The pattern of response to anti–interleukin-1 treatment distinguishes two subsets of patients with systemic-onset juvenile idiopathic arthritis

Authors


Abstract

Objective

To assess the clinical response to interleukin-1 (IL-1) blockade and in vitro IL-1β and IL-18 secretion in patients with systemic-onset juvenile idiopathic arthritis (JIA).

Methods

Twenty-two patients with systemic-onset JIA were treated with the IL-1 receptor antagonist (IL-1Ra) anakinra. Monocytes from 18 patients and 20 healthy donors were activated by different Toll-like receptor ligands. Intracellular and secreted IL-1β and IL-18 were analyzed by Western blotting and enzyme-linked immunosorbent assay.

Results

Ten patients with systemic-onset JIA exhibited a dramatic response to anakinra and were classified as complete responders. Eleven patients had an incomplete response or no response, and 1 patient could not be classified in terms of response. Compared with patients who had an incomplete response or no response, complete responders had a lower number of active joints (P = 0.02) and an increased absolute neutrophil count (P = 0.02). In vitro IL-1β and IL-18 secretion in response to various stimuli was not increased and was independent of treatment efficacy. Likewise, secretion of IL-1Ra by monocytes from patients with systemic-onset JIA was not impaired. An overall low level of IL-1β secretion upon exposure to exogenous ATP was observed, unrelated to treatment responsiveness or disease activity.

Conclusion

Two subsets of systemic-onset JIA can be identified according to patient response to IL-1 blockade. The 2 subsets appear to be characterized by some distinct clinical features. In vitro secretion of IL-1β and IL-18 by monocytes from patients with systemic-onset JIA is not increased and is independent of both treatment outcome and disease activity.

Treatment with anakinra, a recombinant interleukin-1 receptor antagonist (IL-1Ra), is the most effective therapy for hereditary autoinflammatory syndromes related to mutations in the inflammasome component cryopyrin/NALP3 (1–4), which result in partial loss of control of IL-1β processing and in secretion of large amounts of IL-1β (4–7). Anakinra has also shown promising results in the treatment of systemic-onset juvenile idiopathic arthritis (JIA) (8, 9) and adult-onset Still's disease (10), 2 severe inflammatory diseases characterized by arthritis along with systemic features such as high spiking fever, rash, hepatosplenomegaly, and serositis (11). The clinical benefit observed after treatment of systemic-onset JIA with IL-1Ra and the finding of enhanced in vitro production of IL-1β by patient monocytes has suggested that this disease could also be related to dysregulated production and secretion of IL-1β (9).

IL-1β is a powerful proinflammatory cytokine and plays a key role in innate and adaptive immunity (12). Secretion of IL-1β occurs through a non-classical pathway (13, 14) and requires 2 signals (13, 14). First, Toll-like receptor (TLR) signaling induces gene expression and synthesis of the inactive IL-1β precursor (proIL-1β). Then, a second signal drives processing and secretion of the cytokine. A highly effective second signal is exogenous ATP, which binds P2X7 receptors expressed on the surface of monocytes (15–17). Processing requires the assembly of the inflammasome, a multiprotein intracellular complex responsible for the activation of caspase 1. Active caspase 1 in turn converts proIL- 1β into the biologically active cytokine (7, 18, 19). Generation of mature IL-1β is associated with its secretion.

In this study, we evaluated the clinical response to IL-1 blockade in a cohort of 22 patients with systemic-onset JIA and studied the levels of IL-1β and IL-18 secreted in vitro by patient monocytes. We found that ∼40% of patients with systemic-onset JIA respond very satisfactorily to IL-1 blockade, with a pattern strongly reminiscent of that observed in autoinflammatory diseases related to cryopyrin mutation. In vitro IL-1β and IL-18 secretion in response to various stimuli was not increased and was independent of treatment efficacy. A lower number of involved joints and a higher absolute neutrophil count were associated with a greater likelihood of response to anakinra.

PATIENTS AND METHODS

Patients.

The study was approved by the Ethics Board of G. Gaslini Institute. Twenty-two patients with systemic-onset JIA (20) (11 girls and 11 boys) were selected for treatment with anakinra (Table 1). All patients had received long-term corticosteroid therapy, mostly in association with 1 or more second-line agents. Patients previously treated with etanercept or infliximab underwent a washout period of at least 2 weeks or 8 weeks, respectively.

Table 1. Baseline characteristics of the patients with systemic-onset juvenile idiopathic arthritis*
PatientAge, years/sexDisease duration, yearsNo. of active jointsCRP, mg/dlWBCs (neutrophils), × 103/mm3TreatmentPrednisone dosage, mg/kg/day
  • *

    CRP = C-reactive protein; WBCs = white blood cells; MTX = methotrexate; AZA = azathioprine; etan. = etanercept; NSAID = nonsteroidal antiinflammatory drug; CSA = cyclosporin A; inflix. = infliximab; thal. = thalidomide.

S17.8/M3.9615.325.8 (23.8)Steroid, MTX, AZA, etan.2
S28.5/M1.5223.423.4 (22.3)NSAID, steroid, MTX, CSA3
S314.3/M3.245.929.4 (27.9)NSAID, steroid, MTX, etan.0.5
S412.0/F2.0184.114.9 (9.7)Steroid, MTX, CSA0.5
S59.1/F2.4311.58.9 (6.1)NSAID, steroid, MTX, CSA0.8
S611.7/F2.2710.313.4 (9.3)NSAID, steroid, MTX, inflix.0.4
S712.5/M2.638.718.0 (16.3)Steroid, MTX0.2
S810.4/F0.6214.826.3 (23.7)Steroid, CSA0.5
S913.2/F10.91014.935.8 (30.9)NSAID, steroid1.1
S107.5/F0.515.2714.8 (9.5)Steroid, MTX0.4
S1116.8/M3.348.917.3 (14.6)NSAID, steroid0.1
S1212.6/F0.531.622.0 (19.7)Steroid, MTX1.5
S136.0/M3.0411.916.2 (8.1)NSAID, steroid, MTX, etan.0.3
S1412.4/M0.95513.911.9 (8.1)Steroid, MTX, inflix.0.15
S1517.0/M10.8914.718.0 (15.7)Steroid, CSA, AZA0.15
S1618.7/M4.8451.311.4 (8.8)NSAID, steroid0.2
S173.58/M2.1277.117.1 (8.9)NSAID, steroid, etan.0.5
S184.3/F0.7612.713.6 (9.1)Steroid, CSA0.78
S193.3/M1.62012.914.3 (10.1)Steroid, thal., MTX, etan.1
S200.9/F0.31616.520.3 (7.3)NSAID, steroid, thal.2.5
S2112.8/F10.5415.013.0 (8.2)NSAID, steroid, etan.0.5
S2211.7/F5.5102.818.2 (15.7)Steroid, etan.0.2

Anakinra treatment was started after informed consent was provided by parents and, when applicable, by patients. The starting dosage of anakinra was 1 mg/kg/day, subcutaneously (maximum 100 mg). Response to treatment was evaluated based on findings in a number of clinical and laboratory parameters (fever, rash, number of active joints, erythrocyte sedimentation rate [ESR], C-reactive protein [CRP] level, white blood cell count, hemoglobin level) during followup.

Six additional patients with active systemic-onset JIA who had not received steroid treatment were also analyzed. Their mean age was 8.1 years (range 3.5–13.3 years), and their mean disease duration was 3.2 years (range 0.4–10.8 years).

In vitro production and secretion of IL-1β, IL-18, and IL-1Ra.

Fresh monocytes from 9 of the patients with systemic-onset JIA (patients 4, 5, 6, 8, 10, 12, 17, 19, and 22) were analyzed for secretion of IL-1β, IL-1Ra, and IL-18 before and after anakinra treatment. Monocytes from the 6 additional, non–steroid-treated patients with active systemic-onset JIA were also analyzed. Monocytes from 20 age-matched healthy individuals attending our clinic for routine examinations prior to minor surgical procedures or from other young donors were used as controls, after informed consent was obtained.

Cell preparation and in vitro treatment.

Monocytes isolated from patients with systemic-onset JIA and healthy controls were enriched by adherence in RPMI 1640 medium containing 10% fetal bovine serum (Sigma-Aldrich, Milan, Italy). Purity was assessed by fluorescence-activated cell sorter analysis. The proportion of CD14+ cells was consistently ≥75%, and that of CD3+ cells was ≤5%. Peripheral blood mononuclear cells (PBMCs) or enriched monocytes were activated with different stimuli at 37°C in RPMI 1640 supplemented with 1% Nutridoma-HU (Roche, Milan, Italy), for 3 hours or 24 hours as previously described (14, 21, 22). The stimuli used were as follows: lipopolysaccharide (LPS) (1 μg/ml; Sigma-Aldrich), muramyldipeptide (3 μg/ml; Calbiochem, La Jolla, CA), Staphylococcus aureus (107 heat-killed S aureus/ml; InvivoGen, San Diego, CA), the yeast cell wall derivative zymosan (50 μg/ml; kindly provided by Dr. Guido Frumento, Genoa, Italy), phorbol myristate acetate (PMA) (50 ng/ml; Sigma-Aldrich), and ionomycin (1 μg/ml; Sigma-Aldrich).

In some experiments, after 3 hours of stimulation with LPS, supernatants were replaced with RPMI 1640–1% Nutridoma-HU to which 1 mM ATP (Sigma-Aldrich) had been added, with further incubation for 15 minutes. At the end of each experiment, supernatants were collected and cells lysed in 1% Triton X-100 lysis buffer.

Enzyme-linked immunosorbent assay (ELISA).

To determine the concentrations of IL-1β, IL-1Ra (R&D Systems, Minneapolis, MN), and IL-18 (MBL, Nagoya, Japan), ELISAs were performed on supernatants of monocyte cultures.

Western blot analysis.

Cell lysates and trichloroacetic acid–concentrated supernatants, prepared as previously described (22), were boiled in reducing Laemmli sample buffer, resolved by sodium dodecyl sulfate–12% polyacrylamide gel electrophoresis, and electrotransferred as described (14, 21, 22). Filters were probed with anti–IL-1β monoclonal antibody 3ZD (IgG1; obtained from the National Cancer Institute Biological Resources Branch, Frederick, MD) followed by horseradish peroxidase–conjugated goat anti-mouse IgG (Dako, Glostrup, Denmark) and developed with ECL-Plus (GE Healthcare, Milan, Italy).

Multicytokine determination.

Sera from 16 patients with systemic-onset JIA were collected before and 1 week after the beginning of treatment with anakinra and immediately stored at −80°C until analysis. Serum levels of 28 different soluble molecules (IL-1β, IL-1Ra, IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, IL-10, IL-12 [p70], IL-13, IL-15, IL-17, total IL-18, basic fibroblast growth factor, eotaxin, granulocyte colony-stimulating factor [G-CSF], granulocyte–macrophage colony-stimulating factor [GM-CSF], interferon-γ [IFNγ], interferon-inducible protein 10 [IP-10], monocyte chemotactic protein 1, macrophage inflammatory protein 1α [μIP-1α], MIP-1β, platelet-derived growth factor, RANTES, tumor necrosis factor α [TNFα], vascular endothelial growth factor [VEGF]) were analyzed by Bio-Plex cytokine multiplexable bead assay (Bio-Rad, Richmond, CA). Soluble IL-1 decoy receptor and pentraxin 3 (PTX3) were also measured by ELISA, as previously described (23, 24). Sera from 10 age-matched healthy subjects were used as controls, after informed consent was obtained.

Genetic analysis of CIAS1 mutations and P2X7 receptor polymorphisms.

DNA was extracted from peripheral blood by standard methods. All coding regions and intronic flanking sequences of CIAS1 were amplified by polymerase chain reaction using specific primers designed with Primer Express 2.0 software (Applied Biosystems, Foster City, CA), as previously described (4).

Four primer pairs designed with Primer Express 2.0 were used to selectively amplify exons 5, 9, 10, 11, and 13 of the P2X7 gene, as previously described (4). DNA analysis of each exon was also performed in 50 Italian control subjects.

Statistical analysis

Comparisons of clinical and laboratory parameters before and after treatment were performed using Wilcoxon's matched pairs test and McNemar's chi-square test for continuous and categorical variables, respectively. Differences in serum cytokine levels and in vitro cytokine secretion between patients with systemic-onset JIA and healthy controls or between disease subgroups (responders versus nonresponders) were analyzed by Mann-Whitney U test.

RESULTS

Heterogeneity of clinical response to anti–IL-1 treatment among patients with systemic-onset JIA.

At baseline, all 22 patients with systemic-onset JIA exhibited active arthritis and elevated levels of acute-phase reactants. High fever was present in 17, and rash in 12. Soon after the initiation of anakinra treatment, a general amelioration of the disease was observed, with normalization of the systemic features (fever and rash) and decrease in the levels of acute-phase reactants in the vast majority of the patients (details on clinical and laboratory findings at baseline and over time available from the authors upon request).

However, as early as the first week of treatment, it was clear that there were 2 distinct patterns of response to anakinra among the patients with systemic-onset JRA (Figure 1). One group of 10 patients (patients S1–S10) showed prompt control of systemic and articular manifestations, with normalization of acute-phase reactant levels and persistence of complete control of the disease after a mean followup of 1.36 years (range 0.3–2.59 years). These patients were designated complete responders (Figure 1A). Four months after the start of therapy, they were able to discontinue all medications except anakinra. Two of these patients have now been receiving anakinra for >2.5 years, and 5 others have received anakinra for >1 year.

Figure 1.

Response to anakinra in patients with systemic-onset juvenile idiopathic arthritis who were designated good responders (A) and those designated incomplete responders or nonresponders (B). At the time of statistical analysis, the duration of followup did not differ between the 2 groups. Number of active joints and C-reactive protein (CRP) level are shown as representative of clinical and laboratory parameters, respectively. The course of findings for other acute-phase reactants (white blood cell count and erythrocyte sedimentation rate) and hemoglobin levels was highly consistent with that for CRP. Numbers preceded by the letter S are individual patient numbers. P values (versus baseline) were determined by Wilcoxon's nonparametric matched pairs test. NS = not significant.

A second group of 11 patients (patients S11–S20 and S22 [incomplete responders or nonresponders]) exhibited a variable response to anakinra treatment, with general improvement soon after the initiation of the therapy, but with a tendency toward recurrence of the disease manifestations, especially arthritis and elevation of acute-phase reactant levels, during followup, despite increases in the daily dosage to 3 or 4 mg/kg (Figure 1B). Conversely, the systemic features in these patients, such as fever and rash, were generally well controlled over time. Seven patients in this group (patients S11–S17 [incomplete responders]) are still receiving anakinra treatment with a mean followup of 1.34 years (range 0.3–2.1). However, they are also being treated with second-line agents (5 of 7 patients) and/or oral steroids (6 of 7 patients). Due to the persistence of active disease even at higher doses of anakinra and poor compliance with the daily injection protocol, patients S18, S19, S20, and S22 discontinued the treatment after 6 months, 3 months, 6 months, and 1 month, respectively (nonresponders).

Apart from variable skin reactions at the injection site, no major adverse events were observed. However, on day 13 of treatment, 2 patients (patients S11 and S21) presented with laboratory features consistent with macrophage activation syndrome (MAS), i.e., marked elevation of liver enzyme, ferritin, and triglyceride levels, and decreased platelet count. Both patients discontinued anakinra and were treated with steroids and cyclosporin A, with rapid control of the MAS. Six months later, patient S11 was re-treated with anakinra for a relapse of his underlying disease, without any further sign of MAS. The parents of patient S21 did not allow reinstitution of the treatment with anakinra after complete control of MAS was achieved. This patient therefore was not classified as a responder or a nonresponder.

Lack of increase in IL-1β secretion by monocytes from both responders and nonresponders after stimulation with LPS and ATP.

To investigate whether the different responses to anakinra among patients with systemic-onset JIA were related to different levels of IL-1β secretion, spontaneous and LPS-induced production and secretion of IL-1β were studied in monocytes from 9 anakinra-treated patients with active systemic-onset JIA and 20 healthy controls. IL-1β is an inducible cytokine that is not expressed by resting monocytes (12). Findings in monocytes from healthy donors confirmed that in most cases (70%), no intracellular proIL-1β could be detected (Figure 2A). In contrast, monocytes from 6 of 9 patients with systemic-onset JIA (67%) exhibited intracellular proIL-1β in the absence of stimulation, indicating a certain degree of in vivo preactivation. Very little, if any, IL-1β was spontaneously secreted by monocytes from either controls or JIA patients during 3 hours of incubation in the absence of stimuli (Figures 2A and B).

Figure 2.

Interleukin-1β (IL-1β) production, processing, and secretion by monocytes from healthy controls and patients with systemic-onset juvenile idiopathic arthritis (SoJIA). A and B, Monocytes from healthy controls (A) or from representative patients with systemic-onset JIA (B) were left unstimulated or were cultured with lipopolysaccharide (LPS) for 3 hours or with LPS for 3 hours and ATP for 15 minutes, and the presence of IL-1β in cell lysates and supernatants was analyzed by Western blotting. Numbers preceded by the letter S are individual patient numbers. SF = steroid free. C, Supernatants of monocytes from 20 healthy controls, 9 patients with systemic-onset JIA before and after anakinra treatment, and 6 patients with active JIA who had not received steroids were treated with LPS or with LPS and ATP as described above, and IL-1β levels were measured. Data are presented as box blots, where the boxes represent the 25th to 75th percentiles, the lines within the boxes represent the median, and the lines outside the boxes represent the highest and lowest values. ∗ = P < 0.05 versus controls, by Mann-Whitney U test.

Treatment with LPS induced a strong increase in levels of intracellular proIL-1β in all controls (Figure 2A), whereas only variable induction was observed in patient monocytes (Figure 2B), with no response to LPS in some cases. IL-1β secretion detected by ELISA after 3 hours of LPS stimulation was variable but overall was low in all patients with systemic-onset JIA compared with healthy controls, in supernatants both from purified monocytes (P = 0.004 by Mann-Whitney U test) (Figure 2C) and from PBMCs (data not shown). No differences in IL-1β secretion were found between anakinra responders and incomplete responders or nonresponders. Secretion of IL-1β was strongly enhanced by extracellular ATP, which triggers the purinergic receptor P2X7 expressed on the surface of monocytes (Figure 2).

Interestingly, patients with active systemic-onset JIA exhibited defective responses to ATP (Figure 2B), with levels of ATP-induced IL-1β secretion significantly lower than those observed in healthy donors (P = 0.004) (Figure 2C). Again, no differences were found between responders and incomplete responders or nonresponders. ATP stimulation of monocytes from healthy donors results in caspase 1 activation and secretion (21, 25). In accordance with the low IL-1β secretion induced by ATP in monocytes from patients with systemic-onset JIA, caspase 1 activation or secretion was undetectable (results not shown). We have recently observed a severe impairment of the response to ATP in patients carrying CIAS1 mutations (4). However, none of our patients with systemic-onset JIA had mutations in the CIAS1 gene (results not shown).

Reduced secretion of IL-1β in response to ATP could be due to expression of nonfunctional variants of the highly polymorphic P2X7 receptor (26). The P2X7 gene was analyzed in the 9 patients with systemic-onset JIA represented in Figure 2C. Patient S8 expressed only the 2 variants T357S and E496A, a combination found in 3% of healthy subjects and considered the most common cause of near-absent P2X7 function in PBMCs (27) (Table 2). Patient S12 was heterozygous for E496A (26) and I568N (28), 2 mutations that, when associated, are indicative of nonfunctional or low-functional P2X7 (27). The other patients did not express variants known to result in loss of function (27).

Table 2. Functionally relevant P2X7 receptor polymorphisms in 9 patients with systemic-onset juvenile idiopathic arthritis
PatientP2X7 receptor genotype
Nucleotide 489 C>T (His155Tyr)*Nucleotide 946 G>A (Arg307Gln)Nucleotide 1096 C>G (Thr357Ser)Nucleotide 1513 A>C (Glu496Ala)Nucleotide 1729 T>A (Ile568Asn)
  • *

    The effect of the His155Tyr substitution is controversial: Shemon et al reported this mutation as nonfunctional (27), whereas Cabrini et al proposed that it increases P2X7 receptor activity (41).

  • Arg307Gln and Ile568Asn substitutions have been reported to confer complete loss of function also in heterozygous dosage (27).

  • The most common substitutions, at amino acid residues 357 and 496 of P2X7, have been demonstrated to induce complete loss of receptor function when present in homozygous dosage or in combination with another loss-of-function polymorphism (27).

S4T/TG/GC/CA/AT/T
S5T/TG/GC/GA/AT/T
S6C/CG/GC/CA/CT/T
S8C/TG/GC/GA/CT/T
S10C/TG/GC/CA/AT/T
S12C/TG/GC/CA/CT/A
S17C/CG/GC/CA/AT/T
S19C/TG/GC/CA/AT/T
S22C/CG/GC/CA/AT/T

Effects of ongoing steroid treatment and disease activity on IL-1β secretion after LPS and ATP stimulation.

All 22 patients with systemic-onset JIA in the responder analysis were treated with steroids, which are known to down-regulate IL-1β (29, 30), raising the possibility of interference of the therapy with IL-1β production and secretion. Monocytes from these patients were therefore compared with monocytes from 6 additional patients with active systemic-onset JIA who had not received steroids. Although there was variation in the amount of proIL-1β constitutively produced by resting monocytes from non–steroid-treated patients, LPS induced an increase in proIL-1β synthesis (Figure 2B) and levels of IL-1β secretion (Figure 2C) to levels comparable with those observed in healthy individuals. In contrast, levels of ATP-induced IL-1β secretion were significantly lower than in healthy controls (Figure 2C), indicating that, although steroid treatment may be involved in the low IL-1β production and secretion detected in patients with active systemic-onset JIA, it does not explain the defective response to ATP stimulation.

Monocytes from the anakinra-treated patients with systemic-onset JIA were also analyzed after 1 week of treatment. Despite a significant improvement in disease activity, spontaneous and induced IL-1β production and secretion did not change significantly (Figure 2C), with no evident differences between responders and nonresponders (data not shown).

IL-1β secretion induced by different inflammatory stimuli in monocytes from patients with systemic-onset JIA.

The kinetics of IL-1β secretion and the response to different pathogen-associated molecular patterns were investigated in 7 patients with active systemic-onset JIA and 12 healthy controls. As shown in Figure 3A, 24 hours after plating in the presence of the TLR-4 ligand LPS, the levels of IL-1β secreted by monocytes from patients with systemic-onset JIA before anakinra treatment were still lower than those detected in normal individuals, confirming the observations made at 3 hours. Similarly, the levels of IL-1β secretion induced on monocytes from the same JIA patients by muramyldipeptide (a ligand for NOD-2), S aureus, and zymosan (both more active on TLR-2) (31, 32) were consistently comparable with or lower than those in healthy controls, after either 3 hours or 24 hours of exposure. The effect of PMA plus ionomycin, which was previously shown to trigger IL-1β secretion by PBMCs from patients with systemic-onset JIA (9), was also investigated. In the present investigation, no increased secretion of IL-1β by either monocytes (Figure 3A) or PBMCs (data not shown) from the 7 JIA patients studied was observed after PMA/ionomycin treatment.

Figure 3.

Comparison of the secretion of IL-1β, IL-18, and IL-1 receptor antagonist (IL-1Ra) by monocytes from healthy controls (open bars) and patients with systemic-onset JIA (shaded bars). A, IL-1β detected in supernatants from monocytes cultured for 3 hours or 24 hours in the presence of zymosan (Zym), muramyldipeptide (MDP), Staphylococcus aureus (Staph), phorbol myristate acetate and ionomycin (PMA/Iono), or LPS. ∗ = P < 0.05; ∗∗ = P < 0.01 versus controls, by Mann-Whitney U test. B, IL-18 detected in supernatants from monocytes cultured for 24 hours in the absence (medium) or presence of 1 μg/ml LPS. C, IL-1Ra detected by enzyme-linked immunosorbent assay in supernatants from monocytes cultured for 24 hours in the absence or presence of 1 μg/ml LPS. No significant differences in levels of IL-18 or IL-1Ra between patients and controls were observed. Data are presented as box blots, where the boxes represent the 25th to 75th percentiles, the lines within the boxes represent the median, and the lines outside the boxes represent the highest and lowest values. See Figure 2 for other definitions.

Secretion of IL-18 and IL-1Ra by monocytes from patients with systemic-onset JIA in levels comparable with those in healthy controls.

Unlike IL-1β, IL-18 is constitutively produced by monocytes. However, like IL-1β, IL-18 is processed by caspase 1, and its secretion requires LPS priming (33). The secretion of IL-18 by monocytes from patients with systemic-onset JIA, without stimulation or activated with LPS, was investigated. IL-18 was barely detectable after 3 hours of exposure to LPS, in monocytes from both healthy controls and patients (data not shown). Detectable amounts of the secreted mature cytokine were found after longer exposure to LPS (24 hours) and were comparable with or even lower than those in healthy subjects (Figure 3B).

Activated human monocytes simultaneously synthesize both IL-1β and its specific inhibitor IL-1Ra (34). Therefore, we wished to investigate whether the good clinical response to anakinra among patients in the responder group could be due to defective production of IL-1Ra (rather than to increased release of IL-1β) by monocytes from these patients. However, the levels of IL-1Ra secreted by monocytes from the JIA patients before and after stimulation with LPS were comparable with those observed in healthy controls (Figure 3C).

Analysis of clinical and laboratory parameters to characterize the complete responder subset of systemic-onset JIA.

To try to identify parameters that could distinguish complete responders from incomplete responders/nonresponders to anakinra, several baseline clinical and biologic features were investigated. Among the clinical variables, complete responders had a significantly lower number of actively involved joints (median 3.5 [range 1–10]) compared with incomplete responders and nonresponders (median 7 [range 3–55]) (P = 0.02). Conversely, no significant differences in systemic features, i.e., presence of fever, rash, hepatosplenomegaly, and serositis, were observed.

Among laboratory parameters measured at baseline, no differences were observed in levels of acute-phase reactants (CRP, ESR, fibrinogen, ferritin) or hemoglobin between the 2 groups. Conversely, complete responders had a significantly higher number of circulating neutrophils (median 19.3 × 103/mm3 [range 6.1–30.9]) compared with incomplete responders and nonresponders (9.1 × 103/mm3 [range 7.3–19.7]) (P = 0.02). This difference was not related to ongoing steroid treatment since there were no significant differences in the daily dosage of steroids between the group of complete responders (median 0.5 mg/kg [range 0.2–3]) and the incomplete responders and nonresponders (median 0.5 mg/kg [range 0.1–2.5]) (Table 1).

As noted above, no significant differences in the amounts of secreted IL-1β and IL-18 were observed between responders (patients S4, S5, S6, S8, and S10) and incomplete responders/nonresponders (patients S12, S17, S19, and S22).

With the aim of identifying possible serum markers, 30 different soluble molecules in sera from 16 patients with systemic-onset JIA (7 responders and 9 incomplete responders or nonresponders), collected the day before the beginning of treatment, were analyzed and compared with those in 10 age-matched healthy controls (details available from the authors upon request). Levels of a number of soluble mediators were found to be increased in serum from patients with active systemic-onset JIA. Most of the soluble mediators displaying the highest degree of difference in comparison with healthy controls (P < 0.0001 or P < 0.001) were IL-1–related cytokines (IL-18, IL-6, GM-CSF, G-CSF, PTX3, and soluble IL-1Ra). High levels of IL-17 were also detected (P < 0.001). Levels of IL-10, IL-7, IP-10, IFNγ, VEGF, IL-4, TNFα, IL-12, and IL-13 were modestly, but significantly, elevated (P < 0.01 or P < 0.05). Serum IL-1β was detectable in only 4 patients with systemic-onset JIA.

The multi-analysis of the soluble molecules at baseline in responders and incomplete responders/nonresponders did not lead to identification of a specific serologic pattern distinguishing the 2 subgroups of patients. However, the group of patients with an incomplete response or nonresponse to anakinra exhibited increased levels of G-CSF (P = 0.04) and undetectable levels of IL-9, the latter of which were increased in patients with complete response to anti–IL-1 treatment (P = 0.02) (Figure 4).

Figure 4.

Levels of serum cytokines and other soluble molecules in patients with systemic-onset juvenile idiopathic arthritis who were designated good responders to anakinra (n = 7) (shaded bars) and those designated incomplete responders or nonresponders (n = 9) (open bars). Values were determined at baseline (pre) and after 1 week of treatment (post). Results are expressed as the mean and SD pg/ml (logarithmic scale). ∗ = P < 0.05 versus pretreatment level, by Wilcoxon's nonparametric matched pairs test. P values of 0.04 and 0.02 shown on the figure (determined by Mann-Whitney U test) refer to the difference between the group of responders and the group of incomplete responders or nonresponders. IL-1Ra = interleukin-1 receptor antagonist; GM-CSF = granulocyte–macrophage colony-stimulating factor; G-CSF = granulocyte colony-stimulating factor; IP-10 = interferon-inducible protein 10; IFNγ = interferon-γ; VEGF = vascular endothelial growth factor; TNFα = tumor necrosis factor α.

Serum levels of the different soluble mediators were also analyzed after 1 week of treatment. Among the soluble molecules that were overexpressed before anakinra treatment, a significant down-modulation after 1 week of treatment was observed for IL-6 only (mean ± SD serum concentration after treatment 79.9 ± 66.8 pg/ml; P = 0.009). When responders and nonresponders were analyzed separately, significant decreases in the levels of IL-6 (P = 0.02) and GM-CSF (P < 0.05) were observed in the responder group only, with no significant decreases in levels of the other soluble molecules. Conversely, increased levels of soluble IL-1Ra (likely due to the ongoing treatment) were observed in both responders and nonresponders (P = 0.02) (Figure 4).

DISCUSSION

Results of the present study show that systemic-onset JIA can be divided into 2 subsets according to the type of response to anakinra. One subset, accounting for ∼40% of patients, had a dramatic and persistent response to IL-1 blockade, which allowed the rapid discontinuation of all other treatments. In the other group of patients, the treatment, although showing an effect on systemic manifestations, did not control arthritis and inflammation and was either withdrawn or continued in combination with second-line agents or steroids. These findings support the hypothesis that systemic-onset JIA is a heterogeneous condition, as also indicated by marked differences in outcome (11). Indeed, in approximately half of patients the disease is characterized by a monocyclic or intermittent course, in which arthritis tends to remit when systemic features are controlled. In the other half, the disease follows an unremitting course, and chronic arthritis often leads to joint damage.

The proportion of patients who were responders to anakinra in the present study was lower than that observed in a previous investigation (9). This discrepancy may be related to differences in patient selection. In fact, in the present study we found that complete responders had a higher absolute neutrophil count and a lower number of active joints at baseline compared with patients who were incomplete responders or nonresponders. In the study by Pascual et al (9), however, 8 of 9 patients with systemic-onset JIA who showed good response to anakinra also had mild articular involvement, with a mean number of active joints and total white blood cell count that were very similar to those observed in our group of responder patients.

Only minor differences between the 2 response groups were observed when serum levels of 30 different cytokines and other soluble molecules were analyzed in this study. Consistently, lower levels of G-CSF were found in responders, in association with higher neutrophil counts. This apparent discrepancy may relate to a negative feedback mechanism. Indeed, G-CSF receptor expressed by neutrophils has been found to mediate uptake and clearance of G-CSF (35), indicating that plasma levels of the growth factor are related to the number of neutrophils.

Anakinra is highly effective in the treatment of the cryopyrinopathies (1–4, 36), which are characterized by dramatically increased secretion of IL-1β (4–7), raising the possibility that patients with systemic-onset JIA who are complete responders to anakinra have similar alterations in IL-1β secretion despite the absence of mutations in cryopyrin. This hypothesis was, however, ruled out by our data showing that in vitro IL-1β secretion by monocytes from patients with systemic-onset JIA was comparable with or even lower than that by monocytes from healthy controls. This low level of secretion was independent of the clinical response to anakinra as well as of the pathogen-associated molecular patterns used to induce IL-1β secretion.

The low levels of in vitro IL-1β release detected in patients with systemic-onset JIA could be due to ongoing steroid treatment (29, 30). Monocytes from a number of patients with systemic-onset JIA who had not received steroids secreted IL-1β in levels comparable with those in controls. Interestingly, however, IL-1β secretion in response to exogenous ATP was defective in patients with systemic-onset JIA regardless of whether they had taken steroids, suggesting a potential alteration of ATP signaling. Several nonfunctional variants of the ATP membrane receptor P2X7 have been described (27) and might account for the low degree of ATP-induced IL-1β secretion observed in patients with systemic-onset JIA. However, variants of the gene that are possibly associated with abnormal function (27) were expressed by only 2 of 9 patients. This suggests that, even if mutations of P2X7 may have a role in the decreased responsiveness to ATP in some patients with systemic-onset JIA, in most patients P2X7 polymorphism is not responsible for the low IL-1β secretion observed in vitro following exposure to exogenous ATP. Thus, the mechanism underlying this resistance to ATP is as yet unelucidated and warrants further investigation.

Our data differ from those reported by Pascual et al, who found that blood cells from patients with systemic-onset JIA secreted more IL-1β than those from healthy individuals (9). The high variability of the disease, together with the different methods of cell activation, may account in part for this discrepancy. In our group of patients, PMA/ionomycin-induced secretion (by either PBMCs or enriched monocytes) was comparable with or lower than that observed in healthy controls. Similarly, IL-1β secretion was consistently not increased in systemic-onset JIA patient monocytes compared with control monocytes after stimulation with different pathogen-associated molecular patterns that trigger different pathogen-sensing receptors (31, 32). This suggests that monocytes from patients with systemic-onset JIA are resistant to induction of IL-1β secretion in vitro, independent of the stimulus used and of the support of other cell types.

The finding that patient monocytes did not show increased IL-1β secretion in vitro was unexpected in a disease with many IL-1–related symptoms. A possible explanation may be that in systemic-onset JIA, unlike in diseases in which there is a mutated cryopyrin, the increase in IL-1β secretion occurs at privileged sites, possibly controlled by the local microenvironment. Similarly, IL-18, one of the inflammation-related cytokines that is significantly increased in sera from patients with systemic-onset JIA (37, 38), was secreted in vitro by monocytes from patients in both systemic-onset JIA subgroups (responders and nonresponders) at levels falling into the range observed in healthy controls. Thus, there is also an apparent discrepancy between in vivo and in vitro production of IL-18. Interestingly, IL-18 overexpression has been observed in the bone marrow of a patient with systemic-onset JIA who had MAS, suggesting that bone marrow may be the source not only of increased serum IL-18 but also of other proinflammatory cytokines (39). In any case, our data showing efficacy of anakinra treatment in spite of low levels of IL-1β reinforce the concept that in cytokine-mediated diseases, causation may be established by observing the results of specific receptor blockade or specific cytokine neutralization, rather than by elevated levels of the cytokine (40).

In conclusion, we have shown that a subset of patients with systemic-onset JIA has a dramatic response to anakinra. This finding suggests a major role of IL-1 in this disease subset and provides for the first time a parameter with which to investigate the heterogeneity of systemic-onset JIA. The 2 response subsets are characterized by some distinct clinical features. In vitro secretion of IL-1β and IL-18 by monocytes from patients with systemic-onset JIA is not increased and is independent of both treatment outcome and disease activity.

AUTHOR CONTRIBUTIONS

Dr. Gattorno had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Gattorno, Martini, Rubartelli.

Acquisition of data. Piccini, Lasigliè, Tassi, Brisca, Carta, Delfino, Ferlito, Pelagatti, Caroli, Buoncompagni, Viola, Loy, Sironi, Vecchi, Ravelli,

Analysis and interpretation of data. Gattorno, Piccini, Buoncompagni, Viola, Loy, Vecchi, Ravelli,

Manuscript preparation. Gattorno, Martini, Rubartelli.

Statistical analysis. Gattorno.

Ancillary