Rotation or change of biotherapy after TNF blocker treatment failure for axial spondyloarthritis: the ROC-SpA study, a randomised controlled study protocol

Abstract

Introduction

Axial spondyloarthritis (axSpA) is a chronic inflammatory disease characterised by inflammatory low back pain. Non-steroidal anti-inflammatory drugs (NSAIDs) are recommended as a first treatment in axSpA. In case of inadequate response to NSAIDs, biological disease-modifying antirheumatic drugs (bDMARDs) should be introduced according to the recommendations of the European League Against Rheumatism (EULAR) and the American College of Rheumatology. Until 2015, only bDMARD was recommended for axSpA in case of failure to anti-tumour necrosis factor (TNF). The 2022 Assessment of SpondyloArthritis International Society (ASAS)-EULAR recommendation proposed to start an alternative bDMARD but without advocating a switch in mode of action as proposed in rheumatoid arthritis. Since 2015, the inhibition of interleukin (IL)-17 has demonstrated efficacy in axSpA. Then, we designed a randomised multicentre clinical trial to identify the more effective treatment after a first anti-TNF failure in axSpA, comparing an anti-IL-17 to a second anti-TNF.

Methods and analysis

The ROC-SpA (Rotation Or Change of biotherapy after first anti-TNF treatment failure in axSpA patients) study is a prospective, randomised, multicentre, superiority open-label phase IV trial comparing an anti-IL-17 strategy (secukinumab or ixekizumab) to a second TNF blocker in a 1:1 ratio. Patients with an active axSpA (Bath Ankylosing Spondylitis Disease Activity Index >4 or ankylosing spondylitis disease activity score (ASDAS) >3.5) with inadequate 3 months response to a first anti-TNF and with a stable dose of conventional synthetic DMARDs, oral corticosteroids and/or NSAIDs for at least 1 month are included in 31 hospital centres in France and Monaco. The primary outcome is the ASAS40 response at week 24. The secondary outcomes are ASAS40 at weeks 12 and 52, other clinical scores (ASAS20, partial remission rate, ASDAS major improvement rate) at weeks 12, 24 and 52 with the drugs and anti-drugs concentrations at baseline, weeks 12, 24 and 52. The primary analysis is performed at the end of the study according to the intent-to-treat principle.

Ethics and dissemination

Ethics approval was obtained from the committee for the protection of persons (Comité de protection des personnes Ouest IV #12/18_1, 6 February 2018) and registered in ClinicalTrials.gov and in EudraCT. Results of this study, whether positive or negative, will be presented at national and international congresses, to national axSpA patient associations and published in a peer-reviewed journal. It could also impact the international recommendation to manage patients with axSpA.

Trial registration number

NCT03445845 and EudraCT2017-004700-22.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• The ROC-SpA (Rotation Or Change of biotherapy after first anti-TNF treatment failure in axSpA patients) study is the first prospective, randomised, multicentre, superiority open-label phase IV trial comparing two strategies in axial spondyloarthritis (axSpA) with inadequate 3 months response to a first anti-tumour necrosis factor (TNF).

• The primary endpoint was Assessment of SpondyloArthritis International Society 40 at week 24 which is representative to physician expectation.

• We compared anti-interleukin-17 to a second TNF blocker without other strategies as Janus kinase (JAK) inhibitors.

• International recommendations for axSpA management will be updated based on our results.

Introduction

Axial spondyloarthritis (axSpA) is a chronic inflammatory disease characterised by inflammatory low back pain. Heterogeneity of its prevalence reported over the world could be due to the importance of environmental and genetic factors in this disease with the haplotype human leucocyte antigen-B27 which is the most important and the best-known factor. The clinical features of axSpA¹ are inflammatory back pain often due to sacroiliitis and enthesitis. Non-steroidal anti-inflammatory drugs (NSAIDs) are the cornerstone of treatment but do not adequately control inflammation in all patients.

A better knowledge of axSpA pathology and the discovery of the strong involvement of two families of cytokines allowed the development of biological disease-modifying antirheumatic drugs (bDMARDs). Tumour necrosis factor (TNF) was the first cytokine targeted in axSpA.² Since 2003, pharmaceutical companies investigated the efficacy of TNF blockers already used in rheumatoid arthritis (RA). Etanercept is a fusion protein with TNF receptor type II p75 and IgG1 Fc fragment whereas adalimumab, infliximab and golimumab are monoclonal antibodies. Certolizumab is a fusion between a fab fragment targeting TNF and a pegol fraction. All demonstrated efficacy versus placebo in a randomised double-blinded study in radiographic (r)-axSpA, previously named ankylosing spondylitis (AS),3,7 and non-radiographic axSpA,7,10 except for infliximab.

Since 2010, interleukin (IL)-23/17 pathway was suspected to play a role in axSpA physiopathology since the locus of IL-23 receptor was associated with axSpA in genome-wide studies.¹¹ To date, two anti-IL17, secukinumab and ixekizumab, are available on the French market and demonstrated efficiency in randomised double-blinded study versus placebo in AS^{12 13} and nr-SpA.^{13 14}

In clinical trials, response to therapy was assessed by Assessment of SpondyloArthritis International Society (ASAS) scores with ASAS20 and ASAS40 denoting roughly 20% and 40% of improvement, respectively. Overall, the ASAS20 response rate was around 60% for all anti-TNFs. These drugs have now been approved by the European Medicines Agency (EMA) and are currently available in France in case of inadequate response to NSAIDs.^{15 16} Inadequate response is defined by Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) >4 despite treatment. Given that around one-third of patients with axSpA fail to the first anti-TNF within the first 2 years,^{17 18} many patients with axSpA receive a second bDMARDs. Various international societies proposed some recommendations to treat axSpA. In axSpA with persistent high disease activity despite NSAIDs, anti-TNF should be started according to the European League Against Rheumatism (EULAR)¹⁹ and the American College of Rheumatology recommendations.²⁰ In case of failure to a first anti-TNF, the 2022 ASAS-EULAR recommendation²¹ proposed to re-evaluate the diagnosis and to start a new bDMARDs without recommending to switch mode of action as proposed in RA.²² This is in line with the ‘Treat-to-Target’ principle which has not yet been validated in axSpA since the TICOSPA study failed to achieve its primary objective but did show efficacy in some secondary endpoints including ASAS40.²³ This approach already demonstrated its benefit in RA²⁴ or in psoriatic arthritis (PsA).²⁵ This concept was also suggested for axSpA with low levels of evidence.²⁶

Thus, we hypothesised that anti-IL-17 may provide a better clinical response at 24 weeks when compared with a second anti-TNF after the failure of a first anti-TNF in axSpA. In a second step, we hypothesised that anti-IL-17 compared with a second anti-TNF may provide a better clinical response after failure of a first anti-TNF in axSpA at weeks 12 and 52 and a better maintenance of the second bDMARD. We also plan to compare the tolerance, the incremental cost-effectiveness ratio (ICER) and the quality-adjusted life year (QALY) between the anti-IL-17 strategy and the second anti-TNF strategy. The pharmacokinetic parameters of bDMARDs are assessed to find out whether they are predictive of a clinical response at week 24 or associated with low disease.

Methods and analysis

Study design

The ROC-SpA (Rotation Or Change of biotherapy after first anti-TNF treatment failure in axSpA patients) study is a prospective, randomised, multicentre, open-label superiority phase IV trial, comparing two strategies in patients with axSpA with inadequate response to a first anti-TNF: Anti-IL-17 compared with a second anti-TNF. A total of 31 centres over France and Monaco participate in this trial. The sponsor is Centre Hospitalier Universitaire de Saint-Etienne (Délégation à la Recherche Clinique et à l’Innovation). The ROC-SpA study is conducted by the CIC1408 of Saint-Etienne in collaboration with IMIDIATE-CRI (Immune-Mediated Inflammatory Disease Alliance for Translational and Clinical Research part of French Clinical Research Infrastructure Network). The trial is funded by a grant from Programme Hospitalier de Recherche Clinique 2018 (French Ministry of Health). The study is conducted in accordance with the Declaration of Helsinki and has been approved by an ethics committee (Comité de protection des personnes Ouest IV #12/18_1, 6 February 2018). The protocol follows the Consolidated Standards of Reporting Trials and the Standard Protocol Items: Recommendations for International Trails guidelines,^{27 28} and was recorded in ClinicalTrials.gov and in EudraCT. Patients are randomised to receive anti-IL-17A (secukinumab or ixekizumab (since its reimbursement in the French market on September 2020); experimental group) or a second anti-TNF (etanercept, adalimumab, certolizumab, golimumab or infliximab; control group) for 52 weeks with no crossover design. A visit is scheduled for each patient at baseline and at 12, 24 and 52 weeks (figure 1). At each visit, the regular follow-up of patients with axSpA treated with bDMARDs includes BASDAI,²⁹ ankylosing spondylitis disease activity score (ASDAS)^{30 31} assessments and tolerance. In addition to the daily practice, a blood sample (5 mL) is collected and the questionnaires EuroQol,³² Short Form 36³³ and cost effectiveness are performed. To ensure a blinded evaluation of the main criterium ASAS40 response at week 24, the patients are contacted by phone call to fulfil the questionnaires (BASDAI, Bath Ankylosing Spondylitis Functional Index (BASFI)³⁴ and verbal analogue scale). Disease activity is measured by BASDAI and ASDAS. Low disease activity is defined by BASDAI <4 and ASDAS <2.1.

Figure 1. Design of ROC-SpA study in order to guarantee the blindness of the evaluation of the main judgement criterion, a telephone call was made to collect the information needed to calculate it. ASAS, Assessment of SpondyloArthritis International Society; axSpA, axial spondyloarthritis; ROC-SpA, Rotation Or Change of biotherapy after first anti-TNF treatment failure in axSpA patients; TNF, tumour necrosis factor.

Study participants

Patients with AxSpA are recruited during outpatient visits in each centre. The inclusion of patients is performed by a large number of clinicians involved in the IMIDIATE-CRI network. During a visit, a screening examination is performed to evaluate whether the patient is probably eligible for the study. During regular disease assessment of axSpA treated by a first anti-TNF, in case of inadequate response to the first anti-TNF, the physician checks inclusion criteria and exclusion criteria. Patients who fulfil inclusion criteria are informed about the trial and the protocol. If patients agree to participate, a consent form is distributed to them and an inclusion visit is planned after 7–21 days. After written informed consent, the inclusion visit is conducted by a medical doctor and includes a clinical assessment, filling questionnaires to assess the disease activity and the quality of life. Blood samples (5 mL) are collected to measure bDMARDs concentration with a biological bank.

Inclusion and exclusion criteria

The study includes patients aged ≥18 years with active axSpA (BASDAI >4 or ASDAS >3.5) with an inadequate response after at least 3 months to a first anti-TNF and with a stable dose of a conventional synthetic DMARDs, oral corticosteroids and/or NSAIDs for at least 1 month before inclusion. Additionally, patients must give informed written consent (online supplemental file 1) and must be affiliated with a social security scheme and able to complete questionnaires. Exclusion criteria are in table 1.

Table 1. Exclusion criteria of ROC-SpA study.

Due to the biological drugs	Any contraindication to anti-TNF and/or anti-IL-17
	Inflammatory bowel diseases
	Existing pregnancy, lactation, or intended pregnancy within the next 15 months
	Active tuberculosis or other severe infections such as sepsis or opportunistic infections
	Active infections, including chronic or localised infections
	Moderate-to-severe heart failure (NYHA classes III/IV)
General	Impossibility to give informed consent
	Impossibility to be followed for 12 months

ILinterleukinNYHANew York Heart AssociationROC-SpARotation Or Change of biotherapy after first anti-TNF treatment failure in axSpA patientsTNFtumour necrosis factor

Interventions

In this study, the patients included in the control group or experimental group are treated according to the current clinical practice and national recommendations. The experimental group (anti-IL-17) is treated with secukinumab or ixekizumab in compliance with the marketing authorisation regimen. The control group is switched to a second anti-TNF (bio-originator or biosimilar), different from the first anti-TNF used before the inclusion.

Anti-TNF and anti-IL-17 are prescribed in compliance with the marketing authorisation specifications and reimbursement conditions. The treatment is provided by the French Health Insurance in the daily practice. Compliance, batch number, expiration date and the date and time of drug administration are checked by the patients on a booklet. The reasons of non-administration are documented on the electronic case report froms (eCRF).

Randomisation

The randomisation is done centrally by an interactive web response system (IWRS): The IWRS allocates a unique randomisation number to all eligible subjects. The method of sequence generation is a computerised random number generator using permuted block sizes of 2 and 4 in SAS software. Randomisation is stratified by the centre and the cause of failure to the first anti-TNF: primary non-response, secondary non-response and any side effect related to the first anti-TNF. Primary non-response is defined by no improvement during the treatment with a BASDAI over 4 and ASDAS over 2.1 at any time during the first anti-TNF whereas secondary non-response is defined by a flare-up of disease after a phase of improvement with a BASDAI over 4 and ASDAS over 2.1.

Blinding

Patients and physicians are not blinded to the group. However, to assess a blinded assessment, a clinical research assistant (CRA) contacts the patient by phone call to fulfil the questionnaires (BASDAI, BASFI and verbal analogue scale) allowing ASAS40 response determination just before the visit.

Outcomes

Primary outcome

The primary outcome is the ASAS40 response at week 24 after beginning the anti-IL-17 or the second anti-TNF. ASAS40 response is defined as a ≥40% improvement and an absolute improvement from baseline of ≥2 units (range 0–10) in ≥3 of the 4 domains (patient global assessment, spine pain, function represented by BASFI and morning stiffness) without any worsening in the remaining domain.³¹

Secondary outcomes

The secondary outcomes are summarised in the table 2.

Table 2. Secondary outcomes which will be compared between the IL-17 strategy versus a second anti-TNF strategy.

Clinical response	ASAS40 at weeks 12 and 52
	ASAS20 at weeks 12, 24 and 52
	Partial remission rates at weeks 12, 24 and 52
	ASDAS major improvement rates at weeks 12, 24 and 52
Maintenance rates of the TNF and IL-17 blockers at weeks 12, 24 and 52
Tolerance of the treatment	Specific questions at each visit
	Notes on booklet by the patient
Medical direct and indirect costs	Incremental cost-effectiveness ratio (ICER)
Health benefits	Quality-adjusted life years (QALYs) at baseline and every 6 months
	Short Form 36 (SF-36) at baseline and every 6 months
Drugs and anti-drugs concentrations	TNF blockers and anti-drug antibody concentrations at the inclusion
	TNF and IL-17 blockers concentration at weeks 12, 24 and 52

ASAS, Assessment of SpondyloArthritis International Society; ASDAS, ankylosing spondylitis disease activity score; bDMARD, biological disease-modifying antirheumatic drugsILinterleukinTNFtumour necrosis factor

The ASAS20 response is based on the same principle than ASA40 response but with an improvement of at least 20%. Partial remission is defined by values lower than 20/100 in each four criteria (patient global assessment, spine pain, function represented by BASFI and morning stiffness). ASDAS major improvement was defined by a variation of ASDAS-C reactive protein (CRP) ≥2.

Tolerance of the treatment is collected actively (specific questions on the different expected outcomes in the eCRF) at each visit with a focus on any kind of infection. Moreover, patients note on a booklet any side effects during the treatment to reduce the bias of memorisation.

The ICER of the anti-IL-17 strategy versus the second anti-TNF strategy is defined as follows:

Δ(Cost_anti-IL/17 – Cost_{2nd TNF}) / Δ(Efficacy _anti-IL-17 – Efficacy_{2nd TNF})

in which direct and indirect costs will be elicited from the payer (national health insurance) perspective and effectiveness expressed QALY and derived from the EQ-5D-5L using French norms.³⁵

Drugs and anti-drugs concentrations are assessed by specific Lisa Tracker kits from Theradiag (Marne-La-Vallée, France) performed in a central laboratory (Laboratoire d’Immunologie of Saint-Etienne).

Statistical power

Registry data suggest an ASAS40 response of around 30% at 12 weeks of a second anti-TNF.³⁶ We expect in our study an ASAS40 response around 20% at 24 weeks. The ASAS40 response to 24 weeks of secukinumab after failure to a first anti-TNF was around 40%.³⁷ At the time this study was written, no data were available for ixekizumab. However, ASAS response rates from different studies with different designs are not recommended. We therefore hypothesised that the ASAS40 response of patients treated with secukinumab would be 40% after 24 weeks compared with 20% in patients treated with a second anti-TNF. We thus used a superiority model to determine whether the ASAS40 response of patients treated with an anti-IL-17 was superior to the ASAS40 response of patients treated with a second anti-TNF. With these assumptions, it is necessary to include 119 patients with axSpA per group for an alpha risk of 5% and a statistical power of 90%. To compensate for the risk of lost to follow-up, we included 150 patients per group for a total of 300 patients with axSpA.

Data collection and management

Data are recording in an eCRF: a secured electronic system via a web navigator with a control system. The investigator is in charge of the accuracy, quality and relevancy of all collected data. A CRA appointed by the sponsor is in charge of the study, data collection in writing, data recording, data saving and reporting in accordance with the Saint-Etienne CHU Standardised Operating Procedures as well as the Good Clinical Practice guidelines and the in-force legislation and laws. Patients participating in this research are informed of the right to access and rectification of data concerning them. Furthermore, only authorised personnel by the sponsor (investigator, CRA, technician research assistant) and representatives of health authorities have access to this information. People who have direct access to data, all the safety measures are taken to assure the confidentiality of information in regard to the experimental drugs, the trial, the individuals undergoing the trial, the patient’s identity and the obtained results. During the biomedical research, the collected data on the individuals undergoing the trial are anonymised. The sponsor guarantees that each individual undergoing the trial has given his/her informed consent through writing for the right to access their data for the purposes of clinical research quality control. Data are saved on the internal network of the University Hospital in a directory dedicated to the study, a directory accessible only to the statistician. At the end of the research, all documents (case report files, investigator files…) are archived and stored for 15 years in each centre. Once data processing is completed, the computer data is temporarily stored in a library to which access is restricted to only authorised personnel.

Data analysis

A statistical analysis plan will be written blinded to the data. No intermediate statistical analysis was planned. Statistical analysis will be performed on the available information after dataset freezing. The statistical analysis will be performed according to the intention to treat (ITT) principle that is to say on all the randomised patients according to the group to which they will be randomly assigned regardless of what treatment they receive. A sensitivity analysis will be performed on the modified-ITT population which comprises all randomised subjects who received at least one dose of a study medication, if more than 5% of patients do not receive at least one dose. If the rate of missing data is 2% or more, an imputation of missing data will be performed for ITT analysis, according to two methods:

• The multiple imputation method,

• and in the worst case scenario: a patient with missing data on his primary endpoint will be considered like non-respondent (ASAS40) at week 24 for the anti-IL-17 group and respondent the second anti-TNF group.

Qualitative data will be presented as the number and the percentage of patients in each treatment group. Quantitative data will be presented as mean and SD, range, median and IQR by treatment group. The absence of imbalance between the groups will be clinically checked on the demographic and initial variables. No statistical test will be performed, in order to avoid a multiplicity of tests, the differences to be evaluated in clinical terms. All statistical tests will be performed with the use of a two-sided type I error rate of 5%. All analyses will be performed using SAS software (SAS-Windows on computer PC) and R software.

Primary analysis

The second anti-TNF strategy will be considered as the reference strategy and the anti-IL-17 as the alternative one. The primary endpoint will be the proportion of patients with axSpA with a clinical response ASAS40 at week 24. This proportion and the corresponding 95% CI will be estimated by the treatment group. Treatment effect will be estimated by the OR and its 95% CI. For that, an adjusted analysis of the stratification factors will be performed for the primary endpoint with logistic regression (dependent variable will be the clinical response at week 24 and the treatment group and the stratification factors as independent variables). The resulting p value will be reported. In case of imbalance between groups at inclusion, an adjusted analysis on the unbalanced covariate will be performed for the primary endpoint.

Secondary analysis

For analysis of other secondary endpoints, no p value will be done except for safety outcomes, only the 95% CI of the efficacy index will be estimated.³⁸

For qualitative secondary endpoints, treatment effect will be estimated by the OR and its 95% CI using an adjusted analysis on the stratification factors with logistic regression. For quantitative secondary endpoints, the effect of treatment will be estimated by the least-square means difference between the two groups and its 95% CI.

To determine if the first TNF blocker and anti-drug antibody concentrations at inclusion predict the clinical response at 24 weeks, a multivariate logistic regression will be made, with:

• The clinical response at 24 weeks as interest event,

• the first TNF blocker and anti-drug antibody concentrations at inclusion as potential predictors of clinical response at 24 weeks,

• and, the study drug as adjustment factor.

A variable will be considered as an independent risk factor of clinical response when the p value is inferior to 0.05 in the multivariate model.

The same methods will be used to determine if the bDMARDS concentrations and anti-drug antibody concentrations at inclusion will be associated with the low disease activity (ie, BASDAI score strictly inferior to 4 and ASDAS-CRP strictly inferior to 2.1).

Regarding cost-effective comparison, costs will be presented in Euros currents in the form of averages (±SD) and median. The 95% CI will be determined by the bootstrap technique. A sensitivity analysis will be conducted to consider:

• Differences in rates of medical or paramedical consultations by geographic location and conditions of professional practice.

• Differences in valuation of indirect costs (frictional or human capital approach, considering only paid or unpaid activities). The primary cost-utility analysis was conducted using a complete-case analysis sample. We will test differences in costs and outcomes between the two strategies using parametric on non-parametric tests (χ² test, Fisher’s Exact Test for count data, McNemar χ² test and 2-sample Mann-Whitney Wilcoxon test) as appropriate. The incremental ICER will be calculated by dividing the difference in costs by the difference in QALYs between the groups.³⁹

A probabilistic sensitivity analysis will be performed by estimating 5000 ICERs using 5000 samples generated by bootstrap in order to plot a cost-effectiveness plane.³⁹ Sensitivity analyses will also be performed on costs by increasing or decreasing the study drug up to 130% or down to 70% of the current cost.⁴⁰ The impact of missing data on the results will be investigated for costs and outcomes.⁴¹ An imputed ICER will be estimated by means of multiple imputations by chained equations (30 datasets generated).^{42 43}

To help deciding which strategy should be promoted, the probability of each strategy to be cost-effective for different willingness to pay thresholds will be computed.³⁹

Patient recruitment strategies

Recruitment of patients started in June 2018 and finished in July 2023. In total, 31 French recruitment centres were involved: Rheumatologic departments in 31 academic hospitals including 7 departments of Paris Assistance Publique des Hôpitaux de Paris hospitals.

We used several communication channels to communicate about the study (French Society of Rheumatology) and publications in social network and the general media. CRI-IMIDIATE was used to communicate about the study.

Adverse events

Informations about adverse events, serious and unexpected adverse events are declared on eCRF.

The investigator reports to the sponsor immediately any serious adverse event by filling out a specific form of eCRF. The initial announcement can be followed by forwarding all relevant additional information: Within 8 days in case of a fatal or life-threatening event and within 15 days in all other cases. All unexpected serious adverse event (SAE) for which it has not been able to exclude a link with the experimental drug will be sent to (agence nationale de sécurité du médicament et des produits de santé) ANSM by the sponsor. The frequency of expected SAE will be followed by the vigilance of the sponsor. In case of an increase in the frequency of SAEs expected that will be reclassified as unexpected and will be a statement of new fact to the committee for the protection of persons (CPP) and ANSM. The independent data monitoring committee (iDMC) monitored the patients’ safety during the study and gave recommendations to the steering committee. Study treatment could be permanently stopped if any of the following events occur: Safety reason, pregnancy, withdrawal of consent, loss of follow-up, treatment failure, sponsor and/or steering committee following an advice on safety aspects of the study by the iDMC.

In some situations, it might be necessary to interrupt the study medication temporarily. Patients should be encouraged to restart the study drug after a short interruption (1–7 days) except when absolutely contraindicated. In the event that the administration of the study drug is temporarily or permanently discontinued, follow-up, including assessment of any clinical event, will be pursued. The reasons for premature discontinuation of study drugs will be recorded.

Patient and public involvement

There is no patient or public involvement.

Ethics and dissemination

Ethics approval was obtained from the CPP and ANSM. A core information and informed consent form is provided to all patients. Modifications to the protocol is implemented neither by the sponsor nor the investigator without agreement by both parties. However, the investigator may implement a deviation from or a change of the protocol to eliminate an immediate hazard(s) to the trial subjects without prior ethics committee (EC)/institutional review board (IRB)/sponsor approval/favourable opinion. As soon as possible, any implemented deviation or change, reasons for it and if appropriate the proposed protocol amendment should be submitted to the EC/IRB/Bayer. Any deviations from the protocol must be fully explained and documented by the investigator. Any change to the study should be written and filed as an amendment to this protocol. Results of this study, whether positive or negative, will be presented at national and international congresses, to national axSpA patient associations and published in a peer-reviewed journal.

Discussion

This is the first randomised study evaluating the efficacy of an anti-IL-17 compared with a second anti-TNF after a failure of a first anti-TNF in axSpA. The present objective is to investigate whether one of these strategies might be more effective in a pragmatic trial, in a current practice setting. Current trials in axSpA performed by pharmaceutical companies generally aim to extend the indications of a particular drug. We chose a subset of patients with an inadequate response to a first anti-TNF. We expect that anti-IL-17 will provide a better clinical response at 24 weeks when compared with a second anti-TNF, after failure of a first anti-TNF in axSpA.

We did not include in our trial anti-IL-23 and JAK inhibitor. At the time of the design of the study, we expected a putative efficacy of ustekinumab, a fully human monoclonal antibody targeting the subunit p40 sharing by IL-12 and IL-23. This bDMARDs was first developed and approved by EMA for PsA.^{44 45} In a post hoc analysis including only patients with inflammatory back pain observed an efficacy assessed by BASDAI 50 response at 30% with ustekinumab versus 11% with placebo.⁴⁶ Despite this result, three randomised control studies failed to demonstrate the efficacy of ustekinumab in axSpA.⁴⁷ Recently, guselkumab, a fully human monoclonal antibody targeting the subunit p19 IL-23 showed an improvement in axial symptoms of PsA.⁴⁸ A specific clinical trial is ongoing to assess the guselkumab efficacy in active PsA axial disease by assessing reduction in axial symptoms and inflammation (NCT04929210).⁴⁹ In another hand, efficacy of upadacitinib, a JAK inhibitor, was demonstrated since 2019^{50 51} with the approbation of the drug in 2023.

Limited data exist on secukinumab efficacy in case of anti-TNF failure but a trend for a lower efficacy has been observed (ASAS20 response at 68% vs 50% at 16 weeks).¹² To date, few studies compared the efficacy between secukinumab and an anti-TNF. A randomised head-to-head study showed no efficacy difference between secukinumab and an anti-TNF inhibitor in naïve patients with AS.⁵² But no randomised study has been conducted in patients with axSpA after a first anti-TNF failure. A comparative study has reported a comparable efficacy between secukinumab and an anti-TNF after a first anti-TNF failure in patients with axSpA.⁵³ Another study first showed no difference of efficacy between secukinumab and an anti-TNF after a first anti-TNF failure.⁵⁴ However, when dividing patients into two groups with primary and secondary non-responders to anti-TNF, the second anti-TNF showed a better efficacy than secukinumab in secondary non-responders.⁵⁵ But these studies are observational, limited by confounding factors, selection biases and missing data. For this reason, a randomised trial is needed to determine whether there is a difference in efficacy between secukinumab and an anti-TNF after first anti-TNF failure.

In 2005, EMA proposed to use ASAS40 response criteria at 12 or 24 weeks in clinical trials. Accordingly, clinical trials assess short efficacy of new drugs focused mainly on ASAS20 response at 12 or 14 weeks. In the daily practice, physicians expect to have a durable clinically significant response. In our study, response to the ASAS40 (primary objective) is assessed at week 24 to be more representative of patient and physician expectations, especially as there is no placebo group.

Analysis of efficiency from registries needs to be interpreted carefully due to many biases inherent in this kind of study. Interpretation of efficacy by indirect comparison of randomised phase III studies is also difficult due to different population at inclusion with some bias. In our study, all drugs are self-injected subcutaneously in order to avoid the bias based on treatment adherence. To reduce the assessment bias due to the open label, a blinded nurse expert in axSpA assessment contacts by call 2 weeks before the primary outcome measurement to assess perform questionnaires allowing ASAS criteria calculation. Concentration of bDMARDs and anti-drugs antibody performed in a central laboratory (Laboratoire d’Immunologie, CHU Saint-Etienne) in a blinded manner excepted for the treatment to assess. To reduce confusion bias, stratification is performed according to the centre and the reason to stop the first anti-TNF (primary or secondary failure, side effect). To reduce the memorisation bias, all patients note every infection to a booklet. To stay in a pragmatic approach, the second anti-TNF in the control group or the anti-IL-17 in the experimental group was chosen by the investigator who decided on the best suitable treatment for the patient after discussing with him or her.

As all new therapies or strategies, the balance risk-benefit to the patient needs to be evaluated. With an anti-TNF, serious side effects can occur including serious infections with pulmonary and extrapulmonary tuberculosis, pulmonary sepsis (including Legionella), skin and joints infections as well as excess of some cancers (in particular skin cancers).56,58 With an anti-IL-17, data on longstanding safety are limited. During phase III study, some candida infections without systemic form were reported.¹²

This pragmatic study will therefore offer opportunity to identify the best strategy in patients with axSpA in failure to a first anti-TNF.

Trial status

The recruitment started in June 2018 and ended in July 2023. The trial will end in July 2024, 52 weeks after the last inclusion.