Elevated serum levels of free interleukin-18 in adult-onset Still’s disease

Charlotte Girard^1,*, Ju¨rgen Rech^2,*, Michael Brown³, Danie`le Allali⁴, Pascale Roux-Lombard^4,5, Franc¸ois Spertini⁶, Eduardo J. Schiffrin⁷, Georg Schett², Bernhard Manger², Sylvette Bas^1,5, Greg del Val⁷ and Cem Gabay¹

Abstract

Objective. IL-18 is a pro-inflammatory cytokine of the IL-1 family that is naturally inhibited by IL-18 binding protein (IL-18BP). High levels of IL-18 have been described in the serum of adult-onset Still’s disease (AOSD) patients, but only total IL-18 levels (including inactive IL-18 bound to IL-18BP) have been measured. With a specific immunoassay, we aimed to measure free IL-18 serum levels in AOSD patients and other rheumatic diseases.

Methods. An ELISA was developed to measure free IL-18. Its sensitivity and specificity were tested by spiking recombinant IL-18 or IL-18BP in serum and PBS supplemented with 5% BSA. The binding affinity of IL-18 to IL-18BP was calculated by titration experiments using the ELISA and by Biacore analysis. Sera of 37 AOSD patients and 138 controls (40 healthy controls, 30 RA, 29 SLE, 21 AS and 18 PsA) were assayed for free IL-18, IL-18BP, total IL-18 and other cytokines. Correlations were performed between free IL-18 and markers of disease activity in AOSD patients.

Results. Free IL-18 serum levels were significantly higher in AOSD patients (median 8.89 pg/ml) than in healthy and disease controls (1.37 pg/ml; P<0.01). Free IL-18 serum levels correlated with AOSD activity.

The affinity of IL-18 to IL-18BP was found to be much higher than previously described, with a dissoci-ation constant ranging from 30 to 50 pM.

Conclusion. Free IL-18 levels are specifically elevated in AOSD compared with other inflammatory dis-eases, suggesting that IL-18 represents a potential target for the treatment of AOSD.

Key words: adult-onset Still’s disease, free IL-18, IL-18BP

Rheumatology key messages

. The affinity of IL-18 binding protein for IL-18 is stronger than previously reported.

. Serum free IL-18 levels are higher in adult-onset Still’s disease patients than in healthy and disease controls.

. Free IL-18 serum levels correlate with clinical and biological markers of adult-onset Still’s disease activity.

Introduction

Adult-onset Still’s disease (AOSD) is a rare inflammatory sys-temic disease, the aetiology and pathogenesis of which remain unknown. An AOSD diagnosis is challenging and based on validated criteria and exclusion of differential diag-noses including inflammatory, infectious and neoplastic dis-eases [1–4]. IL-18 has been considered among the possible inflammatory agents involved in AOSD pathogenesis [5]. IL-18 is a pro-inflammatory cytokine of the IL-1 family that is produced by various cell types including monocytes/macro-phages [6, 7]. The biological activity of IL-18 is tightly

1Division of Rheumatology, Department of Internal Medicine Specialties, University Hospitals of Geneva, Geneva, Switzerland,

2Department of Internal Medicine 3, University Erlangen-Nuremberg, Erlangen, Germany,³Lyndin Group, Inc., Baltimore, MD, USA,

4Division of Immunology and Allergy, University Hospitals of Geneva,

5Department of Genetics and Laboratory Medicine, University of Geneva, Geneva,⁶Division of Immunology and Allergy, Centre Hospitalier Universitaire Vaudois (CHUV) and⁷AB2 Bio, Innovation Park, Ecole Polytechnique Fe´de´rale de Lausanne, Lausanne, Switzerland

*Charlotte Girard and Ju¨rgen Rech contributed equally to this study.

Correspondence to: Cem Gabay, Division of Rheumatology, University Hospitals of Geneva, 26 Avenue de Beau Se´jour, CH-1211 Geneva 14, Switzerland. E-mail: cem.gabay@hcuge.ch

Submitted 20 November 2015; revised version accepted 11 July 2016

!The Author 2016. Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For Permissions, please email: journals.permissions@oup.com

RHEUMATOLOGY

Rheumatology 2016;55:2237–2247 doi:10.1093/rheumatology/kew300 Advance Access publication 10 September 2016

BASIC SCIENCE

Downloaded from https://academic.oup.com/rheumatology/article-abstract/55/12/2237/2631579 by University de Geneve user

on 27 July 2018

controlled by IL-18 binding protein (IL-18BP), a naturally occurring inhibitor that binds IL-18 with high affinity and, by preventing its receptor interaction, blocks IL-18-induced nuclear factor-kappa B (NF-kB) activation and IFN-g produc-tion [8]. IL-18BP is constitutively present in the circulaproduc-tion, thus preventing excessive IL-18-mediated inflammatory re-sponses. The humanIL-18BPgene encodes four different isoforms (IL-18BPa, b, c, d) that are produced by alternative mRNA splicing and differ primarily in their C-terminal regions.

IL-18BPa is the predominant isoform and exhibits the great-est affinity for IL-18 [9].

High levels of IL-18 have been found in the serum of patients with many inflammatory and/or rheumatic dis-eases such as RA, SLE and psoriasis [10–12]. Even higher serum levels of IL-18 have been described in AOSD and its childhood counterpart, systemic onset JIA (SoJIA), and may be associated with a risk of developing macrophage activation syndrome (MAS), a severe compli-cation of these two diseases [13, 14]. In AOSD, circulating levels of IL-18 correlated with clinical status and laboratory markers of disease activity [15–25]. However, all previous studies were performed with immunoassays that do not distinguish 18 bound to 18BP from free bioactive IL-18, thus limiting interpretation of the results regarding the role of IL-18 in AOSD pathogenesis.

The aim of this study was to measure serum levels of free IL-18 in AOSD compared with other rheumatic dis-eases and healthy controls. With this purpose, an im-munoassay was specifically designed to selectively detect free IL-18 levels.

Patients and methods

Patients and samples

Sera from 37 AOSD patients and 138 controls [30 RA, 29 SLE, 21 AS, 18 PsA and 40 healthy subjects (HS)] were included in the study. All AOSD patients fulfilled Yamaguchi’s criteria [2] and were diagnosed as adults.

They were recruited in three University Hospitals [Erlangen (Germany), Geneva and Lausanne (Switzerland)]. All RA, SLE, AS and PsA patients were followed at the University Hospital of Geneva. They satisfied the 1987 ACR criteria for RA [26], the 1982 ACR criteria for SLE [27], the 2009 Assessment of SpondylArthritis international Society criteria for AS [28] and the classification criteria for PsA [29].

Several serum samples were obtained from AOSD pa-tients at different time points during follow-up. A single serum sample was available for controls. All serum sam-ples were stored at 80C until use for cytokine determination.

Demographic, clinical, biological and therapeutic data of AOSD and controls were retrospectively recorded from the patient’s files. Disease activity was assessed using validated scores for RA [28-joint DAS with CRP (DAS28-CRP) [30]], SLE (SLEDAI [31]), AS (BASDAI [32]) and PsA (DAS28-CRP) whenever possible. AOSD patients were considered clinically active if they had fever and/or inflam-matory arthralgia/arthritis and/or any suggestive cutane-ous lesions and/or sore throat. The disease was otherwise

considered inactive. CRP was not part of the criteria for disease activity. Table 1 shows the main characteristics of patients and HS at the time of the first blood sampling.

The study was approved by the Geneva Ethics Committee on Research Involving Humans (protocol number 15-012). According to Swiss law, due to the retro-spective design of our study, informed patient consent was not needed from AOSD patients and controls.

Determination of free IL-18 levels and IL-18/IL-18BP binding affinity

Serum levels of free IL-18 were determined by a sandwich ELISA method developed by AB2 Bio (Lausanne, Switzerland). A detailed description is provided in the sup-plementary Methods section ‘Free IL-18 ELISA’, available atRheumatologyOnline. The IL-18/IL-18BP dissociation constant was measured either by IL-18 titration experi-ments or surface plasmon resonance technology (Biacore, Uppsala, Sweden), as described in the supple-mentary Methods section ‘Measurement of IL-18/IL-18BP dissociation constant’, available atRheumatologyOnline.

Determination of serum levels of other cytokines and soluble receptors

Serum concentrations of IL-18BPa were assayed by a sandwich ELISA (DuoSet kit, R&D Systems, Minneapolis, MN, USA) according to the manufacturer’s protocol.

Serum levels of total IL-18, IL-6, IL-8, IL-12p70, IL-17A, IFN-g, TNF-a, IL-1 receptor antagonist (IL-1Ra) and sol-uble IL-2 receptor (sIL-2R) were measured using a unique blend of magnetic bead–based multiplex immunoassay (Bio-Plex Pro Human cytokines; Bio-Rad Laboratories, Hercules, CA, USA) as described in the supplementary Methods section ‘Multiplex assay for the determination of serum levels of other cytokines and soluble receptors’, available atRheumatologyOnline.

Statistical methods

Statistical analyses were performed using the software GraphPad Prism 6 (GraphPad Software, La Jolla, CA, USA) and R v.3.2.2 (R Foundation, Vienna, Austria).

Continuous variables were expressed as median [interquar-tile range (IQR)] and categorical variables were expressed as number and percentage. Non-parametric Kruskal–Wallis test was used to analyse differences between all patient groups and HS, followed by Mann–Whitney test to com-pare two groups. Correlations were performed by the Spearman rank correlation analysis. Receiver operating characteristic (ROC) curves were used to evaluate and compare the diagnostic performance of two variables. P-values<0.05 were considered significant.

Results

Development and validation of ELISA to determine free serum IL-18 levels in patient blood

We used a specific ELISA to measure free IL-18 and tested its sensitivity with either complete or IL-18BP-depleted serum from a healthy individual or in

2238 www.rheumatology.oxfordjournals.org

Charlotte Girardet al.

Downloaded from https://academic.oup.com/rheumatology/article-abstract/55/12/2237/2631579 by University de Geneve user

on 27 July 2018

5% BSA PBS buffer supplemented or not with 33 ng/ml recombinant IL-18BPa, a concentration equivalent to that found in the tested serum. Human IL-18 was spiked in the tested fluids at different concentrations ranging from 2.7 to 2000 pg/ml. Free IL-18 was not or barely detectable in complete serum, remaining at the assay background level at all spiked 18 concentrations (Fig. 1A). In contrast, IL-18BP-depleted serum and 5% BSA PBS buffer control had a linear recovery of>90% for all spiked IL-18 con-centrations58.2 pg/ml.

In order to verify that free IL-18 can be detected in com-plete serum, a serum sample from a second healthy donor, containing 38 ng/ml endogenous IL-18BPa (corres-ponding to 2.16 nM using the calculated molecular weight of 17.6 kDa), was tested by spiking high levels of IL-18 ranging from 2 to 100 ng/ml. In parallel, the same IL-18 concentrations were spiked in 5% BSA PBS supple-mented with 38 ng/ml recombinant IL-18BPa. Free IL-18 was detected by ELISA at5–10 ng/ml spiked IL-18, cor-responding to a molarity of 0.28–0.56 nM, assuming that the molecular weight of mature IL-18 is 18 kDa (Fig. 1B).

IL-18/IL-18BPa binding affinity

The IL-18 binding affinity to IL-18BPa was calculated using the results obtained with IL-18 titration in serum

from healthy individuals and PBS supplemented with equivalent levels of IL-18BPa, as well as by IL-18BPa titration in 5% BSA PBS supplemented by a constant concentration of IL-18 (Fig. 1C). The resulting calcu-lated dissociation constant (KD) of 50 pM shows a stronger endogenous IL-18BPa affinity to IL-18 than that previously reported (400 pM) [9]. The data col-lected in 5% BSA PBS supplemented by recombinant IL-18BPa gave even lower calculated values of 35 pM. To verify our titration results, we performed Biacore measurements using microchips coated with biotinylated recombinant IL-18BPa (Fig. 1D). The ob-tained Biacore KDwas 25.9 pM (S.D. 4.8) with an asso-ciation constant of 5.3 10⁵/Ms (S.D. 1.2) and a dissociation constant of 13.3 10⁶/s (S.D. 2.7), con-firming the strong endogenous IL-18BPa binding affi-nity to IL-18 that was obtained by serum titration.

Published serum levels of IL-18BP in HS ranged from 0.5 to 7 ng/ml [33]. When titrating IL-18BPa in BSA PBS buffer at physiological concentrations ranging between 312 pg/ml and 12 ng/mL, free IL-18 became undetectable with the addition of IL-18BPa at about 20- to 25-fold larger molarity, suggesting that a significant excess of 18BPa is necessary to inhibit the biological activity of IL-18 (Fig. 1C).

TABLE1 Baseline characteristics of patients and healthy controls

Characteristic AOSD^a RA SLE

SpA

3.96 (3.23–4.44)^b 3.48 (2.53–4.19)^c

SLEDAI,

Synthetic DMARDs include HCQ, SSZ, LEF, AZA, ciclosporin, MMF and CYC.^aActivity measures for AOSD are defined in the Patients and Methods section.^bDAS28-CRP available for 26 patients at the time of blood collection.^cDAS28-CRP available for 18 patients at the time of blood collection.^dSLEDAI available for 28 patients at the time of blood collection.^eBASDAI available for 8 patients at the time of blood collection.

www.rheumatology.oxfordjournals.org 2239

Free IL-18 in AOSD

Downloaded from https://academic.oup.com/rheumatology/article-abstract/55/12/2237/2631579 by University de Geneve user

on 27 July 2018

Clinical and laboratory characteristics of AOSD patients

Thirty-seven AOSD patients were included in the study. At the time of diagnosis, all but 1 (97%) had fever, 34 (92%) had arthralgia and/or arthritis, 29 (78%) had typical evan-escent rash, 20 (54%) had a sore throat and 27 (73%) had lymphadenopathies and/or splenomegaly and/or hepato-megaly. All patients had elevated CRP and/or ESR levels.

Twenty-seven of 37 (73%) had leukocytosis and 32 (86%) had elevated transaminases. Serum ferritin levels were elevated in all but one patient (97%), with a median value of 2396mg/l (IQR 850–8000). None of the patients had MAS at the time of blood collection, but two of them had a past history of MAS.

The first blood collection occurred at 0–312 months [median 0 (IQR 0–29)] after the AOSD diagnosis.

Twenty-four of 37 (65%) patients had clinically active dis-ease at the time of the first blood collection and 10 (27%) were free of any anti-inflammatory and/or immunosup-pressive drug. Among the 196 AOSD blood samples col-lected after the first blood collection, 37 (19%) were taken during a disease flare, and a specific treatment was on-going in 173 (88%) cases. The total number of samples per patient varied from 1 to 27 (average of 6.4 per patient).

Twenty-seven of 37 (73%) patients had more than one sample collected during follow-up. Twelve of 37 (32%) patients had samples collected during active and inactive disease, 11 (30%) during inactive disease only and 14 FIG. 1 Determination of free IL-18 levels and measurement of IL-18/IL-18BPa binding affinity

(A) Increasing concentrations of recombinant IL-18 were added to a complete or IL-18BP-depleted human serum and to 5% BSA PBS buffer supplemented or not with IL-18BPa. (B) Free IL-18 levels were determined in a second complete human serum and in 5% BSA PBS buffer supplemented with the same amount of IL-18BPa by spiking higher concen-trations of IL-18. (C) Free IL-18 levels were also measured in 5% BSA PBS buffer supplemented with 400 pg/ml re-combinant IL-18 followed by the addition of increasing concentrations of IL-18BPa. (D) The results of surface plasmon resonance experiments. RU = resonance units.

2240 www.rheumatology.oxfordjournals.org

Charlotte Girardet al.

Downloaded from https://academic.oup.com/rheumatology/article-abstract/55/12/2237/2631579 by University de Geneve user

on 27 July 2018

(38%) during active disease only. Considering a CRP cut-off of 10 mg/l, CRP levels were high in 21% (36/174) of sera samples collected during inactive disease vs 88%

(49/56) during active disease.

Serum levels of free IL-18, total IL-18 and IL-18BPa in AOSD patients and controls

Serum levels of free IL-18 were determined in patients with AOSD and in controls (Fig. 2A). Levels were significantly higher in AOSD patients [median 8.89 pg/ml (IQR 1.37–39.8)] compared with HS [1.37 pg/ml (1.37–1.37), P<0.0001] as well as with patients with other inflammatory rheumatic diseases, including RA [1.37 pg/ml (1.37–2.11), P = 0.0041], SLE [1.37 pg/ml (1.37–1.37), P<0.0001], AS [1.37 pg/ml (1.37–1.37), P = 0.004] and PsA [1.37 pg/ml (1.37–1.59), P = 0.0094]. To confirm that elevated values

detected by ELISA in AOSD patients corresponded specif-ically to free IL-18, sera from two AOSD patients with ele-vated concentrations of free IL-18 were supplemented with increasing concentrations (from 7.81 to 125 ng/ml) of re-combinant IL-18BPa. The results showed a gradual dis-appearance of detected free IL-18, with complete inhibition at 125 ng/ml IL-18BPa (supplementary Table S1, available atRheumatologyOnline).

Serum levels of total IL-18 (Fig. 2B) were also signifi-cantly higher in AOSD patients [median 1.23 ng/ml (IQR 0.29–15.56)] than in all control groups [HS: 0.05 ng/ml (0.04–0.08), P<0.0001; RA: 0.09 ng/ml (0.06–0.15), P<0.0001; SLE: 0.06 ng/ml (0.04–0.15), P<0.0001; AS:

0.07 ng/ml (0.04–0.11), P<0.0001; PsA: 0.05 ng/ml (0.04–0.09), P<0.0001]. Serum concentrations of IL-18BPa (Fig. 2C) were significantly higher in AOSD patients FIG. 2 Serum levels of IL-18 and IL-18BPa in AOSD and controls

Free IL-18 (A), total IL-18 (B), IL-18BPa (C), total IL-18/IL-18BPa ratio (D) were measured in the serum of patients with AOSD from the first blood collection and compared to controls. Corresponding P-values appear above each group (ns=no significant difference; *P<0.05; **P<0.01; ***P<0.001). The correlation between free and total IL-18 (E) and free IL-18 and total IL-18/IL-18BPa ratio (F) was performed in AOSD patients. The coefficient of correlation (r) and P-value were obtained by Spearman rank analysis.

www.rheumatology.oxfordjournals.org 2241

Free IL-18 in AOSD

Downloaded from https://academic.oup.com/rheumatology/article-abstract/55/12/2237/2631579 by University de Geneve user

on 27 July 2018

[median 18.5 ng/ml (IQR 8.6–49.7)] than in HS [11 ng/ml (6.9–18.5), P = 0.0254] but not in comparison with other diseases [RA: 16.9 ng/ml (12.6–20.5); SLE: 26.2 ng/ml (21.9–35.7); AS: 19.4 ng/ml (15.9–24.4); PsA: 15.7 ng/ml (14.1–17.4)]. The calculated total IL-18:IL-18BPa ratio (Fig. 2D) was significantly higher in AOSD than in all con-trol groups (P<0.001). Total IL-18 (Fig. 2E) correlated well with free IL-18 serum levels (r = 0.9335, P<0.0001), but the total IL-18:IL18BPa ratio correlated only moderately with free IL-18 (r = 0.4134, P = 0.0122) (Fig. 2F), thus indi-cating this ratio is not a good estimate of free IL-18 levels in human samples.

Serum levels of IL-6, IL-8, IL-12p70, IL-17, TNF-a, IFN-g, IL-1Ra, sIL-2R and CRP in AOSD patients and controls

A set of pro-inflammatory cytokines (IL-6, IL-8, IL-12p70, IL-17A, TNF-a, IFN-g, IL-1Ra, sIL-2R) was measured in the same samples by multiplex assay. IL-12p70, IL-17, TNF-a and IFN-g were below the detection limit of the assay (2.3, 1.6, 3.8 and 2.7 pg/ml, respectively). Serum levels of IL-6, IL-1Ra, IL-8 and sIL-2R, as well as CRP, are shown in supplementary Fig. S1, available at Rheumatology Online. No statistical difference was observed regarding serum levels of IL-6, IL-1Ra, IL-8 and sIL-2R between AOSD and RA patients. Serum

levels of IL-6 and IL-1Ra were significantly higher in AOSD than SLE patients, but there was no difference in IL-8 and sIL-2R levels between these two groups.

Cytokine levels were globally higher in AOSD than in spondylarthropathies (AS and PsA). Serum levels of IL-8 were not elevated in AOSD patients compared with HS.

CRP levels were higher in AOSD patients than in controls except SLE.

Serum levels of free IL-18 were higher in AOSD patients with active disease and correlated with other biomarkers of disease activity

Serum levels of free IL-18 and other cytokines were mea-sured in 233 available blood samples collected from the 37 AOSD patients. Sixty-one samples were obtained during clinically active disease and 172 were collected during inactive disease. Free IL-18 serum levels were sig-nificantly higher in active than in inactive disease [median 29.9 pg/ml (IQR 9.07–63.1) vs 1.37 (1.37–1.37), P<0.0001] (Fig. 3A). The same results were obtained for total IL-18 (data not shown). In contrast, serum levels of IL-18BPa, IL-6, IL-8, IL-1Ra and sIL-2R were not different in either group (data not shown).

When stratifying patients according to their free IL-18 serum levels at the time of the first blood collection (<or

>20 pg/ml) (supplementary Fig. S2, available at

FIG. 3 Serum levels of free IL-18 and other biomarkers in patients with active and inactive AOSD

(A) Free IL-18 levels were measured by ELISA. (B) Serum CRP, (C) circulating leukocytes, (D) serum ferritin and (E) alanine aminotransferase (ALT) levels were obtained from patients’ files. Disease was considered inactive when patient had no fever, joint manifestations, rash or sore throat at the time of blood sampling and active if any of the above-mentioned symptoms were present. All sera available were tested (for some patients, two or more samples collected at different time points). N corresponds to the number of samples assayed in each group. Mann–Whitney test was used to compare the active and inactive groups. Results are expressed as P-values (ns = no significant difference; *P<0.05; **P<0.01; ***P<

0.001).

2242 www.rheumatology.oxfordjournals.org

Charlotte Girardet al.

Downloaded from https://academic.oup.com/rheumatology/article-abstract/55/12/2237/2631579 by University de Geneve user

on 27 July 2018

Rheumatology Online), we observed that patients with high baseline free IL-18 levels had higher subsequent free IL-18 levels than those with low baseline levels, re-gardless of their disease activity. Furthermore, levels of free IL-18 remained significantly higher in active than in inactive disease in both patients groups [free IL-18 level

Rheumatology Online), we observed that patients with high baseline free IL-18 levels had higher subsequent free IL-18 levels than those with low baseline levels, re-gardless of their disease activity. Furthermore, levels of free IL-18 remained significantly higher in active than in inactive disease in both patients groups [free IL-18 level