Clinical Rheumatology (2020) 39:2307–2315 https://doi.org/10.1007/s10067-020-04970-3

ORIGINAL ARTICLE

A systematic review and network meta-analysis of current and investigational treatments for active

A. Deodhar1 & S. D. Chakravarty2,3 & C. Cameron4 & S. Peterson5 & R. Hensman4 & S. Fogarty4 & P. Spin4 & S. Kafka2 & S. Nair6 & L. S. Gensler7

Received: 8 October 2019 /Revised: 17 January 2020 /Accepted: 31 January 2020 / Published online: 27 February 2020 # The Author(s) 2020

Abstract Objective To compare the relative efficacy of current and investigational biologic and oral small molecule (OSM) treatments for active ankylosing spondylitis (AS). Methods A systematic literature review was conducted to identify all phase 2/3 randomized trials of interest in patients with AS. Outcomes assessed were ≥ 20% improvement in the Assessment of Spondyloarthritis International Society Criteria (ASAS20) and change from baseline in Bath Ankylosing Spondylitis Functional Index (BASFI) and C-reactive protein (CRP) at weeks 12– 16. Bayesian network meta-analyses were conducted for outcomes using a random effects model. Baseline-risk adjustment was also conducted to account for differences in placebo response across studies. Surface Under the Cumulative Ranking curve (SUCRA) values are reported, reflecting the relative probability that intervention was the best of all interventions. Results The investigational agent tofacitinib 5 mg was the top-ranked treatment (SUCRA, 93%) for ASAS20 response, followed by intravenous (IV) 2 mg/kg (90%). Golimumab IV 2 mg/kg and 5 mg/kg were the top two ranked treatments for change from baseline in BASFI (golimumab IV, 81%; infliximab, 80%) and change from baseline in CRP (infliximab, 90%; golimumab IV, 82%). Conclusions Two approved therapies (golimumab IV, infliximab) and one investigational product ranked highest for efficacy in AS.

Key Points • Although golimumab IV, infliximab, and tofacitinib ranked highest for efficacy in AS, differences in efficacy between approved and investigational therapies were not statistically significant.

Keywords Ankylosing spondylitis . Biologic . Intravenous golimumab . Network meta-analysis . Systematic review

Electronic supplementary material The online version of this article Introduction (https://doi.org/10.1007/s10067-020-04970-3) contains supplementary material, which is available to authorized users. Ankylosing spondylitis is a chronic, immune-mediated, in- flammatory condition within the family of spondyloarthritis * A. Deodhar [email protected] (SpA) with several shared clinical, genetic, and immunologic features [1]. Active ankylosing spondylitis is characterized by inflammation at the sacroiliac joint and insertion sites of ten- 1 Oregon Health & Science University, Portland, OR, USA dons and ligaments (i.e., enthesitis), as well as increased risk 2 Janssen Scientific Affairs, LLC, Horsham, PA, USA of fusion of the sacroiliac joint and spine [2, 3]. Patients ex- 3 Drexel University College of Medicine, Philadelphia, PA, USA perience severe back pain, spinal stiffness, and reduced spinal 4 Cornerstone Research Group Inc., Burlington, Ontario, Canada mobility, which may lead to severe deformity (i.e., bamboo 5 Janssen Global Services, LLC, Horsham, PA, USA spine) in some. Current guidelines recommend non-steroidal anti-inflam- 6 Janssen Pharmaceutica NV, Turnhoutseweg 30, 2340 Beerse, Belgium matory drugs (NSAIDs) and physical therapy as first-line treatment; however, some patients still experience active 7 University of California San Francisco, San Francisco, CA, USA 2308 Clin Rheumatol (2020) 39:2307–2315 disease [1]. For these patients, biologic therapies targeting two removed from the results. Randomized controlled trials pub- cytokine pathways are available, including five tumor necrosis lished before 2016 were identified through review of bibliog- factor (TNF) inhibitors [4–9] and one (IL)-17 in- raphies of relevant SLRs and meta-analyses, such as Wang hibitor [10]. Janus kinase (JAK) inhibitors [11, 12], two IL-17 et al. [16] A supplementary search of clinicaltrials.gov was inhibitors [13, 14], and one IL-23 inhibitor [15] are currently conducted on November 26, 2018, to obtain any additional under investigation for the treatment of AS. relevant references. Because of the lack of head-to-head studies that directly compare many of the available biologic therapies and oral small molecules (OSM) for active ankylosing spondylitis, sys- Selection criteria tematic literature reviews (SLRs) and Bayesian network meta- analyses (NMAs) have been conducted to evaluate relative Eligibility criteria were developed as follows and used to efficacy. Recent SLRs and NMAs have compared efficacy screen all identified studies. The study population was defined ≥ within a class of therapies (i.e., TNF inhibitors) and across as adults ( 18 years) diagnosed with active ankylosing spon- all then-available biologics for the treatment of ankylosing dylitis. Interventions of interest were (ADA), spondylitis [16, 17]. Since these publications, the intravenous apremilast (APR), (CZP), etanercept (IV) formulation of the TNF inhibitor golimumab (GOL IV) (ETN), filgotinib, infliximab (IFX), , GOL (both wasapprovedbytheUSFoodandDrugAdministration IV and subcutaneous (SC) formulations), placebo (PBO), (FDA) for the treatment of adults with active ankylosing spon- , (SEC), tofacitinib, and dylitis. Additionally, data from recently published phase 3 . Outcomes of interest were predefined as an im- ≥ clinical trials for ixekizumab (IXE) [18, 19] and ustekinumab provement of 20% in the Assessment of Spondyloarthritis (UST) [19], as well as phase 2 data for risankizumab (RIS) International Society Criteria (ASAS20), change in Bath [15], tofacitinib (TOF) [12], and filgotinib (FIL) [11], have Ankylosing Spondylitis Functional Index (BASFI), and become available. change in C-reactive protein (CRP) at weeks 12 to 16. All The objective of this study was to conduct a comprehensive included studies were published full-text RCTs. Although comparison of all current and investigational treatments for the IL-17A and IL-17F inhibitor bimekizumab is in clinical active ankylosing spondylitis based on all available phase 2/ development, it was not included because a peer-reviewed 3 data for interventions of interest using a Bayesian NMA. publication on the use of this agent in AS is not currently available. Conference abstracts were excluded.

Materials and methods Study selection

Systematic literature review An independent review of identified titles and abstracts was conducted in duplicate to identify studies matching the Search strategy predefined eligibility criteria. Eligible records were reviewed in full text by the same two reviewers. Conflicts between An SLR was conducted to identify all phase 2/3 randomized reviewers were resolved through a consensus meeting. controlled trials (RCTs) that compared the efficacy of bio- The study selection process was illustrated in a modified logics and OSMs in the treatment of active ankylosing spon- Preferred Reporting Items for Systematic Reviews and Meta- dylitis. A strategy was developed by an information specialist Analyses (PRISMA) flow diagram. Data from eligible articles and searches were conducted on May 28, 2018, and were collected using a data extraction form. In addition to the November 5, 2018, using the OVID platform to search outcomes of interest listed above, study design and patient OVID MEDLINE, including Epub Ahead of Print, In- baseline characteristics were extracted to assess the compara- Process and Other Non-Indexed Citations, Embase, and the bility of studies and identify the presence of heterogeneity. CENTRAL Database of the Cochrane Library (Wiley version) Data extraction was performed by a single reviewer and con- (Appendix S1). The search strategies used a combination of firmed by a second reviewer. The quality of each study was “ ” controlled vocabulary ( Spondylitis, Ankylosing ) and key- assessed using the Cochrane Risk of Bias Assessment Tool for “ ”“ words (e.g., ankylosing spondylitis, ankylopoietic Randomized Controlled Trials. spondylarthritis”, “Bechterew”). Vocabulary and syntax were adjusted across databases and results were limited from April 1, 2016, to November 5, 2018. An amended version of the Ethics 2008 Cochrane Highly Sensitive Search Strategy (sensitivity, and precision-maximizing version) was applied to the Ovid Ethics was not a requirement given analyses based on publicly searches. Where applicable, animal-only records were available summary-level data. Clin Rheumatol (2020) 39:2307–2315 2309

Network meta-analysis Results

NMAs were conducted using Bayesian methods. In a Study selection Bayesian NMA, a posterior distribution for each treatment effect is derived by combining a prior probability distribu- A total of 1466 published manuscripts were identified from tion (i.e., a prior belief of the relative efficacy of a treat- database searches and 104 from supplementary searches. A ment) with a likelihood (i.e., a statistical model) for the total of 1570 records were screened at the title and abstract effect estimate. A non-informative prior, which assigns stage and 276 were included for full-text review. Of those an equal probability to any theoretically plausible effect reviewed in full text, 244 records were excluded with reason estimate, was used to avoid undue influence of an overly and 32 records, representing 30 unique studies, were included informative prior distribution. Given the data and likeli- in the final review (Appendix S1). hood, a posterior distribution consisting of 15,000 effect estimates (i.e., mean difference (MD) for continuous end- Study and patient characteristics points or odds ratio (OR) for binary endpoints) was esti- mated using Markov Chain Monte Carlo re-sampling The 30 unique RCTs enrolled a total of 6711 patients. A sum- methods. This distribution was then used to obtain the me- mary of study and patient baseline characteristics is presented in dian effect estimate (i.e., a point estimate) and the 95% Appendix S2. The mean age of patients ranged from 27.4 to credible interval (CrI), which includes all posterior values 48.0 years and the mean disease duration ranged from 5.2 to between the 2.5th and 97.5th percentiles of the posterior 23.0 years. The proportion of male patients ranged from 52.6 to distribution. 100% and the proportion of HLA-B27-positive patients ranged Bayesian NMAs were conducted for ASAS20 response, from 65.0 to 97.4%. Mean baseline scores for Bath Ankylosing change from baseline in BASFI, and change from baseline Spondylitis Disease Activity Index (BASDAI) ranged from 4.4 in CRP using a random effects (RE) model. ASAS20 re- to 7.6 cm, for BASFI ranged from 3.2 to 7.4 cm, and for CRP sponse was defined as a ≥ 20% relative improvement from ranged [23] from 6.2 to 32.0 mg/l. Of the 30 studies, 23 allowed baseline and absolute improvement from baseline of ≥ 10 concomitant use of conventional synthetic disease-modifying units on a 0–100mm scale in ≥ 3 of the following domains: antirheumatic drugs (csDMARDs), four prohibited concomi- patient global assessment, spinal pain assessment, function tant use, and three did not report this information. (BASFI), and inflammation (the last two questions of The risk of bias was generally low across trials for random BASDAI) [20]. Pairwise comparisons between interven- sequence generation, allocation concealment, incomplete out- tions were reported as median OR with 95% CrIs for comes, and selective reporting (Appendix S4). ASAS20 and as MD with 95% CrIs for change from base- line in BASFI and change from baseline in CRP. The NMA results Surface Under the CumulativeRankingcurve(SUCRA) values, reported as percentages, were calculated to reflect An NMA was conducted for each outcome of interest: attain- the relative probability of an intervention being among the ment of ASAS20 response, change from baseline in BASFI, best options. Each NMA was performed in accordance with and change from baseline in CRP. The network diagram for the methodology recommended by the National Institute ASAS20 is provided in Fig. 1. Each intervention is represent- for Health and Care Excellence (NICE). To adjust for dif- ed by a node, and randomized comparisons are shown as links ferences in placebo response across studies, a baseline risk- between the nodes. Results of each NMA are presented in adjusted sensitivity analysis was conducted using code re- Figs. 2, 3,and4 for the models with the best fit. An odds ratio ported in the NICE Decision Support Unit (DSU) greater than one or a mean difference less than 0 implies that Technical Support Document (TSD) 3. Deviance informa- the reference treatment was better than the comparator. Model tion criterion (DIC) and posterior residual deviance were fit statistics for each analysis are provided in Appendix S6.A reported to assess model fit for each analysis. The estimat- significant β for a baseline risk-adjusted NMA suggested that ed regression coefficient (β) and 95% CrIs were reported the adjusted model was a better fit than the unadjusted model. to assess model fit for baseline risk-adjusted analyses. All analyses were performed using WinBUGS [21]andR[22] ASAS20 with burn-in and sampling durations ≥ 50,000 iterations. Comparisons of baseline risk-adjusted and unadjusted Twenty-six studies were included in the comparison of NMAs were used to assess risk of bias from between- ASAS20 at weeks 12 to 16 (Fig. 1). The baseline risk- study heterogeneity. Inconsistency models were developed adjusted model had the best fit because the 95% CrI for β and compared to consistency models to assess inconsisten- excluded 0 (Appendix S6). Tofacitinib 5 mg had superior cy in the networks. ASAS20 response compared to GOL IV 2 mg/kg and TOF 2310 Clin Rheumatol (2020) 39:2307–2315

Fig. 1 Evidence network for ASAS20 NMA. Each intervention is represented by a node and randomized comparisons are shown as links between the nodes. The size of each node represents the number of patients randomized to each treatment, and the width of connections is reflective of number of RCTs. Abbreviations: ADA, adalimumab; APR, apremilast; ASAS20, improvement of ≥ 20% in the Assessment of Spondyloarthritis International Society Criteria; CZP, certolizumab pegol; ETN, etanercept; FIL, filgotinib; GOL, golimumab; IFX, infliximab; IV, intravenous; IXE, ixekizumab; LD, loading dose; NMA, network meta-analysis; PBO, placebo; Q2W, every 2 weeks; Q4W, every 4 weeks; RIS, risankizumab; TOF, tofacitinib; SC, subcutaneous; SEC, secukinumab; UST, ustekinumab

10 mg, and both TOF 5 mg and GOL IV 2 mg/kg were supe- Change from baseline in CRP rior to TOF 2 mg, PBO, RIS 90 mg, SEC SC 75 mg, UST 90 mg, RIS 180 mg, APR 30 mg, and UST 45 mg (Fig. 2). Nineteen studies were included in the comparison of the Tofacitinib 5 mg and GOL IV 2 mg/kg were of greater or change from baseline in CRP at weeks 12 to 16 (Appendix equal efficacy compared to the other treatments. Rankings S5B). The unadjusted model had the best fit because the 95% based on SUCRA values for ASAS20 response were highest CrI for β excluded 0 (Appendix S6). Infliximab 5 mg/kg was for TOF 5 mg (93%), GOL IV 2 mg/kg (90%), and FIL ranked highest among all treatments, followed by GOL IV 200 mg (86%) (Fig. 5). 2 mg/kg. Infliximab 5 mg/kg showed superior reduction from baseline in CRP compared to UST 90 mg, UST 45 mg, and PBO and GOL IV 2 mg/kg was superior to PBO (Fig. 4). Change from baseline in BASFI SUCRA values for the change in CRP NMA were highest for IFX 5 mg/kg (90%), GOL IV 2 mg/kg (82%), and IXE Twenty-three studies were included in the comparison of 80 mg Q4W (76%) (Fig. 5). the change from baseline in BASFI at weeks 12 to 16 (Appendix S5A). The unadjusted model had the best fit Additional analyses because the 95% CrI for β excluded 0 (Appendix S6). Golimumab IV 2 mg/kg was ranked highest among all Between-study heterogeneity was assessed for each outcome treatments, followed by IFX 5 mg/kg. Both interven- using the I2 statistic for pairwise comparisons to placebo with tions had superior reductions from baseline for BASFI ≥ 2 independent studies (Appendix S10)[23]. The I2 statistic compared to PBO and UST 45 mg, and IFX 5 mg/kg was low (< 40%) or moderate (30 to 60%) for most compari- wasalsosuperiortoUST90mg(Fig.3). SUCRA sons, suggesting that heterogeneity was unlikely to bias the values for the change from baseline in BASFI were results. Substantial heterogeneity (I2 > 60%) was observed only highest for GOL IV 2 mg/kg (81%), IFX 5 mg/kg in pairwise comparisons of treatments with low SUCRA values (80%),andGOLSC100mg(69%)(Fig.5). which suggests heterogeneity is unlikely to alter conclusions. A Clin Rheumatol (2020) 39:2307–2315 2311

League Table − Adjusted (Odds Ratio) Credible interval includes 1 Ye s N o TOF 5 mg 1.2 GOL IV 2 (0.34,4.23) mg/kg 1.34 1.12 FIL 200 mg (0.38,4.89) (0.35,3.56) 1.5 1.26 1.11 ETN 50 mg (0.45,5.01) (0.45,3.44) (0.37,3.38) 1.68 1.41 1.24 1.13 IFX 5 mg/ (0.5,5.62) (0.53,3.57) (0.4,3.76) (0.43,2.86) kg 1.86 1.56 1.4 1.25 1.11 ETN 25 mg (0.63,5.44) (0.65,3.63) (0.52,3.66) (0.62,2.48) (0.53,2.35) 2.24 1.89 1.66 1.5 1.32 1.19 GOL 100 mg (0.67,7.53) (0.67,5.07) (0.53,5.18) (0.55,4.1) (0.56,3.34) (0.55,2.74) 2.27 1.88 1.66 1.5 1.33 1.19 1 CZP 400 mg (0.63,7.83) (0.58,5.82) (0.51,5.56) (0.51,4.33) (0.46,4.04) (0.47,3.14) (0.33,3.04) 2.45 2.06 1.83 1.65 1.45 1.31 1.1 1.1 ADA 40 mg (0.89,7.25) (0.91,4.84) (0.71,4.92) (0.78,3.68) (0.77,3.13) (0.79,2.41) (0.52,2.44) (0.45,2.86) 2.47 2.06 1.83 1.63 1.47 1.32 1.1 1.1 1 SEC IV 75 (0.78,8.23) (0.76,5.74) (0.62,5.77) (0.64,4.44) (0.59,3.79) (0.62,2.97) (0.42,2.97) (0.39,3.26) (0.46,2.13) mg 2.55 2.12 1.89 1.71 1.52 1.36 1.14 1.14 1.03 1.03 SEC IV 150 (0.86,7.9) (0.85,5.48) (0.7,5.36) (0.73,4.03) (0.68,3.47) (0.71,2.66) (0.47,2.7) (0.42,3.09) (0.53,1.88) (0.5,2.11) mg 2.55 2.14 1.92 1.71 1.53 1.36 1.15 1.14 1.04 1.04 1 SEC IV 300 (0.76,8.68) (0.72,6.16) (0.6,5.86) (0.63,4.63) (0.58,4.02) (0.59,3.17) (0.4,3.11) (0.38,3.44) (0.43,2.29) (0.39,2.61) (0.45,2.13) mg 2.66 2.23 1.98 1.78 1.59 1.42 1.19 1.19 1.08 1.08 1.04 1.04 GOL 50 mg (0.87,8.09) (0.89,5.52) (0.68,5.75) (0.73,4.3) (0.73,3.52) (0.74,2.82) (0.59,2.34) (0.42,3.23) (0.55,1.96) (0.45,2.52) (0.49,2.19) (0.43,2.63) 2.74 2.29 2.02 1.82 1.64 1.45 1.23 1.22 1.11 1.11 1.08 1.07 1.03 IXE 80 mg (0.9,8.16) (0.9,5.72) (0.73,5.6) (0.77,4.16) (0.72,3.63) (0.76,2.79) (0.51,2.82) (0.44,3.18) (0.59,1.93) (0.46,2.51) (0.51,2.17) (0.43,2.58) (0.48,2.12) Q2W 2.78 2.35 2.09 1.88 1.67 1.5 1.25 1.24 1.14 1.14 1.1 1.1 1.06 1.02 SEC SC 150 (0.96,8.42) (0.88,6.05) (0.76,5.79) (0.79,4.51) (0.7,4.12) (0.76,3.06) (0.48,3.11) (0.46,3.4) (0.55,2.15) (0.47,2.64) (0.51,2.33) (0.45,2.7) (0.47,2.33) (0.5,2.22) mg 2.81 2.38 2.11 1.89 1.68 1.51 1.27 1.25 1.16 1.15 1.12 1.09 1.07 1.04 1.01 SEC SC 150 (0.88,9.12) (0.79,6.72) (0.69,6.38) (0.72,4.99) (0.61,4.82) (0.66,3.56) (0.44,3.5) (0.43,3.73) (0.48,2.57) (0.42,3.03) (0.45,2.68) (0.41,3.09) (0.4,2.71) (0.43,2.51) (0.49,2.1) mg − no LD 2.92 2.47 2.17 1.97 1.74 1.56 1.32 1.31 1.19 1.19 1.14 1.14 1.1 1.07 1.05 1.04 IXE 80 mg (0.97,8.7) (0.97,5.95) (0.78,5.97) (0.82,4.53) (0.78,3.94) (0.82,3.01) (0.53,3.02) (0.49,3.4) (0.63,2.02) (0.49,2.7) (0.55,2.33) (0.47,2.79) (0.52,2.32) (0.6,1.95) (0.48,2.17) (0.43,2.47) Q4W 3.11 2.55 2.25 2.03 1.81 1.63 1.35 1.36 1.23 1.23 1.2 1.19 1.14 1.12 1.1 1.09 1.04 CZP 200 mg (0.9,10.81) (0.84,7.94) (0.72,7.68) (0.73,5.94) (0.64,5.66) (0.7,4.15) (0.47,4.25) (0.55,3.44) (0.51,3.01) (0.44,3.62) (0.47,3.21) (0.42,3.66) (0.44,3.23) (0.45,3.03) (0.43,2.94) (0.38,3.15) (0.42,2.79) 3.51 2.93 2.64 2.35 2.09 1.87 1.57 1.57 1.43 1.43 1.39 1.38 1.33 1.28 1.26 1.25 1.19 1.15 RIS 18 mg (0.79,14.26) (0.84,9.68) (0.61,10.25) (0.64,7.82) (0.67,6.31) (0.63,5.47) (0.45,4.89) (0.39,5.81) (0.45,3.84) (0.39,4.62) (0.4,4.22) (0.38,4.73) (0.42,3.86) (0.4,3.86) (0.35,3.95) (0.31,4.44) (0.37,3.65) (0.29,4.11) 3.51 2.93 2.6 2.33 2.05 1.86 1.56 1.56 1.42 1.42 1.37 1.36 1.31 1.28 1.24 1.24 1.19 1.14 1 TOF 10 mg (1.21,10.29) (0.89,9.63) (0.78,9.12) (0.8,7.28) (0.71,6.55) (0.75,5.07) (0.51,4.91) (0.49,5.39) (0.54,3.65) (0.47,4.2) (0.51,3.78) (0.46,4.37) (0.48,3.87) (0.49,3.6) (0.47,3.35) (0.43,3.72) (0.45,3.35) (0.36,3.69) (0.26,3.98) 4.06 3.4 2.99 2.69 2.4 2.16 1.81 1.79 1.64 1.64 1.61 1.57 1.52 1.49 1.45 1.43 1.38 1.31 1.15 1.16 TOF 2 mg (1.46,11.87)(1.03,10.74)(0.91,10.26) (0.93,8.38) (0.8,7.56) (0.84,5.81) (0.58,5.7) (0.57,6.18) (0.61,4.14) (0.55,4.95) (0.59,4.29) (0.52,5.07) (0.54,4.38) (0.55,4.06) (0.54,3.98) (0.5,4.32) (0.52,3.82) (0.41,4.44) (0.3,4.77) (0.43,3.07) 4.24 3.55 3.15 2.82 2.49 2.25 1.89 1.89 1.73 1.72 1.66 1.64 1.6 1.54 1.51 1.49 1.44 1.39 1.19 1.21 1.05 PBO (1.55,11.79) (1.32,9.33) (1.26,8.25) (1.28,6.68) (1.01,6.87) (1.2,4.68) (0.71,4.91) (0.77,4.87) (0.84,3.34) (0.73,4.07) (0.81,3.45) (0.73,4.06) (0.67,3.81) (0.78,3.35) (0.83,2.84) (0.73,3.14) (0.74,3.06) (0.56,3.4) (0.35,4.95) (0.48,3.06) (0.42,2.63) 5.81 4.85 4.34 3.91 3.42 3.1 2.6 2.62 2.37 2.35 2.27 2.28 2.16 2.12 2.07 2.06 1.97 1.92 1.65 1.65 1.44 1.39 RIS 90 mg (1.35,23.79)(1.45,15.63)(1.05,16.87)(1.11,12.84)(1.15,10.25) (1.04,8.98) (0.77,8.18) (0.65,9.66) (0.78,6.39) (0.68,7.65) (0.71,6.93) (0.63,7.88) (0.73,6.37) (0.69,6.44) (0.6,6.45) (0.54,7.36) (0.65,5.95) (0.48,6.68) (0.57,4.94) (0.41,6.28) (0.35,5.55) (0.37,4.63) 5.53 4.63 4.12 3.71 3.29 2.96 2.49 2.46 2.26 2.26 2.19 2.17 2.08 2.03 1.99 1.97 1.91 1.81 1.57 1.6 1.36 1.31 0.96 SEC SC 75 (1.69,18.75)(1.59,13.74)(1.35,13.13)(1.37,10.31) (1.25,9.33) (1.27,7.12) (0.89,6.95) (0.83,7.6) (0.96,5.12) (0.81,6.13) (0.89,5.49) (0.78,6.25) (0.84,5.51) (0.85,5.11) (0.91,4.5) (0.75,5.34) (0.77,4.83) (0.62,5.43) (0.46,6.1) (0.5,4.86) (0.45,4.33) (0.55,3.12) (0.28,3.6) mg 5.94 4.97 4.47 3.97 3.51 3.19 2.67 2.68 2.43 2.41 2.33 2.33 2.24 2.18 2.11 2.1 2.02 1.95 1.69 1.72 1.46 1.42 1.02 1.07 UST 90 mg (2,17.89) (1.98,12.75) (1.6,12.19) (1.7,9.48) (1.56,8.73) (1.67,6.43) (1.09,6.58) (1.02,7.2) (1.22,4.5) (1.02,5.69) (1.13,5.04) (0.96,6.02) (1.05,4.97) (1.09,4.65) (1.03,4.54) (0.89,5.24) (1,4.29) (0.74,5.03) (0.54,5.85) (0.63,4.67) (0.55,4.03) (0.71,2.81) (0.34,3.34) (0.44,2.67) 6.83 5.73 5.24 4.57 4.08 3.66 3.07 3.09 2.8 2.79 2.68 2.7 2.59 2.51 2.44 2.44 2.33 2.28 1.98 1.96 1.68 1.63 1.18 1.24 1.16 RIS 180 mg (1.57,29.81) (1.6,19.18) (1.16,20.86) (1.3,15.87) (1.29,12.62) (1.2,10.85) (0.88,9.89) (0.74,11.81) (0.9,7.89) (0.77,9.05) (0.83,8.25) (0.74,9.57) (0.82,7.7) (0.78,7.92) (0.71,8.07) (0.62,8.9) (0.73,7.43) (0.54,8.03) (0.65,5.96) (0.47,7.82) (0.41,6.9) (0.41,5.69) (0.38,3.48) (0.32,4.41) (0.34,3.6) 7.53 6.34 5.67 5.09 4.53 4.05 3.41 3.41 3.09 3.11 2.98 2.99 2.87 2.77 2.71 2.69 2.6 2.5 2.17 2.19 1.86 1.78 1.31 1.37 1.27 1.11 APR 30 mg (2.45,23.82)(2.37,16.36)(1.97,16.06)(2.02,12.23)(1.93,10.68) (1.99,8.39) (1.33,8.42) (1.17,9.24) (1.44,5.95) (1.21,7.33) (1.34,6.43) (1.15,7.64) (1.26,6.4) (1.31,5.97) (1.2,6.02) (1.03,6.86) (1.2,5.64) (0.87,6.49) (0.69,7.35) (0.74,5.99) (0.64,5.29) (0.79,3.88) (0.41,4.3) (0.51,3.45) (0.56,2.81) (0.33,3.79) 7.36 6.19 5.52 4.92 4.38 3.96 3.29 3.3 3 2.99 2.89 2.89 2.79 2.69 2.62 2.6 2.53 2.42 2.11 2.13 1.82 1.75 1.27 1.33 1.25 1.08 0.97 UST 45 mg (2.57,22.61) (2.49,15.9) (2.09,15.56)(2.13,12.27) (2,10.99) (2.09,8.1) (1.38,8.27) (1.26,9.07) (1.57,5.79) (1.28,7.22) (1.42,6.38) (1.19,7.5) (1.32,6.2) (1.36,5.98) (1.29,5.88) (1.12,6.64) (1.27,5.45) (0.93,6.3) (0.7,7.32) (0.79,5.88) (0.69,5.13) (0.88,3.6) (0.44,4.22) (0.55,3.41) (0.69,2.29) (0.35,3.73) (0.45,2.29) Fig. 2 League table of pairwise comparisons for all treatments in the Assessment of Spondyloarthritis International Society Criteria; Crls, ASAS20 NMA. Pairwise comparisons are reported as ORs and 95% credible intervals; CZP, certolizumab pegol; ETN, etanercept; FIL, Crls. Comparators are ordered from largest (top-left) to smallest filgotinib; GOL, golimumab; IFX, infliximab; IXE, ixekizumab; IV, (bottom-right) SUCRA values for the ASAS20 NMA. Please refer to intravenous; LD, loading dose; NMA, network meta-analysis; OR, odds Fig. 5 for SUCRA values for each NMA. Superior improvements in ratio; PBO, placebo; Q2W, every 2 weeks; Q4W, every 4 weeks; RIS, ASAS20 are denoted in bold text and light gray cells. Results are risankizumab; SC, subcutaneous; SEC, secukinumab; SUCRA, Surface shown for the baseline risk-adjusted model for ASAS20. Please refer to Under the Cumulative Ranking curve; TOF, tofacitinib; UST, Appendix S6 for the model fit statistics. Abbreviations: ADA, ustekinumab adalimumab; APR, apremilast; ASAS20, improvement of ≥ 20% in the

League Table − Unadjusted (Mean Difference)

GOL IV 2 Credible interval includes 0 Ye s N o mg/kg −0.1 IFX 5 mg/ (−1.88,1.73) kg −0.35 −0.26 GOL 100 mg (−2.37,1.59) (−2.01,1.39) −0.4 −0.31 −0.04 SEC SC 150 (−2.44,1.6) (−2.14,1.43) (−2.01,1.91) mg −0.44 −0.34 −0.08 −0.03 GOL 50 mg (−2.18,1.3) (−1.84,1.08) (−1.47,1.3) (−1.75,1.7) −0.43 −0.33 −0.08 −0.05 0 APR 30 mg (−2.67,1.77) (−2.4,1.66) (−2.24,2.08) (−2.21,2.17) (−1.98,1.98) −0.48 −0.39 −0.13 −0.09 −0.05 −0.04 SEC IV 150 (−2.53,1.56) (−2.19,1.4) (−2.05,1.85) (−2.06,1.91) (−1.77,1.73) (−2.27,2.22) mg −0.59 −0.49 −0.23 −0.19 −0.15 −0.14 −0.1 IXE 80 mg (−2.31,1.17) (−1.95,0.96) (−1.86,1.44) (−1.86,1.54) (−1.53,1.29) (−2.04,1.86) (−1.82,1.63) Q2W −0.59 −0.5 −0.24 −0.19 −0.16 −0.14 −0.1 −0.01 SEC IV 75 (−2.58,1.44) (−2.26,1.27) (−2.11,1.76) (−2.15,1.83) (−1.85,1.58) (−2.36,2.12) (−1.61,1.42) (−1.73,1.72) mg −0.68 −0.57 −0.31 −0.28 −0.23 −0.24 −0.2 −0.1 −0.09 FIL 200 mg (−2.77,1.38) (−2.4,1.21) (−2.28,1.71) (−2.26,1.78) (−2.01,1.56) (−2.49,2.04) (−2.24,1.87) (−1.86,1.64) (−2.14,1.92) −0.7 −0.59 −0.34 −0.3 −0.26 −0.26 −0.22 −0.11 −0.11 −0.02 IXE 80 mg (−2.39,1.01) (−2.05,0.82) (−1.98,1.32) (−1.97,1.43) (−1.61,1.13) (−2.17,1.73) (−1.9,1.48) (−1.25,0.96) (−1.81,1.59) (−1.78,1.73) Q4W −0.74 −0.64 −0.38 −0.33 −0.31 −0.3 −0.25 −0.15 −0.14 −0.06 −0.05 ADA 40 mg (−2.35,0.82) (−1.99,0.58) (−1.9,1.09) (−1.93,1.24) (−1.52,0.92) (−2.18,1.55) (−1.87,1.32) (−1.3,0.91) (−1.75,1.41) (−1.7,1.53) (−1.16,1.03) −0.8 −0.68 −0.44 −0.39 −0.36 −0.34 −0.31 −0.2 −0.21 −0.11 −0.09 −0.05 CZP 400 mg (−2.81,1.27) (−2.53,1.14) (−2.39,1.6) (−2.39,1.7) (−2.11,1.45) (−2.56,1.96) (−2.33,1.75) (−1.95,1.56) (−2.25,1.8) (−2.17,1.98) (−1.81,1.67) (−1.65,1.6) −0.88 −0.79 −0.52 −0.48 −0.43 −0.44 −0.39 −0.3 −0.28 −0.21 −0.18 −0.13 −0.09 TOF 5 mg (−2.99,1.2) (−2.68,1.04) (−2.56,1.49) (−2.58,1.6) (−2.3,1.4) (−2.77,1.83) (−2.55,1.65) (−2.18,1.5) (−2.45,1.74) (−2.36,1.89) (−2.04,1.57) (−1.85,1.57) (−2.26,1.99) −0.92 −0.8 −0.55 −0.52 −0.47 −0.47 −0.42 −0.32 −0.31 −0.22 −0.21 −0.17 −0.12 −0.03 CZP 200 mg (−2.93,1.16) (−2.64,1.03) (−2.5,1.48) (−2.52,1.57) (−2.18,1.33) (−2.67,1.82) (−2.46,1.63) (−2.11,1.46) (−2.34,1.74) (−2.32,1.89) (−1.95,1.55) (−1.76,1.52) (−1.7,1.56) (−2.09,2.18) −0.89 −0.81 −0.53 −0.5 −0.47 −0.47 −0.41 −0.31 −0.3 −0.21 −0.19 −0.15 −0.09 −0.03 0.01 SEC SC 75 (−2.94,1.1) (−2.6,0.9) (−2.47,1.39) (−2.04,1.02) (−2.17,1.26) (−2.68,1.8) (−2.42,1.56) (−2.07,1.39) (−2.31,1.68) (−2.29,1.76) (−1.92,1.45) (−1.74,1.42) (−2.19,1.9) (−2.09,2.11) (−2.06,2.01) mg −1.12 −1.01 −0.77 −0.73 −0.67 −0.68 −0.64 −0.52 −0.52 −0.43 −0.41 −0.38 −0.32 −0.23 −0.2 −0.22 TOF 10 mg (−3.19,0.97) (−2.91,0.83) (−2.74,1.28) (−2.78,1.37) (−2.47,1.14) (−2.92,1.62) (−2.74,1.48) (−2.37,1.27) (−2.61,1.51) (−2.51,1.66) (−2.21,1.38) (−2.02,1.33) (−2.49,1.8) (−1.94,1.53) (−2.36,1.91) (−2.28,1.91) −1.27 −1.17 −0.9 −0.87 −0.82 −0.82 −0.78 −0.68 −0.68 −0.58 −0.56 −0.52 −0.48 −0.37 −0.36 −0.37 −0.15 ETN 25 mg (−2.82,0.51) (−2.48,0.33) (−2.41,0.85) (−2.4,0.95) (−2.07,0.65) (−2.66,1.19) (−2.37,1.04) (−1.92,0.75) (−2.28,1.09) (−2.24,1.23) (−1.78,0.84) (−1.56,0.72) (−2.15,1.34) (−2.05,1.53) (−1.99,1.46) (−1.93,1.41) (−1.87,1.7) −1.38 −1.28 −1.02 −0.97 −0.94 −0.94 −0.9 −0.8 −0.77 −0.7 −0.68 −0.63 −0.59 −0.5 −0.47 −0.48 −0.27 −0.12 TOF 2 mg (−3.5,0.69) (−3.24,0.53) (−3.09,1.01) (−3.15,1.13) (−2.8,0.86) (−3.27,1.38) (−3.02,1.18) (−2.69,0.97) (−2.94,1.23) (−2.9,1.39) (−2.56,1.08) (−2.35,1.06) (−2.82,1.5) (−2.23,1.26) (−2.67,1.57) (−2.6,1.61) (−1.99,1.43) (−2.1,1.55) −1.59 −1.5 −1.25 −1.19 −1.17 −1.16 −1.11 −1.02 −1 −0.93 −0.9 −0.85 −0.81 −0.73 −0.69 −0.71 −0.48 −0.34 −0.22 UST 90 mg (−3.35,0.12) (−2.99,−0.07) (−2.91,0.44) (−2.94,0.57) (−2.58,0.26) (−3.16,0.8) (−2.86,0.62) (−2.44,0.38) (−2.76,0.69) (−2.75,0.84) (−2.28,0.47) (−2.06,0.38) (−2.62,0.97) (−2.53,1.15) (−2.51,1.04) (−2.48,1.06) (−2.3,1.32) (−1.84,0.95) (−2.04,1.63) −1.75 −1.65 −1.39 −1.35 −1.31 −1.32 −1.28 −1.17 −1.16 −1.08 −1.06 −1.01 −0.96 −0.87 −0.84 −0.84 −0.63 −0.5 −0.38 −0.15 UST 45 mg (−3.5,−0.03) (−3.17,−0.23) (−3.09,0.26) (−3.1,0.39) (−2.74,0.11) (−3.28,0.66) (−2.99,0.46) (−2.62,0.21) (−2.9,0.56) (−2.89,0.7) (−2.45,0.32) (−2.22,0.23) (−2.77,0.77) (−2.67,0.98) (−2.67,0.92) (−2.61,0.89) (−2.5,1.18) (−1.98,0.81) (−2.19,1.47) (−1.3,0.97) −1.9 −1.8 −1.54 −1.5 −1.46 −1.46 −1.41 −1.31 −1.31 −1.22 −1.2 −1.16 −1.11 −1.01 −0.98 −1 −0.78 −0.64 −0.52 −0.3 −0.15 PBO (−3.34,−0.49) (−2.9,−0.75) (−2.86,−0.22)(−2.89,−0.05)(−2.46,−0.45) (−3.15,0.24) (−2.85,0.01) (−2.3,−0.35) (−2.74,0.09) (−2.7,0.23) (−2.17,−0.25)(−1.83,−0.44) (−2.6,0.34) (−2.55,0.54) (−2.48,0.45) (−2.38,0.41) (−2.34,0.73) (−1.64,0.2) (−2.02,1.07) (−1.31,0.72) (−1.16,0.86) Fig. 3 League table with pairwise comparisons for all treatments in the Ankylosing Spondylitis Functional Index; Crls, credible intervals; CZP, BASFI NMA. Pairwise comparisons are reported as MDs and 95% Crls. certolizumab pegol; ETN, etanercept; FIL, filgotinib; GOL, golimumab; Comparators are ordered from largest (top-left) to smallest (bottom-right) IV, intravenous; IFX, infliximab; IXE, ixekizumab; MD, mean SUCRA values for the BASFI NMA. Please refer to Fig. 5 for SUCRA difference; NMA, network meta-analysis; PBO, placebo; Q2W, every values for each NMA. Superior improvements in BASFI are denoted in 2 weeks; Q4W, every 4 weeks; SC, subcutaneous; SEC, secukinumab; bold text and light gray cells. Results are shown for the unadjusted model SUCRA, Surface Under the Cumulative Ranking curve; TOF, tofacitinib; for BASFI. Please refer to Appendix S6 for the model fit statistics. UST, ustekinumab Abbreviations: ADA, adalimumab; APR, apremilast; BASFI, Bath 2312 Clin Rheumatol (2020) 39:2307–2315

League Table − Unadjusted (Mean Difference) Credible interval includes 0 Ye s N o IFX 5 mg/ kg −0.31 GOL IV 2 (−1.89,1.19) mg/kg −0.51 −0.2 IXE 80 mg (−1.86,0.83) (−1.49,1.16) Q4W −0.56 −0.25 −0.05 IXE 80 mg (−1.93,0.79) (−1.58,1.14) (−1.01,0.88) Q2W −0.6 −0.3 −0.09 −0.04 ADA 40 mg (−1.85,0.63) (−1.49,0.93) (−1.02,0.79) (−0.96,0.85) −0.63 −0.32 −0.12 −0.08 −0.03 ETN 25 mg (−2.05,0.84) (−1.69,1.17) (−1.31,1.15) (−1.29,1.23) (−1.04,1.13) −0.95 −0.63 −0.43 −0.39 −0.34 −0.32 FIL 200 mg (−2.55,0.6) (−2.21,0.93) (−1.84,0.91) (−1.79,1) (−1.64,0.92) (−1.84,1.09) −1.04 −0.74 −0.54 −0.49 −0.45 −0.41 −0.1 GOL 50 mg (−2.58,0.43) (−2.21,0.8) (−1.86,0.81) (−1.8,0.85) (−1.59,0.78) (−1.87,0.96) (−1.6,1.42) −1.08 −0.77 −0.57 −0.53 −0.49 −0.45 −0.14 −0.04 SEC SC 150 (−2.61,0.42) (−2.25,0.73) (−1.91,0.74) (−1.85,0.78) (−1.65,0.68) (−1.92,0.89) (−1.65,1.44) (−1.53,1.41) mg − no LD −1.1 −0.79 −0.59 −0.53 −0.5 −0.47 −0.15 −0.06 −0.01 GOL 100 mg (−2.64,0.38) (−2.25,0.72) (−1.91,0.72) (−1.89,0.78) (−1.67,0.69) (−1.95,0.88) (−1.7,1.35) (−1.13,0.99) (−1.47,1.46) −1.11 −0.8 −0.6 −0.55 −0.5 −0.49 −0.16 −0.06 −0.04 −0.02 TOF 10 mg (−2.75,0.47) (−2.37,0.83) (−2.07,0.81) (−1.99,0.87) (−1.81,0.77) (−2.07,1) (−1.83,1.49) (−1.64,1.52) (−1.6,1.59) (−1.61,1.57) −1.21 −0.9 −0.7 −0.65 −0.61 −0.58 −0.26 −0.17 −0.13 −0.11 −0.11 TOF 5 mg (−2.82,0.33) (−2.47,0.66) (−2.14,0.67) (−2.08,0.71) (−1.87,0.65) (−2.13,0.84) (−1.9,1.34) (−1.73,1.37) (−1.71,1.41) (−1.66,1.43) (−1.42,1.18) −1.35 −1.04 −0.83 −0.78 −0.74 −0.7 −0.4 −0.29 −0.26 −0.24 −0.24 −0.13 SEC IV 75 (−3.31,0.56) (−2.95,0.9) (−2.67,0.94) (−2.6,0.99) (−2.44,0.94) (−2.62,1.08) (−2.38,1.6) (−2.23,1.58) (−2.18,1.65) (−2.17,1.65) (−2.24,1.77) (−2.1,1.86) mg −1.37 −1.05 −0.86 −0.8 −0.76 −0.75 −0.41 −0.32 −0.26 −0.26 −0.24 −0.14 −0.03 SEC IV 150 (−3.35,0.56) (−2.98,0.87) (−2.66,0.96) (−2.6,0.98) (−2.46,0.96) (−2.65,1.06) (−2.4,1.53) (−2.19,1.6) (−2.21,1.61) (−2.17,1.67) (−2.28,1.77) (−2.09,1.82) (−2.05,2.05) mg −1.32 −1.01 −0.81 −0.76 −0.72 −0.69 −0.37 −0.27 −0.23 −0.22 −0.21 −0.1 0.03 0.05 TOF 2 mg (−2.93,0.25) (−2.56,0.59) (−2.25,0.6) (−2.18,0.67) (−2,0.58) (−2.24,0.77) (−2,1.24) (−1.8,1.27) (−1.8,1.33) (−1.75,1.35) (−1.54,1.09) (−1.38,1.15) (−1.94,2.03) (−1.91,2.02) −1.33 −1.03 −0.82 −0.77 −0.73 −0.7 −0.39 −0.29 −0.25 −0.24 −0.22 −0.12 0.01 0.04 −0.01 SEC SC 150 (−2.8,0.11) (−2.42,0.41) (−2.06,0.36) (−2,0.46) (−1.8,0.36) (−2.08,0.57) (−1.84,1.09) (−1.68,1.12) (−1.31,0.81) (−1.61,1.18) (−1.72,1.31) (−1.58,1.38) (−1.88,1.91) (−1.8,1.91) (−1.48,1.47) mg −1.38 −1.07 −0.87 −0.83 −0.78 −0.76 −0.44 −0.33 −0.3 −0.29 −0.28 −0.17 −0.05 −0.03 −0.06 −0.05 UST 90 mg (−2.85,−0.05) (−2.47,0.29) (−2.11,0.26) (−2.04,0.31) (−1.82,0.21) (−2.14,0.43) (−1.86,0.94) (−1.75,0.95) (−1.67,1) (−1.69,1.02) (−1.78,1.15) (−1.65,1.22) (−1.91,1.79) (−1.87,1.78) (−1.52,1.36) (−1.34,1.17) −1.45 −1.14 −0.95 −0.89 −0.86 −0.82 −0.5 −0.41 −0.36 −0.35 −0.34 −0.23 −0.1 −0.08 −0.12 −0.11 −0.06 SEC SC 75 (−3.22,0.23) (−2.83,0.6) (−2.52,0.6) (−2.45,0.65) (−2.29,0.6) (−2.46,0.76) (−2.26,1.23) (−2.11,1.28) (−2.04,1.34) (−2.04,1.31) (−2.13,1.5) (−2.01,1.55) (−2.2,2.02) (−2.2,2.01) (−1.9,1.63) (−1.69,1.44) (−1.64,1.55) mg −1.6 −1.29 −1.09 −1.04 −1 −0.97 −0.65 −0.55 −0.51 −0.49 −0.49 −0.38 −0.25 −0.24 −0.28 −0.26 −0.21 −0.15 UST 45 mg (−3.03,−0.26) (−2.64,0.11) (−2.3,0.07) (−2.25,0.12) (−2.03,0.04) (−2.3,0.24) (−2.06,0.76) (−1.93,0.79) (−1.82,0.79) (−1.9,0.85) (−1.97,0.93) (−1.82,1.03) (−2.09,1.57) (−2.06,1.6) (−1.76,1.16) (−1.52,0.95) (−1.18,0.8) (−1.73,1.44) −1.8 −1.5 −1.29 −1.24 −1.2 −1.17 −0.86 −0.75 −0.71 −0.71 −0.69 −0.6 −0.47 −0.44 −0.48 −0.47 −0.42 −0.35 −0.21 PBO (−2.93,−0.72) (−2.55,−0.41) (−2.11,−0.5) (−2.07,−0.45) (−1.74,−0.66) (−2.16,−0.29) (−1.98,0.27) (−1.82,0.25) (−1.76,0.31) (−1.75,0.33) (−1.89,0.51) (−1.73,0.56) (−2.05,1.17) (−2.09,1.15) (−1.66,0.67) (−1.42,0.42) (−1.24,0.46) (−1.7,1) (−1.08,0.66) Fig. 4 League table with pairwise comparisons for all treatments in the Abbreviations: ADA, adalimumab; CRP, C-reactive protein; Crls, CRP NMA. Pairwise comparisons are reported as MDs and 95% Crls. credible intervals; ETN, etanercept; FIL, filgotinib; GOL, golimumab; Comparators are ordered from largest (top-left) to smallest (bottom-right) IFX, infliximab; IV, intravenous; IXE, ixekizumab; LD, loading dose; SUCRA values for the CRP NMA. Please refer to Fig. 5 for SUCRA MD, mean difference; NMA, network meta-analysis; PBO, placebo; values for each NMA. Superior improvements in CRP are denoted in bold Q2W, every 2 weeks; Q4W, every 4 weeks; TOF, tofacitinib; SC, text and light gray cells. Results are shown for the unadjusted model for subcutaneous; SEC, secukinumab; SUCRA, Surface Under the CRP. Please refer to Appendix S6 for the model fit statistics. Cumulative Ranking curve; UST, ustekinumab

SUCRA ASAS20 BASFI CRP (Adjusted) (Unadjusted) (Unadjusted) TOF 5 mg 93% 48% 41% GOL IV 2 mg/kg 90% 81% 82% FIL 200 mg 86% 55% 55% ETN 50 mg 84% NA NA IFX 5 mg/kg 80% 80% 90% ETN 25 mg 77% 31% 71% GOL 100 mg 65% 69% 47% CZP 400 mg 64% 51% NA ADA 40 mg 60% 53% 73% SEC IV 75 mg 59% 59% 38% SEC IV 150 mg 57% 63% 37% SEC IV 300 mg 57% NA NA Legend GOL 50 mg 54% 67% 50% 100% IXE 80 mg Q2W 53% 60% 74%

SEC SC 150 mg 52% 67% 35% 50%

Treatment SEC SC 150 mg − no LD 51% NA 48% IXE 80 mg Q4W 48% 55% 76% 0% CZP 200 mg 46% 46% NA RIS 18 mg 41% NA NA TOF 10 mg 40% 38% 47% TOF 2 mg 32% 29% 37% PBO 28% 8% 11% RIS 90 mg 19% NA NA SEC SC 75 mg 18% 46% 32% UST 90 mg 15% 19% 32% RIS 180 mg 14% NA NA APR 30 mg 8% 63% NA UST 45 mg 8% 14% 23%

Fig. 5 Heat map of SUCRA values for all treatments for each NMA. International Society Criteria; BASFI, Bath Ankylosing Spondylitis Results are color coded such that comparators with larger SUCRA Functional Index; CRP, C-reactive protein; CZP, certolizumab pegol; values are in green and those with a lower SUCRA values are shown in ETN, etanercept; FIL, filgotinib; GOL, golimumab; IFX, infliximab; IV, red. Results of best-fitting model are presented. Results are shown for the intravenous; IXE, ixekizumab; LD, loading dose; NA, not available; baseline risk-adjusted model for ASAS20 and the unadjusted models for NMA, network meta-analysis; PBO, placebo; Q2W, every 2 weeks; change in BASFI and change in CRP. Please refer to Appendix S6 for the Q4W, every 4 weeks; RIS, risankizumab; SC, subcutaneous; SEC, model fit statistics. Abbreviations: ADA, adalimumab; APR, apremilast; secukinumab; SUCRA, Surface Under the Cumulative Ranking curve; ASAS20, improvement of ≥ 20% in the Assessment of Spondyloarthritis TOF, tofacitinib; UST, ustekinumab Clin Rheumatol (2020) 39:2307–2315 2313 comparison of unadjusted and baseline risk-adjusted analyses study had non-radiographic axial spondyloarthritis with low for each outcome showed that results were similar between functional impairment [8]. No difference in the reduction in models overall (Appendices S6 and S7). For ASAS20, the CRP from baseline was found among compared treatments. baseline risk-adjusted NMAyielded smaller heterogeneity com- These results align with those presented in the current analy- paredtotheunadjustedNMA.ForchangeinBASFIandCRP, ses, where no difference in change from baseline in BASFI or the unadjusted and baseline risk-adjusted models yielded simi- CRP between TNF inhibitor treatments was found. It is im- lar results. A sensitivity analysis including studies reporting portant to note that the results from Wang et al. are not directly results at 10 weeks was conducted for each outcome. This comparable to the analyses presented. First, multiple treatment sensitivity was conducted to align with the most recently pub- doses were combined for each intervention in the Wang et al. lished NMA, Wang et al. [16], which included studies reporting analyses, whereas each intervention dose was evaluated sep- results between 10 to 16 weeks. Inclusion of studies reporting arately in the present study. Second, the Wang et al. analysis results at 10 weeks resulted in the addition of one new study did not restrict studies to phase 2/3 trials and, therefore, in- (Marzo-Ortega 2005 [24]) in the ASAS20, change from base- cluded IFX biosimilars in their analyses. In contrast, data line in BASFI, and change from baseline in CRP networks. informing the analyses presented here were restricted to phase Results of these analyses were similar to those from the refer- 2/3 trials and excluded biosimilars. The third major difference ence analyses on all three outcomes (Appendix S7). from the Wang analysis is that our investigation was not re- stricted to TNF inhibitors, and thus compared a broader scope of treatments. Discussion Strengths and limitations This systematic review and NMA compared the efficacy of all current and investigational treatment options for ankylosing To the best of our knowledge, this study is the most recent and spondylitis by assessing ASAS20 response, change from base- comprehensive SLR and NMA comparing both currently avail- line in BASFI, and change from baseline in CRP. Results of the able and investigational biologic and OSM treatments for ac- best-fitting model for each network showed that TOF 5 mg and tive ankylosing spondylitis. Prior NMAs published in this ther- GOL IV 2 mg/kg were the two top-ranked treatments for apeutic area have been restricted to TNF inhibitors. This study ASAS20 response, GOL IV 2 mg/kg and IFX 5 mg/kg were uniquely provides data on the comparative efficacy of treat- the two top-ranked treatments for change from baseline in ment options with differing mechanisms of action to better BASFI, and IFX 5 mg/kg and GOL IV 2 mg/kg were the two inform treatment decisions. Given the recent advances in re- top-ranked treatment for change from baseline in CRP. search for the treatment of ankylosing spondylitis, this NMA Tofacitinib 5 mg was ranked first in the baseline risk- also provides data on the comparative efficacy of newly studied adjusted model for ASAS20; however, this was likely due to treatments such as GOL IV, FIL, IXE, RIS, and UST. The the higher placebo response rate observed in the TOF study, analyses conducted aligned with best practices for NMAs which shifted the results in favor of TOF 5 mg after the adjust- [25, 26] and used publicly available code to ensure transparen- ment. Further, the only study for TOF was a phase 2 trial [12] cy and reproducibility. Furthermore, an adjusted analysis ac- with a relatively small patient population, whereas data for counting for baseline risk differences between trials was also other treatments were based on phase 3 studies with larger conducted to ensure baseline differences were not confounding populations. Thus, favorable results for TOF 5 mg for the results. The analysis reported here focuses on published ASAS20 should be interpreted with caution. Golimumab IV summary-level data from RCTs. There are indirect treatment 2 mg/kg was associated with a greater reduction in BASFI than comparison methodologies, such as matching-adjusted indirect PBO and UST 45 mg and a greater reduction in CRP from comparisons (MAIC) and propensity score matching (PSM), baseline than PBO. GOL IV 2 mg/kg had comparable or supe- available that can leverage individual participant data (IPD) rior efficacy than all other comparators. Results of the baseline from RCTs to adjust for potential heterogeneity between stud- risk-adjusted NMAwere consistent with the unadjusted results. ies. Although MAICs and PSMs can better adjust for hetero- A recent NMA comparing treatments for ankylosing spon- geneity than NMA by leveraging IPD, there was limited het- dylitis published by Wang et al. [16] compared the efficacy of erogeneity identified in the current analysis. Further, MAICs six TNF inhibitors (ADA, CZP, ETN, GOL SC, IFX, and IFX and PSMs only permit comparison of one treatment at a time. biosimilar) using BASDAI, BASFI, and CRP outcome mea- In contrast, NMA provides a more holistic view of the body of sures [16]. Results of the 12-week analyses found that IFX evidence in ankylosing spondylitis. This is the first NMA to had a superior reduction in BASFI from baseline compared to simultaneously compare current and investigational treatments. CZP, whereas no difference was found for other treatment There were limited clinical trial data for each treatment to comparisons. This reduction may be because approximately inform the comparative efficacy of biologic and OSM thera- half of the patients who received CZP in the included phase 3 pies for active ankylosing spondylitis. This is evident from the 2314 Clin Rheumatol (2020) 39:2307–2315 presence of many single study connections in each network, S.D. Chakravarty, S.Kafka, and S. Peterson are employees and share- which limits the strength of the analysis. Insufficient clinical holders of Johnson and Johnson. S. Nair is an employee of Johnson and Johnson. C. Cameron is an employee and shareholder of Cornerstone data also limited the choice of outcomes presented in this Research Group Inc. S. Fogarty, R. Hensman, and P. Spin are employees analysis. All three outcomes reported represent measures of of Cornerstone Research Group Inc. Cornerstone consults for various change in a patient’s status; however, these measures do not pharmaceutical, medical device, and biotech companies. L.S. Gensler show the absolute status of a patient. Ankylosing Spondylitis has received research grants and/or has been a consultant for and has received honorarium from AbbVie, Amgen, Eli Lilly, Galapagos, Disease Activity Score (ASDAS) inactive disease, defined as Janssen, Novartis, Pfizer, and UCB. ASDAS < 1.3, was investigated as a measure of the absolute status of a patient in a limited number of studies. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, Unfortunately, limited reporting on ASDAS inactive disease adaptation, distribution and reproduction in any medium or format, as among the identified studies prevented assessment of this out- long as you give appropriate credit to the original author(s) and the come. Similarly, most studies did not report change in active source, provide a link to the Creative Commons licence, and indicate if inflammatory lesions (e.g., short-tau inversion recovery, changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated gadolinium-enhanced T1-weighed imaging), though CRP otherwise in a credit line to the material. If material is not included in the was reported broadly and was investigated as a marker for article's Creative Commons licence and your intended use is not change in active inflammation. Another limitation of this anal- permitted by statutory regulation or exceeds the permitted use, you will ysis was that data collected between weeks 12 to 16 were need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. pooled because of the lack of consistent reporting across stud- ies. Given that ankylosing spondylitis is a chronic disease, it was expected that results at these timepoints would be com- parable. Furthermore, safety outcomes such as adverse events, References infection, and toxicity were not examined in this study. Since safety events are typically uncommon, RCTs often lack suffi- 1. Ward MM, Deodhar A, Akl EA, Lui A, Ermann J, Gensler LS, cient follow-up and sample size to detect differences in such Smith JA, Borenstein D, Hiratzka J, Weiss PF, Inman RD, outcomes. Other NMAs examining adverse events have Majithia V, Haroon N, Maksymowych WP, Joyce J, Clark BM, pooled indications to overcome this limitation [27]. Finally, Colbert RA, Figgie MP, Hallegua DS, Prete PE, Rosenbaum JT, Stebulis JA, van den Bosch F, Yu DT, Miller AS, Reveille JD, almost all indications of superior efficacy of GOL IV, IFX, Caplan L (2016) American College of Rheumatology/Spondylitis and TOF were in comparison to failed investigational thera- Association of America/Spondyloarthritis Research and Treatment pies (e.g., UST, APR, RIS) and PBO. Network 2015 recommendations for the treatment of ankylosing spondylitis and nonradiographic axial spondyloarthritis. Arthritis Rheumatol 68(2):282–298. https://doi.org/10.1002/art.39298 2. Garcia-Montoya L, Gul H, Emery P (2018) Recent advances in Conclusions ankylosing spondylitis: understanding the disease and manage- ment. F1000Res:7. https://doi.org/10.12688/f1000research.14956. 1 TheSLRandNMApresentedinthisstudyarethemostup-to- 3. Gorman JD, Sack KE, Davis JC Jr (2002) Treatment of ankylosing date assessment of the comparative efficacy of current and in- spondylitis by inhibition of tumor necrosis factor alpha. N Engl J – vestigational biologic and OSM therapies for ankylosing spon- Med 346(18):1349 1356. https://doi.org/10.1056/NEJMoa012664 4. Deodhar A, Reveille JD, Harrison DD, Kim L, Lo KH, Leu JH, dylitis. Two approved interventions ranked highest among all Hsia EC (2018) Safety and efficacy of golimumab administered treatments for efficacy outcomes. Tofacitinib 5 mg was ranked intravenously in adults with ankylosing spondylitis: results through the highest for ASAS20 (though data was derived from a phase week 28 of the GO-ALIVE study. J Rheumatol 45(3):341–348. 2 study), golimumab IV for change from baseline in BASFI, https://doi.org/10.3899/jrheum.170487 5. Bao C, Huang F, Khan MA, Fei K, Wu Z, Han C, Hsia EC (2014) and infliximab for change from baseline in CRP. Safety and efficacy of golimumab in Chinese patients with active ankylosing spondylitis: 1-year results of a multicentre, randomized, Acknowledgments The authors thank Leena Patel of Cornerstone double-blind, placebo-controlled phase III trial. Rheumatology Research Group Inc. for editorial assistance. (Oxford) 53(9):1654–1663. https://doi.org/10.1093/rheumatology/ keu132 Funding This study was funded by Janssen Scientific Affairs, LLC. 6. van der Heijde D, Dijkmans B, Geusens P, Sieper J, De Woody K, Williamson P, Braun J, Ankylosing Spondylitis Study for the Evaluation of Recombinant Infliximab Therapy Study G (2005) Compliance with Ethical Standards Efficacy and safety of infliximab in patients with ankylosing spon- dylitis: results of a randomized, placebo-controlled trial (ASSERT). Conflict of interest A. Deodhar has received research grants and/or has Arthritis Rheum 52(2):582–591. https://doi.org/10.1002/art.20852 been a consultant for, and has received honorarium from AbbVie, 7. Huang F, Gu J, Zhu P, Bao C, Xu J, Xu H, Wu H, Wang G, Shi Q, Amgen, Boeheringer Ingelheim, Bristol Myers Squibb, Eli Lilly, Andhivarothai N, Anderson J, Pangan AL (2014) Efficacy and Janssen, Novartis, Pfizer, and UCB. Cornerstone received financial sup- safety of adalimumab in Chinese adults with active ankylosing port from Janssen Scientific Affairs, LLC, for the conduct of this study. spondylitis: results of a randomised, controlled trial. Ann Rheum Clin Rheumatol (2020) 39:2307–2315 2315

Dis 73(3):587–594. https://doi.org/10.1136/annrheumdis-2012- 16. Wang R, Dasgupta A, Ward MM (2018) Comparative efficacy of 202533 tumor necrosis factor-alpha inhibitors in ankylosing spondylitis: a 8. Landewe R, Braun J, Deodhar A, Dougados M, Maksymowych systematic review and Bayesian network metaanalysis. J WP, Mease PJ, Reveille JD, Rudwaleit M, van der Heijde D, Rheumatol 45(4):481–490. https://doi.org/10.3899/jrheum.170224 Stach C, Hoepken B, Fichtner A, Coteur G, de Longueville M, 17. Chen C, Zhang X, Xiao L, Zhang X, Ma X (2016) Comparative Sieper J (2014) Efficacy of certolizumab pegol on signs and symp- effectiveness of biologic therapy regimens for ankylosing spondy- toms of axial spondyloarthritis including ankylosing spondylitis: litis: a systematic review and a network meta-analysis. Medicine 24-week results of a double-blind randomised placebo-controlled (Baltimore) 95(11):e3060. https://doi.org/10.1097/MD. phase 3 study. Ann Rheum Dis 73(1):39–47. https://doi.org/10. 0000000000003060 1136/annrheumdis-2013-204231 18. van der Heijde D, Cheng-Chung WJ, Dougados M, Mease P, 9. Barkham N, Coates LC, Keen H, Hensor E, Fraser A, Redmond A, Deodhar A, Maksymowych WP, Van den Bosch F, Sieper J, Cawkwell L, Emery P (2010) Double-blind placebo-controlled trial Tomita T, Landewe R, Zhao F, Krishnan E, Adams DH, Pangallo of etanercept in the prevention of work disability in ankylosing B, Carlier H, Coast V (2018, 2018) Ixekizumab, an interleukin-17A spondylitis. Ann Rheum Dis 69(11):1926–1928. https://doi.org/ antagonist in the treatment of ankylosing spondylitis or radiograph- 10.1136/ard.2009.121327 ic axial spondyloarthritis in patients previously untreated with bio- 10. Marzo-Ortega H, Sieper J, Kivitz A, Blanco R, Cohen M, Martin R, logical disease-modifying anti-rheumatic drugs (COAST-V): 16 Readie A, Richards HB, Porter B, Measure 2 Study G (2017) week results of a phase 3 randomised, double-blind, active- Secukinumab and sustained improvement in signs and symptoms controlled and placebo-controlled trial. Lancet of patients with active ankylosing spondylitis through two years: 19. Deodhar A, Gensler LS, Sieper J, Clark M, Calderon C, Wang Y, results from a phase III study. Arthritis Care Res (Hoboken) 69(7): Zhou Y,Leu JH, Campbell K, Sweet K, Harrison DD, Hsia EC, van 1020–1029. https://doi.org/10.1002/acr.23233 der Heijde D (2018, 2018) Three multicenter, randomized, double- 11. van der Heijde D, Baraliakos X, Gensler LS, Maksymowych WP, blind, placebo-controlled studies evaluating the efficacy and safety Tseluyko V, Nadashkevich O, Abi-Saab W, Tasset C, Meuleners L, of ustekinumab in axial spondyloarthritis. Arthritis Rheumatol Besuyen R, Hendrikx T, Mozaffarian N, Liu K, Greer JM, Deodhar 20. Anderson JJ, Baron G, van der Heijde D, Felson DT, Dougados M A, Landewe R (2018, 2018) Efficacy and safety of filgotinib, a (2001) Ankylosing spondylitis assessment group preliminary defi- selective Janus kinase 1 inhibitor, in patients with active ankylosing nition of short-term improvement in ankylosing spondylitis. spondylitis (TORTUGA): results from a randomised, placebo-con- Arthritis Rheumatol 44(8):1876–1886 trolled, phase 2 trial. Lancet 21. Lunn DJ, Thomas A, Best N, Spiegelhalter D (2000) WinBUGS — 12. van der Heijde D, Deodhar A, Wei JC, Drescher E, Fleishaker D, a Bayesian modelling framework: concepts, structure, and extensi- Hendrikx T, Li D, Menon S, Kanik KS (2017) Tofacitinib in pa- bility. Stat Comput 10:325–337 tients with ankylosing spondylitis: a phase II, 16-week, randomised, 22. R Core Team (2013) R: A language and environment for statistical placebo-controlled, dose-ranging study. Ann Rheum Dis 76(8): computing. R Foundation for Statistical Computing, Vienna 1340–1347. https://doi.org/10.1136/annrheumdis-2016-210322 Available at: http://www.R-project.org/ 13. Deodhar A, Poddubnyy D, Pacheco-Tena C, Salvarani C, 23. Higgins JPTGS (2011) Cochrane handbook for systematic reviews Lespessailles E, Rahman P, Jarvinen P, Sanchez-Burson J, of interventions version 5.1.0. The Cochrane collaboration, Gaffney K, Lee EB, Krishnan E, Santisteban S, Li X, Zhao F, Available from www.handbook.cochrane.org Carlier H, Reveille JD, Coast W (2018, 2018) Efficacy and safety 24. Marzo-Ortega H, McGonagle D, Jarrett S, Haugeberg G, Hensor E, of Ixekizumab in the treatment of radiographic axial O'Connor P, Tan AL, Conaghan PG, Greenstein A, Emery P (2005) spondyloarthritis: 16 week results of a phase 3 randomized, dou- Infliximab in combination with methotrexate in active ankylosing ble-blind, placebo controlled trial in patients with prior inadequate spondylitis: a clinical and imaging study. Ann Rheum Dis 64(11): response or intolerance to tumor necrosis factor inhibitors. Arthritis 1568–1575. https://doi.org/10.1136/ard.2004.022582 Rheumatol 25. Dias S, Sutton AJ, Ades AE, Welton NJ (2013) Evidence synthesis 14. van der Heijde D, Gensler LS, Deodhar A, Baraliakos X, for decision making 2: a generalized linear modeling framework for Poddubnyy D, Farmer MK, Baeten D, Kumke T, Oortgiesen M, pairwise and network meta-analysis of randomized controlled trials. Dougados M (2018) LB0001 dual neutralisation of il-17a and il- Med Decis Mak 33(5):607–617. https://doi.org/10.1177/ 17f with bimekizumab in patients with active ankylosing spondy- 0272989X12458724 litis (AS): 12-week results from a phase 2b, randomised, double- 26. Dias S, Sutton AJ, Welton NJ, Ades AE (2013) Evidence synthesis blind, placebo-controlled, dose-ranging study. J Ann Rheum Dis for decision making 3: heterogeneity–subgroups, meta-regression, 77(Suppl 2):70. https://doi.org/10.1136/annrheumdis-2018-eular. bias, and bias-adjustment. Med Decis Mak 33(5):618–640. https:// 7889 doi.org/10.1177/0272989X13485157 15. Baeten D, Ostergaard M, Wei JC, Sieper J, Jarvinen P, Tam LS, 27. Adverse effects of biologics: a network meta-analysis and Cochrane Salvarani C, Kim TH, Solinger A, Datsenko Y, Pamulapati C, overview (2011) https://www.cochranelibrary.com/cdsr/doi/10. Visvanathan S, Hall DB, Aslanyan S, Scholl P, Padula SJ (2018) 1002/14651858.CD008794.pub2/abstract. Accessed January 10, Risankizumab, an IL-23 inhibitor, for ankylosing spondylitis: re- 2020 sults of a randomised, double-blind, placebo-controlled, proof-of- concept, dose-finding phase 2 study. Ann Rheum Dis 77(9):1295– Publisher’snote Springer Nature remains neutral with regard to jurisdic- 1302 tional claims in published maps and institutional affiliations.