Article Text

Effectiveness of secukinumab versus an alternative TNF inhibitor in patients with axial spondyloarthritis previously exposed to TNF inhibitors in the Swiss Clinical Quality Management cohort
  1. Raphael Micheroli1,
  2. Christoph Tellenbach1,2,
  3. Almut Scherer2,
  4. Kristina Bürki1,
  5. Karin Niederman3,
  6. Michael J Nissen4,
  7. Pascal Zufferey5,
  8. Pascale Exer6,
  9. Burkhard Möller7,
  10. Diego Kyburz8,
  11. Adrian Ciurea1
  1. 1 Department of Rheumatology, Zurich University Hospital, Zurich, Switzerland
  2. 2 Statistics Group, Swiss Clinical Quality Management Foundation, Zurich, Switzerland
  3. 3 School of Health Professions, Institute of Physiotherapy, Zurich University of Applied Sciences, Winterthur, Switzerland
  4. 4 Department of Rheumatology, University Hospital, Geneva, Switzerland
  5. 5 Department of Rheumatology, Centre hospitalier universitaire vaudois, Lausanne, Switzerland
  6. 6 Praxis Rheuma-Basel, Basel, Switzerland
  7. 7 Department of Rheumatology, Clinical Immunology and Allergy, University Hospital of Bern, Bern, Switzerland
  8. 8 Department of Rheumatology, University Hospital Basel, Basel, Switzerland
  1. Correspondence to Professor Adrian Ciurea, Rheumatology, University Hospital Zurich, 8091 Zurich, Switzerland; adrian.ciurea{at}


Objective To compare effectiveness of treatment with secukinumab (SEC) with that of alternative tumour necrosis factor inhibitors (TNFis) in patients with axial spondyloarthritis (axSpA) after withdrawal from one or more TNFis.

Methods Patients diagnosed as having axSpA in the Swiss Clinical Quality Management cohort were included if they had initiated SEC (n=106) or an alternative TNFi (n=284) after experiencing TNFi failure. Drug retention was investigated with matching weights propensity score (PS) analyses and multiple adjusted Cox proportional hazards models. Matching weights PS-based analyses and multiple-adjusted logistic regression analyses were used to assess the proportion of patients reaching 50% reduction in the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI50) at 1 year.

Results SEC was more often used as third-line or later-line biological drug (76% vs 40% for TNFi). Patients starting SEC had higher BASDAI, Bath Ankylosing Spondylitis Functional Index, Bath Ankylosing Spondylitis Metrology Index and C reactive protein levels. A comparable risk of drug discontinuation was found for SEC versus TNFi (HR 1.14, 95% CI 0.78 to 1.68 in the PS-based analysis and HR 1.16, 95% CI 0.79 to 1.71 in the multiple-adjusted analysis). No significant difference in BASDAI50 responses at 1 year was demonstrated between the two modes of biological drug action, with CI of estimates being, however, wide (OR for SEC vs TNFi 0.76, 95% CI 0.26 to 2.18 and 0.78, 95% CI 0.24 to 2.48 in the PS-based and the covariate-adjusted model, respectively).

Conclusion Our data suggest a comparable effectiveness of SEC versus an alternative TNFi after prior TNFi exposure.

  • ankylosing spondylitis
  • anti-TNF
  • DMARDs (biologic)
  • spondyloarthritis
  • treatment

Statistics from

Request Permissions

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.

Key messages

What is already known about this subject?

  • Secukinumab (SEC), an interleukin 17A inhibitor (IL-17i), has been shown to be effective for the treatment of radiographic axial spondyloarthritis (axSpA) and to provide a therapeutic alternative to tumour necrosis factor inhibitors (TNFis).

  • Current international treatment recommendations mention TNFi or IL-17i as the first biological disease-modifying antirheumatic drug (bDMARD) option, as well as in the context of previous TNFi failure.

What does this study add?

  • In this real-life cohort study comparing two treatment options after prior withdrawal from at least one TNFi, patients treated with SEC experienced comparable outcomes as patients treated with an alternative TNFi when assessing drug retention and the proportion of patients reaching a 50% reduction in the Bath Ankylosing Spondylitis Disease Activity Index at 1 year.

How might this impact on clinical practice or future developments?

  • This first comparative study of two modes of biological action in axSpA corroborates current international treatment recommendations.

  • Head-to-head randomised, controlled trials of bDMARDs with different modes of action are needed.


Tumour necrosis factor inhibitors (TNFis) have been the treatment of first choice for patients with active axial spondyloarthritis (axSpA) and particularly ankylosing spondylitis (AS) after insufficient response to non-steroidal anti-inflammatory drugs (NSAIDs) for more than a decade.1 More recent data have demonstrated a comparable effect of inhibition of interleukin 17 (IL-17) by either secukinumab (SEC) or ixekizumab in axSpA.2–4 In the absence of head-to-head comparisons between these two modes of action in axSpA, current recommendations do not favour TNFi over anti-IL17 inhibitors (IL-17i) as first biological disease-modifying antirheumatic drugs (bDMARDs).5 The recommendations stipulate that, if TNFi therapy fails, switching to another TNFi or to an IL-17i should be considered. Data favouring one approach over the other are, however, lacking. There is ample evidence that a second TNFi may be efficacious in axSpA, although to a lower level than in first-line therapy.6 Furthermore, IL-17 inhibition was less effective in patients having failed TNFi in comparison to TNFi-naive patients.7 We sought to compare effectiveness of a subsequent TNFi versus SEC—the first IL-17i approved for AS—in patients with axSpA with prior exposure to TNFi in a large observational cohort.


Study population

We used data from a large observational cohort of patients with a clinical diagnosis of axSpA followed in the Swiss Clinical Quality Management (SCQM) registry.8 Inclusion criteria were (1) prior treatment with at least one TNFi, (2) start of either a TNFi or SEC after inclusion and (3) availability of a baseline visit with clinical assessments according to the Assessment in SpondyloArthritis international Society (ASAS) handbook.9 As patients with non-radiographic axSpA are also treated with SEC, all patients with a clinical diagnosis of axSpA were included, irrespective of the approval status of individual bDMARDs in Switzerland and worldwide. The study period was from February 2015, the date of the first recorded SEC start in SCQM up to the database closure on the 31 December 2019.


The predefined primary outcome of interest was the duration of drug retention with SEC compared with TNFi in TNFi-exposed patients. Interruptions of treatment with the same drug of less than 3 months were ignored. Ongoing treatments were censored at the last visit. As SEC was approved for axSpA much later than TNFi and potentially used more often after multiple bDMARD failures, a sensitivity analysis was performed including only patients treated with a second or third line bDMARD.

The secondary outcomes of interest were the achievement of the Bath Ankylosing Spondylitis Disease Activity Index criteria for 50% improvement (BASDAI50) at 1 year (±6 months), the proportion of patients reaching an Ankylosing Disease Activity Disease Activity Score (ASDAS) <2.19 as well as the ASAS improvement criteria for bDMARD continuation (∆BASDAI ≥2 and ∆ASDAS ≥1.1, respectively) at the same time point. ASDAS was calculated using C reactive protein levels.10 Patients having stopped the drug before the outcome assessment were considered as non-responders (response/tolerance analyses).11 The outcomes were also assessed in the subset of patients still on drug at 1 year (completer analyses).

Statistical analyses

Baseline characteristics between patients starting SEC versus TNFi were compared by using the Fisher’s exact test for categorical variables and the Mann-Whitney U test for continuous variables. All tests were two-sided, with a significance level of 0.05. All statistical analyses were performed in the R statistical software (R Development Core Team, 2018, V.3.5.0).

Multiple imputation of missing values

To account for missing baseline covariate values, multiple imputation by chained equations was used12 and was implemented separately for drug retention and response analyses. Full details about the variables included in the imputations are provided in the online supplementary table S1. Since 85% of patients had one or more missing covariates, 85 imputations each with 20 iterations were performed. Predictive mean matching was used to impute continuous variables and logistic regression for binary variables. Individual components of outcome variables (ASDAS) were used for the imputation of the completer and response/tolerance analyses. Convergence of imputations was assessed by visual inspection of the mean and variance changes by iteration and dataset. Pooling of model estimates was performed according to Rubin’s rules. The MICE package version 2.46.0 was used for the imputation.

Propensity score (PS) analyses

Given the observational nature of the dataset used for this comparative effectiveness study between SEC and TNFi, we used PS-based analyses to take potential confounding by indication into account.13 The PS, ranging from 0 to 1, indicates the per patient likelihood of being treated with SEC, given measured patient baseline covariates that may influence treatment allocation. After careful evaluation of different PS approaches, we chose a matched-weight approach over a 1:1 matching approach as it has been shown to potentially have better statistical properties.14 Also, it achieved a better balance of covariates between the two treatments in our datasets in preliminary analyses. In PS-weighted analyses, higher weights are given to observations of patients with more similar probabilities of receiving SEC and TNFi (PS close to 0.5). This approach has the advantage of not excluding observations that cannot be matched. Such observations are included but receive a much lower weight. The bias introduced by arbitrary excluding subjects is therefore reduced.

The following steps were performed to calculate, evaluate and use the PS.13 14 First, 16 covariates were selected based on expert knowledge in terms of potentially influencing treatment effectiveness or both treatment assignment and effectiveness: sex, age, human leucocyte antigen B27 (HLA-B27), ASDAS, Bath Ankylosing Spondylitis Functional Index, Bath Ankylosing Spondylitis Metrology Index, body mass index, presence of enthesitis and of peripheral arthritis, ever psoriasis, uveitis and inflammatory bowel disease (IBD), ASAS classification status as non-radiographic versus radiographic axSpA, current smoking, reason for discontinuation of the previous TNFi (insufficient response vs other reasons) and the number of previous bDMARDs used (either as a continuous variable or as a dichotomous variable (second line vs later line)). Second, standardised mean differences (SMD) of all covariates in SEC versus TNFi treated patients were visualised for all 85 multiply imputed datasets (SMD being the difference in sample mean of a covariate between the two treatment groups, standardised by their respective sample variances). Variables were considered unbalanced if the SMD exceeded 10%.15 Third, binomial PS were fitted for each imputed dataset by logistic regression with treatment category as a dependent variable and the parameters described above as independent covariates. The PS were transformed thereafter to matching weights.14 The level of balance between treatment groups in terms of SMD was rechecked after applying matching weights, and the areas of common support of summed matching weights were visualised for the TNFi and SEC groups, respectively, in mirrored histograms.14 Matching weights PS were implemented for each effectiveness analysis, as baseline populations differed in retention, response/tolerance and completer analyses in the whole axSpA population as well as in the subpopulation fulfilling the ASAS classification criteria. HRs for discontinuation of SEC versus TNFi as well as ORs for different response outcomes were estimated before and after applying the PS matching weights. If SMDs of any variable exceeded 10% in a multiply imputed dataset after weighting, an additional analysis was performed discarding the individual datasets not meeting these criteria. Importantly, every one of the 85 imputed datasets contains data for all patients and hence exclusion of one or more imputed datasets (if necessary) did not exclude individual patients from the analyses. The results from the consecutive steps of the matching weights PS analyses are described in detail in the online supplementary material.

Covariate adjustment

We used covariate adjustment as an alternative method to take into account confounding by indication.16 The same 16 covariates were introduced in a multiple adjusted Cox proportional hazard model to compare drug retention of SEC versus TNFi as well as in logistic regression analyses to estimate the ORs for different response outcomes in patients treated with SEC versus TNFi in response/tolerance and completer analyses.

Indirect patient and public involvement

The database used in this study was developed by a committee that included patient representatives.


Baseline characteristics of patients diagnosed as having axSpA (n=390) fulfilling the inclusion criteria are shown in table 1 (n=284 for alternative TNFi and n=106 for SEC). The proportions of patients fulfilling the ASAS axSpA classification criteria and of HLA-B27 positivity, as well as sex distribution were similar in the two groups. However, a significantly higher number of patients initiating SEC had previously been treated with at least two different TNFi (76.4% vs 40.1%, p<0.001). Accordingly, patients starting SEC had higher baseline disease activity levels and a greater impairment in physical function and mobility. Some numerical differences were observed with regard to the proportions of patients presenting with radiographic sacroiliitis, peripheral manifestations and extramusculoskeletal manifestations (table 1).

Table 1

Baseline characteristics of patients with axSpA initiating treatment with SEC or an alternative TNFi after previous exposure to one or more TNFis

Drug retention

The proportion of patients who stopped the studied bDMARD for insufficient effectiveness, as well as for other reasons was similar in the SEC and the TNFi groups (table 2). In patients with known nature of adverse events, no cases of uveitis were reported, two patients discontinued SEC because of diarrhoea (lymphocytic colitis on biopsy in one patient) and three patients discontinued the bDMARD because of paradoxical psoriasis (one on TNFi and two reactivated while on SEC—having occurred initially during previous TNFi treatment).

Table 2

Reason for discontinuation of the bDMARD in use

After adjusting for potential confounding factors as well as for parameters potentially influencing the choice of treatment, the results revealed a comparable hazard of discontinuation for SEC versus TNFi (HR 1.14, 95% CI 0.78 to 1.68, p=0.50 and HR 1.16, 95% CI 0.79 to 1.71, p=0.45 in PS-based and multiple-adjusted models, respectively). The results for the whole axSpA population as well as for patients fulfilling the ASAS classification criteria are shown in table 3. Similar results were found in a sensitivity analysis including only patients treated with a second-line or third-line bDMARD for both types of analysis (PS-based and multiple-adjusted models; table 4).

Table 3

Cox proportional hazards model for analysis of drug retention of secukinumab versus an alternative TNFi after prior exposure to one or more TNFis

Table 4

Sensitivity analysis: Cox proportional hazards model for analysis of drug retention of secukinumab versus an alternative TNFi after prior exposure to only one or two TNFis

Clinical response

A total of 302 patients (77.4%) had been treated for at least 1 year and had an available follow-up visit within the requested interval (214 patients (75.4%) initiating a next TNFi and 88 patients (83.0%) initiating SEC) to allow for the evaluation of ‘response/tolerance’. Baseline characteristics of these patients at start of the biologic are shown in table 5. The mean±SD length of follow-up was comparable between the two groups (11.5±2.9 vs 11.5±3.2 for alternative TNFi and SEC, respectively). BASDAI at follow-up was available in 225/302 patients (74.5%) with a comparable proportion in the SEC and TNFi groups (70.5% and 76.2%, respectively). No significant difference in BASDAI50 on treatment with SEC versus TNFi was found in PS-based analyses for both response/tolerance and completer analyses (OR 0.76, 95% CI 0.26 to 2.18, p=0.60 and OR 0.70, 95% CI 0.20 to 2.46, p=0.58, respectively). Similar results were obtained in multiple-adjusted analyses (OR 0.78, 95% CI 0.24 to 2.48, p=0.67 and OR 0.68, 95% CI 0.17 to 2.71, p=0.59 for response/tolerance and completer analyses, respectively). No significant differences and broad 95% CI were also observed for all secondary outcomes (proportion of patients improving by at least 2 BASDAI points or at least 1.1 ASDAS points, as well as the percentage of patients achieving an ASDAS <2.1) and both for the whole axSpA population and for patients fulfilling the ASAS classification criteria (figure 1).

Figure 1

Comparison of response rates at 1 year of treatment with secukinumab (SEC) versus an alternative tumour necrosis factor inhibitor (TNFi) for different outcome measures, taking into account potential confounding by indication with two different methods. ORs for treatment with SEC versus an alternative TNFi and 95% CIs are shown for different outcomes in propensity score-based models (black) and multiple-adjusted models (grey). The vertical line represents an OR of 1. (A) Response/tolerance analyses: response in patients with available outcome at 1 year, patients having discontinued the biologic in the meantime being considered non-responders and (B) completer analyses: response in patients still treated with the initial biologic at 1 year. ASAS, Assessment in SpondyloArthritis international Society; ASDAS <2.1, achievement of an Ankylosing Spondylitis Disease Activity Score <2.1, representing low disease activity; ∆ASDAS ≥1.1, improvement in ASDAS of at least 1.1 points, representing a clinically important improvement; axSpA, axial spondyloarthritis; BASDAI50, 50% improvement in the Bath Ankylosing Spondylitis Disease Activity Index; ∆BASDAI ≥2, improvement in BASDAI of at least two points.

Table 5

Baseline characteristics of patients initiating treatment with SEC or an alternative TNFi after prior exposure to one or more TNFis with available follow-up visit at 1 year for different response analyses


The results of our drug retention analyses in patients with axSpA with prior TNFi exposure suggest a comparable survival of SEC versus a further TNFi. The analyses have been adjusted for baseline characteristics known to affect treatment response as well as for other parameters potentially affecting the choice of treatment. SEC is, in contrast to the majority of TNFi, not yet approved for the treatment of non-radiographic axSpA, and differential clinical responses have been demonstrated with regard to extramusculoskeletal manifestations between the two modes of action (SEC being superior to TNFi for the treatment of psoriasis, of unclear relevance for uveitis and not effective/counterproductive for IBD).17–19 Both PS and conventional covariate adjustment methods can be used to adjust for potential confounding in observational studies and selection of the most suitable method remains a subject of debate.16 While we had not intended to compare both techniques, reaching very similar results through both approaches adds to the robustness of the findings.

We were not able to compare retention of SEC versus TNFi as first-line bDMARD, as only 14 TNFi-naive patients in SCQM initiated SEC during the study period. It is known that in comparison to first-line bDMARD therapy, clinical benefit is mitigated in later-line treatments.6 Given important differences in the proportion of patients with previous exposure to two or more bDMARDs at baseline in the SEC versus the TNFi group, the analysis was adjusted for the number of previously used biologics. Inclusion of patients with previous failure of one or more TNFis also affected the sex distribution of our overall population (the proportion of 57% females seeming unusual for axSpA). As response to TNFi is lower in women than in men,20–23 the female gender is enriched in a population with previous unsuccessful exposure to TNFi. While there was no evidence for a differential impact of the reason for discontinuation of the previous TNFi on retention of the two bDMARDs with different modes of action, a potential effect of primary versus secondary ineffectiveness could not be assessed.

Drug retention was used as the primary outcome in our comparative analyses of treatment with SEC versus TNFi in patients having previously failed TNFi, as BASDAI levels to evaluate response outcomes were only available in approximately 75% of the follow-up visits at 1 year (either due to limited follow-up time or due to a missing assessment around 1 year after treatment start). The point estimates of BASDAI50 and ASDAS <2.1 outcomes 1 year after start of treatment were very close to 1, indicating no difference between SEC and TNFi. However, given rather wide CIs the quest for additional data, particularly randomised controlled head-to-head trials, is warranted.

Although SEC was so far only approved for AS,2 it is used in Switzerland in patients diagnosed as having axSpA, irrespectively of the fulfilment of modified New York classification criteria. This was also the case for TNFi before their approval for non-radiographic axSpA, not only in Switzerland8 but also in the USA.24 The question arises whether at least part of the patients not fulfilling the ASAS classification criteria might have been misdiagnosed as having axSpA. Although this cannot be formally excluded, the results were confirmed in the population of patients fulfilling the ASAS classification criteria. Moreover, the proportion of patients reaching an ASDAS <2.1 found here in the response/tolerance and completer analyses in the TNFi group was similar to the outcome achieved after switching TNFi in an SCQM population fulfilling the ASAS axSpA classification criteria published previously.11 The number of patients with multiple TNFi failures was higher in patients starting SEC with potentially an increased risk of misdiagnosis in this group. While it remains unclear whether our statistical adjustments were able to completely compensate for this imbalance, our main results were confirmed in a sensitivity analysis of second- and third-line treatments only.

The observational nature of our study represents its major limitation. Therefore, although we have adjusted our analyses for a multitude of confounders, we cannot exclude residual confounding. In particular, the exact date of MRI information on sacroiliac joint inflammation was not available to allow for correlation of potential differential inflammatory changes at start of TNFi versus IL-17i. Moreover, the issue of concomitant fibromyalgia not responding to immunosuppressive agents confounding our results could not be addressed. With regards to a potential bias by treatment in different calendar periods,25 we have only included patients starting a subsequent TNFi after approval of SEC in Switzerland.

In conclusion, these longitudinal data from a real-life axSpA cohort suggest that after TNFi exposure, switching to SEC is comparably effective to switching to another TNFi. Head-to-head randomised trials would be particularly useful to confirm these results and to potentially identify specific patient subsets who may particularly benefit from switching to IL-17 inhibition.


We thank all patients and their rheumatologists for participation and the entire SCQM staff for data management and support. A list of rheumatology private practices and hospitals that are contributing to the SCQM registries can be found on:


Supplementary materials

  • Lay summary

    Disclaimer : This is a summary of a scientific article written by a medical professional (“the Original Article”). The Summary is written to assist non medically trained readers to understand general points of the Original Article. It is supplied “as is” without any warranty. You should note that the Original Article (and Summary) may not be fully relevant nor accurate as medical science is constantly changing and errors can occur. It is therefore very important that readers not rely on the content in the Summary and consult their medical professionals for all aspects of their health care and only rely on the Summary if directed to do so by their medical professional. Please view our full Website Terms and Conditions.
    Copyright © 2020 BMJ Publishing Group Ltd & European League Against Rheumatism. Medical professionals may print copies for their and their patients and students non commercial use. Other individuals may print a single copy for their personal, non commercial use. For other uses please contact our Rights and Licensing Team.


  • Handling editor Josef S Smolen

  • RM and CT contributed equally.

  • Contributors Study conception and design: RM, CT, AS, AC. Acquisition of data: RM, KB, KN, MJN, PZ, PE, BM, DK, AC. Statistical analysis: CT, AC, AS. All authors contributed to the interpretation of the data. AC wrote the manuscript. All authors critically revised the manuscript. All authors revised and approved the final manuscript to be published.

  • Funding The SCQM Foundation is supported by the Swiss Society of Rheumatology and by AbbVie, Celgene, iQone, Lilly, Merck Sharp & Dohme, Novartis, Pfizer, Roche, Samsung Bioepis, Sandoz, Sanofi Genzyme and UCB. This study was supported by a research grant from Novartis.

  • Competing interests AC has received consulting and/or speaking fees from AbbVie, Celgene, Eli Lilly, Janssen-Cilag, Merck Sharp & Dohme, Novartis, Pfizer and UCB. MJN has received consulting and/or speaking fees from Abbvie, Celgene, Eli Lilly, Novartis and Pfizer.

  • Patient and public involvement Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.

  • Patient consent for publication Not required.

  • Ethics approval The study was approved by the ethics committee of the Kanton of Zurich and written informed consent was obtained from all patients.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement All data relevant to the study are included in the article or uploaded as supplementary information. All data relevant to the study are included in the article or uploaded as supplementary information.