3369–76.

30 ning Y, Manegold pC, Hong YK, et al. Interleukin-8 is associated with proliferation, migration, angiogenesis and chemosensitivity in vitro and in vivo in colon cancer cell line models. Int J Cancer 2011;128:2038–49.

31 Connolly M, Marrelli A, Blades M, et al. Acute serum amyloid A induces migration, angiogenesis, and inflammation in synovial cells in vitro and in a human rheumatoid arthritis/SCId mouse chimera model. J Immunol 2010;184:6427–37.

32 Hoeve MA, Savage nd, de Boer t, et al. divergent effects of IL-12 and IL-23 on the production of IL-17 by human t cells. Eur J Immunol 2006;36:661–70.

33 Corbera-Bellalta M, planas-rigol E, Lozano E, et al. Blocking interferon γ reduces expression of chemokines CxCL9, CxCL10 and CxCL11 and decreases macrophage infiltration in ex vivo cultured arteries from patients with giant cell arteritis. Ann Rheum Dis 2016;75:1177–86.

34 terrier B, Geri G, Chaara W, et al. Interleukin-21 modulates th1 and th17 responses in giant cell arteritis. Arthritis Rheum 2012;64:2001–11.

35 Ciccia F, rizzo A, Guggino G, et al. difference in the expression of IL-9 and IL-17 correlates with different histological pattern of vascular wall injury in giant cell arteritis. Rheumatology 2015;54:1596–604.

36 Shen F, ruddy MJ, plamondon p, et al. Cytokines link osteoblasts and inflammation: microarray analysis of interleukin-17- and tnF-alpha-induced genes in bone cells. J Leukoc Biol 2005;77:388–99.

37 Gaffen SL, Jain r, Garg AV, et al. the IL-23-IL-17 immune axis: from mechanisms to therapeutic testing. Nat Rev Immunol 2014;14:585–600.

38 Wang Y, Wu H, Wu x, et al. Interleukin 17A promotes gastric cancer invasiveness via nF-κB mediated matrix metalloproteinases 2 and 9 expression. PLoS One 2014;9 e96678.

1824 Conway R, et al. Ann Rheum Dis 2018;77:1815–1824. doi:10.1136/annrheumdis-2018-213488

Basic and translational research

39 Zerbini A, Muratore F, Boiardi L, et al. Increased expression of interleukin-22 in patients with giant cell arteritis. Rheumatology 2018;57:64-72.

40 Weyand CM, Goronzy JJ. Arterial wall injury in giant cell arteritis. Arthritis Rheum 1999;42:844–53.

41 planas-rigol E, terrades-Garcia n, Corbera-Bellalta M, et al. Endothelin-1 promotes vascular smooth muscle cell migration across the artery wall: a mechanism

contributing to vascular remodelling and intimal hyperplasia in giant-cell arteritis. Ann Rheum Dis 2017;76:1624–34.

42 Brack A, rittner HL, Younge Br, et al. Glucocorticoid-mediated repression of cytokine gene transcription in human arteritis-SCId chimeras. J Clin Invest 1997;99:2842–50.

43 Ciccia F, Alessandro r, rizzo A, et al. IL-33 is overexpressed in the inflamed arteries of patients with giant cell arteritis. Ann Rheum Dis 2013;72:258–64.

1825Mistry P, et al. Ann Rheum Dis 2018;77:1825–1833. doi:10.1136/annrheumdis-2018-213746

Basic and translational research

ExtEndEd rEport

Dysregulated neutrophil responses and neutrophil extracellular trap formation and degradation in PAPA syndromepragnesh Mistry,1 Carmelo Carmona-rivera,1 Amanda K ombrello,2 patrycja Hoffmann,2 nickie L Seto,1 Anne Jones,2 deborah L Stone,2 Faiza naz,3 philip Carlucci,1 Stefania dell’orso,3 Gustavo Gutierrez-Cruz,3 Hong-Wei Sun,4 daniel L Kastner,2 Ivona Aksentijevich,2 Mariana J Kaplan1

To cite: Mistry p, Carmona-rivera C, ombrello AK, et al. Ann Rheum Dis 2018;77:1825–1833.

Handling editor Josef S Smolen

► Additional material is published online only. to view please visit the journal online (http:// dx. doi. org/ 10. 1136/ annrheumdis- 2018- 213746).

1Systemic Autoimmunity Branch, national Institute of Arthritis and Musculoskeletal and Skin diseases (nIAMS), national Institutes of Health (nIH), Bethesda, Maryland, USA2Inflammatory disease Section, national Human Genome research Institute, nIH, Bethesda, Maryland, USA3Laboratory of Muscle Stem Cells and Gene regulation, nIAMS, nIH, Bethesda, Maryland, USA4Biodata Mining and discovery Section, nIAMS, nIH, Bethesda, Maryland, USA

Correspondence todr Mariana J Kaplan, Systemic Autoimmunity Branch, national Institute of Arthritis and Musculoskeletal and Skin diseases (nIAMS), national Institutes of Health (nIH), Bethesda, MA 20892, USA; mariana. kaplan@ nih. gov

received 10 May 2018revised 17 July 2018Accepted 1 August 2018published online First 21 August 2018

© Author(s) (or their employer(s)) 2018. no commercial re-use. See rights and permissions. published by BMJ.

Key messages

What is already known about this subject? ► Neutrophils are abundant in tissues of patients with pyoderma gangrenosum and acne (PAPA) syndrome but their role in the pathogenesis of this condition remains unclear.

What does this study add? ► This study demonstrates that PAPA syndrome is characterised by dysregulation in neutrophil phenotype and function, including enhanced formation and decreased clearance of neutrophil extracellular traps (NETs).

► These NETs infiltrate affected organs and promote activation of macrophages and exacerbation of inflammation.

How might this impact on clinical practice or future developments?

► As such, targeting aberrant neutrophil responses may be an effective therapeutic strategy in patients diagnosed with PAPA syndrome.

ABsTrACTObjectives pyogenic arthritis, pyoderma gangrenosum and acne (pApA) syndrome is characterised by flares of sterile arthritis with neutrophil infiltrate and the overproduction of interleukin (IL)-1β. the purpose of this study was to elucidate the potential role of neutrophil subsets and neutrophil extracellular traps (nEt) in the pathogenesis of pApA.Methods neutrophils and low-density granulocytes (LdG) were quantified by flow cytometry. Circulating nEts were measured by ELISA and pApA serum was tested for the ability to degrade nEts. the capacity of nEts from pApA neutrophils to activate macrophages was assessed. Skin biopsies were analysed for nEts and neutrophil gene signatures.results Circulating LdGs are elevated in pApA subjects. pApA neutrophils and LdGs display enhanced nEt formation compared with control neutrophils. pApA sera exhibit impaired nEt degradation and this is corrected with exogenous dnase1. recombinant human IL-1β induces nEt formation in pApA neutrophils but not healthy control neutrophils. nEt formation in healthy control neutrophils is induced by pApA serum and this effect is inhibited by the IL-1 receptor antagonist, anakinra. nEts from pApA neutrophils and LdGs stimulate IL-6 release in healthy control macrophages. nEts are detected in skin biopsies of patients with pApA syndrome in association with increased tissue IL-1β, IL-8 and IL-17. Furthermore, LdG gene signatures are detected in pApA skin.Conclusions pApA syndrome is characterised by an imbalance of nEt formation and degradation that may enhance the half-life of these structures in vivo, promoting inflammation. Anakinra ameliorates nEt formation in pApA and this finding supports a role for IL-1 signalling in exacerbated neutrophil responses in this disease. the study also highlights other inflammatory pathways potentially pathogenic in pApA, including IL-17 and IL-6, and these results may help guide new therapeutic approaches in this severe and often treatment-refractory condition.

InTrOduCTIOnAutoinflammatory disorders are characterised by abnormalities in the innate immune system resulting in excessive production of proinflam-matory cytokines and exuberant inflammatory

responses.1 2 One of these conditions is pyogenic arthritis, pyoderma gangrenosum and acne (PAPA) syndrome, a rare autosomal dominant inherited disorder caused by mutations in the proline-ser-ine-threonine phosphatase interacting protein 1 (PSTPIP1/CD2BP1) gene.3 PSTPIP1 is a cytoskel-eton-associated protein that acts as a bridge to proteins that interact with proline, glutamic acid, serine and threonine (PEST)-type protein tyro-sine phosphatase (PTP) (PEST-PTP) and regulates actin reorganisation,4 T cell activation,5 the phos-phorylation status of Wiskott-Aldrich syndrome protein6 and cAbl tyrosine kinase,7 and the release of interleukin (IL)-1β.8 The phosphorylation status of PSTPIP1 is likely important for its function.8 Three missense mutations in PSTPIP1 account for most patients with PAPA syndrome,9 resulting in a hyperphosphorylated protein that interacts more avidly with the pyrin protein inflammasome and leading to dysregulated IL-1β and downstream tumour necrosis factor (TNF)-α production in mononuclear cells.8 10 PAPA syndrome subjects

1826 Mistry P, et al. Ann Rheum Dis 2018;77:1825–1833. doi:10.1136/annrheumdis-2018-213746

Basic and translational research

present with a variable disease expressivity and severity of skin lesions. Within the clinical spectrum of disease, patients with milder presentations can often be effectively treated with the IL-1 receptor antagonist, anakinra.11 12 There are also patients who can achieve sustained clinical remission while taking the anti-TNF agents.10 However, there are patients with highly refractory disease who require combination (anti-IL-1/anti-TNF) or even triple (anti-IL-1/anti-TNF/glucocorticoid) therapy to effectively suppress the intense inflammation effects resulting from this disease.

A putative role for neutrophils has been considered in PAPA syndrome, given the presence of neutrophilic dermatoses and sterile joint effusions with prominent neutrophilic infiltrates.8 9 13 Indeed, patients with PAPA syndrome have elevated circulating levels of various neutrophil serine proteases and antimicrobial proteins including myeloperoxidase (MPO), neutrophil elastase (NE) and lactoferrin, suggesting global neutrophil activation and/or enhanced neutrophil cell death.9 Modular analyses of peripheral blood mononuclear cells (PBMC) from patients with PAPA syndrome previously revealed an overexpression of genes expressed in neutrophils and other myeloid cells.1 Among the upregulated neutrophil-associated genes were those encoding for neutrophil granule proteins including lipocalin 2 (LCN-2) and defensin A4, which are highly expressed in low-density granulocytes (LDG).1 14 LDGs are a proinflammatory subset of neutrophils that were first described in systemic lupus erythe-matosus (SLE) as a population of granulocytes that, due to their ‘low buoyancy’, sediment within the PBMC fraction after density centrifugation.15 In addition to SLE,14 16–18 LDGs have been reported in other conditions such as antineutrophil cyto-plasmic antibody (ANCA)-associated vasculitis19 and psoriasis20 but have yet to be described in autoinflammatory syndromes. LDGs represent a mixed population of neutrophils that express cell-surface markers of terminally differentiated neutrophils (CD15high/CD14lo/CD10+/CD16+)16 despite in some cases having less segmented nuclear morphology that is character-istic of immature neutrophils.17 Transcriptional analysis of lupus LDGs has been consistent with the presence of cells representing immature stages of neutrophil differentiation.14 17

Importantly, LDGs in SLE and other conditions have an enhanced capacity to spontaneously undergo neutrophil cell death through the formation of neutrophil extracellular traps (NET). NETs are an extracellular meshwork of chromatin fibres decorated with immunostimulatory molecules and various enzymes and proteins from the cytosol and neutrophil gran-ules.18 21 NETs can be induced with various infectious and sterile inflammatory stimuli including several cytokines.17 21–24 During NET formation, various histones and other peptides become post-translationally modified by the enzyme peptidyl arginine deiminase 4 through citrullination. Histone citrullination may contribute to chromatin decondensation, a hallmark of death by NET formation.25 26 Besides their antimicrobial role, NETs can cause vascular damage and stimulate immune cells in pleio-tropic ways, including promoting type I interferons (IFN) and proinflammatory cytokine synthesis by target cells if not readily cleared by host DNase enzymes.16 17 27 Given the putative role of neutrophils in PAPA syndrome, we characterised neutrophil subsets in this condition and determined a potential role of NETs in the disease pathogenesis.

MATerIAls And MeTHOdsMaterials and methods are included in the online supplementary material.

resulTsIdentification and characterisation of ldGs in PAPA syndromeThe demographic and clinical characteristics of patients studied are included in online supplementary tables 1 and 2. Among the subjects studied, genetic screening for PAPA mutations identi-fied a family that included five PAPA subjects that all carried the A230T mutation (online supplementary figure 1). To determine if LDGs are present in the circulation of PAPA subjects, we quan-tified by fluorescence-activated cell sorting (FACS) the presence of CD10−/+/CD14lo/CD15 cells in the PBMC fraction as previ-ously reported.14 16 28 When compared with healthy controls, both SLE (n=18; range 3.4%–46%) and PAPA subjects (n=12; range 0.7%–53%) had higher percentage of LDGs within their total PBMC fraction compared with healthy control subjects (n=5; range 0.9%–6.1%) (figure 1A). To determine the absolute number of LDGs, negative selection magnetic bead procedure was performed on PBMCs.14 16 We confirmed that both SLE and PAPA subjects had increased numbers of LDGs compared with healthy control subjects (figure 1B). We developed a scoring model to quantify disease activity in PAPA syndrome to deter-mine its correlation with LDG numbers. PAPA disease activity significantly associated with the percentage (β=0.699, p=0.004) and absolute numbers (β=0.623, p=0.01) of LDGs (figure 1C and data not shown).

Previous reports suggest that lupus LDGs represent, at least in part, a subset of immature neutrophils, given the elevated expression of mRNAs encoding for granule proteins.14 17 We performed RNASeq analysis on PAPA LDGs and compared gene expression with autologous normal-density PAPA neutrophils (NDN) and healthy control neutrophils. Indeed, similar to lupus LDGs, PAPA LDGs have increased gene expression of mRNAs encoding for granule proteins (figure 1D, online supplementary table 3). Findings were confirmed by qRT-PCR, where both PAPA LDGs and SLE LDGs express significantly higher mRNA levels of primary (MPO), secondary (LCN-2) and tertiary (MMP8) granule proteins compared with healthy control neutrophils and autologous normal-density PAPA and lupus neutrophils (figure 1E–H). In summary, elevated levels of LDGs are found in PAPA syndrome at comparable levels to those in SLE, and they associate with PAPA syndrome disease activity. PAPA LDGs upregulate mRNAs encoding for granule proteins, suggesting an immature phenotype.

PAPA neutrophils undergo enhanced neT formationTo further characterise neutrophil subsets in PAPA, we exam-ined their ability to undergo spontaneous NET formation when compared with controls and SLE. Unstimulated LDGs from PAPA and SLE subjects displayed an enhanced ability to form NETs compared with autologous NDNs and healthy control neutro-phils (figure 2A,B). PAPA NDNs also had significantly higher levels of NET formation compared with healthy control neutro-phils (p=0.0043) but to a lesser degree than LDGs (p=0.0022) (figure 2A,B). To further confirm whether enhanced NET formation occurs in vivo in PAPA syndrome, we quantified NET remnants in circulation using an assay previously described by our group and others.18 29 PAPA subjects had significantly higher levels of citrullinated histone H3-DNA complexes compared with healthy control individuals, suggesting that NET formation is enhanced in vivo in PAPA neutrophils (figure 2C).

Previous studies have reported that sera from a subset of patients with lupus degrade NET inefficiently due to disruptions in DNase1 activity.30 We tested the sera of 10 PAPA subjects for its ability to degrade NETs. We determined that 80% of the PAPA

1827Mistry P, et al. Ann Rheum Dis 2018;77:1825–1833. doi:10.1136/annrheumdis-2018-213746

Basic and translational research

Figure 1 Identification and characterisation of LDGs in PAPA syndrome. (A) Percentage of LDGs in the PBMC fraction of healthy controls (HC) (n=5), patients with SLE (n=18) and patients with PAPA syndrome (n=12). (B) Total LDG numbers per mL of whole blood in HC (n=5), patients with SLE (n=18) and patients with PAPA syndrome (n=10). (C) A linear regression analysis of the percentage of LDGs in PBMC fraction and disease activity (n=15). (D) RNASeq analysis of HC normal-density neutrophils (NDN), PAPA NDN and PAPA LDG (n=1/each group) identified upregulated expression of neutrophil granule genes in PAPA LDGs. (E–H) Total RNA was extracted from unstimulated HC NDN, SLE NDN and autologous LDGs, and PAPA NDN and autologous LDGs (n=5/group), and neutrophil granule gene expression was measured by qRT-PCR. Expression values were normalised to the expression of the glyceraldehyde 3-phosphate dehydrogenase (GAPDH) housekeeping gene and fold induction over HC NDN was calculated. The results shown in (A) to (H) represent the mean±SEM (*P<0.05; **P<0.01). LDG, low-density granulocyte; MPO, myeloperoxidase; PAPA, pyogenic arthritis, pyoderma gangrenosum and acne; PBMC, peripheral blood mononuclear cell; SLE, systemic lupus erythematosus.

subjects’ sera (n=8, called ‘PAPA1’) were impaired in the ability to degrade NETs when compared with healthy control sera and a minor subset of PAPA sera (20%, n=2, called ‘PAPA2’) that effi-ciently degraded NETs (figure 3A,B). To determine if this defi-ciency was due to insufficient serum DNase I activity, we added recombinant DNase I to the ‘PAPA1’ sera and this restored the ability to degrade NET efficiently (figure 3C,D). Taken together, both NDNs and LDGs in patients with PAPA syndrome display enhanced NET formation in vitro and in vivo. Elevated levels of circulating NET products may be due to an imbalance in NET formation and degradation caused by insufficient DNase I activity in the sera of PAPA subjects.

serum from PAPA subjects induces neT formation in an Il-1-dependent mannerHealthy control neutrophils were incubated with healthy control or PAPA serum and NET formation was measured. PAPA, but not healthy control, sera induced significant NET formation in healthy control neutrophils (figure 4A). Moreover, this process was dependent on IL-1 signalling as healthy control neutrophils pretreated with the IL-1R antagonist anakinra, prior to addition of PAPA serum, had a significant decrease in NET formation compared with untreated neutrophils (figure 4B). These results suggest that increases in serum IL-1β levels in PAPA contribute to enhanced NET formation.

Since IL-1 blockade abrogates PAPA serum-induced NETosis, we examined if recombinant human IL-1β would induce NETs in PAPA neutrophils. Recombinant IL-1β induced significant NET formation in PAPA neutrophils and this induction was reduced to basal levels with anakinra pretreatment (figure 4C). In contrast, healthy control neutrophils and lupus neutrophils did not undergo enhanced NET formation in response to recombinant IL-1β stim-ulation (figure 4D and online supplementary figure 2C). Previous studies have suggested that priming neutrophils with recombinant TNF-α can enhance their responsiveness to uric acid in gouty arthritis.31 We primed healthy control and lupus neutrophils with recombinant TNF-α and stimulated with recombinant IL-1β to recapitulate the enhanced NETosis in PAPA neutrophils after IL-1β stimulation. We did not observe a significant increase in NETosis in healthy control and lupus neutrophils after TNF-α priming and IL-1β stimulation when compared with unstimulated controls (online supplementary figure 2). To determine if the reason for increased responsiveness to IL-1 in PAPA neutrophils was due to elevated surface expression of IL-1R, we quantified this receptor by FACS in PAPA and healthy control neutrophils and found no significant difference (online supplementary figure 3). Collectively, these results suggest that IL-1β is one of the circulating NET-in-ducing molecules found in patients with PAPA syndrome and that PAPA neutrophils may be specifically primed to respond to IL-1β stimulation.

1828 Mistry P, et al. Ann Rheum Dis 2018;77:1825–1833. doi:10.1136/annrheumdis-2018-213746

Basic and translational research

Figure 2 PAPA neutrophils and LDGs undergo enhanced NET formation. (A) Representative images of unstimulated HC NDN, SLE NDN, SLE LDGs, PAPA NDN and PAPA LDGs isolated from peripheral blood after 1 hour (LDG) or 2 hours (NDN) of incubation at 37°C. NETs were identified by colocalisation of extracellular myeloperoxidase (MPO) (red) and DNA labelled with Hoechst 33342 (blue). Original magnification x40. (B) Quantification of the percentage of cells undergoing NET formation is plotted as mean±SEM (n=6 for HC and PAPA, n=5 for SLE, **p<0.01). (C) Circulating NET remnants in HC and PAPA subjects quantified (n=10 subjects/group) plotted as mean±SEM (*p<0.05). The OD index was calculated by dividing the individual OD450 values by the average of HC OD450 values. HC, healthy control; LDG, low-density granulocyte; NDN, normal-density neutrophil; NET, neutrophil extracellular trap; OP, optical density; PAPA, pyogenic arthritis, pyoderma gangrenosum and acne; SLE, systemic lupus erythematosus.

Proinflammatory cytokines are elevated in PAPA serumIt has previously been shown that proinflammatory cytokines including IL-1β, TNF-α, IL-17A and IFN-γ can induce/prime NET formation in neutrophils and that patients with PAPA syndrome have elevated levels of IL-1β and TNF-α in circula-tion.8–10 24 32 We next investigated if these cytokines were present in the serum of PAPA subjects to help explain the observed enhanced NET formation. PAPA subjects had significant elevations in circu-lating IL-1β, TNF-α, IL-6, IL-17A and IFN-γ while most of these cytokines were largely undetected in healthy controls (figure 5). Even the PAPA subjects who were in clinical remission had detect-able levels of circulating IL-1β and TNF-α (figure 5A,B). IL-17A, a cytokine not previously associated with PAPA but found in skin of patients with psoriasis,33 was elevated in three PAPA subjects with current or recent history of pyoderma gangrenosum (PG) lesions (figure 5C and online supplementary table 1). Linear regression analysis revealed that serum IL-6 levels correlated with circulating NET complexes (β=0.931, p=0.007) (figure 5F). Furthermore, NETs generated by PAPA neutrophils and LDGs significantly induced IL-1β and IL-6 mRNA gene expression and IL-6 protein release in healthy control monocyte-derived macrophages (figure 5G–I). In summary, PAPA is associated with elevated levels of proinflammatory cytokines that may contribute to dysregulated NET formation and these NETs in turn may enhance the produc-tion of proinflammatory mediators by macrophages.

neTs are present in PAPA skin lesionInfiltrating netting neutrophils were previously shown to be present in kidney, skin and placentas of patients with lupus14 34 and have been described in tissues in other diseases.24 29 Immuno-histochemistry was performed on the skin biopsy from a patient with PAPA syndrome with active PG despite combination immu-nosuppressive therapy and in skin tissue from another patient with PAPA syndrome with no active skin manifestations (figure 6A). A

meshwork of extracellular traps (ET) identified by positive extra-cellular citrullinated histone H4 (cit-H4) staining was present in the skin lesion but absent in the non-lesional tissue sample (figure 6B). We confirmed that the ETs in the skin lesion were derived from neutrophils because of colocalisation of extracellular cit-H4 with NE (online supplementary figure 4). RNA was purified from PAPA skin lesion as well as from healthy control skin. Transcripts of neutrophil granule genes previously found to be upregulated in lupus and vasculitis LDGs were quantified by real-time PCR.16 19 Neutrophil granule genes were highly expressed at the mRNA level in PAPA skin lesion and were significantly lower in skin tissue from a healthy control (figure 6F–H). This was confirmed at the protein level by quantifying skin MPO levels (online supplemen-tary figure 5). Expression of IL-1β mRNA and protein and IL-8 and IL-17A mRNA, but not TNF, IL-6 or IL-18, was significantly increased in the PAPA skin lesion (figure 6C–E, online supplemen-tary figure 5 and data not shown), when compared with control skin. RNASeq was performed on PAPA and healthy control skin samples to determine the transcriptional environment in active skin disease (PG). Ingenuity pathway analysis (IPA) revealed that upreg-ulated pathways included IL-17A signalling, and cytotoxic T cell (CTL) and natural killer (NK) cell signalling pathways (figure 6I, online supplementary table 7). Overall, these results suggest that enhanced NET formation by various neutrophil subsets may contribute to tissue injury in PAPA and that mutations in PSTPIP1 may cause dysregulation of CTL and NK cell function that may further contribute to tissue destruction.

dIsCussIOnDespite considerable advances into the molecular mechanism of inflammation in PAPA syndrome,8 the role that neutrophil subsets and NETs may play in the pathogenesis of this condition has yet to be explored. In recent years, some reports have revealed that neutrophils are dysregulated in PAPA syndrome.1 9 One particular

1829Mistry P, et al. Ann Rheum Dis 2018;77:1825–1833. doi:10.1136/annrheumdis-2018-213746

Basic and translational research

Figure 3 Impaired NET degradation in a subset of PAPA subjects. (A, B) HC normal-density neutrophils (NDN) were stimulated with PMA (2.5 µM) for 4 hours to induce NETs and subsequently incubated with 1% serum from either HC (n=5) or PAPA subjects (n=10) for 16 hours. ‘PAPA1’ represents PAPA subjects who did not degrade NETs efficiently and ‘PAPA2’ represents PAPA subjects that were efficient degraders of NETs. (C, D) HC NDNs were stimulated with PMA (2.5 µM) for 4 hours to induce NETs, followed by incubation with or without 1% serum from PAPA1 subjects (n=7) in the presence or absence of 1 U/mL DNase1 for 16 hours. (A) Represents the mean+SEM from three independent experiments (*P<0.05; **P<0.01; ***P<0.001). HC, healthy control; NET, neutrophil extracellular trap; PAPA, pyogenic arthritis, pyoderma gangrenosum and acne.

study identified a granulocyte gene expression signature in PBMCs from patients with PAPA syndrome, suggesting the presence of an abnormal subset of neutrophils called LDGs that were previously described in SLE.1 14 16–18 We sought to determine if LDGs were indeed present in the blood from patients with PAPA syndrome and if so, characterise their phenotype using LDGs from SLE as a bench-mark. The results of these studies suggest that circulating LDGs are elevated in PAPA and at comparable levels relative to LDGs in SLE.

We found that LDGs significantly associate with disease activity in PAPA, similar to what has been described in ANCA-associated vasculitis.19 Previous studies have reported that LDGs express higher levels of transcripts for serine proteases and bactericidal proteins found in azurophilic granules, secondary granules and tertiary granules of maturing neutrophils.14 17 Similar to SLE, this was confirmed in PAPA LDGs in our study.16 35 The expression of these genes is greatest during the myeloblast/promyelocytic stage

1830 Mistry P, et al. Ann Rheum Dis 2018;77:1825–1833. doi:10.1136/annrheumdis-2018-213746

Basic and translational research

Figure 4 PAPA serum induces NET formation in healthy control neutrophils in an IL-1-dependent manner. (A) HC normal-density neutrophils (NDN) were stimulated with 10% serum from either HC (n=4) or PAPA subjects (n=6) for 2.5 hours and NET formation was quantified and reported as mean±SEM (**P<0.01). (B) HC NDNs were pretreated with/without anakinra (149 µg/mL) for 15 min prior to stimulation with 10% PAPA serum (n=6) for 2.5 hours and NET formation was quantified and plotted as mean±SEM (**P<0.01). (C, D) PAPA or HC NDNs were left unstimulated or stimulated with recombinant human IL-1β (100 ng/mL) for 2.5 hours in the absence/presence of anakinra (149 µg/mL) pretreatment, respectively. (C) and (D) are the mean±SEM from four different samples for each group (*P<0.05). HC, healthy control; IL, interleukin; NET, neutrophil extracellular trap; PAPA, pyogenic arthritis, pyoderma gangrenosum and acne.

of neutrophil differentiation suggesting that PAPA LDGs represent an immature neutrophil subset.17 36 It is unclear why neutrophil precursors would be prematurely released from the bone marrow in PAPA but previous studies have reported that proinflammatory cytokines/chemokines produced during an inflammatory episode or infection can trigger the proliferation and release of immature granulocytes.37 Various chemokines including IL-8, macrophage inflammatory protein-2, leukotriene B4 and complement compo-nent C5a have been shown to regulate neutrophil mobilisation during inflammation.38–40 Granulocyte-macrophage colony-stimu-lating factor (GM-CSF) is one of the main cytokines that regulate the production and egress of neutrophils from the bone marrow.41 Indeed, while circulating levels of GM-CSF have yet to be reported for PAPA syndrome, SLE PBMCs produce increased GM-CSF and IL-1 signalling can promote the production of this cytokine.42 43 Future studies should address the role of GM-CSF in neutrophil dysregulation in PAPA.

It was previously described that patients with PAPA syndrome have elevated circulating levels of neutrophil granule enzymes,

suggesting a global neutrophil activation.9 The marked release of neutrophil granule enzymes into the periphery could be a result of enhanced degranulation or dysregulated NET forma-tion.14 17 21 Indeed, spontaneous NET formation was enhanced in PAPA LDGs, similar to what has been observed in SLE. These results were confirmed by identifying circulating NET remnants in PAPA and that the majority of these subjects exhibited enhanced NET formation and decreased ability of their serum to degrade NETs. Unlike SLE, where some serum samples could not rescue the ability to degrade NETs with the addition of recombinant DNase1 (suggesting the presence of DNase1-specific inhibitors),30 all of the poor-degrading PAPA serum samples restored their ability to disassemble NETs with the addition of recombinant DNase1. These results suggest that DNase1 activity in mutant PAPA cells is not sufficient to degrade an overproduction of NETs and that DNase1-specific inhibitors do not play a major role in aberrant NET clearance in PAPA. Future studies should focus on under-standing how DNase1 activity or levels may be perturbed in PAPA syndrome.

1831Mistry P, et al. Ann Rheum Dis 2018;77:1825–1833. doi:10.1136/annrheumdis-2018-213746

Basic and translational research

Figure 5 Proinflammatory cytokines are elevated in PAPA serum. (A–E) Serum concentrations of IL-1β, TNF-α, IFN-γ and IL-6 (A, B, D, E) and OD index for IL-17A (C) in HC and PAPA subjects (n=9/group). Results represent mean±SEM (*P<0.05; **P<0.01). (F) A linear regression analysis of the OD index from citrullinated histone (H3)-DNA complexes in the plasma from PAPA subjects and IL-6 protein in PAPA sera (n=6). (G–I) HC macrophages were stimulated with spontaneously formed NETs from PAPA NDN or LDGs for 2 hours (G, H) or 24 hours (I). (G, H) Total RNA was isolated and qRT-PCR was performed to measure IL-1β and IL-6 gene expression; results represent mean±SEM from four independent experiments (*P<0.05). (I) Culture supernatants were analysed by ELISA for IL-6 protein and results represent the mean±SEM from four independent experiments (*P<0.05). HC, healthy control; IFN, interferon; IL, interleukin; LDG, low-density granulocyte; NDN, normal-density neutrophil; NET, neutrophil extracellular trap; OD, optical density; PAPA, pyogenic arthritis, pyoderma gangrenosum and acne; TNF, tumour necrosis factor.

The enhanced NET formation in PAPA is likely driven by inflam-matory stimuli characteristic of this disease, as serum from these patients stimulated NET formation in healthy control neutrophils in an IL-1-dependent manner. As control neutrophils formed NETs after stimulation with PAPA serum but not recombinant IL-1β, it is possible that PAPA neutrophils are primed due to chronic inflam-mation and other proinflammatory cytokines. It has previously been reported that priming of healthy neutrophils is necessary to induce responses to various stimuli, including toll-like receptor agonists and cytokines.44 To support this hypothesis, the serum of PAPA subjects showed elevated levels of several proinflamma-tory cytokines that may prime neutrophils and make them more prone to NET formation. Furthermore, NETs derived from PAPA neutrophils and LDGs induced the release of IL-6 by macrophages, suggesting a vicious cycle where NETs are induced by proinflam-matory cytokines and these NETs in turn enhance production of proinflammatory mediators. In a subset of PAPA subjects, mark-edly elevated levels of circulating IL-17 were detected. IL-17 has been associated with conditions characterised by skin inflamma-tion including psoriasis and atopic dermatitis.45 Of note, all the PAPA subjects with increased circulating IL-17A had history of cystic acne or PG and we found evidence of marked upregulation of IL-17 transcripts in PAPA skin. These observations should be

systematically studied in future studies to better understand the source of cells contributing to higher IL-17 levels in PAPA, as well as the pathogenic role that this cytokine has in the disease and whether it could represent an important therapeutic target.

One of the most difficult aspects of treating PAPA syndrome is PG refractory to many immunosuppressive and immunomodula-tory therapies.12 46 In proof of concept studies, we found evidence of significant NET formation in PG skin in PAPA but not in unaf-fected skin. This suggests that NETs may contribute to tissue destruction in PAPA syndrome. LDGs have been previously shown to cause vascular damage in lupus and netting neutrophils were reported in skin and kidneys from patients with dermatological and renal manifestations.14 16 Although the study by Villanueva et al was unable to determine whether these cells were neutrophils or LDGs, we made this differentiation in our PAPA skin lesion.14 Neutrophils and LDGs share many similarities including surface markers and the ability to form NETs,16 17 but they can be distin-guished based on their gene expression of granule proteins.14 17 19 Indeed, PAPA skin lesion expressed elevated mRNA levels of genes encoding for granule proteins previously reported for SLE LDGs. Given higher IL-1β, IL-17 and IL-8 expression in PAPA skin, it is possible that local production of these cytokines in the skin microenvironment promotes neutrophil recruitment, in situ skin

1832 Mistry P, et al. Ann Rheum Dis 2018;77:1825–1833. doi:10.1136/annrheumdis-2018-213746

Basic and translational research

Figure 6 Neutrophil extracellular traps (NET) are present in PAPA skin lesion. (A) Pyoderma gangrenosum lesion that developed in a patient with PAPA syndrome with the A230T mutation. The lesion developed while receiving anakinra, infliximab and prednisone at doses up to 60 mg/daily. A skin biopsy taken from this lesion was obtained and used to perform experiments depicted in (B)–(I). (B) Immunohistochemistry was performed on skin lesions from the patient with PAPA syndrome described in (A) and in non-lesional skin sample from another PAPA subject. NETs are identified by colocalisation of citrullinated histone H4 (cit-H4) (red) and DNA labelled with Hoechst 33342 (blue). Main figure scale bar 100 µM, inset scale bar 20 µM. (C–H) Total RNA was extracted from HC skin biopsy and from lesional skin from the patient with PAPA syndrome described in (A). Proinflammatory cytokine and neutrophil granule gene expression were measured by qRT-PCR, normalised to the expression of GAPDH, and fold induction was calculated to HC skin. (I) RNASeq analysis of HC skin (n=1) and PAPA lesional skin (n=1) identified upregulated expression of neutrophil/LDG, IL-17A signalling and cytotoxic T cell and NK cell signalling genes. CTL, cytotoxic T cell; HC, healthy control; IL, interleukin; LDG, low-density granulocyte; MPO, myeloperoxidase; NFAT, nuclear factor of activated T cells; NK, natural killer; PAPA, pyogenic arthritis, pyoderma gangrenosum and acne.

NET formation and tissue damage. Indeed, RNASeq on PAPA skin tissue uncovered numerous pathways related to neutrophil processes including granulocyte adhesion, diapedesis and IL-17 and IL-8 signalling, further supporting the role of neutrophils in skin pathology in PAPA. Of note, we observed that eicosanoid signalling was upregulated in PAPA skin and Denny et al previously reported that SLE LDGs secrete enhanced levels of eicosanoids.16 The RNASeq analysis also revealed that CTL, NK and granzyme B signalling were upregulated in PAPA skin. The role of these cyto-toxic pathways in PAPA skin involvement remains to be better characterised in future studies.

Aberrant neutrophil phenotypes have previously been described in other autoinflammatory conditions47 and we now report that

neutrophil subsets are dysregulated in PAPA syndrome in associa-tion with enhanced NET formation in blood and tissues that may further exacerbate inflammatory processes by other myeloid cells and contribute to the dysregulated immune microenvironment in this disease. Within the field of autoinflammatory diseases, where a number of conditions have had marked expansion in targeted treatment options, patients with PAPA syndrome remain refrac-tory to treatment in a number of cases. At times, this necessi-tates the use of dual biologic treatment such as anti-IL1/anti-TNF agents to try and decrease the inflammatory cascade. This study suggests that other pathways may potentially represent novel ther-apeutic targets. In addition to IL-17, future studies should address the role of specifically targeting NET formation as a therapeutic

1833Mistry P, et al. Ann Rheum Dis 2018;77:1825–1833. doi:10.1136/annrheumdis-2018-213746

Basic and translational research

option in this group of conditions, as recently shown for autoim-mune diseases.48 A better understanding of neutrophil biology in syndromes with aberrant innate immune responses should provide better insight into the pathogenesis of the disease and further ther-apeutic development.Contributors pM, CCr, nLS and pC performed the experiments. AKo, pH, AJ, dLS, dLK and IA characterised the clinical cohort, patient recruitment and genetic analysis/pedigree. Fn, Sdo and GGC were involved in rnA sequencing support. HWS was involved in analysis of rnA sequencing. pM and MJK conceived the study and drafted the manuscript.

Funding this study was supported by the Intramural research program at nIAMS/nIH (ZIAAr041199 and the office of Science and technology nIAMS). this study utilised the high-performance computational capabilities of the Helix Systems at the nIH (http:// helix. nih. gov).

Competing interests none declared.

Patient consent not required.

ethics approval nIddK/nIAMS IrB.

Provenance and peer review not commissioned; externally peer reviewed.

data sharing statement the rnA sequencing data have been submitted to GEo and a GEo accession number will be provided at the time of publication (currently pending).

RefeRences 1 Smith EJ, Allantaz F, Bennett L, et al. Clinical, molecular, and genetic characteristics of

pApA syndrome: a review. Curr Genomics 2010;11:519–27. 2 Masters SL, Simon A, Aksentijevich I, et al. Horror autoinflammaticus: the

molecular pathophysiology of autoinflammatory disease (*). Annu Rev Immunol 2009;27:621–68.

3 Wise CA, Gillum Jd, Seidman CE, et al. Mutations in Cd2Bp1 disrupt binding to ptp pESt and are responsible for pApA syndrome, an autoinflammatory disorder. Hum Mol Genet 2002;11:961–9.

4 Spencer S, dowbenko d, Cheng J, et al. pStpIp: a tyrosine phosphorylated cleavage furrow-associated protein that is a substrate for a pESt tyrosine phosphatase. J Cell Biol 1997;138:845–60.

5 Badour K, Zhang J, Shi F, et al. the Wiskott-Aldrich syndrome protein acts downstream of Cd2 and the Cd2Ap and pStpIp1 adaptors to promote formation of the immunological synapse. Immunity 2003;18:141–54.

6 Côté JF, Chung pL, théberge JF, et al. pStpIp is a substrate of ptp-pESt and serves as a scaffold guiding ptp-pESt toward a specific dephosphorylation of WASp. J Biol Chem 2002;277:2973–86.

7 Cong F, Spencer S, Côté JF, et al. Cytoskeletal protein pStpIp1 directs the pESt-type protein tyrosine phosphatase to the c-Abl kinase to mediate Abl dephosphorylation. Mol Cell 2000;6:1413–23.

8 Shoham nG, Centola M, Mansfield E, et al. pyrin binds the pStpIp1/Cd2Bp1 protein, defining familial Mediterranean fever and pApA syndrome as disorders in the same pathway. Proc Natl Acad Sci U S A 2003;100:13501–6.

9 demidowich Ap, Freeman AF, Kuhns dB, et al. Brief report: genotype, phenotype, and clinical course in five patients with pApA syndrome (pyogenic sterile arthritis, pyoderma gangrenosum, and acne). Arthritis Rheum 2012;64:2022–7.

10 Cortis E, de Benedetti F, Insalaco A, et al. Abnormal production of tumor necrosis factor (tnF) – alpha and clinical efficacy of the tnF inhibitor etanercept in a patient with pApA syndrome [corrected]. J Pediatr 2004;145:851–5.

11 dierselhuis Mp, Frenkel J, Wulffraat nM, et al. Anakinra for flares of pyogenic arthritis in pApA syndrome. Rheumatology 2005;44:406–8.

12 Brenner M, ruzicka t, plewig G, et al. targeted treatment of pyoderma gangrenosum in pApA (pyogenic arthritis, pyoderma gangrenosum and acne) syndrome with the recombinant human interleukin-1 receptor antagonist anakinra. Br J Dermatol 2009;161:1199–201.

13 Callen Jp. neutrophilic dermatoses. Dermatol Clin 2002;20:409–19. 14 Villanueva E, Yalavarthi S, Berthier CC, et al. netting neutrophils induce endothelial

damage, infiltrate tissues, and expose immunostimulatory molecules in systemic lupus erythematosus. J Immunol 2011;187:538–52.

15 Hacbarth E, Kajdacsy-Balla A. Low density neutrophils in patients with systemic lupus erythematosus, rheumatoid arthritis, and acute rheumatic fever. Arthritis Rheum 1986;29:1334–42.

16 denny MF, Yalavarthi S, Zhao W, et al. A distinct subset of proinflammatory neutrophils isolated from patients with systemic lupus erythematosus induces vascular damage and synthesizes type I IFns. J Immunol 2010;184:3284–97.

17 Carmona-rivera C, Kaplan MJ. Low-density granulocytes: a distinct class of neutrophils in systemic autoimmunity. Semin Immunopathol 2013;35:455–63.

18 Lood C, Blanco Lp, purmalek MM, et al. neutrophil extracellular traps enriched in oxidized mitochondrial dnA are interferogenic and contribute to lupus-like disease. Nat Med 2016;22:146–53.

19 Grayson pC, Carmona-rivera C, xu L, et al. neutrophil-related gene expression and low-density granulocytes associated with disease activity and response to treatment in antineutrophil cytoplasmic antibody-associated vasculitis. Arthritis Rheumatol 2015;67:1922–32.

20 Lin AM, rubin CJ, Khandpur r, et al. Mast cells and neutrophils release IL-17 through extracellular trap formation in psoriasis. J Immunol 2011;187:490–500.

21 Brinkmann V, reichard U, Goosmann C, et al. neutrophil extracellular traps kill bacteria. Science 2004;303:1532–5.

22 Kenny EF, Herzig A, Krüger r, et al. diverse stimuli engage different neutrophil extracellular trap pathways. Elife 2017;6.

23 Gupta AK, Hasler p, Holzgreve W, et al. Induction of neutrophil extracellular dnA lattices by placental microparticles and IL-8 and their presence in preeclampsia. Hum Immunol 2005;66:1146–54.

24 Khandpur r, Carmona-rivera C, Vivekanandan-Giri A, et al. nEts are a source of citrullinated autoantigens and stimulate inflammatory responses in rheumatoid arthritis. Sci Transl Med 2013;5:178ra40.

25 Smith CK, Kaplan MJ. the role of neutrophils in the pathogenesis of systemic lupus erythematosus. Curr Opin Rheumatol 2015;27:448–53.

26 Wang Y, Li M, Stadler S, et al. Histone hypercitrullination mediates chromatin decondensation and neutrophil extracellular trap formation. J Cell Biol 2009;184:205–13.

27 Lande r, Ganguly d, Facchinetti V, et al. neutrophils activate plasmacytoid dendritic cells by releasing self-dnA-peptide complexes in systemic lupus erythematosus. Sci Transl Med 2011;3:73ra19.

28 Marini o, Costa S, Bevilacqua d, et al. Mature Cd10+ and immature Cd10- neutrophils present in G-CSF-treated donors display opposite effects on t cells. Blood 2017;129:1343–56.

29 Kessenbrock K, Krumbholz M, Schönermarck U, et al. netting neutrophils in autoimmune small-vessel vasculitis. Nat Med 2009;15:623–5.

30 Hakkim A, Fürnrohr BG, Amann K, et al. Impairment of neutrophil extracellular trap degradation is associated with lupus nephritis. Proc Natl Acad Sci U S A 2010;107:9813–8.

31 Yokose K, Sato S, Asano t, et al. tnF-α potentiates uric acid-induced interleukin-1β (IL-1β) secretion in human neutrophils. Mod Rheumatol 2018;28:513–7.

32 Mitroulis I, Kambas K, Chrysanthopoulou A, et al. neutrophil extracellular trap formation is associated with IL-1β and autophagy-related signaling in gout. PLoS One 2011;6:e29318.

33 Frieder J, Kivelevitch d, Menter A. Secukinumab: a review of the anti-IL-17A biologic for the treatment of psoriasis. Ther Adv Chronic Dis 2018;9:5–21.

34 Marder W, Knight JS, Kaplan MJ, et al. placental histology and neutrophil extracellular traps in lupus and pre-eclampsia pregnancies. Lupus Sci Med 2016;3:e000134.

35 Bennett L, palucka AK, Arce E, et al. Interferon and granulopoiesis signatures in systemic lupus erythematosus blood. J Exp Med 2003;197:711–23.

36 Cowland JB, Borregaard n. the individual regulation of granule protein mrnA levels during neutrophil maturation explains the heterogeneity of neutrophil granules. J Leukoc Biol 1999;66:989–95.

37 Cain dW, Snowden pB, Sempowski Gd, et al. Inflammation triggers emergency granulopoiesis through a density-dependent feedback mechanism. PLoS One 2011;6:e19957.

38 terashima t, English d, Hogg JC, et al. release of polymorphonuclear leukocytes from the bone marrow by interleukin-8. Blood 1998;92:1062–9.

39 Jagels MA, Chambers Jd, Arfors KE, et al. C5a- and tumor necrosis factor-alpha-induced leukocytosis occurs independently of beta 2 integrins and L-selectin: differential effects on neutrophil adhesion molecule expression in vivo. Blood 1995;85:2900–9.

40 Burdon pC, Martin C, rankin SM. the CxC chemokine MIp-2 stimulates neutrophil mobilization from the rat bone marrow in a Cd49d-dependent manner. Blood 2005;105:2543–8.

41 pollmächer t, Korth C, Schreiber W, et al. Effects of repeated administration of granulocyte colony-stimulating factor (G-CSF) on neutrophil counts, plasma cytokine, and cytokine receptor levels. Cytokine 1996;8:799–803.

42 Willeke p, Schlüter B, Schotte H, et al. Increased frequency of GM-CSF secreting pBMC in patients with active systemic lupus erythematosus can be reduced by immunoadsorption. Lupus 2004;13:257–62.

43 Suzuki A, takahashi t, okuno Y, et al. IL-1 production as a regulator of G-CSF and IL-6 production in CSF-producing cell lines. Br J Cancer 1992;65:515–8.

44 Hayashi F, Means tK, Luster Ad. toll-like receptors stimulate human neutrophil function. Blood 2003;102:2660–9.

45 tesmer LA, Lundy SK, Sarkar S, et al. th17 cells in human disease. Immunol Rev 2008;223:87–113.

46 Stichweh dS, punaro M, pascual V. dramatic improvement of pyoderma gangrenosum with infliximab in a patient with pApA syndrome. Pediatr Dermatol 2005;22:262–5.

47 Manukyan G, petrek M, Kriegova E, et al. Activated phenotype of circulating neutrophils in familial Mediterranean fever. Immunobiology 2013;218:892–8.

48 Knight JS, Subramanian V, o’dell AA, et al. peptidylarginine deiminase inhibition disrupts nEt formation and protects against kidney, skin and vascular disease in lupus-prone MrL/lpr mice. Ann Rheum Dis 2015;74:2199–206.

1835Ann Rheum Dis December 2018 Vol 77 No 12

Letters

Table 1 Characteristics of patients presenting with demyelinating disorders during TNF inhibitor (TNFi) therapy

Total n=24

Demographics

Age, years median (range) 40 (30–75)

Female n (%) 16 (67)

Neurological involvement

Central n (%) 15 (62.5)

Among central optic neuritis n (%) 9 (37.5)

Peripheral n (%) 9 (37.5)

Underlying disease

Spondyloarthritis n (%) 13 (54.2)

Rheumatoid arthritis n (%) 8 (33.3)

Ulcerative colitis n (%) 3 (12.5)

Last TNFi before demyelination

Infliximab n (%) 11 (46)

Etanercept n (%) 8 (33)

Adalimumab n (%) 4 (17)

Golimumab n (%) 1 (4)

Relapse of demyelination n (%) 3 (12.5)

Table 2 Percentage of carriage of the different alleles in the case and control population

Single nucleotide polymorphism Allele

Rheumatoid arthritis cases with demyelination N (%)

Rheumatoid arthritis cases with central demyelination N (%)

Rheumatoid arthritis cases with peripheral demyelination N (%)

Multiple sclerosis (MS) population from GWAS ref.5

Rheumatoid arthritis controls N (%)

rs1800693 A 29 (60.5%) 18 (60%) 11 (61.1%) 55% 360 (61.4%)

G (MS risk allele) 19 (39.5%) 12 (40%) 7 (38.9%) 45% 228 (38.6%)

rs4149584 C 47 (98%) 29 (96.7%) 18 (100%) 97.8 227 (98.7%)

T (MS risk allele) 1 (2.1%) 1 (3.3%) 0 (0%) 2.2% 3 (1.3%)

GWAS; genome wide association study.

Frequency of tumour necrosis factor alpha receptor superfamily 1A multiple sclerosis-associated variants in patients with rheumatoid arthritis with anti-tumour necrosis factor therapy-related demyelinating complications

Tumour necrosis factor alpha (TNFα) is a key cytokine in inflam-matory rheumatic diseases but is also implicated in multiple scle-rosis (MS) where TNFα is elevated in demyelinating plaques. However, TNFα inhibitors (TNFi) have opposite effects: they have revolutionised the treatment of chronic rheumatic diseases but may cause flares of MS.1 Several studies have suggested an association between TNFi use and occurrence of demyelin-ating disorders.2 3 Their incidence is estimated up to 4% after 18 months of treatment in a recent prospective study with systematic neurological examination before and after treatment initiation.4

In addition, two TNF receptor superfamily 1 (TNFRSF1A) single nucleotide polymorphisms, (SNPs) rs1800693 and rs4149584 (TNFRSF1A R92Q), have been shown to increase the risk of developing MS.5 The rs1800693*G allele leads to a dysfunctional TNFα soluble receptor that inhibits TNFα signal-ling.6 Rare variants of TNFRSF1A locus have also been implicated in an autoinflammatory disease called TNF receptor-associated periodic syndrome.7

The aim of this study was to test for the association of the TNFRSF1A MS risk variants with TNFi-related demyelinating complications (central or peripheral) in the rheumatoid arthritis population.

Cases were recruited between March 2013 and December 2015, through successive newsletters over 6 months to report cases sent via the Club Rhumatismes et Inflammation (CRI) nationwide network (http://www. CRI- net. com), a section of the French Society of Rheumatology of almost 2400 clinicians using biological treatments for patients with immune-mediated inflammatory diseases. Patients with rheumatoid arthritis treated with TNFi, from the ReAct cohort8 or rheumatoid arthritis with interstitial lung disease,9 and who did not develop demyelin-ating complication constituted the control population (n=294). The local ethics committee approved the study and patients gave their informed consent. Genotypes were established after DNA extraction from peripheral blood mononuclear cells using TaqMan allelic discrimination in cases, with specific probes for the rs1800693 and rs4149584 SNPs (Thermo Fisher Scientific, France) or whole exome sequencing in controls.9 Frequencies of carriage of either of the alleles were compared using a Fisher’s exact test.

Twenty-four cases from 11 centres presented demyelinating disorders. Their median age was 39.7 years (range 30.4–75.8) and 16 (67%) were women. Neurological symptoms occurred after a median of 18.3 months (1–66) of anti-TNF therapy; 15 (62.5%) had central neurological involvement and 9 (37.5%) had peripheral involvement. The median follow-up was 26 months (4–54) for cases and 30 months for controls (table 1).

Genotyping studies revealed no significant difference in the frequency of the rs1800693 and rs4149584 MS risk alleles between cases and controls (table 2).

This study did not find an association between TNFRSF1A MS-associated SNPs and the occurrence of demyelination events related to TNFi. However, as our study was under-powered to detect association signals with an OR lower than 4 for rs1800693, we cannot definitely exclude the contribu-tion of these TNFRSF1A risk variants for lower ORs. Thus, we saw a non-significant increase of the percentage of T MS risk allele of rs4149584 in cases, especially in central demy-elinating involvement (1.3% vs 3.3% for controls and cases, respectively). We acknowledge the effect that mixed ances-tral groups would have on detecting a difference in allele frequencies between cases and controls. Patients of Asian ancestries have a lower frequency of the risk allele for both SNPs compared with Europeans. This might have favoured the absence of difference observed. This effect cannot be measured since gathering information on ethnic groups is not allowed in France.

The reported TNFRSF1A MS risk variants do not seem to be associated with the risk of TNFi-related demyelinating complications.

1836 Ann Rheum Dis December 2018 Vol 77 No 12

Letters

Samuel Bitoun,1 Corinne Miceli-Richard,1 Céline Verstuyft,2 Pierre Antoine Juge,3 Philippe Dieudé,3 Jean-Marie Berthelot,4 Christophe Richez,5 Cécile Cauquil,6 Christelle Sordet,7 Sylvie Melac-Ducamp,8 Laure Gossec,9,10 Beatrice Bouvard,11 Emmanuelle Dernis,12 Eric Houvenagel,13 Marie-Astrid Boutry-Bacle,14 Xavier Mariette,1 Raphaèle Seror1

1Department of Rheumatology, Université Paris-Sud, AP-HP, Hôpitaux Universitaires Paris-Sud, INSERM U1184, Le Kremlin Bicêtre, France2Department of Pharmacogenetics, Hôpital de Bicètre, Le Kremlin Bicetre, France3Department of Rheumatology, CHU Bichat, Paris, France4Department of Rheumatology, C.H.U. Hôtel Dieu, Nantes, France5Department of Rheumatology, C.H.U Pellegrin, Bordeaux, France6Department of Neurology, Hopital de Bicètre, Le Kremlin Bicètre, France7Rhumatologie, C.H.R.U. Hôpitaux Universitaires Strasbourg, Strasbourg, France8Department of Rheumatology, C.H.I, Nevers, France9Sorbonne Universités, UPMC Univ Paris 06, GRC-08, Institut Pierre Louis d’Epidémiologie et de Santé Publique, Paris, France10Rheumatology Department, Pitie-Salpétrière Hospital, AP-HP, Paris, France11Department of Rheumatology, Centre Hospitalier Universitaire d’Angers, Angers, France12Department of Rheumatology, Centre Hospitalier du Mans, Le Mans, France13Department of Rheumatology, Hôpital Saint Philibert, Lomme, France14Department of Rheumatology Internal medicine, Centre Hospitalier, Abbeville, France

Correspondence to Dr Raphaèle Seror, Université Paris-Sud, AP-HP, Hôpitaux Universitaires Paris-Sud, INSERM U1184, and Centre Hospitalier Universitaire de Bicêtre, Hôpital Bicêtre, Le Kremlin Bicêtre, France; raphaele. seror@ aphp. fr

Handling editor Josef S Smolen

Acknowledgements We thank Pfizer for funding this study with the Passerelle Grant.

Contributors SB, CM-R, RS and XM designed the study. All authors acquired the data. SB, CM-R, RS, XM, PAJ, CV and PD analysed the data. All authors gave final approval to the paper.

Funding This study was funded by Pfizer (10.13039/100004319) and grant number: Passerelle.

Competing interests None declared.

Patient consent Not required.

Ethics approval This study was approved by the local institutional review board named CPP Île de France VII under number PP 12-030.

Provenance and peer review Not commissioned; externally peer reviewed.

© Author(s) (or their employer(s)) 2018. No commercial re-use. See rights and permissions. Published by BMJ.

This work has already been presented at the European League Against Rheumatism and published as a conference abstract [10].

To cite Bitoun S, Miceli-Richard C, Verstuyft C, et al. Ann Rheum Dis 2018;77:1835–1836.

Received 4 February 2018Revised 1 July 2018Accepted 3 July 2018Published Online First 20 July 2018

Ann Rheum Dis 2018;77:1835–1836. doi:10.1136/annrheumdis-2018-213183

RefeRences 1 TNF neutralization in MS: results of a randomized, placebo-controlled multicenter

study. The Lenercept Multiple Sclerosis Study Group and The University of British Columbia MS/MRI Analysis Group. Neurology 1999;53:457–65.

2 Dreyer L, Magyari M, Laursen B, et al. Risk of multiple sclerosis during tumour necrosis factor inhibitor treatment for arthritis: a population-based study from DANBIO and the Danish Multiple Sclerosis Registry. Ann Rheum Dis 2016;75:785–6.

3 Seror R, Richez C, Sordet C, et al. Pattern of demyelination occurring during anti-TNF-α therapy: a French national survey. Rheumatology 2013;52:868–74.

4 Kaltsonoudis E, Zikou AK, Voulgari PV, et al. Neurological adverse events in patients receiving anti-TNF therapy: a prospective imaging and electrophysiological study. Arthritis Res Ther 2014;16:R125.

5 De Jager PL, Jia X, Wang J, et al. Meta-analysis of genome scans and replication identify CD6, IRF8 and TNFRSF1A as new multiple sclerosis susceptibility loci. Nat Genet 2009;41:776–82.

6 Gregory AP, Dendrou CA, Attfield KE, et al. TNF receptor 1 genetic risk mirrors outcome of anti-TNF therapy in multiple sclerosis. Nature 2012;488:508–11.

7 Aganna E, Aksentijevich I, Hitman GA, et al. Tumor necrosis factor receptor-associated periodic syndrome (TRAPS) in a Dutch family: evidence for a TNFRSF1A mutation with reduced penetrance. Eur J Hum Genet 2001;9:63–6.

8 Umiċeviċ Mirkov M, Cui J, Vermeulen SH, et al. Genome-wide association analysis of anti-TNF drug response in patients with rheumatoid arthritis. Ann Rheum Dis 2013;72:1375–81.

9 Juge PA, Borie R, Kannengiesser C, et al. Shared genetic predisposition in rheumatoid arthritis-interstitial lung disease and familial pulmonary fibrosis. Eur Respir J 2017;49:1602314.

1836 Ann Rheum Dis December 2018 Vol 77 No 12

Letters

In RA, becoming seronegative over the first year of treatment does not translate to better chances of drug-free remission

In rheumatoid arthritis (RA), it is becoming common to attempt to taper or stop medication, aiming for sustained drug-free remission (SDFR). Autoantibody seropositivity is a poor prog-nostic factor for this treatment goal. However, autoantibody levels may change and patients may become seronegative, some-times termed ‘immunological remission’.1 Understanding how often this occurs and whether it is favourable for achieving SDFR is important to determine whether becoming seronegative is a meaningful prognostic marker for drug tapering decisions. Furthermore, it will elucidate pathways that lead to long-term resolution of the pathophysiology underlying RA.

To that end, we investigated the relationship between serocon-version and SDFR. In baseline and 1-year serum of 381 patients with seropositive RA, we measured 14 RA-associated autoan-tibodies by ELISA2: anti-CCP2 IgG, IgM and IgA; rheumatoid factor IgM and IgA; anti-CarP IgG, IgM and IgA; anti-acetyl-ated lysine vimentin IgG and anti-acetylated ornithine vimentin IgG (Orgentec Diagnostika , Germany); and anti-ci-trullinated vimentin 59–74 IgG, anti-citrullinated fibrinogen β 36–52 and α 27–43 IgG, anti-citrullinated enolase 5–20 IgG. Patients originated from the IMPROVED study,3 a randomised controlled treat-to-target trial of early (<2 years) untreated RA, steered at disease activity score remission (DAS44<1.6) and DFR, with initial treatment of methotrexate and high-dose prednisone. We investigated whether becoming seronegative over the first year of treatment improves chances of long-term SDFR, defined as remission lasting at least 1 year, starting at any time point and held until the last moment of that individual’s follow-up (maximum 5 years).

The prevalence of seroconversion from positive to nega-tive between 0 and 12 months varied substantially depending on the autoantibody from 2% (anti-CCP2 IgG) to 66% (anti-CarP IgA) (figure 1), occurring mostly in low-positive patients (online supplementary table S1). Demographic and clinical characteristics at baseline and 1 year in patients with serocon-version versus those without did not show marked differences (online supplementary table S1). Of the 359 patients who had outcome data available, 48 (13.4%) achieved long-term SDFR. Intriguingly, seroconversion from positive to negative was not associated with a greater chance of achieving long-term SDFR for any of the 14 antibodies tested (figure 1A). To investigate whether these findings were influenced by low-positive patients whose autoantibody levels merely fluctuated around the cut-off, a sensitivity analysis was conducted including only patients

1837Ann Rheum Dis December 2018 Vol 77 No 12

Letters

Figure 1 (A) Number and percentage of baseline seropositive patients seroconverting (‘Pos to Neg’) or non-converting (‘Stable pos’) between 0 and 12 months are listed on the left, and the percentage of each subset subsequently reaching long-term sustained drug-free remission (DFR) is graphically depicted on the right. (B) Number and percentage of baseline seropositive patients reaching long-term sustained DFR, categorised by the amount of antibody reactivities that were lost (i.e. composite of positive-to-negative seroconversion) between 0 and 12 months. Anti-acetyl pept Abs, anti-acetylated peptide antibodies; Anti-cit pept Abs, anti-citrullinated peptide antibodies; RF, rheumatoid factor.

whose baseline autoantibody levels were above the median, with similar results (online supplementary figure S1). Of the 170 seropositive patients with complete antibody data at 0 and 12 months, only six (3.5%) seroconverted to completely sero-negative by 12 months; 33% (2/6) of these completely serocon-verted patients achieved long-term SDFR, compared with 11.6% (19/164) of patients who were positive for at least one antibody (p=0.11). Patients who seroconverted to negative for a larger number of autoantibodies did not achieve long-term SDFR more often than those who seroconverted for fewer (figure 1B). Rela-tive changes in autoantibody levels between 0 and 12 months did not differ between patients with or without long-term SDFR (data not shown).

The clinical significance of seroconversion in RA and espe-cially its relationship with long-term SDFR, an approximation of

disease ‘cure’ of RA, is a topic of major interest.1 Previous studies found no association of seroconversion with remission or radio-graphic damage.4 5 We here investigated the association between seroconversion and the most favourable long-term outcome of RA, SDFR, and found no association. Thus, it appears that sero-conversion (as measured by current standards) does not identify a group of patients in whom the underlying immunopathology has been favourably modulated, that is, patients in true immuno-logical remission, and is not superior to signals of low inflamma-tory load (e.g. by DAS 6) for predicting successful drug tapering. Future studies are needed to identify whether other immunolog-ical parameters such as the numbers or phenotype of circulating autoreactive B or T cells might be a better reflection of disease persistence and markers for immunological remission.

Emma C de Moel,1 Veerle F A M Derksen,1 Leendert A Trouw,1 Holger Bang,2 Yvonne P M Goekoop-Ruiterman,3 Gerda M Steup-Beekman,1,4 Tom W J Huizinga,1 Cornelia F Allaart,1 René E M Toes,1 Diane van der Woude1

1Department of Rheumatology, Leiden University Medical Center, Leiden, The Netherlands2Orgentec Diagnostika, Mainz, Germany3Department of Rheumatology, Haga Hospital, The Hague, The Netherlands4Department of Rheumatology, Haaglanden Medical Center, The Hague, The Netherlands

Correspondence to Dr Emma C de Moel, Department of Rheumatology, Leiden University Medical Center, Leiden, The Netherlands; e. c. de_ moel@ lumc. nl

Handling editor Josef S Smolen

Acknowledgements We thank Gerrie Stoeken and Linda Herb-van Toorn for their expertise and guidance during the experiments.

Contributors ECM, VFAMD and DW designed the research. ECM and VFAMD performed the experiments. HB kindly provided assays for anti-acetylated peptide antibodies. YPMGR, GMSB and CFA were responsible for design and clinical data collection of the IMPROVED study. ECM analysed the data and drafted the manuscript. All authors contributed to the interpretation of the data and preparation of the manuscript.

Funding This project was funded by the ZonMw (the Netherlands Organisation for Health Research and Development)-consortium MOlecular DIagnostics in RA (MODIRA), project number 436001001.

Competing interests None declared.

Patient consent Obtained

Ethics approval This study was conducted with the approval of the regional ethics committee at Leiden University Medical Center. The IMPROVED study was approved by the medical ethics committees of all participating hospitals.

Provenance and peer review Not commissioned; externally peer reviewed.

Data sharing statement Details concerning data analysis and ’data not shown’ are available upon request.

© Author(s) (or their employer(s)) 2018. No commercial re-use. See rights and permissions. Published by BMJ.

► Additional material is published online only. To view please visit the journal online (http:// dx. doi. org/ 10. 1136/ annrheumdis- 2018- 213823).

To cite de Moel EC, Derksen VFAM, Trouw LA, et al. Ann Rheum Dis 2018;77:1836–1838.

Received 25 May 2018Revised 6 July 2018Accepted 8 July 2018Published Online First 25 July 2018

Ann Rheum Dis 2018;77:1836–1838. doi:10.1136/annrheumdis-2018-213823

RefeRences 1 Schett G, Emery P, Tanaka Y, et al. Tapering biologic and conventional DMARD therapy

in rheumatoid arthritis: current evidence and future directions. Ann Rheum Dis 2016;75:1428–37.

1838 Ann Rheum Dis December 2018 Vol 77 No 12

Letters

2 de Moel EC, Derksen V, Stoeken G, et al. Baseline autoantibody profile in rheumatoid arthritis is associated with early treatment response but not long-term outcomes. Arthritis Res Ther 2018;20:33.

3 Heimans L, Akdemir G, Boer KV, et al. Two-year results of disease activity score (DAS)-remission-steered treatment strategies aiming at drug-free remission in early arthritis patients (the IMPROVED-study). Arthritis Res Ther 2016;18:23.

4 Barra L, Bykerk V, Pope JE, et al. Anticitrullinated protein antibodies and rheumatoid factor fluctuate in early inflammatory arthritis and do not predict clinical outcomes. J Rheumatol 2013;40:1259–67.

5 Guzian MC, Carrier N, Cossette P, et al. Outcomes in recent-onset inflammatory polyarthritis differ according to initial titers, persistence over time, and specificity of the autoantibodies. Arthritis Care Res 2010;62:1624–32.

6 Bergstra SA, Allaart CF. What is the optimal target for treat-to-target strategies in rheumatoid arthritis? Curr Opin Rheumatol 2018;30:1–7.

1838 Ann Rheum Dis December 2018 Vol 77 No 12

Letters

Figure 1 Characteristics of the absolute numbers and proportions of Th17 cells and CD4Treg cells in the PB of patients with pSS. Lymphocytes in the PB of the healthy control (n=30), treated pSS (n=169) and new pSS (n=21) groups were assessed by flow cytometry. (A and B) Neither the absolute number nor proportion of Th17 cells was altered significantly. (C and D) The absolute number but not the proportion of CD4Treg cells was significantly decreased. (E) Reduction of CD4Treg cells led to an imbalance between Th17 cells and Treg cells (increased ratio of Th17/CD4Tregs). (F and G) Comparison of the absolute number and percentage of CD4Treg cells and CD4 +CD25+T cells among these three groups. All data below are presented as medians (Q1, Q3). Statistical analyses were performed using the independent-samples Kruskal-Wallis test and one-way analysis of variance. *P<0.05; **P<0.01; ***P<0.001 vs healthy control group.

Short-term and low-dose IL-2 therapy restores the Th17/Treg balance in the peripheral blood of patients with primary Sjögren’s syndrome

Based on evidence that low-dose interleukin-2 (IL-2) increases CD4+CD25+FOXP3+ Treg (CD4Treg) cells in patients with graft-versus-host disease, induction of autoimmune tolerance has been proposed as a treatment.1 However, evidence-based guidelines for the management of primary Sjögren’s syndrome (pSS) are lacking.2

1839Ann Rheum Dis December 2018 Vol 77 No 12

Letters

Figure 2 Relationship between the CD4 +T cell subsets and disease activity and the impact of interleukin (IL) 2 on these cells. (A–C) Correlation of the levels of Th17 cells and CD4Treg cells and the ratio of Th17/Treg cells with disease activity. All patients were divided into low (0≤ESSDAI<5, n=110), medium (5≤ESSDAI≤14, n=67) or high (ESSDAI >14, n=13) disease activity groups. (D–F) After receiving treatment with a short-term low-dose IL-2 (5.0*105 international units (IU) for 5 days) (n=82), CD4Treg cells were amplified more than Th17 cells, and the ratio of Th17/CD4Treg cells was restored. Data are presented as medians (Q1, Q3). Statistical analyses were performed using the independent-samples Kruskal-Wallis test and one-way analysis of variance. *P<0.05; **P<0.01; ***P<0.001.

Using a modified method of flow cytometry, we aimed to revaluate the exact levels of Th17 and CD4Treg cells in the peripheral blood (PB) of patients with pSS and explore the effects of short-term and low-dose IL-2.3 4

A total of 190 patients with pSS (169, treated with immunosup-pressants; 21, new-onset (sampled as treated or new pSS patients below)) consented at enrolment to donate PB samples for compre-hensive immunophenotyping (see online supplementary table S1 and figure S1). In the study, BD Trucount tubes were used to determine the absolute counts of total CD4+ T cells in the PB, and then, the absolute number of CD4+ T subsets, such as CD4Treg cells, were calculated. Detailed methods and statistical analysis is in online version (see online supplementary text).

Among the CD4+ T subsets in the PB of the healthy control group, Th1 cells had the highest absolute number, and CD4Treg cells were sixfold more abundant than Th17 cells (see online supplementary table S3 and figure 1A–D). Both treated and new pSS patients had significantly fewer absolute CD4Treg cells in their PB (p<0.01) than healthy patients (see figure 1C). CD4Treg cells were the only cell subset with a significant differ-ence in absolute numbers. Notably, neither the absolute number nor proportion of Th17 cells was increased (figure 1A,B). In addition, no significant changes were found in CD4 +CD25+T cells or the percentage of CD4Treg cells (figure 1F,G).5

When all patients with pSS were divided into low (0≤ EULAR Sjögren's syndrome disease activity index (ESSDAI)<5), moderate

(5≤ESSDAI≤14) and high (ESSDAI >14) disease activity groups (see online supplementary figure S2), the smallest absolute number of CD4Treg cells in PB were identified in patients with high disease activity (figure 2B). However, no significant difference in levels of Th17 cell was observed among the groups (figure 2A,C).

After low-dose IL-2 treatment, all CD4 +T cells were increased (see online supplementary figure S3), but CD4Treg cells were greatly amplified (fourfold increase) (see online supplemen-tary figure S5, figure 2E). Although the absolute number of Th17 cells also increased, the ratio of Th17/Treg decreased to normal levels (figure 2D,F). Although no obvious improve-ment in disease activity was found in IL-2 group compared with non-IL-2 group during short-term observation (supplementary figure S4), we observed significant reductions in glucocorticoid and disease-modifying antirheumatic drugs (DMARDS) usage (unpublished data) during long-term observation. No obviously abnormal indicators of laboratory examination were observed during treatment, except for a few cases of rashes around the injection sites and influenza-like symptoms (see online supple-mentary tables S5–S7). It is possible that low-dose IL-2 can induce antigen-specific CD4+ Treg cells during the treatment process.1 6

In conclusion, our study demonstrates that the absolute number of circulating CD4Treg cells in patients with pSS is significantly lower than that in healthy patients; however, the absolute number and proportion of Th17 cells are not

1840 Ann Rheum Dis December 2018 Vol 77 No 12

Letters

significantly different. Notably, low-dose IL-2 treatment restores CD4Treg cells and the ratio of Th17/Treg which are helpful for controlling disease activity.7–10 Our findings strengthen the concept of immunoregulation but not immunosuppression in the treatment of pSS.1

Miao Miao,1 Zhenye Hao,1 Yingying Guo,1 Xiaoying Zhang,2 Shengxiao Zhang,1 Jing Luo,2 Xiangcong Zhao,2 Chen Zhang,2 Xiaoqing Liu,1 Xiaoyan Wu,2 Dan Xu,1 Jinfang Zhao,3 Xuechun Lu,4 Chong Gao,5 Xiaofeng Li2

1The Second Hospital of Shanxi Medical University, Taiyuan, China2Department of Rheumatology, The Second Hospital of Shanxi Medical University, Taiyuan, China3Department of Medical Statistics, Shanxi Medical University, Taiyuan, China4Department of Geriatric Hematology, Chinese PLA General Hospital, Beijing, China5Department of Pathology, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts, USA

Correspondence to Professor Xiaofeng Li, Department of Rheumatology, The Second Hospital of Shanxi Medical University, Taiyuan, China; lxf_ 9859@ sxmu. edu. cn

Handling editor Josef S Smolen

Acknowledgements Ruijie Wu collected the information of outpatients. Qi Wu finished procedures of sampling preparation.

Contributors All of the authors were involved in the drafting of the article or critical revision of the important intellectual content, and all of the authors approved the final version to be published.

Competing interests None declared.

Patient consent Obtained.

Ethics approval Ethics Committee of the Second Hospital of Shanxi Medical University.

Provenance and peer review Not commissioned; externally peer reviewed.

Data sharing statement The data that support the findings of this study are available from the corresponding.

© Author(s) (or their employer(s)) 2018. No commercial re-use. See rights and permissions. Published by BMJ.

► Additional material is published online only. To view please visit the journal online (http:// dx. doi. org/ 10. 1136/ annrheumdis- 2018- 213036).

To cite Miao M, Hao Z, Guo Y, et al. Ann Rheum Dis 2018;77:1838–1840.

Received 17 January 2018Revised 29 May 2018Accepted 8 June 2018Published Online First 23 June 2018

Ann Rheum Dis 2018;77:1838–1840. doi:10.1136/annrheumdis-2018-213036

RefeRences 1 Klatzmann D, Abbas AK. The promise of low-dose interleukin-2 therapy for

autoimmune and inflammatory diseases. Nat Rev Immunol 2015;15:283–94. 2 Gottenberg JE, Lavie F, Abbed K, et al. CD4 CD25high regulatory T cells are not

impaired in patients with primary Sjögren’s syndrome. J Autoimmun 2005;24:235–42. 3 Low HZ, Witte T. Aspects of innate immunity in Sjögren’s syndrome. Arthritis Res Ther

2011;13:218. 4 Gullick NJ, Abozaid HS, Jayaraj DM, et al. Enhanced and persistent levels of interleukin

(IL)-17⁺ CD4⁺ T cells and serum IL-17 in patients with early inflammatory arthritis. Clin Exp Immunol 2013;174:292–301.

5 Liu MF, Lin LH, Weng CT, et al. Decreased CD4+CD25+bright T cells in peripheral blood of patients with primary Sjogren’s syndrome. Lupus 2008;17:34–9.

6 Fontenot JD, Rasmussen JP, Gavin MA, et al. A function for interleukin 2 in Foxp3-expressing regulatory T cells. Nat Immunol 2005;6:1142–51.

7 Furuzawa-Carballeda J, Hernández-Molina G, Lima G, et al. Peripheral regulatory cells immunophenotyping in primary Sjögren’s syndrome: a cross-sectional study. Arthritis Res Ther 2013;15:R68.

8 Passerini L, Santoni de Sio FR, Roncarolo MG, et al. Forkhead box P3: the peacekeeper of the immune system. Int Rev Immunol 2014;33:129–45.

9 Ramsdell F, Ziegler SF. FOXP3 and scurfy: how it all began. Nat Rev Immunol 2014;14:343–9.

10 Boyman O, Sprent J. The role of interleukin-2 during homeostasis and activation of the immune system. Nat Rev Immunol 2012;12:180–90.

1840 Ann Rheum Dis December 2018 Vol 77 No 12

Letters

Clinical associations and expression pattern of the autoimmunity susceptibility factor DIORA-1 in patients with primary Sjögren’s syndrome

Sjögren’s syndrome (SS) is characterised by B cell abnormal-ities and immune-mediated destruction of exocrine glands, primarily the salivary and lacrimal glands.1 2 Among the reported genetic polymorphisms associated with primary SS (pSS), the FAM167A-BLK locus distinguishes itself as an interesting candidate for further analysis based on the strong expression quantitative locus effect of pSS-associated polymor-phisms on FAM167A (member A of the Family with sequence similarity 167), contrasted with only moderate or no effect on BLK.3 4 Little is known about the FAM167A gene and its rele-vance to rheumatic disease pathogenesis. We recently explored FAM167A and its encoded protein Disordered autoimmu-nity-1 (DIORA-1),4 and reported that DIORA-1 is conserved in vertebrates, has an intracellular, cytoplasmic localisation and in mice is predominantly expressed in lung and spleen—two organs with a high content of immune cells. In the present study, we investigated the expression of DIORA-1 in human immune cells and in salivary glands of patients with pSS, and assessed DIORA-1 expression in relation to pSS clinical manifestations.

Notably, we observed expression of DIORA-1 in CD19+ B cells, but little or no expression in monocytes or T cells (figure 1A). DIORA-1 expression in CD19+ B cells was similar in patients with pSS and healthy donors (figure 1B). To further define the expression pattern in B cells, we analysed DIORA-1 expression in cell lines representing discrete differentiation stages of B cells. Interestingly, we observed a graded expres-sion of DIORA-1 in these cell lines, with the highest expression found in the two plasma cell myeloma lines and intermediate expression in other B cell lines, and little or no expression in T cells and other investigated cell types (figure 1C).

We next assessed the expression of DIORA-1 in salivary glands, which constitute the main target organs in pSS. Minor salivary gland biopsies from 38 patients taken at the time of diagnosis and from 20 individuals investigated for SS due to sicca symptoms but not fulfilling the classification criteria (non-SS)5 were analysed by immunohistochemistry. The included individuals and their characteristics are summarised in online supplementary table 1. In patients with pSS, we observed expression of DIORA-1 in focal infiltrates and in discrete cells interstitially (figure 1D). Few or no positive cells were observed in biopsies from non-SS individuals (online supple-mentary figure 1). Assessment of the percentage of cells in the focal infiltrates expressing DIORA-1 demonstrated a significant difference between pSS and non-SS individuals (figure 1E, online supplementary figure 2). Quantification of DIORA-1-expressing cells as the percentage of positive inter-stitial cells also revealed the presence of significantly more DIORA-1+ cells in biopsies of patients with pSS compared with non-SS controls (figure 1F). The interstitial DIORA-1-expressing cells had a plasma cell-like morphology, being large cells with a considerable cytoplasm-to-nucleus ratio and an eccentrically positioned nucleus. Double immunofluores-cence staining using CD138 or CD38 as plasma cell surface

1841Ann Rheum Dis December 2018 Vol 77 No 12

Letters

Figure 1 Expression of DIORA-1 and clinical associations in Sjögren’s syndrome (SS). (A) Expression of DIORA-1 in CD19+ B cells, CD14+ monocytes and CD3+ T cells sorted from buffy coats of healthy donors (n=10) assessed by real-time PCR. (B) Expression of DIORA-1 in CD19+ B cells sorted from peripheral blood of healthy donors (n=16) and patients with primary SS (pSS) (n=17) assessed by real-time PCR. (C) Expression of DIORA-1 in panel of cell lines: HeLa (cervical cancer), A375 (malignant melanoma), HEK293 (human embryonic kidney), Jurkat (acute T cell leukaemia), Nalm6 (B cell precursor leukaemia), Raji (B lymphoblast, Burkitt lymphoma), Daudi (B lymphoblast, Burkitt lymphoma), KM3 (plasma cell myeloma), LP-1 (plasma cell myeloma) assessed by real-time PCR. (D) Representative minor salivary gland biopsies of a patient with pSS and a non-SS patient stained for DIORA-1 by immunohistochemistry. (E) Semiquantitative assessment of DIORA-1 staining of inflammatory foci expressed as per cent DIORA-1+ cells of the infiltrate in patients with pSS and non-SS. (F) Semiquantitative assessment of DIORA-1 staining of interstitial cells expressed as per cent stained cells in patients with pSS and non-SS. (G) Representative double immunofluorescence staining of a salivary gland from a patient with pSS by anti-DIORA-1 (green) and anti-CD138 plasma cell surface marker (red). DAPI stain visualising cell nuclei is represented by blue colour. In the merged image white arrows indicate examples of CD138+ cells with intracellular DIORA-1 expression. Correlation between interstitial cells expressing DIORA-1 and discrete pSS parameters: focus score (H), serum IgG levels (I) and presence of Ro/SSA autoantibodies in serum (J). ***P<0.001. Scale bar represents 200 µm.

markers confirmed that plasma cells in salivary gland tissue from patients with pSS express DIORA-1 (figure 1G, online supplementary figure 3). DIORA-1 expression in other B cell populations was confirmed by CD20 staining (online supple-mentary figure 3).

Analysing expression of DIORA-1 in relation to clin-ical manifestations of pSS, we found that the percentage of DIORA-1+ cells positively correlated with salivary gland focus score as well as serum IgG levels and the presence of Ro/SSA autoantibodies (p<0.001) (figure 1H–J). Detailed information on patients and all methods are available in online supplemen-tary methods. We observed no correlation with patient gender, age, salivary flow rate, extraglandular manifestations or adverse predictors of lymphoma (salivary gland enlargement, rheumatoid factor and lymphopenia) (data not shown). Inter-estingly, however, a recent study reported a blood plasmablast

signature correlating with glandular inflammation in Ro/SSA+ patients with pSS.6 Limitations of the study relate to the fact that mechanistic studies were not performed and that additional patient groups were not investigated.

In summary, our study reveals that the novel protein DIORA-1, encoded by the FAM167A gene linked to pSS, is expressed in B cells and plasma cells in the target organs of patients with pSS and positively correlates with focus score, serum IgG levels and the presence of Ro/SSA autoanti-bodies. Our findings indicate a role for DIORA-1 in distinct B cell subsets, and suggest that DIORA-1 may contribute to the inflammatory process and disease pathogenesis in pSS through B cell involvement.

Lara A Aqrawi,1,2,3 Lara Mentlein,1 Lauro Meneghel,1 Albin Björk,1 Gudny Ella Thorlacius,1 Margarita Ivanchenko,1

1842 Ann Rheum Dis December 2018 Vol 77 No 12

Letters

Jorge I Ramírez Sepúlveda,1 Kathrine Skarstein,2,4 Marika Kvarnström,1 Susanna Brauner,5 Alexander Espinosa,1 Marie Wahren-Herlenius1

1Unit of Rheumatology, Department of Medicine, Karolinska Institutet, Karolinska University Hospital, Stockholm, Sweden2The Gade Laboratory for Pathology, Department of Clinical Medicine, University of Bergen, Bergen, Norway3Department of Oral Surgery and Oral Medicine, Institute of Clinical Odontology, University of Oslo, Oslo, Norway4Department of Pathology, Haukeland University Hospital, Bergen, Norway5Department of Clinical Neuroscience, Karolinska Institute, Karolinska University Hospital, Stockholm, Sweden

Correspondence to Professor Marie Wahren-Herlenius, Unit of Rheumatology, Department of Medicine, Karolinska Institute, Stockholm 171 76, Sweden; marie. wahren@ ki. se

Handling editor Josef S Smolen

Acknowledgements We thank Amina Ossoinak, Vijole Ottosson and Sabrina Meisgen for excellent technical assistance, and gratefully acknowledge Aurélie Ambrosi, Karolinska Institutet, for contributing to the writing of the paper.

Contributors LAA, LarM, SB and MWH conceived the study. MK recruited patients and recorded clinical data. LAA, LarM, LauM, AB, GET, MI, JIRS and SB performed the experiments, analysed the data and generated figures and table with input from KS, AE and MWH. LAA, LarM and MWH wrote the first version of the manuscript, and all authors participated in the editing until its final version.

Funding This study was funded by Vetenskapsrådet, Cancerfonden, Hjärt-Lungfonden, Stockholms Läns Landsting, Karolinska Institutet, Reumatikerförbundet, Norges Forskningsråd, Stiftelsen Konung Gustaf V:s 80-årsfond.

Competing interests None declared.

Patient consent Obtained.

Ethics approval The Regional Ethical Committe Stockholm.

Provenance and peer review Not commissioned; externally peer reviewed.

© Author(s) (or their employer(s)) 2018. No commercial re-use. See rights and permissions. Published by BMJ.

► Additional material is published online only. To view please visit the journal online (http:// dx. doi. org/ 10. 1136/ annrheumdis- 2018- 213634).

LAA and LM contributed equally.

To cite Aqrawi LA, Mentlein L, Meneghel L, et al. Ann Rheum Dis 2018;77:1840–1842.

Received 20 April 2018Revised 3 July 2018Accepted 4 July 2018Published Online First 17 July 2018

Ann Rheum Dis 2018;77:1840–1842. doi:10.1136/annrheumdis-2018-213634

RefeRences 1 Ambrosi A, Wahren-Herlenius M. Update on the immunobiology of Sjögren’s

syndrome. Curr Opin Rheumatol 2015;27:468–75. 2 Brauner S, Folkersen L, Kvarnström M, et al. H1N1 vaccination in Sjögren’s syndrome

triggers polyclonal B cell activation and promotes autoantibody production. Ann Rheum Dis 2017;76:1755–63.

3 Lessard CJ, Li H, Adrianto I, et al. Variants at multiple loci implicated in both innate and adaptive immune responses are associated with Sjögren’s syndrome. Nat Genet 2013;45:1284–92.

4 Mentlein L, Thorlacius GE, Meneghel L, et al. The rheumatic disease-associated FAM167A-BLK locus encodes DIORA-1, a novel disordered protein expressed highly in bronchial epithelium and alveolar macrophages. Clin Exp Immunol 2018.

5 Kvarnström M, Ottosson V, Nordmark B, et al. Incident cases of primary Sjögren’s syndrome during a 5-year period in Stockholm County: a descriptive study of the patients and their characteristics. Scand J Rheumatol 2015;44:135–42.

6 Mingueneau M, Boudaoud S, Haskett S, et al. Cytometry by time-of-flight immunophenotyping identifies a blood Sjögren’s signature correlating with disease activity and glandular inflammation. J Allergy Clin Immunol 2016;137:1809–21.

1842 Ann Rheum Dis December 2018 Vol 77 No 12

Letters

Increase in circulating cells coexpressing M1 and M2 macrophage surface markers in patients with systemic sclerosis

Alterations in macrophage polarisation are recognised among the possible immune system abnormalities contributing to systemic sclerosis (SSc) pathogenesis.1

Macrophages have been classified as classically (M1) or alter-natively (M2) activated, although growing evidence indicates that they may exhibit characteristics shared by more than one of the described phenotypes.2–5

An M2 pre-eminent phenotype has been postulated for SSc monocytes/macrophages.2 The aim of the study was to widen a phenotype characterisation (M1, M2 and mixed M1/M2) of circulating monocytes/macrophages in patients with SSc and healthy subjects (HSs) through flow cytometry analysis. Fifty-eight consecutive patients with SSc (38 limited and 20 diffuse SSc) and 27 age-matched and gender-matched HSs were enrolled after signing informed consent. SSc diagnosis was based on the 2013 American College of Rheumatology/European League Against Rheumatism criteria (online supplementary file 1).6 For flow cytometry analysis, peripheral blood was collected and anti-CD14 and anti-CD45 antibodies were used to identify the monocyte/macrophage lineage; macrophage scavenger receptors (CD204, CD163) and mannose receptor 1 (CD206) were used as M2 phenotype markers; and co-stimulatory molecules (CD80, CD86) and Toll-like receptors (TLR4, TLR2) were used as M1 phenotype markers. CD66b was used to distinguish granulocytes (Miltenyi Biotech, Germany). Flow c ytometry analysis was performed using the Navios flow cytometer and Kaluza analysis software (Beckman Coulter). A total of 5 × 10 6 cells were evaluated and more than 30 events were detected in the smallest subset investigated, according to the consensus guidelines for minimal residual disease. Results were expressed in percentages over total circulating leu c ocytes, unless otherwise specified. The non-parametric al Mann-Whitney U test was used for statis-tical analysis and any p value lower than 0.05 was considered statistically significant. Two initial gating strategies were used to study circulating M2-like monocytes/macrophages, the first gated CD14+cells and the second gated CD204+cells.1 2 7

Using CD14+cells for the initial gating strategy, patients with SSc showed a significantly increased percentage of CD14+C-D163+CD206+cells compared with HSs (6.35%±2.8% vs 5.38±1.98%, p=0.047) (figure 1A). In this subset, cells expressing CD204 were significantly higher in patients with SSc compared with HSs (0.23±0.25 vs 0.14±0.13; p=0.02) (figure 1B). In circulating CD14+CD206+CD163+CD204+M2 cells, a signifi-cantly higher percentage of TLR4+cells was observed compared with HSs (0.0091%±0.013% vs 0.003±0.0026%; p=0.003) (figure 1C).

Using CD204+cells for the initial gating strategy, the percentage of CD204+CD163+CD206+M2 cells was significantly higher in patients with SSc compared with HSs (5.6%±6.9% vs 1.4±1.3% of CD204+cells; p<0.0001), representing 0.042% of the leuco-cyte population in patients with SSc compared with 0.01% in HSs (figure 1D). In this cell population, cells expressing TLR4 were significantly higher in patients with SSc than in HSs (4.65%±6.8% vs 0.66±0.6% of CD204+cells; p<0.0001) (figure 1E). In the CD204+CD163+CD206+TLR4+cell subset, CD14+cells made up 1% of the CD204+cells in patients with SSc compared with 0.31% in HSs (p=0.002), whereas the

1843Ann Rheum Dis December 2018 Vol 77 No 12

Letters

Figure 1 Evaluation of circulating cells expressing M1 and M2 phenotype markers in patients with systemic sclerosis (SSc) and healthy subjects (HSs). Representative flow cytometry panels with quadrant regions and box plot representation of the percentage (%) of circulating (A) CD14+CD206+CD163+cells in the CD14+cell population; (B) CD14+CD206+CD163+CD204+cells in the CD14+CD206+CD163+cell subset; (C) CD14+CD206+CD163+CD204+TLR4+cells in the CD14+CD206+CD163+cell subset, detecting CD14+cells in the leucocyte population as the initial gating strategy in HSs and patients with SSc. Representative flow cytometry panels with quadrant regions and box plot representation of the % of circulating (D) CD204+CD163+CD206+cells in the CD204+CD163+cell subset; (E) CD204+CD163+CD206+TLR4+cells in the CD204+CD163+cell subset; (F) CD14+ and CD14cells in the CD204+CD163+CD206+TLR4+cell subset; (G) CD204+CD163+CD206+TLR4+CD80+CD86+cells in the CD204+CD163+CD206+TLR4+cell subset, detecting the CD204+cells in the leucocyte population as the initial gating strategy in HSs and SSc pts. The flow cytometry panels are representative of the mean % that was obtained. The box plots show the median and the whiskers (which represent the ninth and the ninety-first centiles). The results were expressed as mean±SD.

CD14 cells made up 3.65% of CD204+cells in patients with SSc compared with 0.35% in HSs (p<0.0001) (figure 1F). Moreover, 41.6% of the CD204+CD163+CD206+TLR4+-

cells in patients with SSc were characterised by the coexpres-sion of CD80 and CD86 (M1) compared with 26.8% in HSs (0.02%±0.07% vs 0.001±0.002% of the leucocyte popula-tion, p<0.0001) (figure 1G).Interestingly the four outliers with higher percentages of M2 and M1/M2 cells calculated over the total CD204+cells in the patient populations with SSc , are always related to the same patients (figure 1D, E, F, G). These patients are all women of different ages, characterised by Scl70 positivity, interstitial lung disease (ILD) at lung CT scan, reduced diffusing capacity of the lung for carbon monoxide (DLCO), presence of oesophageal or heart involvement. Two of them were treated with glucocorticoids.

No difference in circulating M1 cell percentage was observed between patients with SSc and HSs (online supplementary file 2).

The patient subgroup analysis according to the different treat-ment regimens pointed out that the lowest percentages of M2 and M1/M2 circulating cell populations were observed in HSs, while the highest values were observed in patients taking only gluco-corticoids. Intermediate values were observed for circulating M2 and M1/M2 populations in not treated patients or in patients treated with both drugs. Highly significant differences in the percentages of the circulating M2 and M1/M2 studied populations were constant between patients with SSc not treated with any of the considered medications and HSs. No significant correlation was reported with other treatment regimens (ie, ERAs) (figure 2). Moreover, no correlations were observed with either limited or diffused cutaneous disease form.

Based on a wide flow cytometry surface marker analysis of circu-lating monocytes/macrophages, the study demonstrated that M2 and more significantly cells expressing both M1 and M2 surface markers characterise patients with SSc compared with HSs. No difference was identified when only M1 markers were used. More-over, the initial gating strategy based on the CD204+cells resulted the most efficient to describe monocyte/macrophage phenotype differences between patients with SSc and HSs.

The result is supported by a recent study demonstrating a remarkable plasticity of circulating monocytes/macrophages, resulting in a ‘spectrum’ of activation states.5 Consistently with our data, a recent study from Moreno-Moral et al showed a downregulation of interferon-γ response and IL6/JAK/STAT3 pathway in SSc monocyte-derived macrophages, possibly describing a ‘SSc specific macrophage’.8–10

Stefano Soldano,1 Amelia Chiara Trombetta,1 Paola Contini,2 Veronica Tomatis,1 Barbara Ruaro,1 Renata Brizzolara,1 Paola Montagna,1 Alberto Sulli,1 Sabrina Paolino,1 Carmen Pizzorni,1 Vanessa Smith,3 Maurizio Cutolo1

1Research Laboratory and Academic Division of Clinical Rheumatology, Department of Internal Medicine, University of Genova, Polyclinic San Martino Hospital, Genoa, Italy2Clinical Immunology, Department of Internal Medicine, University of Genova, Polyclinic San Martino Hospital, Genova, Italy3Department of Rheumatology, Ghent University Hospital, Department of Internal Medicine, Ghent University, Ghent, Belgium

Correspondence to Professor Maurizio Cutolo MD, Research Laboratory and Academic Division of Clinical Rheumatology, Department of Internal Medicine, University of Genova, Polyclinic San Martino Hospital, Genoa 16132, Italy; mcutolo@ unige. it

1844 Ann Rheum Dis December 2018 Vol 77 No 12

Letters

Figure 2 Graphic representation of the subgroup analysis according to treatment regimens. For sample size reasons, patients were stratified according to different treatment regimens. Treatments that could theoretically influence the variables of interest, that is, immunosuppressants, glucocorticoids or endothelin-1 receptor antagonists (ERAs), were considered in the analysis. Patients were defined ‘on immunosuppressant’ if they were steadily under treatment with: ciclosporin 3–5 mg/kg per day, methotrexate 7.5 mg per week or 2.5 mg every 12 hours for three sequential doses per week, mycophenolate 1500–2000 mg per day; ‘on glucocorticoids’ if they were steadily under treatment with prednisone 2.5–5 mg per day; ‘on ERAs’ if they were steadily under treatment with bosentan 250 mg per day or macitentan 10 mg per day. Each enrolled patient was under steady treatment regimen from at least 6 months. (A) Means and SD of the percentage of each cell subset are reported for patients treated with glucocorticoids (GCs), immunosuppressants (IMM), immunosuppressants and glucocorticoids (IMM + GCs) or not treated with any of the two drugs (NO IMM/GCs), and healthy subjects (HSs). (B) Graphic representation of the subgroup analysis of M2 and M1/M2 cell percentages in patients with SSc stratified according to treatment regimens and in HSs. Percentages are calculated over total CD204+cells (%CD204+) or over the total leucocyte population (%leucocytes). PTs, patients with SSc.

Correction notice This article has been corrected since it published Online First. The equal contribution statement has been added.

Handling editor Josef S Smolen

Acknowledgements The authors thank Dr Sara De Gregorio for designing and making figure 1 and supplementary file 1 of the manuscript and Fabio Gallo for supporting the statistical analysis.

Contributors SS, ACT and MC designed the study, analysed the data, carried out the statistical analysis and wrote the manuscript; PC performed the flow cytometry analysis; VT and BR enrolled the patients with systemic sclerosis and healthy subjects and collected the demographic and clinical parameters of the patients with SSc; RB and PM managed and prepared the samples for the flow cytometry analysis; AS, SP and CP supervised the enrolment of the patients with SSc and the healthy subjects and the clinical data collection; VS revised the critical and important intellectual contents of the manuscript.

Competing interests None declared.

Patient consent Not required.

Ethics approval Local Ethics committee (protocol number 273-REG-2015).

Provenance and peer review Not commissioned; externally peer reviewed.

© Author(s) (or their employer(s)) 2018. No commercial re-use. See rights and permissions. Published by BMJ.

► Additional material is published online only. To view please visit the journal online (http:// dx. doi. org/ 10. 1136/ annrheumdis- 2018- 213648).

SS and ACT contributed equally.

To cite Soldano S, Trombetta AC, Contini P, et al. Ann Rheum Dis 2018;77:1842–1845.

Received 24 April 2018Revised 28 June 2018Accepted 2 July 2018Published Online First 16 July 2018

Ann Rheum Dis 2018;77:1842–1845. doi:10.1136/annrheumdis-2018-213648

RefeRences 1 Hunzelmann N, Brinckmann J. What are the new milestones in the pathogenesis of

systemic sclerosis? Ann Rheum Dis 2010;69:i52–6. 2 Higashi-Kuwata N, Jinnin M, Makino T, et al. Characterization of monocyte/

macrophage subsets in the skin and peripheral blood derived from patients with systemic sclerosis. Arthritis Res Ther 2010;12:R128.

3 Wynn TA, Vannella KM. Macrophages in tissue repair, regeneration, and fibrosis. Immunity 2016;44:450–62.

4 Mathai SK, Gulati M, Peng X, et al. Circulating monocytes from systemic sclerosis patients with interstitial lung disease show an enhanced profibrotic phenotype. Lab Invest 2010;90:812–23.

5 Mosser DM, Edwards JP. Exploring the full spectrum of macrophage activation. Nat Rev Immunol 2008;8:958–69.

6 van den Hoogen F, Khanna D, Fransen J, et al. classification criteria for systemic sclerosis: an American college of rheumatology/European league against rheumatism collaborative initiative. Ann Rheum Dis 2013;2013:1747–55.

7 Lambert C, Preijers F, Yanikkaya Demirel G, et al. Monocytes and macrophages in flow: an ESCCA initiative on advanced analyses of monocyte lineage using flow cytometry. Cytometry B Clin Cytom 2017;92:180–8.

8 Moreno-Moral A, Bagnati M, Koturan S, et al. Changes in macrophage transcriptome associate with systemic sclerosis and mediate GSDMA contribution to disease risk. Ann Rheum Dis 2018;77:596–601.

9 Lescoat A, Jégo P, Lecureur V. M-CSF and GM-CSF monocyte-derived macrophages in systemic sclerosis: the two sides of the same coin? Ann Rheum Dis 2018. doi: 10.1136/annrheumdis-2018-213112. [Epub ahead of print 9 Feb 2018].

1845Ann Rheum Dis December 2018 Vol 77 No 12

Letters

10 Behmoaras J, Petretto E. Cell function in disease: there are more than two parties at play. Ann Rheum Dis 2018. doi: 10.1136/annrheumdis-2018-213121. [Epub ahead of print 9 Mar 2018].

1845Ann Rheum Dis December 2018 Vol 77 No 12

Letters

Table 1 Demographic and clinical characteristics of patients with very early diagnosis of systemic sclerosis (VEDOSS) and systemic sclerosis (SSc) (whose sera sample collection was used in the study).

Characteristics SSc (n=28) VEDOSS (n=34)

Age (years) 56.56±12.45 50.50±13.66

Male – –

Female 28 34

lcSSc subsets 25 (89%) –

dcSSc subsets 3 (11%) –

Disease duration (months) 91.3±11.73 11.1±4.2

ANA 26 (93%) 31 (91%)

ACA 16 (57%) 17 (50%)

Anti-topoisomerase I (Scl-70) 4 (14%) 5 (15%)

Digital ulcers 4 (14%) –

Normal NVC pattern 1 (4%) 13 (38%)

Early NVC pattern 6 (21%) 11 (32%)

Active NVC pattern 13 (46%) 9 (26%)

Late NVC pattern 8 (29%) 1 (3%)

Interstitial lung disease 6 (21%) –

Age and disease duration were expressed as mean±SD/SE; disease duration was calculated since the first Raynaud phenomenon in VEDOSS and non-Raynaud symptom of SSc. Patients were not receiving corticosteroids, immunosuppressant or other disease-modifying drugs; patients with cardiac disease, pulmonary artery hypertension or active diseases other than SSc were excluded. Interstitial lung disease was determined by high-resolution computed tomography CT scan. ACA, anti-centromere antibodies; ANA, antinuclear antibodies; dcSSc, diffuse cutaneous SSc; lcSSc, limited cutaneous SSc; NVC, nailfold videocapillaroscopy.

Figure 1 CXCL10 and CXCL11 serum-level modifications associated with very early diagnosis of systemic sclerosis (VEDOSS) and systemic sclerosis (SSc) condition. Baseline serum CXCL10 and CXCL11 levels in VEDOSS versus SSc and healthy subjects (A, B); CXCL10 and CXCL11 serum concentration according to capillaroscopic patterns in VEDOSS and SSc (C, D)—normal, early, active and late; CXCL10 and CXCL11 baseline (T0) and follow-up (T1) determination in sera of VEDOSS subjects shifted (grey boxes) or not (empty boxes) to SSc (E, F); receiver operating characteristic (ROC) curves constructed using chemokine baseline levels (T0) to identify CXCL10 and CXCL11 threshold (cut-off value) capable of discriminating VEDOSS subjects shifting or not to SSc (G, H); CXCL10 and CXCL11 serum-level correlation (I). GraphPad Prism V.5 software (GraphPad Software, La Jolla, California, USA) and SPSS V.24.0 software (SPSS) were used for statistical analysis. The Kolmogorov-Smirnov test was used for normal distribution of the data. Groupwise comparisons were performed using the Mann-Whitney U test. ROC curves were to give a graphic representation of the relationship between true-positive fraction (sensitivity, SE) and false-positive fraction (1 – specificity, Sp); ROC curves were assessed by plotting the values of 1 – Sp against Se in a squared box, where the ROC’s area under the curve (AUC) is used to measure the performance of a diagnostic test. The AUC lies in the interval 0.5–1.0, so that the greater the area, the better the performance of the variable being examined. Chemokine association was determined using Pearson’s correlation coefficient. For all tests, a two-sided p value <0.05 was considered significant. NVC, nailfold videocapillaroscopy.

Association of circulating CXCL10 and CXCL11 with systemic sclerosis

Systemic sclerosis (SSc) is an autoimmune disorder defined ‘vascular’ disease1 due to early defective angiogenesis ending in severe multiorgan fibrosis.2 Biomarker(s) mirroring early micro-vascular derangements in SSc,3 potentially useful for very early diagnosis, are lacking.

C-X-C angiostatic chemokines induced by interferon-γ, like CXCL10 and CXCL11, are involved in vasculopathy and asso-ciate with more severe SSc.4 5 We investigated whether the shift from very early diagnosis of SSc (VEDOSS), when vasculopathy and fibrosis are at very low degree, to definite SSc elicits serum CXCL10/CXCL11 modifications.

CXCL10/CXCL11 were measured by multiplatform bead array in 26 healthy subjects; 62 sera from women admitted to the Scleroderma Clinic of Policlinico Umberto I, Sapienza University of Rome: 34 sera were from VEDOSS6 (mean age 50.50±13.66 years, mean disease duration 11.1±4.2 months) and 28 sera from patients with SSc (mean age 56.56±12.45 years, mean disease 91.3±11.73 months) fulfilling the new American College of Rheumatology/European League Against Rheuma-tism 2013 classification7; table 1 reports patient characteristics.

Within VEDOSS, 29 subjects had a second blood sample collected during follow-up (T1, 40.67±5.46 months); for each one, we compared baseline (T0) and T1 serum chemokines.Informed consent was obtained.

1846 Ann Rheum Dis December 2018 Vol 77 No 12

Letters

CXCL10/CXCL11 were significantly higher in SSc (median: CXCL10 497.70 (range: 48.49–2206) pg/mL; CXCL11 107.20 (15.00–882) pg/mL) versus VEDOSS (CXCL10 168.70 (21.66–1202) pg/mL; CXCL11 23.30 (1.47–217.20) pg/mL) or healthy subjects (CXCL10 145.40 (15.86–310.30) pg/mL; CXCL11 10.90 (4.30–189.20) pg/mL) (figure 1A, B). All patients showed serum chemokines stratification according to capillaroscopic patterns (figure 1C, D), while chemokine levels did not signifi-cantly differ when related to anti-centromere antibodies or anti-topoisomerase positivity (not shown). VEDOSS subse-quently shifted to SSc showed higher baseline chemokines (T0: CXCL10 217 (53.68–469.00) pg/mL; CXCL11 40.57 (15.74–92.04) pg/mL) versus subjects persistent in VEDOSS condition (T0: CXCL10 137.70 (21.66–339) pg/mL; CXCL11 17.47 (1.47–32.03) pg/mL) (figure 1E, F). Only VEDOSS shifted to SSc showed CXCL10/CXCL11 increase at T1 (570.80 (53.25–875.20) pg/mL). Both chemokines were able to discrim-inate VEDOSS subjects developing SSc (figure 1G, H) with the following cut-off values, identified by receiver operating char-acteristic (ROC) analysis: CXCL10 ≥165 pg/mL, area under the curve (AUC)=0.70 (95% CI 0.52 to 0.94, p<0.01) with 0.75 sensitivity and 0.70 specificity; CXCL11 ≥29.67 pg/mL, AUC=0.80 (95% CI 0.67 to 0.99, p<0.01) with 0.75 sensitivity and 0.88 specificity. Serum CXCL10 and CXCL11 positively correlated (ρ=0.6, p<0.0001) (figure 1I).

Due to CXCL10/CXCL11 detrimental effects on vessel homeostasis,5 8 9 their higher baseline concentration in VEDOSS thereafter developing SSc likely mirrors the earliest vascular bed alteration(s)/modification(s)/rearrangement(s) occurring when vasculopathy is still at low degree. With vascular damage progression, serum chemokines increased, as suggested by level stratification according to the capillaroscopic patterns. VEDOSS subjects retaining lower CXCL10/CXCL11 overtime (at T0 and T1) did not evolve to definite SSc and maintained normal capil-laroscopic pattern.

The small sample size investigated did not allow more robust combined analysis, completed by several validation stages, required by studies on functional biomarkers.10 Result confirma-tion needs larger sample size, including more clinical variables and closer follow-up. This preliminary cross-sectional/retrospec-tive analysis encourages further and larger works to ascertain if chemokines may be really the turning point from VEDOSS to definite SSc.

Clara Crescioli,1 Clarissa Corinaldesi,1,2 Valeria Riccieri,3 Valeria Raparelli,4 Massimiliano Vasile,3 Francesco Del Galdo,2,5 Guido Valesini,3 Andrea Lenzi,4 Stefania Basili,3 Cristina Antinozzi1

1Department of Movement, Human and Health Sciences, University of Rome "Foro Italico", Rome, Italy2Leeds Institute of Rheumatic and Musculoskeletal Medicine, University of Leeds, Leeds, UK3Department of Internal Medicine and Medical Specialties, Sapienza University of Rome, Rome, Italy4Department of Experimental Medicine, Sapienza University of Rome, Rome, Italy5Scleroderma Programme, NIHR Leeds Musculoskeletal Biomedical Research Centre, Leeds, UK

Correspondence to Professor Clara Crescioli, Department of Movement, Human and Health Sciences, Section of Health Sciences, Università degli Studi di Roma “Foro Italico”, Rome 00135, Italy; clara. crescioli@ uniroma4. it

Handling editor Josef S Smolen

Acknowledgements The authors thank Dr Silvia Giannattasio, University of Rome “Foro Italico”, for her support and assistance.

Collaborators Dr Silvia Giannattasio.

Contributors All authors were involved in the article drafting and crtitical revisions of important intellectual and experimental contents.All authors approved the final version to be published.

Funding This research was supported by the grant Scientific Independence of young Researchers (SIR) (protocol no. RBSI14D5NX).

Competing interests None declared.

Patient consent Obtained.

Ethics approval Sapienza University of Rome Ethic Committee (protocol no. 2567/15).

Provenance and peer review Not commissioned; externally peer reviewed.

Open access This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http:// creativecommons. org/ licenses/ by- nc/ 4. 0/

© Article author(s) (or their employer(s) unless otherwise stated in the text of the article) 2018. All rights reserved. No commercial use is permitted unless otherwise expressly granted.

SB and CA contributed equally.

SB and CA share the last co-authorship.

To cite Crescioli C, Corinaldesi C, Riccieri V, et al. Ann Rheum Dis 2018;77:1845–1846.

Received 16 February 2018Revised 12 April 2018Accepted 20 April 2018Published Online First 14 May 2018

Ann Rheum Dis 2018;77:1845–1846. doi:10.1136/annrheumdis-2018-213257

RefeRences 1 Matucci-Cerinic M, Kahaleh B, Wigley FM. Review: evidence that systemic sclerosis is

a vascular disease. Arthritis Rheum 2013;65:1953–62. 2 Varga J, Abraham D. Systemic sclerosis: a prototypic multisystem fibrotic disorder. J

Clin Invest 2007;117:557–67. 3 Chora I, Romano E, Manetti M, et al. Evidence for a derangement of the microvascular

system in patients with a very early diagnosis of systemic sclerosis. J Rheumatol 2017;44:1190–7.

4 Bellando Randone S, George J, Mazzotta C, et al. Angiostatic and Angiogenic Chemokines in Systemic Sclerosis: An Overview. J Scleroderma Relat Disord 2017;2:1–10.

5 Liu X, Mayes MD, Tan FK, et al. Correlation of interferon-inducible chemokine plasma levels with disease severity in systemic sclerosis. Arthritis Rheum 2013;65:226–35.

6 Avouac J, Fransen J, Walker UA, et al. Preliminary criteria for the very early diagnosis of systemic sclerosis: results of a Delphi Consensus Study from EULAR Scleroderma Trials and Research Group. Ann Rheum Dis 2011;70:476–81.

7 van den Hoogen F, Khanna D, Fransen J, et al. 2013 classification criteria for systemic sclerosis: an American college of rheumatology/European league against rheumatism collaborative initiative. Ann Rheum Dis 2013;72:1747–55.

8 Christmann RB, Hayes E, Pendergrass S, et al. Interferon and alternative activation of monocyte/macrophages in systemic sclerosis-associated pulmonary arterial hypertension. Arthritis Rheum 2011;63:1718–28.

9 Mach F, Sauty A, Iarossi AS, et al. Differential expression of three T lymphocyte-activating CXC chemokines by human atheroma-associated cells. J Clin Invest 1999;104:1041–50.

1 of 2Ann Rheum Dis December 2018 Vol 77 No 12

Calprotectin is not independent from baseline erosion in predicting radiological progression in early rheumatoid arthritis. Comment on ‘Calprotectin as a marker of inflammation in patients with early rheumatoid arthritis’ by Jonsson et al

We have read with great interest the article by Jonsson et al that was recently published online in ARD,1 which suggested that calprotectin, also known as S100A8/S100A9 heterodimer, was associated with radiographic progression in early rheumatoid arthritis (RA). Calprotectin correlates significantly with inflam-matory markers and disease activity score.2 Besides correlations between baseline calprotectin levels, Clinical Disease Activity Index and ultrasonography power Doppler, the authors showed that baseline calprotectin levels correlated with van der Heijde modified Sharp score (SHS) progression (defined as an increase ≥1 unit/year from 0 to 24 months), independently of age, gender, Clinical Disease Activity Index, erythrocyte sedimenta-tion rate (ESR), C reactive protein (CRP) levels and rheumatoid factor positivity.1

We analysed the initial serum calprotectin among patients with early RA fulfilling American College of Rheumatology/European League Against Rheumatism 2010 of the French observational cohort Etude et Suivi des POlyarthrites Indifférenciées Récentes (ESPOIR). Calprotectin serum concentrations were assessed according to manufacturer method (Hycult, Frontstraat, Neth-erlands; standard range from 1.6 to 25 ng/mL). Univariate and multivariate risk Cox models with a backward stepwise were constructed for 615 patients with early RA for whom radio-logical data were available. Outcome measures included in the analysis were gender, CRP, anti-citrullinated peptide antibody (ACPA), Disease Acivity Score, age, smoking status, calprotectin, disease-modifying antirheumatic drugs (DMARDs) treatment and typical initial erosion. The radiological progression was defined as an increase ≥5 of the total SHS score/year.

CRP, ACPA, DMARD treatment and calprotectin were significantly associated with structural evolution in the univar-iate analysis. When baseline erosion was removed from the multivariate analysis, calprotectin was the only predictor of the structural evolution over 3 years (HR 1.06, 95% CI (1.00 to 1.11), P=0.045, table 1). These results confirmed that

calprotectin predicts radiological progression in a large cohort of early RA. When the presence of baseline typical erosion was combined in the multivariate Cox model, calprotectin was not an independent predictor of structural evolution anymore (HR 1.03, 95% CI (0.97 to 1.10), P=0.297).

Calprotectin, which is predominantly expressed by myelo-monocytic cells and constitutes 40% of the polymorphonuclear neutrophil cytosolic proteins,3 was identified as a marker of RA in the synovial fluid and in the serum, with serum concentra-tions differentiating RA from other rheumatic diseases.4 Besides their intracellular functions,5 calprotectin has been introduced as an important proinflammatory factor mainly secreted by acti-vated neutrophils. A direct role in radiological damage has been suggested because of the activation of matrix metalloproteases by S100 proteins.6

We acknowledge the putative role of new biomarkers,2 such as calprotectin, in early RA management. Jonsson et al showed that calprotectin is a better predictor of structural progression than ESR or CRP. In order to know whether calprotectin should be implemented in daily practice, it is critical to determine whether calprotectin is also independent from major predictors of struc-tural evolution in RA, such as ACPA and baseline erosion.7 In ESPOIR cohort, calprotectin is no more associated with struc-tural damage when baseline erosion is considered.

Maxime Chevreau,1 Marie-Hélène Paclet,2,3 Xavier Romand,1,2 Jean-Louis Quesada,4,5 Olivier Vittecoq,6 Philippe Dieudé,7 Bertrand Toussaint,8 Philippe Gaudin,1,2 Athan Baillet1,2

1Department of Rheumatology, Grenoble Alpes University Hospital, Grenoble, France2Univ Grenoble Alpes, GREPI-UGA EA7408, Grenoble, France3Lab. Biochimie des Enzymes et des Protéines, Centre Hospitalier Universitaire Grenoble Alpes, Grenoble, France4INSERM, Clinical Investigation Center CIC P 1406, Grenoble Alpes University Hospital, Grenoble, France5Scientific Department of the Clinical Research Delegation, Grenoble Alpes University Hospital, Grenoble, France6Department of Rheumatology, Rouen Hospital, Bois-Guillaume, France7Department of Rheumatology, Bichat Hospital, Paris, France8Laboratoire TIMC-IMAG-TheREx, UMR 5525 Centre National de la Recherche Scientifique, Univ Grenoble Alpes, Grenoble, France

Correspondence to Dr Athan Baillet, Rheumatology, Grenoble Alpes University Hospital, Université Grenoble Alpes, Echirolles 38434 Cedex, France; abaillet@ chu- grenoble. fr

Acknowledgements We wish to thank Nathalie Rincheval (CHU Montpellier and EA 2415) who did expert monitoring and data management and all the investigators who recruited and followed the patients (F Berenbaum, Paris-Saint Antoine; MC Boissier, Paris-Bobigny; A Cantagrel, Toulouse; B Combe, Montpellier;

Correspondence

Table 1 Univariate analyses and multivariate analyses—risk factors of van der Heijde modified Sharp score (SHS) progression in the first 3 years

N = 615 No radiological progression, n=290

Radiological progression, n=325

Univariate analysis HR (95% CI) P value

Multivariate analysis HR (95% CI) P value

Female gender (%) 79.3% (230) 76.9% (250) 0.95 (0.73 to 1.23) 0.696 – –

CRP (mg/dL) 7 (4 to 18) 12 (5 to 28) 1.03 (1 to 1.06) 0.047 Not selected

ACPA (IU) 0 (0 to 256) 121 (0 to 572) 1.01 (1 to 1.01) 0.040 1.01 (1 to 1.01) 0.093

DAS-28 5.2±1.22 5.3±1.2 1.07 (0.98 to 1.17) 0.148 – –

Age (years) 48.7 (38 to 56.3) 52.4 (41.1 to 58.4) 1.01 (1 to 1.02) 0.082 – –

Current smoking 48.3% (140) 46.8% (152) 1.04 (0.84 to 1.29) 0.732 – –

DMARDs (%) 77.2% (224) 86.8% (282) 1.50 (1.09 to 2.07) 0.013 1.36 (0.99 to 1.89) 0.060

Calprotectin (µg/cL) 3.2 (1.88 to 4.8) 3.8 (2.3 to 5.3) 1.06 (1.01 to 1.12) 0.027 1.06 (1 to 1.11) 0.045

Univariate and multivariate analyses: Cox model; HR (95% CI).Percentage (number); mean ±SD or median (25th, 75th centiles) where appropriate.ACPA, anti-citrullinated protein/peptide antibody; CRP, C reactive protein (mg/dL); DAS-28, Disease Activity Score-28; DMARDs, disease-modifying antirheumatic drugs, that is, methotrexate ≥7.5 mg/week, leflunomide at the visit before radiological evolution or at the last follow-up visit; IU, international unit; Not selected, outcome was excluded from multivariate Cox model because P value for model entry was >0.5 or P value for model retention was>0.10. Radiological progression was defined as an increase ≥5 of the total SHS score.

2 of 2 Ann Rheum Dis December 2018 Vol 77 No 12

Correspondence

M Dougados, Paris-Cochin; P Fardelone et P Boumier, Amiens; B. Fautrel, Paris-La Pitié; RM Flipo, Lille; Ph Goupille, Tours; F Liote, Paris-Lariboisière; O Vittecoq, Rouen; X Mariette, Paris Bicetre; O Meyer et Ph Dieude, Paris Bichat; A Saraux, Brest; Th Schaeverbeke, Bordeaux; J Sibilia, Strasbourg). We thank V Devauchelle and C Lukas for expert X-ray reading and S Martin (Paris Bichat) who did all the central dosages of Creactive protein, IgA rheumatoid factor and IgM rheumatoid factor and anti-citrullinated protein antibodies. The authors thank Ms Sylvie Papacatsis for her contribution to this study.

Contributors MC, M-HP, XR, J-LQ and AB have made substantial contributions to the conception or design of the work, or the acquisition, analysis or interpretation of data. OV, PD, BT and PG have revised the draft critically for important intellectual content and have approved the final version published.

Funding An unrestricted grant from Merck Sharp and Dohme (MSD) was allocated for the first 5 years. Two additional grants from Institut national de la santé et de la recherche médicale (INSERM) were obtained to support part of the biological database. The French Society of Rheumatology, Pfizer, Abbvie and Roche-Chugai also supported the Etude et Suivi des POlyarthrites Indifférenciées Récentes INSERM (ESPOIR) cohort study. This study was funded by the Scientific Department of the Clinical Research Delegation (DRCI), Grenoble Alpes University Hospital.

Competing interests None declared.

Ethics approval The Montpellier ethics committee.

Provenance and peer review Not commissioned; externally peer reviewed.

© Article author(s) (or their employer(s) unless otherwise stated in the text of the article) 2018. All rights reserved. No commercial use is permitted unless otherwise expressly granted.

To cite Chevreau M, Paclet M-H, Romand X, et al. Ann Rheum Dis 2018;77: e84.

Received 8 December 2017Accepted 15 December 2017Published Online First 10 January 2018

► http:// dx. doi. org/ 10. 1136/ annrheumdis- 2017- 212869

Ann Rheum Dis 2018;77: e84. doi:10.1136/annrheumdis-2017-212816

RefeRences 1 Jonsson MK, Sundlisæter NP, Nordal HH, et al. Calprotectin as a marker of

inflammation in patients with early rheumatoid arthritis. Ann Rheum Dis 2017;76:2031–7.

2 Hammer HB, Fagerhol MK, Wien TN, et al. The soluble biomarker calprotectin (an S100 protein) is associated to ultrasonographic synovitis scores and is sensitive to change in patients with rheumatoid arthritis treated with adalimumab. Arthritis Res Ther 2011;13:R178.

3 Baillet A. [S100A8, S100A9 and S100A12 proteins in rheumatoid arthritis]. Rev Med Interne 2010;31:458–61.

4 Baillet A, Trocmé C, Berthier S, et al. Synovial fluid proteomic fingerprint: S100A8, S100A9 and S100A12 proteins discriminate rheumatoid arthritis from other inflammatory joint diseases. Rheumatology 2010;49:671–82.

5 Berthier S, Nguyen MV, Baillet A, et al. Molecular interface of S100A8 with cytochrome b558 and NADPH oxidase activation. PLoS One 2012;7:e40277.

6 Edgeworth J, Gorman M, Bennett R, et al. Identification of p8,14 as a highly abundant heterodimeric calcium binding protein complex of myeloid cells. J Biol Chem 1991;266:7706–13.

7 Baillet A, Gossec L, Paternotte S, et al. Evaluation of serum interleukin-6 level as a surrogate marker of synovial inflammation and as a factor of structural progression in early rheumatoid arthritis: results from a French national multicenter cohort. Arthritis Care Res 2015;67:905–12.

1 of 2Ann Rheum Dis December 2018 Vol 77 No 12

Response to: ‘Calprotectin is not independent from baseline erosion in predicting radiological progression in early rheumatoid arthritis’ by Chevreau et al

We appreciate the additional data regarding calprotectin and radiographic progression provided by Chevreau et al1 as an eLetter addressing our published research paper.2 It is important to explore new biomarkers in different cohorts of patients with early and established rheumatoid arthritis (RA). Calprotectin levels have previously been shown to be associated with joint damage in established RA.3 4 Hammer et al5 have shown that calprotectin was an independent predictor of radiographic joint damage after 10 years of follow-up. Chevreau et al present data on baseline calprotectin as a predictor of rapid radiographic progression (defined as an increase of ≥5 van der Heijde Sharp score units/year) in a large cohort of patients with early RA, and did not find calprotectin to be associated with structural damage when baseline erosions were considered.1

In order to address the issues raised by Chevreau et al1 we performed additional statistical analyses. When introducing baseline van der Heijde modified Sharp erosion score in the multivariate model (including erythrocyte sedimentation rate, C reactive protein, age, gender, Clinical Disease Activity Index and rheumatoid factor (RF)), baseline erosion score was a significant predictor of radiographic progression (OR 1.14, 95% CI 1.02 to 1.28; table 1). Importantly, calprotectin in the highest quartile remained a significant independent predictor of radiographic damage in the multivariate model (OR 3.52, 95% CI 1.15 to 10.72; table 1). RF was a stronger predictor than anticitrullinated peptide antibody (ACPA) in univariate models; thus, we chose to include RF in our initial model.2 When assessing the multivar-iate model including ACPA instead of RF, both calprotectin and

baseline erosions remained significant independent predictors of radiographic progression, while ACPA was not a significant predictor (OR 1.27, 95% CI 0.52 to 3.08, p=0.60).

Comparison between cohorts should be done with caution, and the Aiming for Remission in Rheumatoid Arthritis: a Randomized Trial Examining the Benefit of Ultrasonography in a Clinical TIght Control Regimen (ARCTIC)6 and Etude et Suivi des Polyarthrites Indifférenciées Récentes (ESPOIR) cohorts7 are different. Serologically, 44% vs 39% of patients were positive for RF-IgM and ACPA in the ESPOIR cohort compared with 71% and 82% in the ARCTIC cohort. Treatment strategies were different in the two cohorts; in ESPOIR there were no protocol-based treatment strategies, as opposed to ARCTIC with a structured, aggressive tight control algorithm aiming for remission. In the ARCTIC cohort, 41% of disease-modifying antirheumatic drug-naïve patients with early RA progressed radiographically during the 2 years of follow-up, with radiographic progression defined as ≥1 unit/year from 0 to 24 months.2 However, rapid radio-graphic progression was rare, occurring in only 11 out of the 230 patients included in the full analyses set.

In modern RA care, patients are identified at an early stage and often before radiographic damage is evident. In such a setting, our results indicate that calprotectin may be an independent predictor of radiographic damage, but the role of calprotectin needs to be further investigated in different cohorts before being fully implemented in routine care.

Maria Karolina Jonsson,1,2 Nina Paulshus Sundlisæter,2 Hilde Haugedal Nordal,1,3 Hilde Berner Hammer,2 Anna-Birgitte Aga,2 Désirée van der Heijde,2,4 Tore Kristian Kvien,2 Bjørg-Tilde Svanes Fevang,1,3 Siri Lillegraven,2 Espen Andre Haavardsholm2,5

1Bergen Group of Epidemiology and Biomarkers in Rheumatic Disease, Department of Rheumatology, Haukeland University Hospital, Bergen, Norway2Department of Rheumatology, Diakonhjemmet Hospital, Oslo, Norway

Correspondence response

Table 1 Predictors of radiographic progression ≥1 unit/year from 0 to 24 months (n=215)

Baseline variables

Univariate Multivariate

OR p value OR p value

Age 1.04 (1.02 to 1.07) <0.001 1.02 (1.00 to 1.05) 0.09

Gender (female) 0.61 (0.35 to 1.07) 0.09 0.71 (0.36 to 1.37) 0.30

Calprotectin quartile (range)

First quartile (186–556 µg/L) Ref Ref Ref Ref

Second quartile (567–1028 µg/L) 1.51 (0.66 to 3.46) 0.33 1.55 (0.60 to 4.01) 0.37

Third quartile (1045–2158 µg/L) 1.39 (0.61 to 3.20) 0.44 0.97 (0.36 to 2.60) 0.94

Fourth quartile (2235–48079 µg/L) 6.06 (2.62 to 14.02) <0.001 3.62 (1.20 to 10.93) 0.02

ESR, quartile (range)

First quartile (1–10 mm/hour) Ref Ref Ref Ref

Second quartile (11–18 mm/hour) 1.07 (0.47 to 2.43) 0.87 0.80 (0.31 to 2.04) 0.64

Third quartile (19–31 mm/hour) 1.26 (0.55 to 2.86) 0.59 0.82 (0.30 to 2.19) 0.69

Fourth quartile (32–110 mm/hour) 3.74 (1.64 to 8.52) 0.002 0.99 (0.29 to 3.35) 0.99

CRP, quartile (range)

First quartile (0.3–2.8 mg/L) Ref Ref Ref Ref

Second quartile (3–6 mg/L) 0.69 (0.29 to 1.64) 0.41 0.41 (0.15 to 1.09) 0.07

Third quartile (7–16 mg/L) 1.29 (0.54 to 3.04) 0.57 0.75 (0.25 to 2.19) 0.59

Fourth quartile (18–117 mg/L) 2.85 (1.20 to 6.76) 0.02 0.89 (0.24 to 3.22) 0.85

CDAI (0-76) 1.02 (1.00 to 1.04) 0.08 1.01 (0.98 to 1.04) 0.50

RF positivity 1.86 (0.99 to 3.48) 0.053 1.92 (0.91 to 4.06) 0.09

vdHS erosion score (0–280) 1.18 (1.07 to 1.29) 0.001 1.14 (1.01 to 1.27) 0.03

P values <0.05 in bold.CDAI, Clinical Disease Activity Index; CRP, C reactive protein; ESR, erythrocyte sedimentation rate; h, hour; L, liter; mg, milligram; mm, millimeter; OR, odds ratio; Ref, reference category (lowest quartile as reference); RF, rheumatoid factor; µg, microgram; vdHS, van der Heijde modified Sharp.

2 of 2 Ann Rheum Dis December 2018 Vol 77 No 12

Correspondence response

3Broegelmann Research Laboratory, Department of Clinical Science, University of Bergen, Bergen, Norway4Department of Rheumatology, Leiden University Medical Center, Leiden, The Netherlands5Institute of Health and Society, University of Oslo, Oslo, Norway

Correspondence to Dr Maria Karolina Jonsson, Bergen Group of Epidemiology and Biomarkers in Rheumatic Disease, Department of Rheumatology, Haukeland University Hospital, Bergen 5021, Norway; jonssonmk@ gmail. com

Handling editor Josef S Smolen

Acknowledgements We would like to thank Inge Dale at CalproLab, Marianne Eidsheim at Broegelmann Research Laboratory, Ellen Moholt and Camilla Fongen at Diakonhjemmet Hospital, and the ARCTIC study group.

Competing interests MKJ reports grants from Norwegian Extra Foundation for Health and Rehabilitation and non-financial support from Calpro AS during the conduct of the study. TKK has received fees for speaking and/or consulting from AbbVie, Biogen, BMS, Boehringer Ingelheim, Celgene, Celltrion, Eli Lilly, Epirus, Hospira, Merck Serono, MSD, Mundipharma, Novartis, Oktal, Orion Pharma, Hospira/Pfizer, Roche, Sandoz and UCB, and received research funding to Diakonhjemmet Hospital from AbbVie, BMS, MSD, Pfizer, Roche and UCB. EAH has received research funding from Pfizer, UCB, Roche, MSD and AbbVie, honorariums as a speaker from Pfizer, UCB, Roche and AbbVie, and honorariums for development of educational material from Pfizer, and has sat on advisory boards for Pfizer.

Provenance and peer review Commissioned; internally peer reviewed.

© Article author(s) (or their employer(s) unless otherwise stated in the text of the article) 2018. All rights reserved. No commercial use is permitted unless otherwise expressly granted.

To cite Jonsson MK, Sundlisæter NP, Nordal HH, et al. Ann Rheum Dis 2018;77:e85.

Received 22 December 2017Accepted 23 December 2017Published Online First 10 January 2018

► http:// dx. doi. org/ 10. 1136/ annrheumdis- 2017- 212816

Ann Rheum Dis 2018;77:e85. doi:10.1136/annrheumdis-2017-212869

RefeRences 1 Chevreau M, Paclet MH, Romand X, et al. Calprotectin is not independent from

baseline erosion in predicting radiological progression in early Rheumatoid Arthritis. Ann Rheum Dis 2018;77:e84.

2 Jonsson MK, Sundlisæter NP, Nordal HH, et al. Calprotectin as a marker of inflammation in patients with early rheumatoid arthritis. Ann Rheum Dis 2017;76:2031–7.

3 Hammer HB, Odegard S, Fagerhol MK, et al. Calprotectin (a major leucocyte protein) is strongly and independently correlated with joint inflammation and damage in rheumatoid arthritis. Ann Rheum Dis 2007;66:1093–7.

4 Nordal HH, Brun JG, Hordvik M, et al. Calprotectin (S100A8/A9) and S100A12 are associated with measures of disease activity in a longitudinal study of patients with rheumatoid arthritis treated with infliximab. Scand J Rheumatol 2016;45:274–81.

5 Hammer HB, Ødegård S, Syversen SW, et al. Calprotectin (a major S100 leucocyte protein) predicts 10-year radiographic progression in patients with rheumatoid arthritis

.

Ann Rheum Dis 2010;69:150–4. 6 Haavardsholm EA, Aga AB, Olsen IC, et al. Ultrasound in management of rheumatoid

arthritis: ARCTIC randomised controlled strategy trial. BMJ 2016;354:i4205. 7 Combe B, Benessiano J, Berenbaum F, et al. The ESPOIR cohort: a ten-year follow-up of

early arthritis in France: methodology and baseline characteristics of the 813 included patients. Joint Bone Spine 2007;74:440–5.

1 of 1Ann Rheum Dis December 2018 Vol 77 No 12

Simultaneous inhibition of α4/β7 integrin and tumour necrosis factor-α in concomitant spondyloarthritis and inflammatory bowel disease

We read with interest the article by Varkas et al1 suggesting a temporal association between the initiation of vedolizumab (VDZ) for inflammatory bowel disease (IBD) and induction or worsening of spondyloarthritis (SpA). This suggests that in patients with IBD, there may be limited efficacy of α4/β7 inte-grin blockade for management of extraintestinal manifestations, particularly for arthritis and sacroiliitis.2 3 Treatment of SpA in subjects on VDZ for associated IBD is highly challenging for the treating physician. We describe below two patients who were treated with a combination treatment of VDZ and certolizumab pegol (CZP) (table 1).

Patient 1 is a 24 year-old man diagnosed with juvenile-onset SpA in 2008 and Crohn’s disease (CD) in 2011. Sequential treat-ments with numerous tumour necrosis factor (TNF)-α inhibitors were associated with loss of efficacy or intolerance. In 2014, initiation of CZP was able to control articular symptoms. During follow-up, he developed uncontrolled disease activity of his CD, and in 2016 it was decided to add VDZ to the CZP treatment. Excellent responses of both gastrointestinal and musculoskeletal symptoms were noted. After 10 months of ongoing combination therapy, the patient is still without any symptoms of either IBD or SpA. No adverse events were reported.

Patient 2 is a 48 year-old woman diagnosed with ulcerative colitis (UC) in 2012. After inadequate response to numerous TNF-α inhibitors, she was initiated on VDZ. Despite good clinical and endoscopic response of her colitis, she developed inflammatory back pain, peripheral arthritis and enthesitis 5

months following the initiation of VDZ. A pelvic MRI confirmed bilateral sacroiliitis. CZP was initiated in lieu of VDZ to provide adequate control of the arthritis. Nine months after the last infu-sion of VDZ, the patient presented with a flare of UC. It was then decided to reinitiate VDZ in addition to CZP. After the fourth infusion of VDZ, recurrence in the symptoms of arthritis was noted, and it was decided to discontinue CZP given its loss of efficacy. No adverse events were reported neither during the time of combination therapy nor following its cessation.

In spite of a relatively short follow-up, it is reassuring that no serious adverse events or infectious complications were recorded in our patients. Sustained remission was observed in patient 1, whereas recurrence of arthritis was observed in patient 2. Our report suggests that combination of VDZ and TNF-α could be considered in the therapeutic armamentarium for patients with SpA and IBD. Previous reports similarly showed no increased safety signals4–6; however, larger studies are needed in order to assess the outcomes of simultaneous inhibition of α4/β7 integrin and TNF-α.

Nicolas Richard,1,2 Elizabeth M Hazel,2 Nigil Haroon,1 Robert D Inman1

1Division of Rheumatology, University Health Network, Toronto Western Hospital, Toronto, Ontario, Canada2McGill University Health Centre, Montreal General Hospital, Montreal, Quebec, Canada

Correspondence to Dr. Nicolas Richard, Division of Rheumatology, University Health Network, Toronto Western Hospital, Toronto, ON M5T 2S8, Canada; nicolas. richard2@ mail. mcgill. ca

Contributors All coauthors have contributed to this letter and agree with its content.

Competing interests None declared.

Provenance and peer review Not commissioned; internally peer reviewed.

© Article author(s) (or their employer(s) unless otherwise stated in the text of the article) 2018. All rights reserved. No commercial use is permitted unless otherwise expressly granted.

To cite Richard N, Hazel EM, Haroon N, et al. Ann Rheum Dis 2018;77:e86.

Received 8 December 2017Accepted 15 December 2017Published Online First 29 December 2017

► http:// dx. doi. org/ 10. 1136/ annrheumdis- 2017- 212858

Ann Rheum Dis 2018;77:e86. doi:10.1136/annrheumdis-2017-212819

RefeRences 1 Varkas G, Thevissen K, De Brabanter G, et al. An induction or flare of arthritis and/or

sacroiliitis by vedolizumab in inflammatory bowel disease: a case series. Ann Rheum Dis 2017;76:878–81.

2 Vavricka SR, Scharl M, Gubler M, et al. Biologics for extraintestinal manifestations of IBD. Curr Drug Targets 2014;15:1064–73.

3 Fedyk ER, Wyant T, Yang LL, et al. Exclusive antagonism of the α4 β7 integrin by vedolizumab confirms the gut-selectivity of this pathway in primates. Inflamm Bowel Dis 2012;18:2107–19.

4 Fischer S, Rath T, Geppert CI, et al. Long-term combination therapy with anti-TNF plus vedolizumab induces and maintains remission in therapy-refractory ulcerative colitis. Am J Gastroenterol 2017;112:1621–3.

5 Bethge J, Meffert S, Ellrichmann M, et al. Combination therapy with vedolizumab and etanercept in a patient with pouchitis and spondylarthritis. BMJ Open Gastroenterol 2017;4:e000127.

6 Hirten R, Longman RS, Bosworth BP, et al. Vedolizumab and infliximab combination therapy in the treatment of crohn’s disease. Am J Gastroenterol 2015;110:1737–8.

Correspondence

Table 1 Patient characteristics, treatment and outcome of subjects treated with VDZ and CZP combination therapy

Patient 1 Patient 2

Age (years) 24 48

Sex Male Female

Ethnicity Caucasian Caucasian

HLA-B27 Negative Negative

Family history Positive Negative

Modified New York criteria* Yes No

Age at SpA onset (years) 15 47†

Age of onset of IBD (years) 21 43

Biologic use prior to combination therapy

GOL, IFX, ADA, CZP IFX, ADA, GOL, VDZ, CZP

Reason for combination therapy Active colitis‡ Active colitis‡

Treatment VDZ+CZP VDZ+CZP

Duration of follow-up (months) 10 3.5

Infections None None

Outcome Remission Flare of SpA

*Sacroiliitis ≥ bilateral grade 2 or unilateral grade 3.†Symptoms onset 5 months following initiation of VDZ.‡Despite adequate control of SpA on CZP.ADA, adalimumab; CZP, certolizumab pegol; GOL, golimumab; HLA, human leucocyte antigen; IBD, inflammatory bowel disease; IFX, infliximab;  SpA, spondyloarthritis; VDZ, vedolizumab.

1 of 1Ann Rheum Dis December 2018 Vol 77 No 12

Response to: ‘Simultaneous inhibition of α4/β7 integrin and tumor necrosis factor-α in concomitant spondyloarthritis and inflammatory bowel disease’ by Richard et al

We thank Dr Richard et al for their interest in our work and the communication of their experiences in daily practice.1 In patients affected with both spondyloarthritis (SpA) and refractory inflam-matory bowel disease (IBD) despite anti-tumour necrosis factor (TNF) therapy, treatment options are limited. While anti-TNF therapy has shown remarkable efficacy in the treatment of both axial and peripheral SpA, even long term, the results in IBD were less satisfying.2 3 With the introduction of vedolizumab, a new mode of action and hope for refractory patients was installed.4 5 Nevertheless, as we reported earlier, vedolizumab did not show any efficacy in a series of patients with SpA and might even induce SpA-like disease in certain individuals.6

Richard et al describe the use of certolizumab pegol in combi-nation with vedolizumab in two patients with IBD refractory to anti-TNF. Interestingly, while patient 1 represents the lack of efficacy of vedolizumab in SpA, unfortunately patient 2 matches the profile of patients developing SpA with the initiation of vedolizumab.

Although we agree that the combination of drugs is inevitable in these particular cases, several practical implications arise. First, which anti-TNF should be preferred and when should it be installed? In these cases, anti-TNF therapy has failed to confine gut inflammation. Therefore, the use of monoclonal antibodies for treatment of extra-articular manifestations may be redun-dant, unless other EAMs such as psoriasis or uveitis occurred in the disease course. Additionally, in the combination of two biological immunosuppressive agents safety is a key determinant. This has particularly been a point of concern when combining biologics in patients with rheumatoid arthritis.7 Therefore, the use of etanercept with shorter half-life may be preferred in these patients developing SpA after vedolizumab treatment.8 However, in patients already diagnosed with both SpA and IBD, the anti-TNF therapy will usually be withdrawn before initiation of vedolizumab due to inefficacy. This could increase the risk of secondary inefficacy of anti-TNF due to immunogenicity when reinstated for SpA. This presumably may have been the case in patient 2. Thus, in case of predictable flare of SpA in patients displaying both diseases, the withdrawal of anti-TNF should be avoided. However, at present there are no clear predictors avail-able that permit to predict which patients will develop new onset SpA or a relapse on initiation of vedolizumab.

Second, there are also questions regarding reimbursements by insurance companies or legislators. Unfortunately, both vedoli-zumab and anti-TNF therapy are expensive therapies. Although the medical need is urgent, the use of both drugs in the same patients is at present based on expert opinion and therefore not necessarily reimbursed. Eventually, outcome and safety data might persuade legislators to embed this treatment strategy in general practice. However, for now, rheumatologists and gastro-enterologists might have to rely on medical need programmes to treat their patients accordingly.

Therefore, the use of vedolizumab monotherapy after loss of efficacy of anti-TNF at the level of the gut in patients displaying

both SpA and IBD should be taken with caution as flares may be expected short term. Additionally, in patients developing SpA-like disease after treatment with vedolizumab, for now, combination therapy with anti-TNF might be inevitable. There is currently little evidence that vedolizumab should be preferred over anti-TNF therapy in biologic-naive patients. Although several case reports have reported on short-term effects, long-term evaluation should be conducted to determine the safety of such a dual strategy on malignancy and infectious complications.

Gaëlle Varkas,1,2 Dirk Elewaut1,2

1Department of Rheumatology, Ghent University Hospital, Ghent, Belgium2VIB Inflammation Research Center, Ghent University, Ghent, Belgium

Correspondence to Dr Dirk Elewaut, Department of Rheumatology, University Hospital of Ghent, Ghent, 9000, Belgium; dirk. elewaut@ ugent. be

Handling editor Josef S Smolen

Competing interests None declared.

Provenance and peer review Commissioned; internally peer reviewed.

© Article author(s) (or their employer(s) unless otherwise stated in the text of the article) 2018. All rights reserved. No commercial use is permitted unless otherwise expressly granted.

To cite Varkas G, Elewaut D. Ann Rheum Dis 2018;77:e87.

Received 2 January 2018Revised 5 January 2018Accepted 5 January 2018Published Online First 26 January 2018

► http:// dx. doi. org/ 10. 1136/ annrheumdis- 2017- 212819

Ann Rheum Dis 2018;77:e87. doi:10.1136/annrheumdis-2017-212858

RefeRences 1 Richard N, Hazel EM, Haroon N, et al. Simultaneous inhibition of α4/β7 integrin and

tumour necrosis factor-α in concomitant spondyloarthritis and inflammatory bowel disease. Ann Rheum Dis 2018;77:e86.

2 Stidham RW, Lee TC, Higgins PD, et al. Systematic review with network meta-analysis: the efficacy of anti-tumour necrosis factor-alpha agents for the treatment of ulcerative colitis. Aliment Pharmacol Ther 2014;39:660–71.

3 Hazlewood GS, Rezaie A, Borman M, et al. Comparative effectiveness of immunosuppressants and biologics for inducing and maintaining remission in Crohn’s disease: a network meta-analysis. Gastroenterology 2015;148:344–54. quiz e14-5.

4 Sandborn WJ, Feagan BG, Rutgeerts P, et al. Vedolizumab as induction and maintenance therapy for Crohn’s disease. N Engl J Med 2013;369:711–21.

5 Feagan BG, Rutgeerts P, Sands BE, et al. Vedolizumab as induction and maintenance therapy for ulcerative colitis. N Engl J Med 2013;369:699–710.

6 Varkas G, Thevissen K, De Brabanter G, et al. An induction or flare of arthritis and/or sacroiliitis by vedolizumab in inflammatory bowel disease: a case series. Ann Rheum Dis 2017;76:878–81.

7 Weinblatt M, Combe B, Covucci A, et al. Safety of the selective costimulation modulator abatacept in rheumatoid arthritis patients receiving background biologic and nonbiologic disease-modifying antirheumatic drugs: a one-year randomized, placebo-controlled study. Arthritis Rheum 2006;54:2807–16.

8 van Dartel SA, Fransen J, Kievit W, et al. Difference in the risk of serious infections in patients with rheumatoid arthritis treated with adalimumab, infliximab and etanercept: results from the Dutch Rheumatoid Arthritis Monitoring (DREAM) registry. Ann Rheum Dis 2013;72:895–900.

Correspondence response

1 of 2Ann Rheum Dis December 2018 Vol 77 No 12

Standard dose of Ustekinumab for childhood-onset deficiency of interleukin–36 receptor antagonist

We read with interest the letter from Bonekamp et al1 reporting two children with severe deficiency of interleukin-36 receptor antago-nist (DITRA) treated with high dose of Ustekinumab. DITRA, an autosomal recessive disease caused by mutations of IL36RN, is characterised by generalised pustular psoriasis and systemic inflam-mation.2 We report here a paediatric case successfully treated with standard dose of Ustekinumab, a monoclonal antibody against the p40 subunit of interleukin (IL)-12/IL-23.

We reported in 2015 a boy born from consanguineous Tunisian parents who presented at 1 month of life with diffuse inflamma-tory pustular psoriasis and erythrodermic extension.3 At the age of 2 months, he was referred to us with fever, irritability, severe failure to thrive and recurrent diarrhoea; he had slight microcephaly and triangular chin. Oesophageal and gastrointestinal endoscopy showed minimal gastric ulcers, normal macroscopic colonic aspect but focal inflammatory infiltrates (predominant lymphocytes) in gastric body specimens. Leucocyte/neutrophil/platelet counts were markedly increased during flares, conversely with normal erythrocyte sedi-mentation/C-reactive protein rates. Cytokine profile secretion was unexpected, with high tumour necrosis factor (TNF)-α secretion in blood and high IL-1β secretion in cerebrospinal fluid (normal IL-1β, IL-6 rates in blood; normal TNF-α, IL-6 rates in cerebro-spinal fluid). Cutaneous biopsy was consistent with the diagnosis of pustular psoriasis and homozygous L27P mutation in IL36RN gene was detected. Topical corticosteroids and acitretin were inefficient. We started treatment with Anakinra, at doses up to 6 mg/kg, which initially induced complete response on fever.3 4 A partial cutaneous response was observed but Anakinra quickly failed to control skin rashes (figure 1A). Moreover, he presented recurrent bacterial skin infections, without biological signs of primary immunodeficiency. In addition, his general condition did not globally improve with time and he still presented severe weight and growth retardation. We decided to stop Anakinra and we started Etanercept during 3 months (0.8 mg/kg/week) without any efficacy. After parental consent, Ustekinumab, on monotherapy, was started at 4.5 years old at a dose of 0.75 mg/kg every 3 months (after second induction dose at week 4). A quick complete response was initially observed and sustained at a 15-month follow-up; his rash completely vanished, he never had fever again and he recovered normal growth velocity (figures 1B, C and 2). We maintained a standard dose until now, with an excellent tolerance, and a 12-month discontinuation of topical steroids.

Our experience underlines the interest of the blockade of the IL-23/T helper cell 17 axis in patients with DITRA and standard dose of Ustekinumab, those regularly applied in treating non-monogenic psoriasis may be sufficient. DITRA and non-monogenic psoriasis may have pathophysiolog-ical features in common,5 meaning that complex interac-tions between innate and adaptive immune systems may be involved.6–10 Failure to thrive in DITRA children may be explained by exudative cutaneous protein loss, systemic inflammatory activity, gastrointestinal inflammatory involve-ment and/or the late repercussions of a lack of early maternal attachment.

Bilade Cherqaoui,1,2 Linda Rossi-Semerano,1,2 Maryam Piram,1,2 Isabelle Koné-Paut1,2

1Paediatric Rheumatology, APHP, CHU Bicêtre, University of Paris Sud, Le Kremlin-Bicetre, Ile-de-France, France2CEREMAIA, French Reference Centre for Auto-Inflammatory Diseases and Inflammatory Amyloidosis, Le Kremlin-Bicêtre, Ile-de-France, France

Correspondence to Dr Bilade Cherqaoui Jr, Paediatric Rheumatology, Hopital Bicetre, Le Kremlin-Bicetre, Ile-de-France 94275, France; bilade. cherqaoui@ hotmail. fr

Contributors BC wrote the first draft of the manuscript. Each author listed has seen and approved the submission of this version of the manuscript and takes full responsibility for the manuscript.

Competing interests None declared.

Patient consent Next of kin consent obtained.

Provenance and peer review Not commissioned; internally peer reviewed.

Correspondence

Figure 1 Cutaneous evolution under Anakinra and Ustekinumab. (A) New flare of generalised psoriasis under treatment with Anakinra (January 2012). (B) Clinical picture after 6-month treatment with Ustekinumab (January 2017). (C) Clinical picture after 12-month treatment with Ustekinumab (October 2017).

Figure 2 Our deficiency of interleukin-36 receptor antagonist patient weight and height curves under three types of biological treatments.

2 of 2 Ann Rheum Dis December 2018 Vol 77 No 12

Correspondence

© Article author(s) (or their employer(s) unless otherwise stated in the text of the article) 2018. All rights reserved. No commercial use is permitted unless otherwise expressly granted.

To cite Cherqaoui B, Rossi-Semerano L, Piram M, et al. Ann Rheum Dis 2018;77:e88.

Received 4 December 2017Accepted 10 December 2017Published Online First 19 December 2017

► http://dx.doi.org/10.1136/annrheumdis-2017-212832

Ann Rheum Dis 2018;77:e88. doi:10.1136/annrheumdis-2017-212793

RefeRences 1 Bonekamp N, Caorsi R, Viglizzo GM, et al. High-dose ustekinumab for severe

childhood deficiency of interleukin-36 receptor antagonist (DITRA). Ann Rheum Dis 2018;77:1241–3.

2 Marrakchi S, Guigue P, Renshaw BR, et al. Interleukin-36-receptor antagonist deficiency and generalized pustular psoriasis. N Engl J Med 2011;365:620–8.

3 Rossi-Semerano L, Piram M, Chiaverini C, et al. First clinical description of an infant with interleukin-36-receptor antagonist deficiency successfully treated with anakinra. Pediatrics 2013;132:e1043–7.

4 Podlipnik S, Morgado-Carrasco D, Fustà-Novell X, et al. Dynamics of plasma cytokines in a patient with deficiency of interleukin-36 receptor antagonist successfully treated with Anakinra. Br J Dermatol 2017.

5 Kim J, Krueger JG. Highly effective new treatments for psoriasis target the IL-23/Type 17 T cell autoimmune axis. Annu Rev Med 2017;68:255–69.

6 Johnston A, Xing X, Wolterink L, et al. IL-1 and IL-36 are dominant cytokines in generalized pustular psoriasis. J Allergy Clin Immunol 2017;140:109–20.

7 Tortola L, Rosenwald E, Abel B, et al. Psoriasiform dermatitis is driven by IL-36-mediated DC-keratinocyte crosstalk. J Clin Invest 2012;122:3965–76.

8 Carrier Y, Ma HL, Ramon HE, et al. Inter-regulation of Th17 cytokines and the IL-36 cytokines in vitro and in vivo: implications in psoriasis pathogenesis. J Invest Dermatol 2011;131:2428–37.

9 Cua DJ, Tato CM. Innate IL-17-producing cells: the sentinels of the immune system. Nat Rev Immunol 2010;10:479–89.

10 Gaffen SL, Jain R, Garg AV, et al. The IL-23-IL-17 immune axis: from mechanisms to therapeutic testing. Nat Rev Immunol 2014;14:585–600.

1 of 1Ann Rheum Dis December 2018 Vol 77 No 12

Response to: 'Standard dose of ustekinumab for childhood-onset deficiency of interleukin-36 receptor antagonist’ by Cherqaoui et al

The case reported by Cherqaoui et al1 on ustekinumab in the treat-ment of deficiency of interleukin-36 receptor antagonist (DITRA) is interesting in several aspects.

The authors describe intrathecal elevation of proinflammatory cytokines, which is a novel and intriguing finding. The mutation in their case is identical to that of the Dutch/Moroccan patient we described. The total absence of any clinical disease activity in a genetically affected younger sister of the patient further supports the variable expression of identical IL36RN mutations as reported by Cherqaoui et al.

As observed in our cases, the patient was treated with several drugs with an absent or only transient improve-ment.2 At variance with our previous experience, the use of ustekinumab at standard dose in monotherapy was able to completely control the clinical picture, without the need for any topical steroids. These observations underline the clin-ical spectrum of disease severity associated with DITRA. In our previous experience, both patients displayed an initial response at standard doses but displayed a tendency to mild/moderate relapses in proximity to the scheduled doses. According to the few cases so far described, it is there-fore conceivable that ustekinumab should be considered among the first-line treatments for this severe condition. As suggested by Cherqaoui et al, a standard dose of 0.75 mg/kg every 12 weeks could be sufficient to completely control disease flares and should therefore be used as an initial approach. However, higher doses at more frequent inter-vals could be necessary to achieve a complete remission in patients with a partial clinical response, with a good safety profile.

Nadia Bonekamp,1 Roberta Caorsi,2 Joost Frenkel,1 Marco Gattorno2

1Department of Pediatrics, Wilhelmina Children’s Hospital, University Medical Center Utrecht, Utrecht, The Netherlands2Rheumatology Unit, G Gaslini Institute, Genova, Italy

Correspondence to Dr Marco Gattorno, Clinica Pediatrica e Reumatologia, G. Gaslini Institute, Genova, Italy; marcogattorno@ gaslini. org

Handling editor Josef S Smolen

Contributors All the authors wrote the reply letter.

Competing interests None declared.

Patient consent Obtained.

Ethics approval G Gaslini Institute.

Provenance and peer review Commissioned; internally peer reviewed.

© Article author(s) (or their employer(s) unless otherwise stated in the text of the article) 2018. All rights reserved. No commercial use is permitted unless otherwise expressly granted.

To cite Bonekamp N, Caorsi R, Frenkel J, et al. Ann Rheum Dis 2018;77:e89.

Received 20 December 2017Accepted 23 December 2017Published Online First 13 January 2018

► http:// dx. doi. org/ 10. 1136/ annrheumdis- 2017- 212793

Ann Rheum Dis 2018;77:e89. doi:10.1136/annrheumdis-2017-212832

RefeRences 1 Cherqaoui B, Rossi-Semerano L, Piram M, et al. Standard dose of Ustekinumab for

childhood-onset deficiency of interleukin-36 receptor antagonist. Ann Rheum Dis 2018;77:e88.

2 Bonekamp N, Caorsi R, Viglizzo GM, et al. High-dose ustekinumab for severe childhood deficiency of interleukin-36 receptor antagonist (DITRA). Ann Rheum Dis 2017;77:1241–3.

Correspondence response

1 of 2Ann Rheum Dis December 2018 Vol 77 No 12

Performance of the 2017 European League Against Rheumatism/American College of Rheumatology classification criteria for adult and juvenile idiopathic inflammatory myopathies in clinical practice

The 2017 European League Against Rheumatism/American College of Rheumatology (EULAR/ACR) classification criteria for idiopathic inflammatory myopathies (IIM) are the first vali-dated classification criteria for IIM. The criteria provide a score with a corresponding probability of having IIM. The cut-offs for classifying a patient with IIM were set at ≥5.5 and ≥6.7 in the

probability models without and with available muscle biopsy results, respectively. At these cut-offs, the criteria had an 87% and 93% sensitivity, and 82% and 88% specificity, in the model without and with the muscle biopsy results. The criteria were tested for sensitivity, but not for specificity in two independent cohorts.1

We evaluated the performance of these criteria in a retrospec-tive cohort of consecutively diagnosed IIM cases administered at our secondary/tertiary rheumatology centre between January 2010 and October 2017. The controls were the consecutive patients, without proof of IIM, who had an extensive diagnostic work-up for IIM, including muscle biopsy, during the same time period.

During the 94-month observation period we diagnosed IIM in 95 patients (72.6% female, age range 28–87 years). A muscle biopsy was performed in 87 (91.6%) patients, and was consis-tent with IIM in 97.7% of them. The control group consisted of 72 subjects (55.6% female, age range 18–88 years). The baseline clinical features of the cases and controls, and the clinical diag-noses of the cases are presented in table 1.

We assessed the sensitivity and specificity of the EULAR/ACR IIM criteria using the model that predicted the probability of IIM without muscle biopsy data in all patients, the model that predicted the probability of IIM with muscle biopsy data in all patients that had biopsy results available, and the model with or without biopsy data depending on the availability of the muscle biopsy results, respectively. The results are presented in table 2. The cases that were misclassified were predominantly those with an immune-mediated necrotising myopathy (35%), and those with a mild myopathy or without a myopathy, who had an interstitial lung disease, and myositis-specific antibodies other than a-Jo-1 (40%). In the control group, the cases presenting with severe toxic rhabdomyolysis were most commonly false positive (71.4%).

To summarise, in our retrospective IIM cohort, the 2017 EULAR/ACR IIM criteria showed a lower sensitivity at a comparable speci-ficity to those originally reported. The lower sensitivity might be at least partially explained by the omission of the histological features of a necrotising myopathy in the muscle biopsy item, and the restric-tion of the antibodies’ item solely to the presence of anti-Jo-1. These limitations had already been raised by the authors.

Alojzija Hočevar, Ziga Rotar, Monika Krosel, Saša Čučnik, Matija Tomšič

Department of Rheumatology, University Medical Centre Ljubljana, Ljubljana, Slovenia

Correspondence to Dr Alojzija Hočevar, Department of Rheumatology, Universitiy Medical Centre Ljubljana, Ljubljana 1000, Slovenia; alojzija. hocevar@ gmail. com

Contributors All authors met the authorship requirements. AH and MT designed the study question. AH, SC and MK collected the data. AH and ZR analysed the data. AH prepared the first draft of the manuscript. AH, ZR, MT, MK and SC significantly contributed to the final manuscript.

Funding The study was funded by the Slovenian National Research (grant P3-0314).

Correspondence

Table 1 Clinical characteristics of patients with an inflammatory myopathy and controls

Characteristics IIM Controls

n 95 72

Age (years)* 62.3 (52.3–71.3) 54.5 (45.0–72.4)

Female:male ratio 2.7 1.3

Symptom duration time (months)* 4.0 (2.0–10.0) 7.0 (1.5–24.0)

Proximal muscle weakness 76.8% 47.2%

Dysphagia 26.3% 9.7%

Elevated muscle enzymes 83.2% 69.4%

Myopathic EMG 67.8% (61/90) 36.5% (23/63)

Skin involvement 53.7% 1.4%

Interstitial lung disease 36.8% 4.2%

Arthritis 24.2% 2.8%

Raynaud phenomenon 21.1% 4.2%

MSA and/or MAA† 58.9% 0%

Anti-Jo-1 16.8% 0%

HEP-2 titre ≥1:80 61.1% 9.7%

Muscle biopsy done 91.6% 100%

Muscle biopsy positive 97.7% 0%

Clinical diagnosis

Dermatomyositis 27.1%

Antisynthetase syndrome 22.1%

Immune-mediated necrotising myopathy

14.7%

Myositis overlap syndrome 14.7%

Polymyositis 10.5%

Cancer-associated myositis 7.4%

Inclusion body myositis 1.1%

Unspecified myositis 2.1%

*Median (IQR).†MSA (myositis- specific antibodies) or MAA (myositis-associated antibodies): anti-Jo-1, anti-PL7, anti-PL12, anti-Mi-2, anti-SRP, anti-HMGCR, anti-MDA5; anti-TIF1γ; anti-NXP-2; anti-SAE; anti-Ku; anti-PM-Scl; anti-Ro; anti-U1RNP; anti-Scl-70.EMG, electromyography; IIM, idiopathic inflammatory myopathies.

Table 2 The sensitivity and specificity of the new EULAR/ACR classification criteria for idiopathic inflammatory myopathies

Sensitivity %(95% CI)

Specificity %(95% CI)

Positive predictive value % (95% CI)

Negative predictive value % (95% CI)

Using the probability model without muscle biopsy resultsncases=95; ncontrols=72

74.7 (64.8 to 83.1) 80.6 (69.5 to 88.9) 83.5 (75.8 to 89.2) 70.7 (62.7 to 77.7)

Using the probability model with muscle biopsy resultsncases=87; ncontrols=72

80.5 (70.6 to 88.2) 90.3 (81.0 to 96.0) 90.9 (83.1 to 95.3) 79.3 (71.3 to 85.5)

Using the probability model depending on the availability of the muscle biopsy results ncases=95; ncontrols=72

78.9 (69.4 to 86.6) 90.3 (81.0 to 96.0) 91.5 (84.0 to 95.6) 76.5 (68.6 to 82.8)

ACR, American College of Rheumatology; EULAR, European League Against Rheumatism.

2 of 2 Ann Rheum Dis December 2018 Vol 77 No 12

Correspondence

Competing interests None declared.

Ethics approval Slovenian National Medical Ethics Committee.

Provenance and peer review Not commissioned; internally peer reviewed.

© Article author(s) (or their employer(s) unless otherwise stated in the text of the article) 2018. All rights reserved. No commercial use is permitted unless otherwise expressly granted.

To cite Hočevar A, Rotar Z, Krosel M, et al. Ann Rheum Dis 2018;77:e90.

Received 30 November 2017Accepted 3 December 2017

Published Online First 13 December 2017

► http:// dx. doi. org/ 10. 1136/ annrheumdis- 2017- 212786

Ann Rheum Dis 2018;77:e90. doi:10.1136/annrheumdis-2017-212774

RefeRence 1 Lundberg IE, Tjärnlund A, Bottai M, et al. 2017 European league against rheumatism/

American college of rheumatology classification criteria for adult and juvenile idiopathic inflammatory myopathies and their major subgroups. Ann Rheum Dis 2017;76:1955–64.

1 of 1Ann Rheum Dis December 2018 Vol 77 No 12

Response to: ‘Performance of the 2017 European League Against Rheumatism/American College of Rheumatology classification criteria for adult and juvenile idiopathic inflammatory myopathies in clinical practice’ by Hočevar et al

We read with interest the letter titled ‘Performance of the 2017 European League Against Rheumatism/American College of Rheumatology classification criteria for adult and juve-nile idiopathic inflammatory myopathies in clinical practice’ by Hočevar et al published in the Annals of the Rheumatic Diseases.1 In the letter the authors report the sensitivity and specificity of the recently published ‘2017 European League Against Rheumatism/American College of Rheumatology clas-sification criteria for adult and juvenile idiopathic inflamma-tory myopathies and their major subgroups’2 observed in a retrospective cohort of 95 patients with idiopathic inflamma-tory myopathies (IIM) collected between 2010 and 2017 in a Slovenian rheumatology centre. As comparators, they used 72 subjects who had had a work-up for IIM, including muscle biopsy, during the same time period as the patients with IIM but without proof of IIM. The sensitivity and specificity were obtained for both the model with and that without muscle biopsy data. The sensitivity in the Slovenian rheumatology cohort was 74.7% without muscle biopsy data and 80.5% with muscle biopsy data. The corresponding figures reported in the original paper2 were 87% and 93%, respectively. The speci-ficity observed in the Slovenian rheumatology centre cohort was 80.6% without muscle biopsy data and 90.3% with muscle biopsy data. The corresponding figures reported in the orig-inal paper2 were 82% and 88%, respectively. The positive and negative predictive values observed in the Slovenian cohort were also smaller than in the original paper2: positive predic-tive values of 84%–91% vs 90%–94%, and negative predictive values of 71%–79% vs 79%–85%.

The cases that were identified as the newly defined subgroup of IIM called immune-mediated necrotising myopathies in the Slove-nian cohort were misclassified by the new criteria. This subgroup was very small (n=11) in the cohort used to derive the criteria. The subjects with a mild or no myopathy but with interstitial lung disease and myositis-specific autoantibodies (MSAs) other than anti-Jo-1 were also misclassified. This group was also small in the cohort that was used to derive the new criteria due to missing infor-mation on a majority of the new MSAs. It is not clear from the report by Hočevar et al how many cases had missing data and how this problem was handled. It is unclear to us how many patients were unclassifiable and how these were handled in the calculations. Unclassifiable patients are those whose predicted probability range spans across the cut-off value. Incorrect handling of these may help

explain the slightly smaller sensitivity and specificity observed in the Slovenian cohort as compared with the classification criteria cohort.

The report in the letter emphasises the need to validate the criteria in external cohorts including all aspects of the hetero-geneous IIM population and also the need to revise the ‘2017 European League Against Rheumatism/American College of Rheu-matology classification criteria for adult and juvenile idiopathic inflammatory myopathies and their major subgroups’, including the newly identified subgroups of IIM like the immune-mediated necrotising myopathy and the patients with predominating extra-muscular manifestations such as interstitial lung disease associated with the newly identified MSAs using validated antibody assays. These opportunities for improvement have been emphasised in the original paper2 and are on the research agenda.

We welcome this report and all others that may appear in the future that will contribute to the validation of the new criteria in external cohorts of patients.

Ingrid E Lundberg,1 Matteo Bottai,2 Anna Tjärnlund1

1Rheumatology Unit, Department of Medicine, Karolinska University Hospital, Karolinska Institutet, Stockholm, Sweden2Unit of Biostatistics, Institute for Environmental Medicine, Karolinska Institutet, Stockholm, Sweden

Correspondence to Professor Ingrid E Lundberg, Rheumatology Unit, Karolinska University Hospital, Stockholm S-171 76, Sweden; ingrid. lundberg@ ki. se

Handling editor Josef S Smolen

Contributors All authors have contributed to the drafting and finalisation of the response letter.

Competing interests None declared.

Provenance and peer review Commissioned; internally peer reviewed.

© Article author(s) (or their employer(s) unless otherwise stated in the text of the article) 2018. All rights reserved. No commercial use is permitted unless otherwise expressly granted.

To cite Lundberg IE, Bottai M, Tjärnlund A. Ann Rheum Dis 2018;77:e91.

Received 20 December 2017Accepted 23 December 2017Published Online First 9 January 2018

► http:// dx. doi. org/ 10. 1136/ annrheumdis- 2017- 212774

Ann Rheum Dis 2018;77:e91. doi:10.1136/annrheumdis-2017-212786

RefeRences 1 Hočevar A, Rotar Z, Krosel M, et al. Performance of the 2017 European League Against

Rheumatism/American College of Rheumatology classification criteria for adult and juvenile idiopathic inflammatory myopathies in clinical practice. Ann Rheum Dis 2018;77:e90.

2 Lundberg IE, Tjärnlund A, Bottai M, et al. 2017 European League Against Rheumatism/American College of Rheumatology classification criteria for adult and juvenile idiopathic inflammatory myopathies and their major subgroups. Ann Rheum Dis 2017;76:1955–64.

Correspondence response