Withdrawal of medical in axial spondyloarthritis: what would be the optimal trial design? U. Kiltz, X. Baraliakos, J. Braun, D. van der Heijde

Rheumazentrum Ruhrgebiet, ABSTRACT bowel disease (1). Ankylosing spondy- St. Josefs-Krankenhaus, Herne, Germany. Remission or low disease activity is litis (AS) is the main subgroup of SpA Uta Kiltz, MD achievable in patients with axial spon- and is characterised by established Xenofon Baraliakos, MD dyloarthritis (SpA), and remission radiographic changes in the sacroiliac Jürgen Braun, MD has been defined as one of the main joints (SIJ) according to the modified Désiréé van der Heijde, MD, PhD targets in treating patients with axial New York criteria (2). Because radio- Please address correspondence to: SpA. However, it is unclear what ac- graphic damage reflects chronic struc- Dr Uta Kiltz, tions should be taken once remission Rheumazentrum Ruhrgebiet, tural changes, these criteria have con- St. Josefs-Krankenhaus, has occurred. Very little data are avail- siderable limitations in early disease Landgrafenstraße 15, able concerning the effect of TNF in- stages. Therefore, new classification 44652 Herne, Germany. hibitors (TNFi) dosage adjustment or criteria for axial and peripheral SpA E-mail: [email protected] on withdrawal strategies in patients have been recently developed (3, 4). Received on August 19, 2013; accepted in with axial SpA and/or in patients with The Assessment of SpondyloArthritis revised form on August 26, 2013. (AS). Most is- international Society (ASAS) classifi- Clin Exp Rheumatol 2013; 31 (Suppl. 78): sues relating to withdrawal of treat- cation criteria for axial SpA require a S47-S50. ment in patients who are in remission history of chronic back pain ≥3 months © Copyright Clinical and cannot be addressed with traditional and an age at onset <45 years as en- Experimental Rheumatology 2013. randomised placebo-controlled trials try criteria. Next, either on (RCT). Facing these challenges, there x-rays or MRI in addition to at least Key words: ankylosing spondylitis, is a need for conducting trials with an one typical SpA feature or presence of spondyloarthritis, remission, study innovative trial design to reflect real- HLA-B27 in addition to at least two design life practice. Possible strategies upon typical SpA features need to be present. remission include continuation, dose Using this set of criteria, two groups of reduction or withdrawal of the effective patients can be classified:a ) having es- . Future scenarios should rec- tablished radiographic changes in the ognise heterogeneity in patients with SIJ, i.e. classified as AS; or b) having axial SpA, which makes it questionable not developed radiographic changes whether different trial designs will be in the SIJ, i.e. classified as non-radio- applicable for the whole group of axial graphic axial SpA (nr-axSpA). At this SpA. Several questions should be ad- time, however, most of the knowledge dressed before conducting a trial to about disease course and effects of study remission in patients with axial treatment options is available for pa- SpA: definition of remission (clinical tients with AS. and/or imaging remission), duration of The disease course of axial SpA is remission as a defining inclusion crite- characterised by ongoing axial inflam- rion, predictors of remission, definition mation and radiographic progression of subgroups (e.g. TNFi naïve patients associated with restricted mobility of or patients who will most likely remain the spine and decreased function (5). in remission), when to restart and fi- Treatment with non-steroidal anti-in- nally dose-adjustment after restart of flammatory drugs (NSAIDs) and exer- the therapy. cise are recommended as first line ther- apies in patients with axial SpA (6). For Introduction those patients who have persistent ac- Disease course and remission in tive disease, the introduction of tumour Competing interests: U. Klitz has received patients with axial spondyloarthritis necrosis factor inhibitors (TNFi) was consultancy fees from AbbVie and MSD; The concept of spondyloarthritis (SpA) a major advance in the management X. Baraliakos has received consultancy fees from AbbVie, Amgen, Janssen, MSD, characterises a disease group with of axial SpA (7). The good clinical re- Novartis, and Pfizer; chronic spinal inflammation and - ex sponse to this therapy in the majority of the other co-authors have declared no traspinal manifestations such as arthri- patients correlates with improvements competing interests. tis, enthesitis, uveitis and inflammatory in physical function and health-related

S-47 Withdrawal of medical therapies in SpA / U. Kiltz et al. quality of life (HRQoL) (8, 9). In many the threshold for major improvement, the pragmatic trial (also called strat- trials a 20–50% reduction in disease ac- a validated tool has been endorsed to egy or comparative effectiveness trial) tivity could be demonstrated in around assess the degree of improvement in (25). The explanatory approach seeks 60% of patients (10). Moreover, it has individual patients. to understand a biological process by been shown that remission in patients Although the field of remission in pa- testing the hypothesis that the speci- with SpA is achievable. The rate of tients with axial SpA is growing, sev- fied biological response is explained remission depends on the patient char- eral questions cannot yet be answered, by exposure to a particular treatment. acteristics such as duration of disease, due to lack of evidence. At the mo- By contrast, the pragmatic approach levels of inflammatory markers, and ment, there is no agreement about the seeks to compare two treatments under functional status – but studies of this best definition of remission for patients the conditions in which they would be topic are limited to date (11). with axial SpA. Although the ASAS applied in practice. In contrast to (RA), definition of partial remission often is Most of the issues relating to with- clinical remission and low disease ac- used, its applicability in clinical prac- drawal of treatment in patients who tivity has until recently not been de- tice is limited. A major possible limi- are in remission cannot be addressed fined very well in SpA. Clinical remis- tation is the fact that physical function with explanatory randomised placebo- sion/inactive disease should be a major is included. Consequently, patients controlled trials (RCT). There is little treatment target, as has been defined in with inactive but longstanding disease argument that the classic parallel RCT the treat-to-target (T2T) recommenda- are not labelled as fulfilling the ASAS with placebo remains the gold standard tions for patients with axial SpA (12). partial remission criteria. However, for establishing the efficacy and short- Frequently, the term ‘partial remission’ the T2T initiative for SpA defined re- term safety of an experimental agent. is used as defined by the ASAS remis- mission as the combination of a low But since information from an explana- sion criteria, i.e. as defined by the do- BASDAI and normal CRP or inactive tory trial often is unlikely to inform a mains of patient global, pain, function ASDAS (ASDAS cut-off <1.3) (12). pragmatic question, RCTs need to be and morning stiffness (13). When these This was based on the notion that nor- adapted for choosing between treat- criteria are applied, remission can be malisation of inflammation is impor- ment options (withdrawal or dose re- achieved in about 12–15% of patients tant. Furthermore, it is not clear for duction strategies). with AS treated with NSAIDs, in about how long the patients should be in re- Clinical trials with an innovative trial 25% of patients with AS treated with mission before stoping a treatment. design to reflect real-life practice are TNFi and in about 50% of patients with needed to improve knowledge concen- early axial SpA treated with TFNi (14- Controlled trials ing possible withdrawal of therapies in 18). While remission is possible in patients AS. In patients with juvenile idiopathic In addition to clinical remission, nor- with axial SpA, it is unclear what ac- arthritis (JIA), a RCT with a withdrawal malisation of inflammatory markers tions should be taken once remission design is often used to test a new drug. or complete clearance of magnetic has occurred. Possible strategies include In such trials, eligible patients are resonance imaging (MRI) inflamma- continuation of therapy without change, treated in an open-label manner with tion (‘imaging remission’) might be dose reduction by longer intervals of the experimental therapy to be tested in an important outcome in treatment lower doses, or withdrawal of the effec- the trial for a few months, after which trials as well. It has been shown that tive therapy. If successful, strategies to responders are randomised in a double- CRP is elevated in one third of patients reduce/withdraw treatment are impor- blind manner either to continue the ex- with AS, and that elevated CRP can be tant not only from the patient’s point of perimental therapy or to switch to pla- used as a predictor for development view (reduced drug exposure risk) but cebo. In this double-blind withdrawal of structural changes in patients with also because of ethical considerations phase, patients who demonstrated a pre- axial SpA (19). CRP is included in the (reducing costs with no worsening of defined definition of disease worsening ASAS-endorsed ankylosing spondyli- patient quality of life). Very few data are withdrawn from the double-blind ties disease activity score (ASDAS), are available on the effect of TNFi dos- withdrawal phase and usually re-treat- a new measurement tool for assessing age adjustment or on withdrawal strate- ed with the experimental therapy in an disease activity in patients with axial gies in patients with axial SpA and/or in open-label manner. Although this trial SpA. Recently, ASDAS thresholds for patients with AS (21-24). design is patient- and -friend- disease activity states, including inac- The issue of withdrawal of an effec- ly, the disadvantage is a bias towards tive disease (equivalent to remission), tive treatment is frequently discussed responders. Nevertheless, this trial de- have been defined (20). In therapeutic in usual care, but seldom addressed in sign has proven to be very effective and trials, it is important to know not only prospective controlled trials because has been used especially in nearly all the actual disease status but also the the optimal trial design is not known. recent trials of biologic agents in chil- change in disease status, which can be Schwartz and Lellouch suggested that dren with JIA (26). Until now, no such assessed by using improvement scores. controlled trials can be categorised into trial conducted in patients with axial With the ASDAS threshold for clini- two different approaches: the explana- SpA has been reported. Nevertheless, cally important improvement as well as tory (also called the efficacy trial) and RCT with a withdrawal design can be

S-48 Withdrawal of medical therapies in SpA / U. Kiltz et al. used in patients with axial SpA to study lenges for the trial design (correct pa- At this time, there is no agreement on at least two issues: (i) which patients tient population, inaccuracy of disease the period of time a patient should be are most likely to flare with discontinu- phenotype in heterogeneous diseases, in remission when considering to stop ation of therapy, and (ii) whether dose variability in patient profiles), the use of or to reduce the medication. Again, this reduction is an option in patients with adaptive designs and biomarker-driven is an important topic for research: does axial SpA. Using this trial design, at- studies is not established in patients the duration of being in remission pre- tention should be directed to the pos- with SpA. Examples from studies in dict the propability of sustained remis- sible differences in an optimal strategy illustrate the potential benefit sion after withdrawal? between patients with early axial SpA in matching drugs with predictive bio- A definition of when to restart treat- versus those with longstanding AS, the markers for future application to small- ment is also an important consideration likelihood of remission is much higher er but more focused phase III studies in a withdrawal design. In principal, for patients with early axial SpA than (31). But the biomarkers reported so far this is the primary outcome of the trial: for patients with longstanding disease. in patients with SpA are far behind pre- the number of patients failing sustained dictive properties to be fulfillled if used remission. If ASDAS is the definition Comparative and adaptive trials in adaptive trials in patients with SpA. of remission, no longer being in inac- Comparative and adaptive trials are tive disease (ASDAS >1.3), reaching examples of new approaches to com- Future scenarios high disease activity (ASDAS >2.1), or plement the classical efficacy RCT. Numerous questions should be ad- showing a worsening of 1.1 (the recip- Examples of comparative studies with dressed before conducting a trial to rocal of a clinical important improve- randomised arms representing two study aspects of remission in patients ment) are all options with advantages treatments (head-to-head comparisons) with axial SpA: definition of remission and disadvantages. For patients who have been published in RA (27, 28). (clinical and/or imaging remission), relapse after being in remission, ad- Both studies illustrate the motivating duration of remission as a defining in- equate trial designs should address the feature for clinician and patient that a clusion criterion, predictors of remis- questions regarding the optimal dose previously tried and failed therapy was sion, how to deal with patients with low after restart of a specific treatment, and not used as the comparator. Studies in disease activity, definition of subgroups investigators should examine whether SpA are pending. The enthusiasm for (e.g. TNFi naïve patients or patients a specific treatment is efficacious again more comparative studies is countered who will most likely remain in remis- after restart. Trials comparing three by the recognition of the impact of the sion), definition of when to restart, and arms – continuation, dose reduction, required sample size when moving finally dose-adjustment after restart of or withdrawal of the effective thera- from placebo to active comparator arm the therapy. One of the first topics for py – appear to be a valid withdrawal with a non-inferiority design. A reduced research seems to be the predictors of design in patients with axial SpA. dosage of an effective drug as an ac- subgroups of patients who are most Ideally, these trials should be placebo tive comparator is an option to further likely to remain in remission after dose controlled so they can be double-blind. evaluate. A withdrawal trial with three reduction or withdrawal. Such a pos- However, a strategy, “adaptive” trial, arms (continuation of medication, re- sible subgroup might be patients naïve with prespecified protocol, can be more duced dose of a medication, and stop of to TNF inhibitors compared to patients informative – perhaps double-blind is a medication) may be considered as a who have already failed to respond to less needed with patient outcomes. trend-setting trial design. TNF inhibitors. In a non-controlled/ Adaptive trials allow adaptation to a non-randomised trial with rituximab Conclusion trial after its initiation without under- in patients with AS, the subgroup of The investigation of withdrawal strate- mining the validity and integrety of TNFi-naïve patients performed better gies is important for patients with axial the trial (29). Adaptive trials provide a compared to the TNFi-failure group SpA but the concept is not easily as- prospectively planned opportunity for (16). In the future there might be some sessed in this patient population. RCTs modification of one or more specific other subgroups of patients who will with a placebo-arm are useful to assess aspects of the study design. In patients be able to reduce or to stop a treatment efficacy of drugs but are not aligned to with RA, a DAS-driven treatment adap- regimen. follow a strategic approach. Thus, in- tation used in randomised trials showed One way to address this topic is to novative approaches to trial design are a significantly greater DAS reduction define predictors of the benefit ofa needed to complement the efficacy tri- and higher likeliness to achieve remis- treatment. Some trials that have been als. RCTs with a withdrawal design are sion in the intensive disease manage- published recently showed that short a step forward to test withdrawal in pa- ment compared to the standard of care disease duration, signs of inflamma- tients with axial SpA. Comparative and treatment (30). There are various adap- tory activity and early achievement of adaptive trials offer the opportunity of tive trial designs of which biomarker- remission are promising variables (32- realising more targeted management driven studies could be important also 34). These factors could also be investi- in explanatory and pragmatic trials. in patients with SpA. Since biomarker- gated as predictors for sustained remis- In that light, well-designed and effec- driven studies face substantial chal- sion after dose reduction/withdrawal. tive pragmatic trials offer an additional

S-49 Withdrawal of medical therapies in SpA / U. Kiltz et al. opportunity to optimise use of current 11. SIEPER J: How to define remission in anky- Ann Rheum Dis 2012; 71: 1212-5. and future treatments. Future scenarios losing spondylitis? Ann Rheum Dis 2012; 71 23. CANTINI F, NICCOLI L, CASSARÀ E, (Suppl. 2): i93-5. KALOUDI O, NANNINI C: Duration of remis- should recognise heterogeneity in pa- 12. SMOLEN JS, BRAUN J, DOUGADOS M et al.: sion after halving of the etanercept dose in tients with axial SpA, define subgroups Treating spondyloarthritis, including anky- patients with ankylosing spondylitis: a ran- of patients who are more likely to re- losing spondylitis and psoriatic arthritis, to domized, prospective, long-term, follow-up spond better to a specific treatment target: recommendations of an international study. Biologics 2013; 7: 1-6. task force. Ann Rheum Dis 2013 Jun 8 [Epub 24. BREBAN M, RAVAUD P, CLAUDEPIERRE P strategy, and might use a three-arm ahead of print]. et al.: Maintenance of infliximab treatment double-blind placebo-controlled trial 13. ANDERSON JJ, BARON G, van der HEIJDE in ankylosing spondylitis: results of a one- design comparing continuation, dose D, FELSON DT, DOUGADOS M: Ankylosing year randomized controlled trial comparing spondylitis assessment group preliminary systematic versus on-demand treatment. reduction, and withdrawal of the effec- definition of short-term improvement in an- Arthritis Rheum 2008; 58: 88-97. tive therapy in patients with axial SpA. kylosing spondylitis. Arthritis Rheum 2001; 25. ZWARENSTEIN M, TREWEEK S: What kind of 44: 1876-86. randomized trials do we need? CMAJ 2009; References 14. van der HEIJDE D, DIJKMANS B, GEUSENS 180: 998-1000. 1. BRAUN J, SIEPER J: Ankylosing spondylitis. P et al.: Efficacy and safety of infliximab in 26. RUPERTO N, GIANNINI EH, PISTORIO A, Lancet 2007; 369: 1379-90. patients with ankylosing spondylitis: results BRUNNER HI, MARTINI A, LOVELL DJ: Is it 2. van der LINDEN S, VALKENBURG HA, CATS of a randomized, placebo-controlled trial time to move to active comparator trials in A: Evaluation of diagnostic criteria for an- (ASSERT). Arthritis Rheum 2005; 52: 582-91. juvenile idiopathic arthritis?: a review of cur- kylosing spondylitis. A proposal for modi- 15. BARKHAM N, KEEN HI, COATES LC et al.: rent study designs. Arthritis Rheum 2010; 62: fication of the New York criteria. Arthritis Clinical and imaging efficacy of infliximab 3131-9. Rheum 1984; 27: 361-8. in HLA-B27-Positive patients with magnetic 27. GABAY C, EMERY P, van VOLLENHOVEN R 3. RUDWALEIT M, van der HEIJDE D, LANDEWÉ resonance imaging-determined early sacro- et al.: Tocilizumab monotherapy versus adal- R et al.: The development of Assessment of iliitis. Arthritis Rheum 2009; 60: 946-54. imumab monotherapy for treatment of rheu- SpondyloArthritis international Society clas- 16. SONG IH, HERMANN K, HAIBEL H et al.: matoid arthritis (ADACTA): a randomised, sification criteria for axial spondyloarthritis Effects of etanercept versus sulfasalazine in double-blind, controlled phase 4 trial. Lancet (part II): validation and final selection. Ann early axial spondyloarthritis on active inflam- 2013; 381: 1541-50. Rheum Dis 2009; 68: 777-83. matory lesions as detected by whole-body 28. WEINBLATT ME, SCHIFF M, VALENTE R 4. RUDWALEIT, M, van der HEIJDE D, LANDEWÉ MRI (ESTHER): a 48-week randomised con- et al.: Head-to-head comparison of subcu- R et al.: The Assessment of SpondyloArthritis trolled trial. Ann Rheum Dis 2011; 70: 590-6. taneous abatacept versus adalimumab for International Society classification criteria for 17. van der HEIJDE D, KIVITZ A, SCHIFF MH et rheumatoid arthritis: findings of a phase peripheral spondyloarthritis and for spondy- al.: Efficacy and safety of adalimumab in IIIb, multinational, prospective, randomized loarthritis in general. Ann Rheum Dis 2011; patients with ankylosing spondylitis: results study. Arthritis Rheum 2013; 65: 28-38. 70: 25-31. of a multicenter, randomized, double-blind, 29. GALLO P, ANDERSON K, CHUANG-STEIN C 5. DORAN MF, BROPHY S, MacKAY K, TAYLOR placebo-controlled trial. Arthritis Rheum et al.: Viewpoints on the FDA draft adaptive G, CALIN A: Predictors of longterm outcome 2006; 54: 2136-46. designs guidance from the PhRMA working in ankylosing spondylitis. J Rheumatol 2003; 18. BRANDT J, LISTING J, HAIBEL H et al.: Long- group. J Biopharm Stat 2010; 20: 1115-24. 30: 316-20. term efficacy and safety of etanercept after 30. SCHOELS M, KNEVEL R, ALETAHA D et al.: 6. BRAUN J, van den BERG R, BARALIAKOS readministration in patients with active anky- Evidence for treating rheumatoid arthritis X et al.: 2010 update of the ASAS/EULAR losing spondylitis. (Oxford) to target: results of a systematic literature recommendations for the management of an- 2005; 44: 342-8. search. Ann Rheum Dis 2010; 69: 638-43. kylosing spondylitis. Ann Rheum Dis 2011; 19. PODDUBNYY D, HAIBEL H, LISTING J et 31. KIM ES, HERBST RS, WISTUBA II et al.: The 70: 896-904. al.: Baseline radiographic damage, elevated BATTLE trial: personalizing therapy for lung 7. BRAUN J, BRANDT J, LISTING J et al.: acute-phase reactant levels, and cigarette cancer. Cancer Discov 2011; 1: 44-53. Treatment of active ankylosing spondylitis smoking status predict spinal radiographic 32. SIEPER J, van der HEIJDE D, DOUGADOS with infliximab: a randomised controlled progression in early axial spondylarthritis. M, BROWN LS, LAVIE F, PANGAN AL: Early multicentre trial. Lancet 2002; 359: 1187-93. Arthritis Rheum 2012; 64: 1388-98. response to adalimumab predicts long-term 8. BRAUN J, McHUGH N, SINGH A, WAJDULA 20. MACHADO P, LANDEWÉ R, LIE E et al.: remission through 5 years of treatment in JS, SATO R: Improvement in patient-re- Ankylosing Spondylitis Disease Activity patients with ankylosing spondylitis. Ann ported outcomes for patients with ankylos- Score (ASDAS): defining cut-off values Rheum Dis 2012; 71: 700-6. ing spondylitis treated with etanercept 50 for disease activity states and improvement 33. SIEPER J, LENAERTS J, WOLLENHAUPT J et mg once-weekly and 25 mg twice-weekly. scores. Ann Rheum Dis 2011; 70: 47-53. al.: Maintenance of biologic-free remission Rheumatology (Oxford) 2007; 46: 999-1004. 21. BARALIAKOS X, LISTING J, BRANDT J et with naproxen or no treatment in patients 9. van der HEIJDE DM, REVICKI DA, GOOCH al.: Clinical response to discontinuation of with early, active axial spondyloarthritis: re- KL et al.: Physical function, disease activity, anti-TNF therapy in patients with ankylos- sults from a 6-month, randomised, open-label and health-related quality-of-life outcomes ing spondylitis after 3 years of continuous follow-up study, INFAST Part 2. Ann Rheum after 3 years of adalimumab treatment in pa- treatment with infliximab. Arthritis Res Ther Dis 2013 Jun 5. [Epub ahead of print]. tients with ankylosing spondylitis. Arthritis 2005; 7: R439-44. 34. SIEPER J, van der HEIJDE D, DOUGADOS M Res Ther 2009; 11: R124. 22. SONG IH, ALTHOFF CE, HAIBEL H et al.: et al.: Efficacy and safety of adalimumab in 10. BRAUN J, BARALIAKOS X, BRANDT J, SIE- Frequency and duration of drug-free remis- patients with non-radiographic axial spondy- PER J: Therapy of ankylosing spondylitis. Part sion after 1 year of treatment with etanercept loarthritis: results of a randomised placebo- II: biological therapies in the spondyloarthriti- versus sulfasalazine in early axial spondy- controlled trial (ABILITY-1). Ann Rheum des. Scand J Rheumatol 2005; 34: 178-90. loarthritis: 2 year data of the ESTHER trial. Dis 2013; 72: 815-22.

S-50