Add-on tocilizumab versus conventional treatment for systemic sclerosis, and cytokine analysis to identify an endotype to tocilizumab therapy

Abstract

Objectives: To evaluate the anti-interleukin (IL)-6 receptor antibody tocilizumab (TCZ) as a treatment of systemic sclerosis (SSc), a randomised parallel group study was conducted, and compared their results and baseline cytokine/chemokine profiles.

Methods: Patients were assigned to a TCZ add-on group (TCZ group, n = 7) and a conventional therapy group (Conv group, n = 6). TCZ (8 mg/kg/month) for 6 months, and the modified Rodnan total skin score (mRSS) were used to compare the efficacy. The association of medical history, baseline pulmonary function tests, blood cell counts, serum C-reactive protein (CRP) and 26 cytokines/chemokines and decrease in mRSS were analysed.

Results: The mean change in mRSS was larger in the TCZ group (6.3) than in the Conv group (3.4), but the difference was not statistically significant because of high variance in the TCZ group. Patients with shorter disease histories and higher CRP had larger decreases in mRSS, and the decrease in mRSS was negatively correlated with IL-13 and C–C motif chemokine ligand (CCL)5.

Conclusion: Although significant between-group differences were not observed, some patients had a decrease in mRSS. Short disease duration, high CRP, low IL-13 and low CCL5 may represent an SSc endotype responsive to TCZ therapy.

Keywords:

• Systemic sclerosis
• tocilizumab
• IL-6
• IL-13
• CCL5

Introduction

Systemic sclerosis (SSc) is a connective tissue disease that presents as sclerosis of the skin and visceral organs. Although there is no specific treatment, interleukin (IL)-6 is known to contribute to SSc pathogenesis. IL-6 was found to be present in the serum of SSc patients but not healthy controls [1], and IL-6 levels were higher in supernatants from cultures of skin tissue isolated from SSc patients than in isolates from healthy donors [2]. Sato et al. found a correlation between IL-6 levels and mRSS [3], and Kawaguchi et al. showed that production of type-1 procollagen by skin fibroblasts isolated from SSc patients was suppressed by anti-IL-6 antibody [4]. We previously demonstrated an in vivo effect of anti-IL-6 receptor antibody in a bleomycin-induced mouse SSc model [5] and reported the successful treatment of two SSc patients with tocilizumab (TCZ), a humanised anti-IL-6 receptor antibody [6].

TCZ is a monoclonal antibody against the IL-6 receptor, and it diminishes the cytokine effects of IL-6 by preventing receptor binding. The clinical effects of TCZ have been demonstrated in Castleman’s disease and rheumatoid arthritis (RA). If IL-6 is involved in the pathogenesis of SSc, then TCZ should have a therapeutic effect in patients with the disease. We have reported decreases in skin scores in SSc patients treated with TCZ [6,7], but several randomised placebo control trials showed spontaneous remission in some SSc patients [8,9]. Therefore, to avoid the effect of spontaneous remission on the results, parallel group comparisons are needed to evaluate TCZ treatment. In a randomised placebo controlled (faSScinate) trial conducted in Europe and North America, the mean skin score of the TCZ group patients was lower than that in the placebo group, but the difference was not statistically significant [10]. We hypothesised that some patients respond well and some respond poorly to TCZ treatment. This randomised parallel group comparison study (UMIN0000055550) was designed to identify the characteristics of SSc patients suitable for TCZ therapy.

Methods

Patients and TCZ administration

The study protocol was approved by the institutional review boards of the Keio University Hospital, Tokyo Women’s Medical University Hospital and Osaka University Hospital, and was registered as (UMIN0000055550) at the University Hospital Medical Information Network Clinical Trials Registry (UMIN-CTR, https://upload.umin.ac.jp/cgi-open-bin/ctr_e/ctr_view.cgi?recptno=R000006582) as a randomised, open label, multicentre trial to assess humanised anti-IL-6 receptor antibody, TCZ, for the treatment of SSc. Patients with diffuse cutaneous SSc were recruited and all study participants gave their written informed consent. Patients were eligible for inclusion if they were 20–65 years of age, had clinical symptoms meeting the 1980 SSc criteria of the American Rheumatology Association, had a skin score >16 and no more than a 10% change within two months of the screening period. Patients participating another clinical trial currently or within the past 6 months, receiving intravenous cyclophosphamide within the past 6 months, using biological drugs including infliximab, etanercept, or adalimumab, or contraindicated or subject to careful administration of TCZ according to the prescribing information, were excluded. Interstitial pneumonitis is a symptom flagged for careful administration, but was omitted from the exclusion criteria because it is a common symptom in SSc patients [11].

This was an open label parallel group comparison study. Participants were randomly assigned to TCZ add-on treatment (TCZ group) or continuing conventional therapy (Conv group). TCZ was administered parenterally at the dosage used for RA (8 mg/kg/month), and the administration period was set for 6 months because the decrease of mRSS was observed within 6 months in the results of former study [6]. Conventional therapy, including steroids and methotrexate, continued without modification. The modified Rodnan total skin score (mRSS) was measured at the start point and 6 months later to compare treatment efficacy. To avoid the inter-examiner variability, mRSS was evaluated by the same examiner of each hospital. To determine the effects of TCZ on visceral organs, the percent vital capacity (%VC) and percent diffuse capacity of lung carbon monoxide (%DLco) were measured at the start and the end of the study. Following the study protocol, serum samples were collected for evaluation as described below. All patients in the TCZ group were observed for 6 months even if they terminate TCZ administration because the influence of a single infusion of TCZ in SSc patients was not known. The last observation carried forward method was used for evaluation of patients with incomplete TCZ data. TCZ was supplied by Chugai Pharmaceutical Co., Ltd. (Tokyo, Japan), with the permission of the Japan Pharmaceutical Manufacturers Association.

Stratification factors

Patient characteristics including disease duration, age, %VC, %DLco, C-reactive protein (CRP) and platelet count (PLT) in initial blood samples, cytokines, chemokines and soluble forms of adhesion molecules [IL-1β, IL-1 receptor antagonist (IL-1Ra), IL-4, IL-5, IL-6, soluble IL-6 receptor (sIL-6R), IL-7, IL-8, IL-9, IL-10, IL-12, IL-13, basic fibroblast growth factor (FGF-2), granulocyte colony-stimulating factor (G-CSF), interferon (IFN)-γ, tumour necrosis factor (TNF)-α, vascular endothelial growth factor (VEGF), C–C motif chemokine ligand (CCL)2, CCL3, CCL4, CCL5, CCL11, C-X-C motif chemokine ligand (CXCL)10, CXCL13 and soluble intercellular adhesion molecule (sICAM)1] were assayed to analyse the treatment response. Cytokines, chemokines, sIL-6R and sICAM-1 were assayed with the Bio-Plex system (Bio-Rad Laboratories, Hercules, CA).

Statistical analysis

The association of laboratory values with decrease in skin scores was determined by the Pearson correlation coefficients (r), |r| > 0.4 was assumed to be significant, and the test statistic (T) was evaluated by Student’s t-test.

Results

TCZ administration

The patient characteristics are shown in Table 1. The mean age, daily steroid dose, and baseline mRSS, PLT and CRP values of the two groups were not significantly different. Four of the seven patients in the TCZ group and three of the six patients in the Conv group presented with a %VC <80, but the mean baseline %VC and %DLco of the groups were not significantly different. No patients underwent right heart catheterisation because echocardiography did not find cases of suspected pulmonary hypertension. All patients in the TCZ group and five of six patients in the Conv group received steroid at the dose listed in Table 1, and two of the Conv group patients received methotrexate (median 9 mg/week) during the study period. Five of the TCZ group patients completed the scheduled TCZ administration, but two withdrew, one voluntarily and the other because of fever. These patients were monitored for the entire study period, but one of the two patients refused serum collection.

Table 1. Baseline profiles of the patients.

	Mean values of
	TCZ group	Conv group
n	7	6
Age	48.7	47.7
Sex (female:male)	4:3	6:0
Disease duration (years)	6.2	7.8
Steroid (prednisolone mg/day)	8.2	7.5
Baseline skin score	23.4	27.8
Baseline %VC	76.1	76.5
Baseline %DLco	50.4	52.9
Baseline PLT (×10^4/mm^3)	24.3	27.3
Baseline CRP (mg/L)	1.3	1.5

%VC: vital capacity as percent of predicted; %DLco: diffusing capacity of the lung carbon monoxide as percent of predicted; PLT: platelet count; CRP: C-reactive protein.

At 6 months, the mean decrease in mRSS (ΔmRSS) was larger in the TCZ group (6.3) than in the Conv group (3.4), but because the variance was large, the difference was not statistically significant. As shown in Figure 1, no patients experienced a worsening of mRSS. However, the ΔmRSS of TCZ patients varied. Two patients had ΔmRSSs of 55% and 48% within 6 months, one had no reduction and the others had reductions comparable to those in the Conv group patients. The spirometric test values did not change significantly in either the TCZ or the Conv group. The mean %VC changed from 74.44% to 73.51% in the TCZ group, and from 80.83% to 81.17% in the Conv group. The mean %DLco changed from 50.6% to 49.4% in the TCZ group, and from 53.6% to 50.1% in the Conv group. After starting TCZ, elevations were seen in total cholesterol in four patients, triglycerides in three patients, serum amylase and uric acid in one patient each, in alanine aminotransferase in two, and in white blood cell count in one. Most increases were transient or intermittent, but two cases of total cholesterol elevation and two of triglyceride elevation did not recover within the study period.

Figure 1. Change in skin scores (A) in the TCZ (red) and Conv (blue) group. At 6 months, the change in mRSS was larger in the TCZ group (6.3) than in the Conv group (3.4), but the variance was large and the difference was not statistically significant (B). Vertical bar indicates standard deviation.

Stratification factors

The correlation among disease duration, age and baseline CRP, PLT, cytokines and chemokines and ΔmRSS were calculated in the TCZ group and also in the Conv group, which served as a control of the effects of steroid or methotrexate (Figure 2). Disease duration (r = 0.4) and baseline CRP level (r = − 0.48) were correlated with ΔmRSS. Shorter disease duration and higher CRP level were associated with a larger ΔmRSS. As shown in Figure 3, some cytokines or chemokines were correlated with ΔmRSS. Baseline IL-13 and CCL5 were inversely correlated with ΔmRSS (both p < .1). Patients with the highest baseline IL-13 levels did not have a decrease in skin score; those with IL-13 <15 pg/mL had a ΔmRSS >10 within 6 months. IL-6 levels in the TCZ group were <20 pg/mL and did not correlate with ΔmRSS. However, a scatterplot including the data obtained from both the TCZ and Conv groups (Figure 4) shows that patients with higher IL-6 levels had lower IL-13 levels and those with lower IL-6 levels had higher IL-13 levels. The values of these two cytokines were inversely proportional (R² = 0.1505) following the relationship y = 80.682x^−0.678 as shown in the figure.

Figure 2. Correlations of skin score and disease duration, age at entry, initial CRP level, and platelet count. Patients with shorter disease duration and higher CRP levels had larger reductions in skin score. The skin score reduction is shown on the x-axis. All study patients received steroids; the Conv group thus served as a study control. Red circles indicate the TCZ group, and blue diamonds indicate the Conv group.

Figure 3. Correlations of skin score and 21 cytokines or chemokines, sICAM-1 and sIL-6R. IL-13 and CCL5 were negatively correlated with skin score reduction. Horizontal axis indicates skin score reduction within the study period. Red circles indicate the TCZ group, and diamonds rhombi indicate the Conv group. One sample data of the TCZ group were missing at CCL2, sICAM-1 and sIL-6R due to out of range or amount of sample volume.

Figure 4. Relationship between baseline IL-6 and IL-13. Patients with higher IL-13 presented with lower IL-6.

Discussion

The decreases in mean mRSS in the TCZ and Conv groups were not significantly different. However, the variance in the response of TCZ group patients, as indicated by the error bars in Figure 1(B), was >3-fold larger than that in the Conv group. Notably, two patients in the TCZ group presented almost 50% reduction of mRSS (55% and 48%). As the mean ratio of decrease of mRSS in the TCZ group was 31.7%, these two patients presented greater changes than the others. As shown in Figure 2, these two patients did not have special conditions such as age, baseline %VC, %DLco, or PLTs. They had a tendency for shorter disease duration. However, disease duration alone did not dictate ΔmRSS. Assuming that the variability in the TCZ group was caused due to the varying sensitivity to TCZ among SSc patients, patient baseline cytokines and chemokines were evaluated for possible association with response to TCZ and to identify an endotype suitable for TCZ treatment.

TCZ is approved for the treatment of RA, and several factors can predict reactivity to TCZ therapy. The ADACTA study [12] found that RA patients with high CXCL13 and low sICAM1 had an increased ACR50 response rate to TCZ [13]. As CXCL13 reflects lymphoid cell activity and sICAM1 reflects myeloid cell activity, the studies concluded that the TCZ response reflected the heterogeneity of synovial tissue in RA. In this study, neither baseline CXCL13 nor sICAM1 was correlated with ΔmRSS. The cells that infiltrate the skin of early-stage SSc patients are predominantly T lymphocytes, and infiltration proceeds the fibrotic changes of SSc [14], but according to our data it is possible that CXCL13 and sICAM1 do not play a major role in the cell infiltration that occurs in SSc.

Soluble IL-6R has also been reported to predict RA treatment response. It can transmit a signal following binding with IL-6, and synovial cell proliferation in RA is dependent on IL-6 and sIL-6R signalling [15]. As sIL-6R can bind to TCZ in the serum, patients with high levels of sIL-6R may have decreased reactivity to TCZ. Consistent with this, high baseline sIL-6R levels have been found to inhibit clinical remission of RA with TCZ treatment [16]. In this study, neither baseline sIL-6R nor IL-6 was correlated with ΔmRSS, but baseline CRP and ΔmRSS were correlated even though CRP is thought to be induced by serum IL-6 [17]. This seemingly contradictory result might reflect IL-6 levels in the TCZ group patients that were too low to confirm its physiological function. All patients in the TCZ group had IL-6 levels <20 pg/mL, but some patients in the Conv group presented with high IL-6 levels, perhaps because of a coincidental sample bias. Another possibility is that the serum IL-6 assays were not representative of the levels in the involved tissues or might depend upon differences of the involved organ. Scala et al. reported that serum IL-6 levels were correlated with the existence of pulmonary fibrosis [18], but baseline IL-6 levels did not correlate with %VC or %DLco in this study (data not shown). Serum IL-6 level might have been affected by disorders of various organs other than the lung or skin.

A negative correlation was between baseline IL-13 and ΔmRSS, and it is possible that IL-13 is associated with fibrosis. Hasegawa et al. reported increased serum IL-4, IL-10 and IL-13 concentrations in SSc patients [19] and Fuschiotti et al. found that increased serum IL-13 in SSc patients was correlated with skin fibrosis [20,21]. Both studies demonstrated the presence of IL-13-producing CD8+ cells in the skin tissue of early-stage SSc and serum IL-13 concentrations that might depend on the extent of disease progression. In a survey of 13 cytokines, Gourh et al. reported increased serum IL-6 and IL-13 levels in SSc patients with pulmonary hypertension [22], but that was not evaluated in this study because all participants showed normal echocardiogram results, and none of the patients underwent the right heart catheterisation.

CCL5 (RANTES) is a chemotactic factor for T cells and eosinophils [23]. Expression of both CCL5 mRNA and protein has been detected in involved skin and the upper gastrointestinal tract [24], and might be associated with the involvement of the upper gastrointestinal disease of SSc. Hasegawa et al. found that serum CCL5 was elevated in early SSc [25]. All patients in this study used proton-pump inhibitors to relieve the symptoms of reflux esophagitis, but no data on the degree of upper gastrointestinal involvement were available, and it is not clear whether oesophageal symptoms affect the response to TCZ treatment.

Scatterplots showed trends toward positive correlation IL-8 and FGF-2 and a negative correlation of CCL11 with ΔmRSS, but none reached significance. The correlations might become clear in a larger population sample. This study was conducted at three facilities over a period of 6 months, and the results warrant larger study with a longer period of evaluation to reveal a more clearly identify an SSc endotype suitable for TCZ therapy. A global double-blinded trial of TCZ for patients with SSc (NCT02453256, Clinicaltrials.gov) is ongoing. In this global study, there are over 200 participants and administration period is 96 weeks (48 weeks of double-blind treatment and 48 weeks of open-label treatment). Such a large scale study with longer period of evaluation might succeed in describing such a subset of patients and clarify the efficacy of TCZ in SSc.

In summary, this parallel group comparison study did not find significant differences in the responses to add-on TCZ and conventional SSc treatment. Some patients in the TCZ group had a decrease in mRSS during the administration period, and those with low serum IL-13 and CCL5 had larger reductions in the mRSS than other patients.

Conflict of interest

Y.S. has received speaker fees from Chugai Pharmaceutical Co. Ltd. and consulting fees from F. Hoffmann-La Roche, Ltd. The TCZ used in this study was supplied by Chugai Pharmaceutical Co. Ltd. All other authors have declared no conflicts of interest.

Acknowledgements

We are very grateful to Professor Toshio Tanaka and Professor Atushi Kumanogoh for their support and insightful comments. We thank Dr Hiroyuki Murota for his important comments on the clinical evaluation. We also thank Mrs. Hiromi Nishinaka for her outstanding assistance.

Additional information

Funding

This work was supported by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (#14591444).