Systemic sclerosis RMD Open: first published as 10.1136/rmdopen-2022-002477 on 3 November 2022. Downloaded from http://rmdopen.bmj.com/ on October 9, 2023 at Sant Paul Hospital Consortia. ORIGINAL RESEARCH Effectiveness and safety of tocilizumab in patients with systemic sclerosis: a propensity score matched controlled observational study of the EUSTAR cohort Simon Kuster,1 Suzana Jordan,1 Muriel Elhai,1,2 Ulrike Held,3 Klaus Steigmiller,3 Cosimo Bruni,1,4 Fabio Cacciapaglia ‍ ‍,5 Serena Vettori,6 Elise Siegert,7,8 Simona Rednic,9 Veronica Codullo ‍ ‍,10 Paolo Airo,11 Yolanda Braun-­Moscovici,12 Nicolas Hunzelmann,13 Maria Joao Salvador,14 Valeria Riccieri,15 Ana-­Maria Gheorghiu,16 Juan José Alegre Sancho,17 Katarzyna Romanowska-­Prochnicka,18,19 Ivan Castellví ‍ ‍,20 Ina Kötter,7,21 Marie-­Elise Truchetet,22 FJ López-­Longo,23 Pavel I Novikov,23 Alessandro Giollo,24 Yuichiro Shirai,25 Laura Belloli,26 Elisabetta Zanatta,27 Eric Hachulla,28 Vanessa Smith ‍ ‍,29,30 Chris Denton,31 Ruxandra M Ionescu,32 Tim Schmeiser ‍ ‍,33 Joerg H W Distler,34 Armando Gabrielli,35 Protected by copyright. Anna-­Maria Hoffmann-­Vold ‍ ‍,36 Masataka Kuwana ‍ ‍,25 Yannick Allanore,2 Oliver Distler ‍ ‍,1 on behalf of the EUSTAR collaborators To cite: Kuster S, Jordan S, ABSTRACT Elhai M, et al. Effectiveness WHAT IS ALREADY KNOWN ABOUT THIS Objectives Tocilizumab showed trends for improving and safety of tocilizumab in skin fibrosis and prevented progression of lung SUBJECT? patients with systemic sclerosis: ⇒ Two placebo-­controlled randomised controlled clin- fibrosis in systemic sclerosis (SSc) in randomised a propensity score matched controlled clinical trials. We aimed to assess safety and ical trials (RCTs) with tocilizumab have been con- controlled observational study of the EUSTAR cohort. RMD Open effectiveness of tocilizumab in a real-­life setting using ducted in systemic sclerosis (SSc). Both RCTs were 2022;8:e002477. doi:10.1136/ the European Scleroderma Trial and Research (EUSTAR) recruiting a highly enriched population of patients rmdopen-2022-002477 database. with inflammatory, early, diffuse, skin-­progressive Methods Patients with SSc fulfilling the American College SSc. Main messages from these two RCTs were ► Additional supplemental of Rheumatology (ACR)/EULAR 2013 classification criteria, trend for improving skin fibrosis and prevention of material is published online only. with baseline and follow-­up visits at 12±3 months, worsening of lung fibrosis over placebo. To view, please visit the journal receiving tocilizumab or standard of care as the control online (http://​dx.d​ oi.​org/​10.​ group, were selected. Propensity score matching was WHAT DOES THIS STUDY ADD? 1136/r​ mdopen-​2022-​002477). applied. Primary endpoints were the modified Rodnan skin ⇒ No significant effectiveness of tocilizumab was score (mRSS) and FVC at 12±3 months compared between shown in this broader, multicentre, propensity the groups. Secondary endpoints were the percentage of score matched, controlled observational, hetero- Received 18 May 2022 progressive/regressive patients for skin and lung at 12±3 geneous, non-­enriched real-­life SSc population Accepted 14 September 2022 months. from the large European Scleroderma Trial and Results Ninety-­three patients with SSc treated with Research registry. tocilizumab and 3180 patients with SSc with standard of ⇒ The consistency of direction in all predefined prima- © Author(s) (or their care fulfilled the inclusion criteria. Comparison between ry and secondary endpoints generates hypothesis employer(s)) 2022. Re-­use groups did not show significant differences, but favoured for potential effectiveness in a broader SSc popula- permitted under CC BY-­NC. No tocilizumab across all predefined primary and secondary tion rather than in highly selective RCT. commercial re-­use. See rights endpoints: mRSS was lower in the tocilizumab group and permissions. Published HOW MIGHT THIS IMPACT ON CLINICAL by BMJ. (difference −1.0, 95% CI −3.7 to 1.8, p=0.48). Similarly, FVC % predicted was higher in the tocilizumab group PRACTICE OR FUTURE DEVELOPMENTS? For numbered affiliations see (difference 1.5 (−6.1 to 9.1), p=0.70). The percentage ⇒ Adds important information from real-­life to the ex- end of article. of progressive/regressive patients favoured tocilizumab isting RCTs with tocilizumab by generating hypoth- Correspondence to over controls. These results were robust regarding the esis that should be confirmed in a prospective RCT Professor Oliver Distler; sensitivity analyses. Safety analysis confirmed previously with broader SSc population. ​ liver.​Distler@u​ sz.​ch O reported adverse event profiles. Kuster S, et al. RMD Open 2022;8:e002477. doi:10.1136/rmdopen-2022-002477    1 RMD Open RMD Open: first published as 10.1136/rmdopen-2022-002477 on 3 November 2022. Downloaded from http://rmdopen.bmj.com/ on October 9, 2023 at Sant Paul Hospital Consortia. Conclusion Although this large, observational, controlled, real-­life MATERIAL AND METHODS EUSTAR study did not show significant effectiveness of tocilizumab Study design on skin and lung fibrosis, the consistency of direction of all predefined This study was designed as a multicentre, propen- endpoints generates hypothesis for potential effectiveness in a broader sity score matched no-­ treatment controlled observa- SSc population. tional study. Data for tocilizumab-­treated patients were requested from the EUSTAR network using a case report form (CRF) designed by the lead investigators.20–22 The treating physicians made the treatment decision INTRODUCTION according to their local practice and routine. Requested Systemic sclerosis (SSc) is a rare but potentially lethal data included demographics, clinical characteristics, autoimmune connective tissue disease characterised by treatment details and adverse events (see online supple- inflammation, fibrosis and microvasculopathy.1 It is a mental file for CRF). Queries were sent to centres for multiorgan disease involving the skin and various internal missing data and data clarification. The matched control organs.2 3 Mortality is high, especially if lungs, heart or group was formed from patients prospectively registered kidneys are involved.4 5 Even though new trials on various in the EUSTAR database, not treated with tocilizumab. disease-­modifying drugs have been conducted over the Database extraction was done on 30 October 2017. last years, therapy is still mainly based on treatment of The local ethic committees of the participating centres organ-­specific complications.6–9 approved the data collection. All patients signed informed Several preclinical and translational studies have shown consent forms when required by the local ethics commit- that interleukin 6 (IL-­6) might play an important role tees. This study was conducted in accordance with the in SSc, in particular when inflammation is driving the guidelines for good clinical practice and the principles of disease process.10 11 IL-­6 serum concentrations in patients the Declaration of Helsinki and registered at www.drks.de with diffuse cutaneous (dc) SSc are significantly higher under DRKS00015537, including a predefined detailed than in healthy individuals.12 Serum IL-­6 levels correlate statistical analysis plan (provided in online supplemental file). with disease severity and mortality, and are associated Protected by copyright. with higher C reactive protein (CRP) levels and platelet Inclusion criteria and selection of control patients counts.12 13 Inhibition of IL-­ 6 prevented the develop- Inclusion criteria for the treatment group were defi- ment of inflammation-­driven dermal fibrosis induced by nite SSc according to the American College of Rheu- bleomycin in mice, but did not show effects in the non-­ matology (ACR)/EULAR 2013 criteria, treatment with inflammatory TSK-­1 model.14 15 tocilizumab and age ≥17 years.23 24 For analysis of primary Tocilizumab (TCZ) is a humanised monoclonal and secondary outcomes, only patients with at least three antibody against the IL-­6 receptor.16 Two randomised applications of tocilizumab and follow-­up at 12±3 months placebo-­ controlled clinical trials (RCTs) with tocili- were included. Safety data were analysed for all patients. zumab have been conducted in SSc. In the phase II Additional/different inclusion criteria for the control faSScinate trial, a trend for improving skin fibrosis group were absence of tocilizumab therapy, disease dura- over placebo was found. Exploratory analysis revealed tion <35 years to meet the tocilizumab group and date a possible stabilisation of FVC.17 18 The phase III of observations after 1 January 2010 (start of the online focuSSced study confirmed the trend on skin fibrosis EUSTAR database). If multiple visits existed for one without reaching statistical significance. Stabilisation control patient, we used the most recent one, and this of lung fibrosis, this time with FVC as a key secondary approach was revisited in the sensitivity analysis. endpoint and additional HRCT quantification, was observed.19 Both RCTs were recruiting a highly enriched Primary and secondary outcomes population of patients with inflammatory, early, diffuse, The primary outcomes were the difference at 12±3 skin-­progressive SSc. months of follow-­up between the tocilizumab and the control group in the modified Rodnan skin score (mRSS) Thus, while these data are promising and have resulted for skin fibrosis and the FVC for pulmonary function in the approval of tocilizumab for SSc-­associated inter- compared. Both outcomes were addressed separately, but stitial lung disease (SSc-­ILD) by the FDA, little is known a single matched set of patient treated-­control pairs was about the effects of tocilizumab in a broader SSc popu- used for analysis. lation. Data from large ‘real-­life’ registries could better Secondary outcomes were the percentage of progres- determine the effects of tocilizumab on a more heteroge- sive patients for skin fibrosis (increase in mRSS of 5 neous, non-­enriched population. The aim of the present points and 25%), lung fibrosis (decrease in either FVC study was to estimate the treatment effect and safety ≥10% or FVC ≥5% and diffusing capacity of the lung for of tocilizumab in patients with SSc, as compared with carbon monoxide DLCO ≥15%), and the percentage patients not treated with tocilizumab in a large real-­life of regressive patients for skin and for lung (decrease observational cohort study using the European Sclero- in mRSS of 5 points and 25% or increase in either FVC derma Trials and Research (EUSTAR) database. ≥10% or FVC ≥5% and DLCO ≥15%).25–27 Presence of 2 Kuster S, et al. RMD Open 2022;8:e002477. doi:10.1136/rmdopen-2022-002477 Systemic sclerosis RMD Open: first published as 10.1136/rmdopen-2022-002477 on 3 November 2022. Downloaded from http://rmdopen.bmj.com/ on October 9, 2023 at Sant Paul Hospital Consortia. ILD was defined as evidence for ILD on HRCT or X-­ray methotrexate, azathioprine, mycophenolate mofetil and as judged be the local investigator. Safety measures were rituximab (biologic) within 6 months before baseline assessed as percentage of patients suffering from adverse were imputed via logistic regression and co-­therapy was events. passively imputed. Missing data in the primary outcomes (mRSS and FVC) in potential control patients led to Subgroup analysis listwise exclusion of these patients, whereas none of the Predefined subgroup analyses included mRSS ≥10 versus patients in the TCZ group were excluded due to missing mRSS <10 at baseline and dcSSc versus limited cutaneous parameters in the primary outcomes. Covariates with a SSc (lcSSc) (for the outcome mRSS) and FVC ≥80% percentage of missing values of more than 50% were a versus FVC <80% at baseline and FVC <80% and X-­ray priori defined to be excluded from the analysis. For the or high-­resolution CT (HRCT) positive versus FVC ≥80% remaining variables, missingness patterns were assessed. and X-­ray or HRCT negative (for the outcome FVC). As we assumed that data were missing completely at Exploratory subgroup analysis included disease dura- random or missing at random, multiple imputation using tion ≤3 years versus disease duration >3 years and C reac- chained equations (MICE) was used for confounders and tive protein (CRP) ≤5 mg/L versus CRP >5 mg/L and outcomes at baseline as well as follow-­up. The number of HRCT and/or X-­ray positive versus HRCT and/or X-­ray multiply imputed data sets was set to 60 (m=60). negative. Subgroup analyses were preceded by a test for inter- Statistical analysis action. Subgroup results were only reported if there Descriptive statistics included mean and SD or median was evidence for differential treatment effect between and IQR for continuous variables, and number and subgroups (if p<0.05). percentage of total for categorical variables. The primary outcomes, mRSS and FVC, were compared Propensity score matching with linear mixed-­effects models to account for the correla- Based on expert opinion and considering the published tion between the matched samples. The results from literature, the following variables were identified as multiply imputed data sets were combined using Rubin’s confounders in an interdisciplinary team discussion Protected by copyright. rule. We used a generalised linear mixed model with bino- (SJ, OD, SK, UH, KS): age at diagnosis, gender, disease mial family with a random intercept accounting for pair subtype (diffuse or limited), baseline mRSS, FVC, DLCO, membership of matched treated and control patients. co-­therapy with immunosuppressive disease-­ modifying Binary secondary outcomes were compared with ORs antirheumatic drugs (either one of prednisone >10 and 95% CIs between treatment groups; again, these were mg/day, methotrexate, azathioprine or mycophenolate corrected for correlation in matched pairs due to matching. mofetil), rituximab within 6 months before baseline, Between-­group differences for continuous outcomes were disease duration in years, and year of treatment. estimated and reported with 95% CIs. The effect measures The propensity score was the estimated probability of a for the binary outcomes are marginal ORs. The signifi- patient in the EUSTAR database to receive tocilizumab. cance level for confirmatory p values was set to 0.05. To estimate the propensity score, a logistic regression A sensitivity analysis was conducted to address pre-­ model was fitted to the confounders.28 29 We used a processing decisions (selection of most recent vs random nearest neighbour 1:1 propensity score matching (PSM) observation for control patients with multiple suitable algorithm.30 Balancing of baseline characteristics before time intervals), as well as matching algorithm (nearest and after matching was assessed with descriptive statistics, neighbour vs exact matching) and robustness of the the standardised mean difference (SMD) and explor- results. For further details and specifications, see the atory p values. If the SMD was smaller than 0.1, the distri- statistical analysis plan (online supplemental file). bution of confounders was assumed to be balanced.31 All data analyses were conducted using R, V.4.1.2 for Windows32 and the packages tableone, mice, MatchIt, Handling of missing values reshape2, dplyr, mitools, lsmeans, VIM, cobalt, ggplot2, The procedure for handling of missing values of poten- lmer and glmer. Results of the study were reported tial control patients and treated patients varied. The according to the STROBE guidelines.33 following variables were used in the multiple imputation model as predictor variables: age in years (numeric), gender (binary), subtype (binary), prednisone (binary), methotrexate (binary), azathioprine (binary), myco- RESULTS phenolate mofetil (binary), rituximab (biologic) within Baseline characteristics of tocilizumab and control patients 6 months before baseline (binary), disease duration in Data from 109 patients with SSc treated with tocilizumab years (numeric), year of treatment (numeric) as well were collected from 25 EUSTAR centres. Of these 109 as baseline and follow-­ up of mRSS, FVC and DLCO patients, 12 were excluded for the effectiveness anal- (numeric). Age, disease duration in years, baseline ysis due to missing follow-­up at 12±3 months and 4 due and follow-­up mRSS, FVC and DLCO were imputed to absence of at least three tocilizumab applications. using predictive mean matching. Subtype, prednisone, Route of administration was intravenous in 65 (69.9%), Kuster S, et al. RMD Open 2022;8:e002477. doi:10.1136/rmdopen-2022-002477 3 RMD Open RMD Open: first published as 10.1136/rmdopen-2022-002477 on 3 November 2022. Downloaded from http://rmdopen.bmj.com/ on October 9, 2023 at Sant Paul Hospital Consortia. Figure 1 Flow diagram. *Matching criteria: age at diagnosis, gender, subtype (limited/diffuse), baseline modified Rodnan skin score, baseline FVC, baseline diffusing capacity of the lung for carbon monoxide, co-­therapy immunosuppressive disease-­ modifying antirheumatic drugs (either one of prednisone >10 mg/day, methotrexate, azathioprine and mycophenolate mofetil), rituximab (biologic) within 6 months before baseline, disease duration (years), year of treatment. TCZ, tocilizumab; EUSTAR, European Scleroderma Trial and Research. subcutaneous in 13 (14.0%) and no information avail- Effects of tocilizumab on skin and lung fibrosis Protected by copyright. able in 15 (16.1%). Follow-­up mRSS as a measure of skin fibrosis after 12±3 From the EUSTAR database, 3180/10 426 patients were months of therapy was lower in the tocilizumab group fulfilling the inclusion criteria for the control group. (mean estimate of 11.2, 95% CI 9.1 to 13.3) compared There were 14.0% missing data in baseline mRSS and with the control group (12.2, 9.7 to 14.6, p=0.48). This 40.9% at follow-­up. For FVC, the corresponding percent- effect was stable regardless of the pre-­processing deci- ages were 25.8% at baseline and 45.2% at follow-­ up. sions and the matching algorithm (see sensitivity anal- Through application of MICE and PSM on these patients, yses). 60 imputed sets of 93 matched sets of treated and control Similar to skin fibrosis, we could consistently see a patients were generated (figure 1). tendency towards a benefit of tocilizumab therapy for lung Baseline demographics and clinical characteristics of fibrosis as measured by FVC per cent predicted, which the tocilizumab group and the control group before and did not reach statistical significance. FVC at follow-­up was after multiple imputation and propensity score matching 88.7 (83.7 to 93.7)% predicted in the tocilizumab group can be found in table 1. As expected, the proportion versus 87.2 (80.8 to 93.6)% predicted in the control group of patients with the diffuse subtype, the prevalence of (p=0.70). Results from the pre-­ defined main analysis increased inflammation markers (CRP/erythrocyte sedi- of primary outcomes are shown in table 3 and figure 2. mentation rate above normal limits) and patients with Accordingly, the mean estimated difference (with 95% CI) arthritis, and overlap to rheumatoid arthritis (online between groups for skin fibrosis measured by mRSS was supplemental table S1) was high, reflecting the typical lower in TCZ −1.0 (−3.7 to 1.8) and higher in TCZ for lung real-­life indications for tocilizumab at time of the fibrosis measured by FVC (% predicted) 1.5 (−6.1 to 9.1). study. Indeed, the majority of physicians selected tocili- zumab as a treatment because of SSc-­associated arthritis (table 2). After multiple imputation and propensity Secondary outcomes score matching, the pooled SMD of baseline covari- The percentage of progressive patients for skin fibrosis ates between the two groups was <0.1 for all matching as well as for lung fibrosis (online supplemental table variables. Therefore, baseline covariates were consid- S3) was lower under tocilizumab therapy as compared ered balanced and further adjustment was not needed with the control group without reaching statistical signifi- (online supplemental table S2). In addition, online cance. The OR for mRSS was 0.7 (0.1 to 4.8, p=0.74) and supplemental table S4 shows baseline characteristics 0.8 (0.3 to 2.2, p=0.63) for progression of ILD as meas- with SMD and p values for a randomly drawn data set ured by decline in FVC. after multiple imputation. No significant effectiveness of tocilizumab could be No covariates had a percentage of missing values >50%; shown for percentage of regressive patients under therapy, therefore, no covariate needed to be excluded. with an OR of 1.1 (0.4 to 2.7, p=0.86) for regression of 4 Kuster S, et al. RMD Open 2022;8:e002477. doi:10.1136/rmdopen-2022-002477 Systemic sclerosis RMD Open: first published as 10.1136/rmdopen-2022-002477 on 3 November 2022. Downloaded from http://rmdopen.bmj.com/ on October 9, 2023 at Sant Paul Hospital Consortia. Table 1 Baseline characteristics before and after multiple imputation and propensity score matching TCZ-­treated patients Patients without TCZ, potential controls Randomly drawn data set Before MICE and PSM after MICE and PSM N=93 N=3180 SMD N=93 SMD Age (mean±SD; years) 50.9±13.5 56.8±13.7 0.43 48.4±15.1 0.18 Sex  Female (n, %) 73 (78.5) 2637 (82.9) 0.11 73 (78.5) <0.001 Systemic sclerosis subtype  Diffuse (n, %) 49 (57.6.) 1319 (41.6) 0.33 52 (55.9) 0.02 Immunosuppressive co-­therapy  Yes 70 (80.5) 882 (29.4) 1.20 75 (80.6) <0.001 Prednisone ≥10 mg/day (n, %) 41 (48.8) 214 (7.5) 1.04 16 (17.2) 0.68 Cyclophosphamide (n, %) – 146 (5.0) – 6 (7.2) – Methotrexate (n, %) 36 (50.0) 336 (11.3) 0.92 31 (33.3) 0.22 Azathioprine (n, %) 6 (9.2) 198 (6.7) 0.09 20 (21.5) 0.45 Mycophenolate mofetil (n, %) 4 (6.7) 232 (7.8) 0.05 18 (19.4) 0.39 D-­Penicillamine (n, %) – 21 (0.7) – 0 (0.0) – Rituximab within 6 months (n, %) 1 (1.1) 42 (1.3) 0.02 1 (1.1) <0.001 Imatinib (n, %) 0 (0.0) 1 (0.0) – 0 (0.0) – TNF-­alpha antagonist (n, %) 0 (0.0) 10 (0.3) 0.08 0 (0.0) <0.001 Abatacept (n, %) 1 (1.1) – – 1 (1.1) – Protected by copyright. Disease duration (mean±SD, years) 6.4±5.4 10.6±7.5 0.65 6.2±4.9 0.04 Autoantibodies positive  ANA (n, %) 73 (92.4) 2789 (95.7) 0.14 78 (96.3) 0.17  ACA (n, %) 12 (16.7) 1054 (38.0) 0.49 10 (13.5) 0.09  Anti-­Scl-­70 (n, %) 54 (65.1) 1013 (36.3) 0.60 42 (53.8) 0.23 CRP ≥5 mg/L (n, %) 49 (56.3) 250 (8.0) 1.21 11 (12.1) 1.054 ESR >25 mm/h (n, %) 38 (54.3) 867 (30.4) 0.50 29 (34.9) 0.40 Baseline mRSS (median, IQR) 14.0 (6.0, 22.2) 6.0 (2.0, 11.0) 0.79 11.0 (6.0, 21.0) 0.07 Baseline FVC % predicted (mean±SD) 84.9±19.6 95.8±21.6 0.52 88.0±22.8 0.01 Baseline DLCO % predicted (mean±SD) 62.2±22.4 67.5±19.9 0.25 65.1 (19.0) 0.12 HRCT or X-­ray positive for ILD (n, %) 49 (73.1) 1276 (48.3) 0.53 37 (47.4) 0.54 Digital ulcers (n, %) 16 (17.8) 274 (12.2) 0.16 12 (18.5) 0.02 Joint synovitis (n, %) 44 (62.0) 320 (10.2) 1.28 14 (15.4) 1.09 Tendon friction rubs (n, %) 25 (31.2) 184 (6.0) 0.69 11 (12.1) 0.48 Demographics and clinical characteristics are defined according to EUSTAR criteria.39 The standardised mean difference (SMD) is a measure for assessing balance of distributions, values <0.1 indicate balanced covariates between matched samples. ACA, anti-­centromere antibodies; ANA, antinuclear antibodies; anti-­Scl-­70, anti-­topoisomerase antibodies; CRP, C reactive protein; DLCO, diffusing capacity for carbon monoxide; ESR, erythrocyte sedimentation rate; FVC, forced vital capacity; ILD, interstitial lung disease; MICE, multiple imputation using chained equations; mRSS, modified Rodnan skin score; PSM, propensity score matching; RNA-­pol III, anti-­polymerase III; TCZ, tocilizumab. mRSS and 1.5 (0.6 to 3.8, p=0.41) for increase of FVC FVC ≥80% versus FVC <80% and FVC ≥80% and X-­ray (table 4). or HRCT positive versus FVC <80% and X-­ray or HRCT Interaction tests for differential treatment effects negative (for the outcome FVC). between the subgroups showed no evidence for subgroup effects (online supplemental table S5), suggesting that Sensitivity analysis either the results are not different or the sample size is The sensitivity analysis addressed multiple observations too small for patients with mRSS ≥10 versus mRSS <10, in patients with more than one suitable time interval dcSSc versus lcSSc (for the outcome mRSS) as well as as well as different matching algorithms. The sensitivity Kuster S, et al. RMD Open 2022;8:e002477. doi:10.1136/rmdopen-2022-002477 5 RMD Open RMD Open: first published as 10.1136/rmdopen-2022-002477 on 3 November 2022. Downloaded from http://rmdopen.bmj.com/ on October 9, 2023 at Sant Paul Hospital Consortia. show significant effectiveness of tocilizumab on skin and Table 2 Indication for treatment with TCZ lung fibrosis. Subgroup analysis did not show evidence n=93 for differences in effectiveness across subpopulations, Joints 67 (72.0%) although sample sizes might have been too small to detect Skin 25 (26.9%) differences in subgroups. However, a remarkable finding Lung 20 (21.5%) of this study was the consistent, although not significant, Myositis 2 (2.2%) point estimates in favour of tocilizumab across all prede- fined primary and secondary endpoints. Furthermore, Heart 1 (1.1%) additional sensitivity analyses were consistent. The data Tendinitis 2 (2.2%) used in this study were from a large registry including Vasculitis 1 (1.1%) a general population of patients with SSc. Although the Coexisting Castelman-­like disease 1 (1.1%) estimated effects of tocilizumab were not significant, the Joints only 50 (53.8%) results of the study may be seen as hypothesis generating for a potential effectiveness of tocilizumab in broader Joints and skin 6 (6.5%) patient populations than studied in the highly selected Joints and lung 7 (7.5%) and enriched RCT populations. The hypothesis that Joints and myositis 1 (1.1%) tocilizumab might also be effective in broader patient Joints, skin and lung 3 (3.2%) populations needs now to be tested in further large Skin only 8 (8.6%) prospective RCTs. Skin and lung 4 (4.3%) Our study has to be interpreted in light of the results from the two RCTs conducted in SSc with a high evidence Skin and heart 1 (1.1%) level.17 19 In both RCTs, there was a consistent trend for Skin and tendovaginitis 2 (2.2%) the primary endpoint mRSS favouring tocilizumab. Skin, lung and myositis 1 (2.2%) Considering that in these RCTs, the study population Lung only 5 (5.4%) was strongly enriched for mRSS dynamics, the current Vasculitis only 1 (1.1%) confirmation of these results in a much less selected Protected by copyright. Coexisting Castelman-­like disease only 1 (1.1%) observational real-­life cohort is encouraging. Regarding lung fibrosis, both RCTs showed strong effects of tocili- NA 3 (3.2%) zumab on FVC as a secondary or exploratory endpoint. NA, not available; TCZ, tocilizumab. An important result of both trials was the successful enrichment for a strong decline of FVC in the placebo groups using a combination of inclusion criteria such as analyses were not significant, but showed consistency of early dcSSc, increased inflammatory markers and recent direction in the differences between tocilizumab and progression of skin fibrosis.34 35 The resulting progres- control observed in the main analysis (summarised in sion of FVC was comparable with that seen in idiopathic online supplemental tables S6–S9). pulmonary fibrosis and allowed demonstrating the strong Safety of tocilizumab difference of FVC in the tocilizumab group compared Safety parameters of all 109 patients were assessed at 0, 3, with placebo with a stabilised FVC. This strong difference 6 and 12 months of follow-­up. Assessed exposed patient of FVC would be more difficult to be shown with less years (PY) were 93.8. A total of 90 adverse events (AEs) enriched patient populations and slower progression of (96 per 100 PY) and 17 serious adverse events (SAEs) FVC in the placebo/control group such as in the Senscis (18.1 per 100 PY) were registered. SAEs are summarised trial7 or in the present study. in table 5. Among the SAEs, there was one death and Considering safety, our study did not reveal signifi- eight events resulted in discontinuation of tocilizumab. cant new potential threats of tocilizumab. The profile of Most frequently observed adverse events were disorders adverse events was similar to that of other studies investi- of the blood and lymphatic system like leucopenia and gating safety parameters in patients with tocilizumab.17 36 37 thrombocytopenia (32 events, 34.2 per 100 PY). However, There was a predominance of infections among the SAEs 31 of 32 were mild (leucocytes >1500/µL, thrombocytes (5/17) and well-­known laboratory abnormalities such as >50 000/µL). Twenty-­five infections were registered (26.7 thrombocytopenia or elevated transaminases among the per 100 PY). Superinfection of digital ulcers was reported AEs (49/90). However, while previous studies suggested in four patients. Elevated transaminases were reported that serious infections might be higher in patients with (18 events, 19.2 per 100 PY), yet only five were >2× upper SSc treated with tocilizumab than in patients with rheu- limit of normal. matoid arthritis, our study could not confirm this.17 38 Of particular interest are infections of digital ulcers, but with only four infections associated with digital ulcers DISCUSSION (two considered serious) observed in 93.8 patient years, Our multicentre, propensity score matched, controlled application of tocilizumab does not seem to dramatically observational study in the EUSTAR database did not increase ulcer infections. 6 Kuster S, et al. RMD Open 2022;8:e002477. doi:10.1136/rmdopen-2022-002477 Systemic sclerosis RMD Open: first published as 10.1136/rmdopen-2022-002477 on 3 November 2022. Downloaded from http://rmdopen.bmj.com/ on October 9, 2023 at Sant Paul Hospital Consortia. Table 3 Primary outcomes at follow-­up (12±3 months) mRSS FVC (% predicted) TCZ, n=93 Mean estimate (95% CI) 11.2 (9.1 to 13.3) 88.7 (83.7 to 93.7) Controls Mean estimate (95% CI) 12.2 (9.7 to 14.6) 87.2 (80.8 to 93.6) Between-­group difference Mean estimate (95% CI) −1.0 (−3.7 to 1.8) 1.5 (−6.1 to 9.1) P value 0.48 0.70 These results represent our main analysis with a nearest neighbour matching algorithm and selection of most recent observation in control patients with multiple possible baseline observations. CI, confidence interval; FVC, forced vital capacity; mRSS, modified Rodnan skin score; TCZ, tocilizumab. Our study has limitations. Despite the relatively large associated with disease progression, such as increased number of patients in the EUSTAR database, there were inflammatory markers and arthritis, were more common much less patients available who received treatment in the tocilizumab patients than in controls. However, with tocilizumab. Therefore, uncertainty increased and this should lead to more progression in the tocilizumab the power of the study was not high enough to allow group and supports a potential positive effect of tocili- definite conclusions. In general, a higher number of zumab in this study. Due to a small number of patients patients is needed to show a treatment effect in obser- in the subgroups, it was not possible to include all vari- vational cohorts than in RCTs because there is more ables into a single imputation model. Therefore, the heterogeneity. In addition, the leading indication to subgroup results are biased towards null and should be treat patients with SSc with tocilizumab was presence of interpreted with caution. Another potential limitation arthritis and/or overlap to rheumatoid arthritis at time is that we matched for any immunosuppressive treat- of the present study and the associated disease character- ment, but not for single immunosuppressive drugs. If istics with, for example, longer disease duration further the immunosuppressive drugs are having very different decrease the likelihood that a significant treatment effect effects on outcomes than others, this matching would Protected by copyright. can be observed, at least for mRSS. After publication not have been perfect. However, numbers would have of the two tocilizumab RCTs, this practice pattern has been too small for matching of single immunosuppres- likely changed and many more patients with SSc-­ ILD sive drugs and results would have not been meaningful. are treated nowadays with tocilizumab. Despite propen- Moreover, the absolute numbers of AE observed in this sity scoring matching for key parameters, some features trial should be considered with caution, as they were Figure 2 Modified Rodnan skin score (mRSS) and FVC at follow-­up (12±3) months. (A) Primary outcome mRSS estimate (95% CI) tocilizumab (TCZ) vs control between-­group mean difference p=0.48. (B) Primary outcome FVC% predicted estimate (95% CI) TCZ vs control between-­group difference p=0.70. Kuster S, et al. RMD Open 2022;8:e002477. doi:10.1136/rmdopen-2022-002477 7 RMD Open RMD Open: first published as 10.1136/rmdopen-2022-002477 on 3 November 2022. Downloaded from http://rmdopen.bmj.com/ on October 9, 2023 at Sant Paul Hospital Consortia. Table 4 Secondary outcomes: progression/regression of mRSS and decline/increase of FVC mRSS FVC Progression Regression Decline Increase Estimated treatment effect of TCZ with OR (95% CI) 0.7 (0.1 to 4.8) 1.1 (0.4 to 2.7) 0.8 (0.3 to 2.2) 1.5 (0.6 to 3.8) P values 0.74 0.86 0.63 0.41 These results represent our main analysis with a nearest neighbour matching algorithm and selection of most recent observation in control patients with multiple possible baseline observations. CI, confidence interval; FVC, forced vital capacity; mRSS, modified Rodnan skin score; OR, odds ratio; TCZ, tocilizumab. collected retrospectively, are therefore likely underes- study reports the largest number of patients with SSc timated, and a control group for the AE was missing. treated with tocilizumab with similar numbers as in Finally, it must be strongly emphasised that observational the phase III focuSSced study.19 In addition, data were studies can provide important signals, which may be used collected from a real-­life setting avoiding over-­enrichment for hypothesis generation and might contribute to gener- as in standard RCTs. Controls were derived from the very alisation of trial results to real-­life setting, but are never large, prospectively collected EUSTAR database. of high enough evidence to prove drug effectiveness. Still, the current study adds important supportive data for effectiveness of tocilizumab in SSc in real-­life setting CONCLUSION consistent with data from the two RCTs. Taken together, in this large, propensity score matched, Strengths of the study include the robust, predefined controlled observational real-­ life EUSTAR study, we and preregistered study protocol (www.drks.de), with a could not show significant effectiveness of tocilizumab detailed statistical analysis plan involving expert biostat- for skin and lung fibrosis across all predefined primary isticians and applying propensity score matching and and secondary endpoints. However, the consistency of using optimised matching procedures, while accounting direction of all predefined endpoints generates hypoth- Protected by copyright. for missing data with a multiple imputation approach. All esis for potential effectiveness in a broader SSc popula- major decisions regarding methodological approaches tion than the highly selective population included in the were re-­ evaluated in sensitivity analyses. The results RCTs. This hypothesis needs to be confirmed by prospec- remained stable indicating that the methodology or tive RCTs with broader patient populations. Safety anal- assumptions did not affect the results of this study. This ysis confirmed previous AE profiles without new signals. Table 5 Summary of serious adverse events over the total follow-­up duration Age Report at Discontinuation Patient n years visit (month) Event of tocilizumab Hospitalisation Patient 5 42 12 Severe thrombocytopenia No No Patient 39 45 6 Toe necrosis and infection Yes Yes Patient 39 45 12 Pneumonitis (after stop of TCZ) – Yes Patient 43 68 6 Atrial flutter Yes No Patient 49 32 12 Digital ulcer with osteomyelitis No NA Patient 50 41 3 Allergic reaction Yes No Patient 50 41 3 Influenza Yes Yes Patient 69 41 3 Renal crisis Yes Yes Patient 69 41 6 Thrombophlebitis (after stop of TCZ) – Yes Patient 70 58 6 Acute heart failure Yes NA Patient 70 58 12 Amputation last phalanx Dig I foot (after stop of TCZ) – NA Patient 71 76 6 Deep vein thrombosis No NA Patient 80 56 0 Severe allergic reaction Yes NA Patient 83 58 3 Bilateral keratitis Yes No Patient 92 63 3 Pneumonia No Yes Patient 92 63 12 Pneumonia No Yes Patient 97 18 12 Sudden cardiac death – NA –, Information is not applicable; NA, data are not available; TCZ, tocilizumab. 8 Kuster S, et al. RMD Open 2022;8:e002477. doi:10.1136/rmdopen-2022-002477 Systemic sclerosis RMD Open: first published as 10.1136/rmdopen-2022-002477 on 3 November 2022. Downloaded from http://rmdopen.bmj.com/ on October 9, 2023 at Sant Paul Hospital Consortia. Author affiliations (Hungary); Michele Iudici, Geneva (Switzerland); Eugene J Kucharz, Katowice 1 (Poland); Bernard Coleiro, Balzan (Malta); Dominique Farge Bancel, Paris (France); Department of Rheumatology; University Hospital Zurich, University of Zurich, Zurich, Switzerland Roger Hesselstrand, Lund (Sweden); Mislav Radic, Split (Croatia); Raffaele Pellerito, 2 Department of Rheumatology, Paris Descartes University, Cochin Hospital, APHP, Torino (Italy); Nemanja Damjanov, Belgrade (Serbia); Jörg Henes, Tübingen (Germany); Vera Ortiz-­Santamaria, Granollers Barcelona (Spain); Stefan Heitmann, Paris, France 3 Stuttgart (Germany); Paul Hasler, Aarau (Switzerland); Bojana Stamenkovic, Niska Department of Biostatistics at Epidemiology, Biostatistics and Prevention Institute, Banja (Serbia); Carlo Francesco Selmi, Rozzano, Milano (Italy); Mohammed Tikly, University of Zurich, Zurich, Switzerland Johannesburg (South Africa); Lidia P Ananieva, Moscow (Russia); Ulf Müller-­ 4 Department of Experimental and Clinical Medicine, Division of Rheumatology, Ladner, Bad Nauheim (Germany); Merete Engelhart, Hellerup (Denmark); Carlos University of Florence, Florence, Italy de la Puente, Madrid (Spain); Cord Sunderkötter, Münster (Germany); Francesca 5 Rheumatology Unit, DETO, University Hospital Polyclinic of Bari, Bari, Italy Ingegnoli, Milano (Italy); Luc Mouthon, Paris (France); Francesco Paolo Cantatore, 6 Department of Precision Medicine, University of Campania 'Luigi Vanvitelli', Foggia (Italy); Susanne Ullman, Copenhagen (Denmark); Maria Rosa Pozzi, Monza Naples, Italy (Italy); Piotr Wiland, Wroclaw (Poland); Marie Vanthuyne, Brussels (Belgium); 7 Clinic for Rheumatology and Immunology, Bad Bramstedt, Germany Juan Jose Alegre-­Sancho, Valencia (Spain); Brigitte Krummel-­Lorenz, Frankfurt 8 (Germany); Kristine Herrmann, Dresden (Germany); Ellen De Langhe, Leuven Berlin Institute of Health at Charité, Berlin, Germany 9 (Belgium); Branimir Anic, Zagreb (Croatia); Sule Yavuz, Altunizade-­Istanbul (Turkey); Clinica Reumatologie, University of Medicine and Pharmacy, Cluj-­Napoca, Romania Carolina de Souza Müller, Curitiba (Brazil); Svetlana Agachi, Chisinau (Republic 10 Department of Rheumatology, IRCCS Foundation Policlinico San Matteo, Pavia, of Moldova); Thierry Zenone, Valence (France); Simon Stebbings, Dunedin (New Italy Zealand); Alessandra Vacca, Monserrato (CA)(Italy); Lisa Stamp, Christchurch 11 UOC Rheumatology and Clinical Immunology, Spedali Civili di Brescia, Brescia, (New Zealand); Kamal Solanki, Hamilton (New Zealand); Douglas Veale, Dublin Italy (Ireland); Esthela Loyo, Santiago (Dominican Republic); Mengtao Li, Beijing (China); 12 B. Shine Department of Rheumatology, Rambam Health Care Campus and Walid Ahmed Abdel Atty Mohamed, Alexandria (Egypt); Edoardo Rosato, Roma Technion, Haifa, Israel (Italy); Cristina-­Mihaela Tanaseanu, Bucharest (Romania); Rosario Foti, Catania 13 University of Cologne, Cologne, Germany (Italy); Codrina Ancuta, Iasi (Romania); Britta Maurer, Bern (Switzerland); Paloma 14 García de la Peña Lefebvre, Madrid (Spain); Jean Sibilia, Strasbourg (France); Ira Department of Rheumatology, Centro Hospitalar e Universitário de Coimbra, Litinsky, Tel-­Aviv (Israel); Francesco Del Galdo, Leeds (UK); Goda Seskute, Vilnius Coimbra, Portugal (Lithuania); Lesley Ann Saketkoo, New Orleans (USA); Eduardo Kerzberg, Buenos 15 Department of Clinical Internal, Anesthesiological and Cardiovascular Sciences, Aires (Argentina); Doron Rimar, Haifa (Israel); Camillo Ribi, Lausanne (Switzerland); Sapienza University of Rome, Rome, Italy Vivien M. Hsu, New Brunswick (USA); Thierry Martin, Strasbourg (France); Lorinda 16 Department of Internal Medicine and Rheumatology, Cantacuzino Hospital, Carol S Chung, Stanford (USA); Tim Schmeiser, Wuppertal-­Elberfeld (Germany); Dominik Davila University of Medicine and Pharmacy, Bucharest, Romania Majewski, Poznan (Poland); Vera Bernardino, Lisboa (Portugal); Piercarlo Sarzi 17 Department of Rheumatology, Hospital Peset, Valencia, Spain Puttini, Milano (Italy). 18 Department of Biophysics, Physiology and Pathophysiology, Medical University of Contributors The manuscript has been seen and approved by all coauthors Warsaw, Warsaw, Poland and contributors. Prof. Dr. med. Oliver Distler is a guarantor and accepts full Protected by copyright. 19 Department of Connective Tissues Diseases, Institute of Rheumatology, Warsaw, responibility for the work and the conduct of the study, has access to the data and Poland controlled the decision to publish. 20 Department of Rheumatology, Hospital de la Santa Creu i Sant Pau, Barcelona, Funding Data analysis was supported by F. Hoffmann-­La Roche Ltd; Basel, Spain Switzerland. 21 University Hospital Hamburg-­Eppendorf, Division of Rheumatology and Systemic Disclaimer Roche did not have any access to primary data, and was not involved Inflammatory Diseases, Hamburg, Germany in data analysis, data interpretation and writing of this article. 22 Department of Rheumatology, Hospital General Universitario Gregorio Marañón, Competing interests SK, SJ, ME, FC, ES, SR, VC, YB-­M, MS, AMG, KR-­P, FJL-­L; Madrid, Spain PIN, AG, YS, LB, VS, TS, AG: none. CB reports consulting fees from Boehringer 23 Department of Rheumatology, National Reference Center for Systemic Ingelheim, Eli-­Lilly. Research Grants from Gruppo Italiano Lotta alla Sclerodermia Autoimmune Rare Diseases, Hopital Pellegrin, Bordeaux, France (GILS), Foundation for Research in Rheumatology (FOREUM), European 24 Rheumatology Section, Department of Medicine, University of Verona, Verona, Scleroderma Trial and Research (EUSTAR), New Horizon Fellowship. AMG has Italy received research funding and/or consulting fees from Abbvie, GenevaRompharm, 25 Department of Allergy and Rheumatology, Nippon Medical School, Tokyo, Japan Novartis and Boehringer Ingelheim. SV received consultancy for BI Global and 26 Rheumatology Unit, ASST Grande Ospedale Metropolitano Niguarda, Milano, Italy BI Italy spa. PA reports personal fees (consultancies) from Bristol Myers Squibb, 27 Rheumatology Unit, Department of Clinical and Experimental Medicine, University Boehringer Ingelheim, non-­financial support from CSL Behring, SOBI, Janssen, Roche, Sanofi, Pfizer. NH has received lecture fees from Boehringer Ingelheim, of Padova, Padova, Italy 28 Pfizer, Sanofi, Roche. VR has/had personal fees or received research funding Referral Centre for Rare Systemic Auto-­immune Diseases for North and North-­ from Bayer, Boehringer Ingelheim, MSD, Corbus, Inventiva, Roche. JJAS reports West of France, Department of Internal Medicine, Claude Huriez Hospital, University personal fees and/or research funding from Janssen, Boehringer Ingelheim, Eli of Lille, Lille, France Lilly, Pfizer, Roche, MSD, Corbus, Inventiva, and GSK, outside of the submitted 29 Department of Internal Medicine, Department of Rheumatology, Ghent University work. IC reports personal fees or research funding from Boehringer Ingelheim, Hospital, Ghent, Belgium Janssen, Roche, Kern, BMS, Sanofi-­Aventis, Roche, UCB and BioCAT. MET has had 30 Unit for Molecular Immunology and Inflammation, VIB Inflammation Research consultancy relationships and/or has received research funding from Boehringer Centre (IRC), Ghent, Belgium Ingelheim, Genentech/Roche and Sanofi in the area of potential treatments of 31 scleroderma and its complications. EZ reports speaker honoraria and consultancy Centre for Rheumatology, University College London Medical School—Royal Free fees from Boehringer Ingelheim, Janssen, GlaxoSmithKline. CD reports personal Campus, London, UK 32 fees or research funding from Janssen, GlaxoSmithKline, Bayer, Sanofi-­Aventis, Internal Medicine and Rheumatology Department—St. Maria Hospital, Carol Galapagos, Boehringer Ingelheim, Roche, CSL Behring, Corbus, Acceleron, Horizon, Davila University of Medicine and Pharmacy, Bucharest, Romania ARXX Therapeutics. RMI had received consultancy and speaker fee from Abbvie, 33 Krankenhaus St. Josef, Wuppertal-­Elberfeld, Germany Boehringer Ingelheim, Novartis, Bristol Myers Squibb, Pfizer, Roche. JHWD has 34 Department of Internal Medicine III, Erlangen University Hospital, Erlangen, consultancy relationships with Active Biotech, Anamar, ARXX, AstraZeneca, Bayer Germany Pharma, Boehringer Ingelheim, Celgene, Galapagos, GSK, Inventiva, Janssen, 35 Dipartimento di Scienze Cliniche e Molecolari, Università Politecnica delle Novartis, Pfizer and UCB. JHWD has received research funding from Anamar, Marche, Ancona, Italy ARXX, BMS, Bayer Pharma, Boehringer Ingelheim, Cantargia, Celgene, CSL 36 Behring, Galapagos, GSK, Inventiva, Kiniksa, Sanofi-­Aventis, RedX, UCB. JHWD Department of Rheumatology, Rikshospitalet University Hospital Oslo, Oslo, is stock owner of 4D Science. AMHV has received research funding and/or Norway consulting fees and/or other remuneration from Actelion, Boehringer Ingelheim, Roche, Bayer, Merck Sharp & Dohme, ARXX, Lilly, Janssen and Medscape. MK has Collaborators EUSTAR Collaborators: Principal Investigator Affiliation; Ulrich received grants or personal fees from AbbVie, Asahi Kasei Pharma, Astellas, Bayer, Walker, Basel (Switzerland): Radim Becvar, Prague (Czech Republic); Maurizio Boehringer Ingelheim, Chugai, Corbus, Eisai, Galapagos, Janssen, Kissei, MBL, Cutolo, Genova (Italy); Patricia E Carreira, Madrid (Spain); László Czirják, Pecs Mochida, Nippon Shinyaku, Ono Pharmaceuticals, Pfizer and Tanabe-­Mitsubishi Kuster S, et al. RMD Open 2022;8:e002477. doi:10.1136/rmdopen-2022-002477 9 RMD Open RMD Open: first published as 10.1136/rmdopen-2022-002477 on 3 November 2022. Downloaded from http://rmdopen.bmj.com/ on October 9, 2023 at Sant Paul Hospital Consortia. Pharma. YA has had consultancy relationships and/or has received research 10 Muangchan C, Pope JE. Interleukin 6 in systemic sclerosis funding from Alpine, Boehringer Ingelheim, Genentech/Roche, Medsenic, Menarini and potential implications for targeted therapy. J Rheumatol and Sanofi in the area of potential treatments of scleroderma and its complications. 2012;39:1120–4. OD has/had consultancy relationship with and/or has received research funding 11 Muangchant C, Pope JE. The significance of interleukin-­6 and C-­ from and/or has served as a speaker for the following companies in the area of reactive protein in systemic sclerosis: a systematic literature review. potential treatments for systemic sclerosis and its complications in the last three Clin Exp Rheumatol 2013;31:122–34. calendar years: Abbvie, Acceleron, Alcimed, Amgen, AnaMar, Arxx, AstraZeneca, 12 Khan K, Xu S, Nihtyanova S, et al. Clinical and pathological Baecon, Blade, Bayer, Boehringer Ingelheim, Corbus, CSL Behring, 4P Science, significance of interleukin 6 overexpression in systemic sclerosis. Ann Rheum Dis 2012;71:1235–42. Galapagos, Glenmark, Horizon, Inventiva, Janssen, Kymera, Lupin, Medscape, 13 De Lauretis A, Sestini P, Pantelidis P, et al. Serum interleukin 6 is Miltenyi Biotec, Mitsubishi Tanabe, MSD, Novartis, Prometheus, Roivant, Sanofi predictive of early functional decline and mortality in interstitial and Topadur. Patent issued “mir-­29 for the treatment of systemic sclerosis” lung disease associated with systemic sclerosis. J Rheumatol (US8247389, EP2331143). 2013;40:435–46. Patient consent for publication Obtained. 14 Desallais L, Avouac J, Fréchet M, et al. Targeting IL-­6 by both passive or active immunization strategies prevents bleomycin-­ Ethics approval This study involves human participants and was approved by the induced skin fibrosis. Arthritis Res Ther 2014;16:R157. ethic committee of the Canton of Zurich (KEK-­ZH), Switzerland (BASEC Nr. 2017-­ 15 Kitaba S, Murota H, Terao M, et al. Blockade of interleukin-­6 01935). receptor alleviates disease in mouse model of scleroderma. Am J Provenance and peer review Not commissioned; externally peer reviewed. Pathol 2012;180:165–76. 16 Hashizume M, Tan S-­L, Takano J, et al. Tocilizumab, a humanized Data availability statement Data are available on reasonable request. anti-­IL-­6R antibody, as an emerging therapeutic option for Anonymised data might be available from OD at the Department of Rheumatology, rheumatoid arthritis: molecular and cellular mechanistic insights. Int University Hospiztal Zurich, University of Zurich, Switzerland on reasonable request. Rev Immunol 2015;34:265–79. Supplemental material This content has been supplied by the author(s). It has 17 Khanna D, Denton CP, Jahreis A, et al. Safety and efficacy of subcutaneous tocilizumab in adults with systemic sclerosis not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been (faSScinate): a phase 2, randomised, controlled trial. Lancet peer-­reviewed. Any opinions or recommendations discussed are solely those 2016;387:2630–40. of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and 18 Distler O, Distler JHW. Tocilizumab for systemic sclerosis: responsibility arising from any reliance placed on the content. Where the content implications for future trials. Lancet 2016;387:2580–1. includes any translated material, BMJ does not warrant the accuracy and reliability 19 Khanna D, Lin CJF, Furst DE, et al. Tocilizumab in systemic sclerosis: of the translations (including but not limited to local regulations, clinical guidelines, a randomised, double-­blind, placebo-­controlled, phase 3 trial. terminology, drug names and drug dosages), and is not responsible for any error Lancet Respir Med 2020;8:963–74. and/or omissions arising from translation and adaptation or otherwise. 20 Jordan S, Distler JHW, Maurer B, et al. Effects and safety of Open access This is an open access article distributed in accordance with the rituximab in systemic sclerosis: an analysis from the European Creative Commons Attribution Non Commercial (CC BY-­NC 4.0) license, which Scleroderma Trial and Research (EUSTAR) group. Ann Rheum Dis 2015;74:1188–94. permits others to distribute, remix, adapt, build upon this work non-­commercially, Protected by copyright. 21 Castellví I, Elhai M, Bruni C, et al. Safety and effectiveness of and license their derivative works on different terms, provided the original work is abatacept in systemic sclerosis: the EUSTAR experience. Semin properly cited, appropriate credit is given, any changes made indicated, and the Arthritis Rheum 2020;50:1489–93. use is non-­commercial. See: http://creativecommons.org/licenses/by-nc/4.0/. 22 Elhai M, Boubaya M, Distler O, et al. Outcomes of patients with systemic sclerosis treated with rituximab in contemporary ORCID iDs practice: a prospective cohort study. Ann Rheum Dis Fabio Cacciapaglia http://orcid.org/0000-0001-7479-4462 2019;78:979–87. Veronica Codullo http://orcid.org/0000-0003-2557-8514 23 van den Hoogen F, Khanna D, Fransen J, et al. 2013 classification Ivan Castellví http://orcid.org/0000-0002-5410-5807 criteria for systemic sclerosis: an American College of Vanessa Smith http://orcid.org/0000-0001-6271-7945 Rheumatology/European League Against Rheumatism collaborative Tim Schmeiser http://orcid.org/0000-0001-7342-0020 initiative. Arthritis Rheum 2013;65:2737–47. Anna-­Maria Hoffmann-­Vold http://orcid.org/0000-0001-6467-7422 24 van den Hoogen F, Khanna D, Fransen J, et al. 2013 classification Masataka Kuwana http://orcid.org/0000-0001-8352-6136 criteria for systemic sclerosis: an American College of Oliver Distler http://orcid.org/0000-0002-0546-8310 Rheumatology/European League Against Rheumatism collaborative initiative. Ann Rheum Dis 2013;72:1747–55. 25 Maurer B, Graf N, Michel BA, et al. Prediction of worsening of skin fibrosis in patients with diffuse cutaneous systemic sclerosis using REFERENCES the EUSTAR database. Ann Rheum Dis 2015;74:1124–31. 1 Varga J, Trojanowska M, Kuwana M. Pathogenesis of systemic 26 Wu W, Jordan S, Becker MO, et al. Prediction of progression of sclerosis: recent insights of molecular and cellular mechanisms interstitial lung disease in patients with systemic sclerosis: the SPAR and therapeutic opportunities. J Scleroderma Relat Disord model. Ann Rheum Dis 2018;77:1326–32. 2017;2:137–52. 27 Hoffmann-­Vold A-­M, Allanore Y, Alves M, et al. Progressive interstitial 2 McMahan ZH, Hummers LK. Systemic sclerosis-­challenges for lung disease in patients with systemic sclerosis-­associated clinical practice. Nat Rev Rheumatol 2013;9:90–100. interstitial lung disease in the EUSTAR database. Ann Rheum Dis 3 Denton CP. Systemic sclerosis: from pathogenesis to targeted 2021;80:219–27. therapy. Clin Exp Rheumatol 2015;33:S3–7. 28 Brookhart MA, Schneeweiss S, Rothman KJ, et al. Variable selection 4 Rubio-­Rivas M, Royo C, Simeón CP, et al. Mortality and survival in for propensity score models. Am J Epidemiol 2006;163:1149–56. systemic sclerosis: systematic review and meta-­analysis. Semin 29 Shadish WR, Clark MH, Steiner PM. Can nonrandomized Arthritis Rheum 2014;44:208–19. experiments yield accurate answers? A randomized experiment 5 Elhai M, Meune C, Boubaya M, et al. Mapping and comparing random and nonrandom assignments. J Am Stat Assoc predicting mortality from systemic sclerosis. Ann Rheum Dis 2008;103:1334–44. 2017;76:1897–905. 30 Ho DE, Imai K, King G. MatchIt: nonparametric preprocessing for 6 Khanna D, Distler JHW, Sandner P, et al. Emerging strategies parametric causal inference. J Stat Softw 2011;42:28. for treatment of systemic sclerosis. J Scleroderma Relat Disord 31 Austin PC. An introduction to propensity score methods for reducing 2016;1:186–93. the effects of confounding in observational studies. Multivariate 7 Distler O, Highland KB, Gahlemann M, et al. Nintedanib for systemic Behav Res 2011;46:399–424. sclerosis-­associated interstitial lung disease. N Engl J Med 32 The R project for statistical computing. Available: https://www.r-​ 2019;380:2518–28. project.org 8 Kowal-­Bielecka O, Fransen J, Avouac J, et al. Update of EULAR 33 von Elm E, Altman DG, Egger M, et al. The Strengthening the recommendations for the treatment of systemic sclerosis. Ann Reporting of Observational Studies in Epidemiology (STROBE) Rheum Dis 2017;76:1327–39. statement: guidelines for reporting observational studies. Lancet 9 Hoffmann-­Vold A-­M, Maher TM, Philpot EE, et al. The identification 2007;370:1453–7. and management of interstitial lung disease in systemic sclerosis: 34 Wu W, Jordan S, Graf N, et al. Progressive skin fibrosis is evidence-­based European consensus statements. Lancet Rheumatol associated with a decline in lung function and worse survival 2020;2:e71–83. in patients with diffuse cutaneous systemic sclerosis in the 10 Kuster S, et al. RMD Open 2022;8:e002477. doi:10.1136/rmdopen-2022-002477 Systemic sclerosis RMD Open: first published as 10.1136/rmdopen-2022-002477 on 3 November 2022. Downloaded from http://rmdopen.bmj.com/ on October 9, 2023 at Sant Paul Hospital Consortia. European Scleroderma Trials and Research (EUSTAR) cohort. 37 Atsumi T, Fujio K, Yamaoka K, et al. Safety and effectiveness of Ann Rheum Dis 2019;78:648–56. subcutaneous tocilizumab in patients with rheumatoid arthritis in a 35 Hoffmann-­Vold A-­M, Allanore Y, Alves M, et al. Progressive interstitial real-­world clinical setting. Mod Rheumatol 2018;28:780–8. lung disease in patients with systemic sclerosis-­associated 38 Koike T, Harigai M, Inokuma S, et al. Effectiveness and safety interstitial lung disease in the EUSTAR database. Ann Rheum Dis of tocilizumab: postmarketing surveillance of 7901 patients with 2021;80:219–27. rheumatoid arthritis in Japan. J Rheumatol 2014;41:15–23. 36 Brunner HI, Ruperto N, Zuber Z, et al. Efficacy and safety of 39 Walker UA, Tyndall A, Czirják L, et al. Clinical risk assessment of tocilizumab in patients with polyarticular-­course juvenile idiopathic organ manifestations in systemic sclerosis: a report from the EULAR arthritis: results from a phase 3, randomised, double-­blind Scleroderma Trials And Research group database. Ann Rheum Dis withdrawal trial. Ann Rheum Dis 2015;74:1110–7. 2007;66:754–63. Protected by copyright. Kuster S, et al. RMD Open 2022;8:e002477. doi:10.1136/rmdopen-2022-002477 11