Objectives No immunomodulatory drug has been approved for primary Sjögren’s syndrome, a systemic autoimmune disease affecting 0.1% of the population. To demonstrate the efficacy of targeting interleukin 6 receptor in patients with Sjögren’s syndrome-related systemic complications.
Methods Multicentre double-blind randomised placebo-controlled trial between 24 July 2013 and 16 July 2018, with a follow-up of 44 weeks, involving 17 referral centres. Inclusion criteria were primary Sjögren’s syndrome according to American European Consensus Group criteria and score ≥5 for the EULAR Sjögren’s Syndrome Disease activity Index (ESSDAI, score of systemic complications). Patients were randomised to receive either 6 monthly infusions of tocilizumab or placebo. The primary endpoint was response to treatment at week 24. Response to treatment was defined by the combination of (1) a decrease of at least 3 points in the ESSDAI, (2) no occurrence of moderate or severe activity in any new domain of the ESSDAI and (3) lack of worsening in physician’s global assessment on a Visual Numeric Scale ≥1/10, all as compared with enrolment.
Results 110 patients were randomised, 55 patients to tocilizumab (mean (SD) age: 50.9 (12.4) years; women: 98.2%) and 55 patients to placebo (54.8 (10.7) years; 90.9%). At 24 weeks, the proportion of patients meeting the primary endpoint was 52.7% (29/55) in the tocilizumab group and 63.6% (35/55) in the placebo group, for a difference of −11.4% (95% credible interval −30.6 to 9.0) (Pr[Toc >Pla]=0.14).
Conclusion Among patients with primary Sjögren’s syndrome, the use of tocilizumab did not improve systemic involvement and symptoms over 24 weeks of treatment compared with placebo.
Trial registration number NCT01782235.
- Sjogren's syndrome
- biological therapy
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
What is already known about this subject?
No immunomodulatory drug has been approved for primary Sjögren’s syndrome, a systemic autoimmune disease affecting 0.1% of the population.
What is the efficacy of targeting interleukin 6 (IL-6) receptor in primary Sjögren’s syndrome?
What does this study add?
In this randomised clinical trial that enrolled 110 patients, the proportion of patients who had a decrease in the systemic disease activity was 52.7% in the tocilizumab group and 63.6% in the placebo group, a non-significant difference.
In primary Sjögren’s syndrome, inhibition of the IL-6 receptor did not improve the systemic disease activity.
How might this impact on clinical practice or future developments?
These negative results for clinical outcome, patient-reported outcomes and immunological outcomes indicate that IL-6 does not represent a relevant therapeutic target in primary Sjogren’s syndrome.
Primary Sjögren syndrome (pSS) is a systemic autoimmune disease affecting 0.1% of the general population1 that mainly targets the exocrine system, such as salivary and lachrymal glands. The clinical presentation is highly heterogeneous. Fatigue, dryness and pain are hallmarks of the disease; one-third to one-half of patients show systemic involvement (notably, articular, lung involvement, peripheral neuropathy, vasculitis) and 5%–10% mucosa-associated lymphoid tissue lymphoma.2
To date, no specific immunomodulatory drug has demonstrated efficacy for this disease. Hydroxychloroquine, prescribed for symptomatic relief of fatigue and pain for a long time, did not prove efficacy despite the inhibition of interferon α.3 4 Drugs first developed for rheumatoid arthritis (RA), such as tumour necrosis factor inhibitors,5 6 or rituximab, also failed to demonstrate effectiveness, except in one of four trials for rituximab.7–10
New insights into understanding the pathogenesis of the disease, including the role of T helper cell subsets,11 salivary gland lymphoid neogenesis12 and kinases involved in the B-cell receptor transduction pathway,13 have not translated into positive randomised trials.14–17 Reasons for these numerous failures include the clinical heterogeneity of patients and challenges in clinical trial design, recruitment and outcome in such a complex autoimmune disease. Our national network of clinicians dedicated to pSS, and involved in the present clinical trial, previously designed some of these negative trials.3 5 9 In these previous trials, most patients did not have systemic complications and were mainly evaluated with patient-reported outcomes. An internationally validated score for systemic disease activity, the European League Against Rheumatism Sjögren’s Syndrome Disease Activity Index (ESSDAI),18 now allows for defining a threshold for moderate systemic disease activity (ESSDAI≥5) and a clinically relevant improvement in systemic disease activity (a 3-point decrease in ESSDAI).19
Interleukin 6 (IL-6) is suspected to play an important pathogenic role in pSS with its crucial roles in B-cell activation and T-cell polarisation.11 20 IL-6 deficiency corrected features of SS in a mouse model of the disease.21 In addition, the safety profile of IL-6 receptor inhibitors has been studied for 10 years in RA.22 Therefore, we investigated the interest of targeting IL-6 in patients with pSS with moderate or high systemic disease activity in a randomised placebo-controlled trial.
The randomisation used a centralised website based on a randomisation list using size-six block generated by an independent statistician. Patients were randomly assigned at a 1:1 ratio stratified by centre to receive 6 monthly infusions of tocilizumab (8 mg/kg) or a placebo between week 0 (W0) and W20. The drug and placebo were indistinguishable in appearance. Treatments were assigned after electronic verification of the correctness of inclusion criteria. Neither the investigators in charge of the study nor the participants were aware of the treatment assignments. A triple-blind procedure was applied, the statistician being unaware of the allocated treatment group during the analyses.
The primary endpoint was assessed at W24. After drug discontinuation, patients had two additional follow-up visits at W32 and W44.
In total, 17 referral clinical centres in France enrolled patients in the Efficacy of TocilizumAb in Primary Sjögren’s syndrome (ETAP) trial between 24 July 2013 and 29 June 2017. The last follow-up date for the last participant was 16 July 2018. To be included, participants had to fulfil the following criteria: American European Consensus Group (AECG) criteria for pSS, ESSDAI≥5, anti-SSA antibody-positive, >18 years old, signed informed consent, no contraindication to tocilizumab and receiving stable doses of non-steroidal anti-inflammatory drugs, oral corticosteroids (prednisone ≤15 mg/day), or pilocarpine, cevimeline, or topical cyclosporine for at least 2 weeks before enrolment, and stable doses of methotrexate, hydroxychloroquine, chloroquine, quinacrine, leflunomide, azathioprine, mycophenolate mofetil or a psychoactive drug for at least 8 weeks before enrolment. We excluded patients who received a biologic, intravenous immunoglobulins, cyclophosphamide or plasmapheresis therapy within 6 months before enrolment and those with severe systemic complications related to pSS at enrolment (vasculitis with renal neurologic, digestive or cardiac involvement, severe interstitial lung disease, symptomatic cryoglobulinemia with severe neurologic involvement, severe renal function impairment, severe myositis).
On 3 October 2015, after the inclusion of 41 patients, because of recruitment difficulties, the protocol was amended to allow the inclusion of anti-SSA antibody-negative patients fulfilling AECG criteria (online supplemental file 2).
The primary endpoint was response to treatment evaluated at week 24. Response to treatment was defined by the combination of (1) a decrease of at least 3 points in ESSDAI, (2) no occurrence of moderate or severe activity in any new domain of the ESSDAI and (3) no worsening in physician’s global assessment on a visual numeric scale ≥1/10, all when compared with enrolment. We chose this composite primary outcome, which is close to the Systemic Lupus Erythematosus (SLE) Response Index in SLE,23 because a 3-point decrease in ESSDAI corresponds to the minimally clinically relevant improvement for individuals with systemic complications19 and to avoid observing a new systemic complication or a worsening of the physician’s global evaluation despite a global improvement in ESSDAI.
Secondary endpoints were each of the three components of the composite primary endpoint; the ESSDAI24; number of tender and swollen joints; the EULAR Sjögren’s Syndrome Patient Reported Index (ESSPRI)25; the Schirmer test score; unstimulated salivary flow; serum IgG, IgA and IgM levels; the Medical Outcomes Survey Short-form 36 (SF-36) quality-of-life index; the Functional Assessment of Chronic Illness Therapy (FACIT)—fatigue scale and Hospital Anxiety and Depression (HAD) Scale score.
With the Casagrande and Pike method,26 we determined that we needed 48 participants in each group to achieve 80% power to detect a difference of 30% between group proportions. The proportion was assumed to be 25% in the placebo group and 55% in the tocilizumab group. The alpha level was set at 5%. The global sample size was increased to 110 to take into account potential missing data and lost to follow-up.
Categorical data are described with number (%) and continuous data with mean (SD) or median (IQR).
Data were analysed with Bayesian methods. Beta distributions Beta(alpha,beta) were used to estimate binary or categorical data and means were estimated with normal distribution (N(μ,σ²)), with specific parameter values depending on the variables (see table 1). The variances were given weakly informative priors.
From a probabilistic point of view, the goal of the study was to compute the probability that the proportion of participants meeting the primary outcome was larger in the tocilizumab than placebo group.
The main outcome was analysed by comparing the proportion of participants meeting the primary outcome with Beta(alpha,beta) distributions. All analyses of the primary outcome were on an intent-to-treat (ITT) basis. For each analysis of the main outcome, we computed the proportion difference (%) (95% credible interval (CrI)) and the probability that the difference was >0 in favour of the tocilizumab group, that is, Pr[Toc >Pla]. This probability must not be confused with the classical p value. Within the ITT analysis, missing data were modelled under the missing at random assumption and managed by using multiple imputation based on treatment group.
A sensitivity analysis was performed on complete data by using different informative priors from previous studies.3 The different priors used in the sensitivity analyses of the main outcome are in online supplemental table 1. The sensitivity analyses of the primary outcome also considered the missing not at random (MNAR) assumption. For this, missing data were imputed under one of two maximum bias patterns: with all missing data replaced by success in the placebo group and failure in the tocilizumab group (MNAR1) or vice versa (MNAR2). A non-responder imputation was also used.
For the secondary outcomes and post hoc analyses, no data imputation was performed for incomplete data. For the secondary outcomes, lowly informative priors were specified for each variable. Posterior estimation of the parameters (difference of means or proportions, OR) were provided with their median and 95% posterior CrI. The different scores, such as ESSDAI, are bounded (ie, with a minimal and maximal possible value) and were modelled first by using beta regression. Comparisons of these models with Gaussian approximations showed no major differences; therefore, results are expressed, based on Gaussian models, as mean (SD) for simplicity. Count data, such as the number of serious adverse events (SAEs), swollen and tender joints were compared by Poisson regression.
Repeated data were analysed by using mixed models with a random effect to take into account the intraindividual correlation. For these repeated measures, results are expressed as after–before mean difference (95% CI) within each group with the addition of the interaction term (with 95% CI) to quantify the mean between-group variation difference that expresses the treatment effect.
For each analysis, a burn-in of 5000 iterations, followed by 100 000 iterations was used for a single Markov chain Monte Carlo (McMC) chain. Convergence of the McMC sample chain was checked graphically and if required, with the Brooks-Gelman-Rubin test. Convergence was observed in each case. Autocorrelation was negligible in each case. All computations involved using R V.3.5.1 and JAGS V.4.3.0 with all the required additional packages.
Patient and public involvement
This research was done without patient involvement. Patients were not invited to comment on the study design and were not consulted to develop patient relevant outcomes or interpret the results. We have invited patients to help us develop our dissemination strategy.
We plan to disseminate the results to study participants and patient organisations.
A total of 110 patients were randomised: 55 to tocilizumab and 55 to placebo. The flow chart of the participants in the study is shown in figure 1. Mean age (SD) was 50.9 (12.4) and 54.8 (10.7) years in the tocilizumab and placebo group, respectively. Median (IQR) disease duration from diagnosis was 4.4 (1.6–9.0) and 4.9 (1.7–7.3) years and median ESSDAI was 11 (8–13.5) and 10 (8–14.8). Baseline characteristics of patients are reported in table 1 and baseline systemic complications in table 2.
At 24 weeks, the proportion of patients meeting the primary endpoint was 52.7% (29/55) in the tocilizumab group and 63.6% (35/55) in the placebo group, for a difference of −11.4% (95% CrI −30.6 to 9.0), (Pr[Toc >Pla]=0.14) after missing data multiple imputation (table 3). In sensitivity analyses, the difference between groups was −24.5 (−41.4 to −6.5), Pr[Toc >Pla]<0.001 (MNAR1) and 7.0% (−11.1 to 24.8), Pr[Toc >Pla]=0.78 (MNAR2).
Mean (SD) ESSDAI at week 24 was 8.3 (5.7) and 7.2 (5.3) in the tocilizumab and placebo groups, respectively, with a similar difference in changes from baseline between groups: 2% (95% CrI −1.2 to 5.2), (Pr[Toc >Pla]=0.89) (figure 2). In a post hoc analysis, at week 24, the mean (SD) clinical ESSDAI (not taking into account the biological domain of the ESSDAI))27 was 5.6 (6.3) and 4.8 (5.9) in the tocilizumab and placebo groups, respectively, with a similar difference in changes from baseline between groups: 1.6% (95% CrI −2.0 to 5.3), (Pr[Toc >Pla]=0.81).
A post hoc analysis restricted to anti-SSA-positive patients showed a similar proportion of patients meeting the primary endpoint (47.5% (19/40) in the tocilizumab group and 64.9% (24/37) in the placebo group; % difference=−16.48% (−37.1 to 4.9) [Pr(diff >0)=0.07]).
In a post hoc analysis, the improvement between W0 and W24 in each of the 12 domains of the ESSDAI, defined as a change from high to moderate or low or no activity, from moderate to low or no activity or from low to no activity, was similar between the tocilizumab and the placebo groups (table 4).
In post hoc analyses of each domain of the ESSDAI, tocilizumab did not improve systemic involvement in patients with moderate or high systemic disease activity compared with placebo (data not shown). In addition, changes were similar between the two groups in all the secondary clinical endpoints (table 5). Mean (SD) tender joint count at W24 was 4.4 (6.2) and 4.6 (7.5) in the tocilizumab and placebo groups (relative risk (RR): 1.1 (95% CI 0.5 to 2.1), (Pr[RR >1]=0.56), respectively. Mean swollen joint count at W24 was 0.5 (1.1) and 1.2 (3.8), respectively (RR=1.2 (0.4 to 2.6), (Pr[RR >1]=0.55). Mean ESSPRI at week 24 was 5.8 (2.0) and 6.2 (2.1), respectively, with a similar difference in changes from baseline between groups (−0.4 (−1.0 to 0.3)), (Pr[Toc >Pla]=0.12) (figure 3).
Changes in ocular and oral dryness assessed by the Schirmer test and unstimulated salivary flow are reported in table 5. Changes were similar between the two groups in fatigue assessed by the FACIT —Fatigue scale, quality of life assessed by the SF-36 and psychological discomfort assessed by the HAD scale (table 5).
A post hoc analysis showed a similar proportion of patients at W24 with a decrease in ESSPRI from baseline ≥1 point (47.9% (23/48) in the tocilizumab group and 42.2% (19/45) in the placebo group; % difference=5.45% (95% CrI −14.2 to 24.9) (Pr[Toc >Pla]=0.71)) and a similar proportion of patients at W24 with a decrease from baseline in ESSDAI≥3 points or a decrease from baseline in ESSPRI≥1 point (tocilizumab: 72% (35/48) vs 77.8% (35/45), % difference=−4.6% (95% CrI −21.7 to 12.7), (Pr[Toc >Pla]=0.3).
Changes were similar in the two groups in serum immunoglobulin and complement levels at W12 (table 5).
Over the first 24 weeks, 14 SAEs occurred in the tocilizumab group and 6 in the placebo group (RR=2.81 (95% CI 0.98 to 6.92)). In the last 24 weeks, 1 SAE occurred in the tocilizumab group and 5 SAEs in the placebo group (table 6). Between W0 and W44, 15 SAEs occurred in the tocilizumab group and 11 in the placebo group (RR=1.53 (0.64 to 3.14)). Three lymphomas occurred in the tocilizumab group and one in the placebo group. In the tocilizumab group, one lymphoma was diagnosed fortuitously thanks to a protocolised salivary gland biopsy before the first infusion of tocilizumab; one lymphoma was diagnosed after the first infusion, but after chart review, could have been suspected before randomisation; one lymphoma was diagnosed 1 month after the last infusion.
The present randomised double-blind placebo-controlled trial did not demonstrate a superiority of tocilizumab over placebo in patients with pSS and moderate or high systemic disease activity. Ineffectiveness was observed on primary and secondary endpoints.
Tocilizumab was prescribed as monthly infusions. Thus, concerns about patients’ adherence to treatment cannot explain the lack of efficacy of tocilizumab. In RA or giant cell arteritis, clinical efficacy is usually observed rapidly, within a few weeks of treatment. In pSS, anti-CD40 antibody treatment and the combination of leflunomide and hydroxychloroquine could improve disease manifestations at weeks 12 and 24, respectively.28 29 Therefore, a longer-term evaluation of the primary endpoint (after 6 months) would not have changed the overall results.
Inhibition of the IL-6 receptor did not improve the systemic disease activity even articular involvement or the main symptoms, fatigue, pain and dryness, of participants. We found no change in objective assessments of dryness, but missing data on Schirmer test and unstimulated salivary flow limit the interpretation of these results.
Moreover, the immunological impact of systemic IL-6 inhibition was unexpectedly low. Before this trial, IL-6 was considered one of the cytokines driving B-cell activation in pSS, along with B-cell activating factor (BAFF) and IL-21.11 30 IL-6 receptor inhibition did not improve serum levels of immunoglobulins or complement at week 12, or the biological domain of the ESSDAI at week 24, in contrast with the effect of other biologics such as rituximab or abatacept.8 31 These results suggest that systemic IL-6 is not a main contributor to peripheral B-cell activation in pSS. Altogether, these negative results for clinical outcome, patient-reported outcomes and immunological outcomes indicate that IL-6 does not represent a relevant therapeutic target in pSS, regardless of the concern regarding the high placebo effect observed, which we discuss below. Of note, pSS and systemic sclerosis are diseases in which systemic inflammation is much less prominent than RA and giant cell arteritis. Tocilizumab did not reach its primary outcome in systemic sclerosis either,32 although some secondary endpoints were reached in that trial.
Limitations of the study mainly include the inclusion criteria, restricted to patients with systemic disease activity and the high placebo effect (decrease of at least 3 points in the ESSDAI observed in more than 60% of the placebo-treated patients).
We included only patients with systemic disease activity. This choice was justified by the preference to use a biologic, with potential adverse events, in patients with more active and severe disease. Such inclusion criteria worsened difficulties in recruitment, already well reported in pSS,33 and thus study duration. In addition, as expected, given that disease activity increases the risk of lymphoma,2 lymphoma was diagnosed in four patients during the study. The present study was the first randomised trial registered in ClinicalTrials.gov to evaluate clinical response according to the ESSDAI. Since then, 16 randomised trials have used the ESSDAI as a primary outcome14 16 28 29 34–45 and similar inclusion criteria (ESSDAI≥5 or 6).46 Of note, the placebo response was very high and concerned all domains of the ESSDAI. The high placebo response was concordant with that observed recently in a negative phase III abatacept trial (51% of patients treated with placebo had a decrease of at least 3 points in the ESSDAI at 24 weeks)47 and in two positive phase IIb trials evaluating ianalumab, an B-cell–depleting BAFF-receptor inhibitor (61.2% of patients treated with placebo had a decrease of at least 3 points in the ESSDAI at 24 weeks)48 and iscalimab, an anti-CD40 antibody (55% of patients treated with placebo had a decrease of at least 3 points in the ESSDAI at 12 weeks).49 Remaining to be determined is whether a time-varying decrease in ESSDAI is related to a natural history of the disease (eg, spontaneous improvement of arthralgias/synovitis, purpura or parotid swelling), heterogeneous clinical assessment in multicentre trials, difficulties to discriminate disease activity from damage or the scoring system itself. Using the ESSDAI, disease activity can be very difficult to differentiate from damage. This might lead to inadequately high ESSDAI scores at enrolment. Moreover, during follow-up, if disease activity is considered as stable, it must not be scored as persistently active after 12 months, according to the ESSDAI scoring system.50 This results in a decrease of the ESSDAI in these patients with high ESSDAI score at enrolment. Given the large number of ongoing or future trials that have based their primary outcome on ESSDAI, re-evaluating the use of this score as a primary outcome criterion might be important. NECESSITY, a European initiative, will combine the data from the present trial with those of previous randomised trials to determine new clinical outcomes in pSS. An initiative from OMERACT on clinical outcomes in pSS is also ongoing to progress on this crucial topic.
Among patients with pSS, the use of tocilizumab did not improve systemic involvement or symptoms over 24 weeks of treatment compared with placebo.
Handling editor Josef S Smolen
Contributors All authors contributed to the study by their substantial contribution to the design of the study, acquisition of data and data interpretation. J-EG, LH and NM had access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. All authors read the manuscript. All individuals included in the Acknowledgements section gave their permission to the corresponding author, who confirms that such permission has been obtained in the Authorship Form.
Funding The study was sponsored by Hôpitaux Universitaires de Strasbourg. Roche Chugai provided tocilizumab and the placebo and a grant to fund the study but had no role in the study design, data collection, analysis, interpretation or manuscript preparation, revision or approval of the manuscript. The French patient’s association (Association Française du Gougerot-Sjögren et des Syndromes Secs, AFGS) gave a grant to fund the study.
Competing interests J-EG received honoraries and research grants from BMS and Pfizer, and honoraries from CSL Behring, Lilly, Janssen, UCB, Roche. All other authors declare no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous 3 years, no other relationships or activities that could appear to have influenced the submitted work.
Patient consent for publication Not required.
Ethics approval The protocol was reviewed and approved by the local institutional review board (Comité de Protection des Personnes Est IV; number: 12/30b). The study was conducted according to the current regulations of the International Conference on Harmonisation guidelines and the principles of the Declaration of Helsinki. Informed consent was obtained from all patients.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available upon reasonable request. Data can be requested from the scientific committee of the trial.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.