FormalPara Key Summary Points

Why carry out this study?

Recent advances in the treatment of ankylosing spondylitis (AS) have produced several advanced therapy options for patients, including tumor necrosis factor, Janus kinase, and interleukin-17 inhibitors. Research comparing clinical effectiveness and economic benefit among these classes is lacking.

This study assessed comparative clinical and economic benefit of approved treatments for AS among patients with AS who were naïve or had demonstrated inadequate response/intolerance to biologic disease-modifying antirheumatic drugs and tumor necrosis factor inhibitors (bio-naïve, bio-IR, and TNF-IR, respectively).

What was learned from this study?

This study showed that rates of clinical response, determined by the Assessment of Spondyloarthritis International Society 40 (ASAS40) criteria, were highest for patients on upadacitinib consistently among bio-naïve, bio-IR, and TNF-IR patients. Likewise, numbers needed to treat (NNT) and cost per clinical responder (CPR) were lowest for patients on upadacitinib across bio-naïve, bio-IR, and TNF-IR patient populations.

This study is the first to compare the clinical and economic benefit among approved advanced therapies for the treatment of AS. Our findings suggest upadacitinib may provide greater clinical and economic benefit in AS regardless of treatment status with biologic therapies.

Introduction

Ankylosing spondylitis (AS) is a spondyloarthropathy characterized by inflammation of the axial skeleton, peripheral joints, and entheses. A hallmark feature of AS is inflammatory spinal pain resulting in back pain and stiffness [1].

According to the European Alliance of Associations for Rheumatology, American College of Rheumatology, and Spondyloarthritis Research and Treatment Network guidelines for the treatment of AS, first-line therapy for AS is non-steroidal anti-inflammatory drugs (NSAIDs), followed by the addition of glucocorticoids and conventional synthetic disease-modifying antirheumatic drugs (DMARD) if disease persists despite NSAID use. Non-medicinal interventions, such as physical therapy, are also recommended [1, 2].

For patients who are refractory to conventional first-line therapies, advanced therapies, such as biologics, are recommended. Historically, tumor necrosis factor (TNF) inhibitors were indicated as first-line advanced therapies, followed by other classes [1]. However, up to 68% of patients do not respond to their initial TNF inhibitor and up to 61% may lose efficacy after an initial period of response [3, 4]. As such, recent updates to the guidelines for AS treatment have also recommended anti-interleukin (IL)-17 inhibitors, such as secukinumab (SEC) or ixekizumab (IXE), and Janus kinase (JAK) inhibitors (e.g., upadacitinib [UPA] and tofacitinib [TOF]) as first-line advanced therapies [1, 2].

Management of AS is associated with significant economic burden by way of high direct medical costs (e.g., pharmacy, inpatient, outpatient, and emergency department [ED] visits) and increasing impairment of work and daily living abilities [5]. Indeed, retrospective claims database studies have demonstrated substantial all-cause medical costs, pharmacy costs, and overall health care resource utilization among patients with active AS [6, 7]. Moreover, patients with greater disease severity at the time of treatment initiation and those not achieving disease control had significantly greater disease-associated costs than patients with milder disease or those who achieved disease control [7, 8].

While TNF and IL-17 inhibitors have long been used in AS, JAK inhibitors are relatively new treatment options for patients with active AS. Indeed, both TOF and UPA were approved for use in patients with AS [9,10,11,12]. As such, there have been no direct comparisons between all approved treatments for patients with active AS, thus highlighting the need for indirect treatment comparisons to be conducted. While a 2016 study compared TNF inhibitors (i.e., infliximab, adalimumab [ADA], golimumab [GOL], etanercept [ETN], certolizumab [CER]) and an IL-17 inhibitor (i.e., SEC) to assess clinical response rates and cost per clinical responder among patients with AS [5], JAK inhibitors were not included in this study thus further highlighting a gap in the literature.

At that time, clinical meaningful improvement in AS was defined as achieving at least 20% improvement in the Assessment of Spondyloarthritis International Society (ASAS20) [5]. In recent years, however, trials have adopted the use of ASAS40, a more stringent outcome requiring at least 40% improvement, which is more likely to reflect a true, clinically meaningful change [13]. More importantly, this measure focuses on improvements that are perceptible to patients and impact their overall health-related quality of life [13, 14]. Thus, in light of the adoption of ASAS40, approval of a new class of therapies, and no head-to-head studies directly comparing the efficacy of these therapies, an update on the clinical and economic benefit of treatment options for AS is warranted. Therefore, this study sought to assess the comparative clinical and economic benefit of approved treatments for AS, stratified by patients who are naïve or have demonstrated inadequate response/intolerance to biologic DMARDs and TNFs (bio-naïve, bio-IR, and TNF-IR respectively).

Methods

Trial Identification

Phase III randomized clinical trials of patients with active AS were identified by a systematic literature review (May 2021). Therapies included in the analysis were JAK inhibitors (i.e., UPA and TOF), anti-IL-17 therapies (i.e., SEC and IXE), and TNF inhibitors (i.e., ADA, ETN, and GOL).

Trials eligible for inclusion in this analysis were conducted among adult patients with active AS who fulfilled the modified New York criteria for AS (i.e., low back pain for at least 3 months improved by exercise but not rest, limitation of lumbar spine, abnormal chest expansion, bilateral sacroiliitis grade 2–4 or unilateral sacroiliitis grade 3–4) [15]; patients with non-radiographic axial spondyloarthritis were excluded. Eligible trials were either placebo (PBO)-controlled or reported results from a head-to-head comparison of two of the advanced therapies included in this study. Moreover, trials were required to report the proportion of patients achieving ASAS40 between weeks 12 and 16. TNF inhibitor therapies infliximab and certolizumab were not included in analyses presented in this manuscript as they did not meet relevant inclusion criteria.

Only those trials using US Food and Drug Administration approved doses of advanced therapies were included (e.g., UPA 15 mg orally once daily [16], IXE 80 mg every 4 weeks [Q4W] [17], GOL 50 mg or 100 mg monthly via subcutaneous injection [SQ] [18], SEC 150 mg [with or without loading dose] and 300 mg by SQ [19], ETN 25 mg twice weekly or 50 mg once weekly by SQ [pooled] [20], TOF 5 mg orally twice daily [21], and ADA 40 mg SQ every other week [22]). Dosing recommendations for SEC are, with a loading dose (LD), 150 mg weekly for 4 weeks and then every 4 weeks thereafter. For SEC without LD, dosing recommendations are 150 mg every 4 weeks or 300 mg every 4 weeks [19]. Trials included in this analysis were selected by two independent researchers; any disagreements were resolved by a third researcher. Data were abstracted by two independent researchers using a standard data abstraction form.

Ethics Compliance

This article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.

Outcome Measures

The ASAS40 responses included in this analysis were evaluated between treatment weeks 12–16. The number needed to treat (NNT), defined as how many patients must be treated to obtain one positive outcome (i.e., clinically meaningful outcome), and cost per responder (CPR), which is the cost needed to achieve a positive outcome, were also assessed. The proportion of patients discontinuing therapy due to adverse events (AEs) as reported from weeks 12–24 was assessed as well.

Costs

This analysis used advanced therapy drug acquisition costs and associated administration costs from a US payer perspective. Unit drug costs as of November 2, 2022 were based on the 2022 US wholesale acquisition cost (WAC) from Merative Micromedex® RED BOOK® [23]. The dosing schedules and unit and administration costs were used to calculate the costs for 16-week treatment for each therapy.

Statistical Methods

A Bayesian network meta-analysis (NMA) was conducted to assess the comparative efficacy and safety, as well as produce inputs to derive NNTs, and CPRs among advanced therapies approved for AS. The NMA approach produces statistical estimations and a framework for probabilistic decision-making which incorporates potential uncertainty in effect estimates.

The proportion of patients achieving ASAS40 at week 12–16 was estimated. To assess outcomes, a random effects model utilizing a logit-link adjusted for baseline risk was used for bio-naïve patients; a fixed effects model utilizing a logit-link adjusted for baseline risk was used for bio-IR patients; and fixed effects models utilizing a risk difference-link were used for TNF-IR patients and for AEs leading to discontinuation. Goodness-of-fit and model diagnostics were assessed on the basis of residual deviance and the deviance information criterion as well as review of leverage plots and the posterior distribution of the between-study heterogeneity parameter in random effects models [24]. Non-informative prior distributions were utilized in the NMA models. Estimated absolute rates were reported as posterior medians and 95% credible intervals (CrI).

NNTs were calculated as the reciprocal of the difference in estimated response rates between the active drug and placebo. CPR was calculated as the 16-week treatment costs for each advanced therapy divided by NMA-estimated ASAS40 absolute response rates. Data from clinical trials are generally from an intent-to-treat population, thus CPR results reported here are based on the intent-to-treat population.

All analyses were performed using Bayesian Markov chain Monte Carlo with bnma 1.1.2 utilizing JAGS 4.3.0 (logit-link models) and R2WinBUGS 2.1–21 utilizing WinBUGS 1.4.3 (risk difference-link models).

Results

Study Selection

A total of 191 records met the inclusion criteria of the clinical systematic literature review, reporting on 63 unique registered clinical trials. After the inclusion/exclusion criteria were applied, 20 registered clinical trials (RCTs) reported by 100 records were included in the NMA. Only records that reported relevant week 12–16 outcomes were used; records reporting long-term (i.e., beyond the randomized phase) outcomes were not used in the NMA. Likewise, conference abstracts with subsequent full-text publication(s) available were not used in the NMA. Thus, a total of 31 records contributed analyzed data for the 21 RCTs that provided efficacy and/or safety data and were used in this NMA (Supplementary Material). Among the trials included, 12 were conducted only among bio-naïve patients (treatments were ADA, GOL, ETN, IXE, and UPA), 2 were only among bio-IR patients (treatments were IXE and UPA), and 7 among both bio-naïve and bio-IR populations (treatments were SEC, TOF, and ETN); for the mixed studies, data stratified by bio-naïve and bio-IR were available and thus eligible to be included in the study.

Assessment of Spondyloarthritis International Society 40 Response Rates at Weeks 12–16

Among bio-naïve patients, the estimated ASAS40 response rate ranged from 17.2% to 53.6%. The highest response rate was observed in patients receiving UPA 15 mg (53.6%); patients receiving PBO had the lowest response rate (Fig. 1a). Most other treatments, including IL-17 inhibitors and TOF, had response rates between 41% and 49% (SEC 150 mg w/o LD, 41.7%; TOF 5 mg, 44.0%; SEC 150 mg, 44.2%; SEC 300 mg, 45.1%; IXE 80 mg Q4W, 48.8%). Among all treatments, TNF inhibitors had the lowest rates of response (ADA 40 mg, 39.0%; GOL 50 mg, 38.6%).

Fig. 1
figure 1

Absolute ASAS40 response rates at week 12–16 in bio-naïve (a) and bio-IR (b) patients with AS. aIXE 80 mg was administered every 4 weeks. bPatients received SEC 150 mg without a loading dose. ADA adalimumab, ARR absolute rate ratio, ASAS40 Assessment of Spondyloarthritis International Society improvement ≥ 40%, Bio biologic disease-modifying antirheumatic drug, CrI credible interval, ETN etanercept, GOL golimumab, IR inadequate response or intolerance, IXE ixekizumab, JAK Janus kinase, PBO placebo, SEC secukinumab, TOF tofacitinib, UPA upadacitinib

Full size image

Among bio-IR patients, ASAS40 response rates were between 8.9% and 39.5%. As seen in bio-naïve patients, UPA 15 mg had the highest estimated ASAS40 response rate (39.5%), followed by SEC 300 mg (35.2%) and SEC 150 mg (35.3%; Fig. 1b). ASAS40 response rates were less than 25% for IL-17 inhibitors IXE 80 mg Q4W (23.1%) and SEC 150 mg without LD (23.2%); the estimated ASAS40 response rate for PBO-treated patients was 8.9%.

In TNF-IR patients, trends in ASAS40 response rates remained similar, with UPA 15 mg (36.3%) and SEC 300 mg (36.0%) demonstrating the highest rates of response (Supplementary Material). Moreover, SEC 150 mg without LD (19.4%) and IXE 80 mg Q4W (23.9%) remained the lowest response rates among active therapies. The estimated ASAS40 response rate for PBO-treated patients was 11.3%.

Rates of Discontinuation Due to Adverse Events at Weeks 12–24 in Overall Populations by Therapy

Estimated absolute rates of AEs leading to discontinuation among overall trial populations varied by therapy (Table 1). Absolute rates of AEs leading to discontinuation, from lowest to highest, were SEC 300 mg (0.0 [0.0, 0.0]), UPA 15 mg (0.0 [0.0, 0.8]), SEC 150 mg with a loading dose (0.6 [0.0, 1.9]), PBO (1.2 [0.5, 1.8]), GOL 50 mg (1.6 [0.0, 4.3]), SEC 150 mg without LD (1.7 [0.0, 5.7]), ADA 40 mg (2.0 [0.4, 3.5]), GOL 100 mg (2.0 [0.0, 6.4]), TOF 5 mg (2.6 [0.0, 6.5]), IXE 80 mg Q4W (3.3 [0.0, 7.8]), and ETN 25 mg/50 mg (3.7 [0.6, 6.3]).

Table 1 Estimated absolute rates of AEs leading to discontinuation in overall populations
Full size table

Cross-Tabulation of Discontinuation Due to Adverse Events and ASAS40

The cross-tabulation of the estimated rate of discontinuation due to AEs and estimated ASAS40 response rates are presented in Fig. 2. UPA was associated with the most favorable outcome assessed for AEs leading to discontinuation and ASAS40 consistently across bio-naïve (Fig. 2a) and bio-IR (Fig. 2b) populations. In bio-naïve patients, TNF inhibitors, such as ADA and GOL, had higher rates of discontinuation due to AEs but lower rates of ASAS40 relative to UPA 15 mg; in bio-IR, IXE and TOF had higher rates of discontinuation due to AEs but lower rates of ASAS40 relative to UPA 15 mg.

Fig. 2
figure 2

Cross-tabulation of estimated absolute rates of discontinuation due to AEs versus rates of ASAS40 among bio-naïve (a) and bio-IR (b) patients with AS. ADA40 adalimumab 40 mg, AE adverse event, ASAS40 Assessment of Spondyloarthritis International Society improvement ≥ 40%, Bio biologic disease-modifying antirheumatic drug, ETN25/30 etanercept 25 mg/50 mg, GOL50 golimumab 50 mg, GOL100 golimumab 100 mg, IL interleukin, IR inadequate response or intolerance, IXE80Q4W ixekizumab 80 mg every 4 weeks, JAK Janus kinase, LD loading dose, PBO placebo, SEC150 secukinumab 150 mg, SEC300 secukinumab 300 mg, TNF tumor necrosis factor, TOF5 tofacitinib 5, UPA15 upadacitinib 15 mg. Rates for any treatment discontinuation due to AEs were assessed at weeks 12–24 and rates for ASAS40 were assessed at weeks 12–16

Full size image

Numbers Needed to Treat to Achieve ASAS40 at Weeks 12–16

Among bio-naïve patients, the NNT to achieve ASAS40 ranged from 2.8 to 4.8 and was lowest for those patients receiving UPA 15 mg (NNT [95% CrI] 2.8 [2.0, 5.5]); patients receiving the TNF inhibitor GOL 50 mg (4.8 [3.0, 11.1]) and ADA 40 mg (4.6 [3.1, 9.6]) had the highest NNT (Fig. 3a). The NNT for all remaining therapies was approximately 3.6.

Fig. 3
figure 3

Number needed to treat for ASAS40 at week 12–16 in bio-naïve (a) and bio-IR (b) patients with AS. aIXE 80 mg was administered every 4 weeks. bPatients received SEC 150 mg without a loading dose. ADA adalimumab, ARR absolute rate ratio, ASAS40 Assessment of Spondyloarthritis International Society improvement ≥ 40%, Bio biologic disease-modifying antirheumatic drug, CrI credible interval, ETN etanercept, GOL golimumab, IR inadequate response or intolerance, IXE ixekizumab, JAK Janus kinase, NNT number needed to treat, PBO placebo, SEC secukinumab, TOF tofacitinib, UPA upadacitinib

Full size image

In bio-IR patients, UPA 15 mg had the lowest NNT (3.3 [2.1, 17.3]), closely followed by SEC 150 mg (3.9 [2.3, 20.6]) and SEC 300 mg (4.0 [1.8, 32.6]; Fig. 3b). NNTs were higher for TOF 5 mg (6.2 [2.1, 63.3]), IXE 80 mg Q4W (7.3 [3.1, 50.4]), and SEC 150 mg without LD (7.3 [2.4, 82.9]).

Among TNF-IR patients, overall trends were consistent, NNTs ranged from 4.0 to 7.8. The lowest NNTs were for UPA 15 mg and SEC 300 mg (4.0 [2.9, 6.5] and 4.0 [1.8, 22.9], respectively; Supplementary Material). IXE 80 mg Q4W (7.8 [4.2, 32.7]) and SEC 150 mg without LD (7.8 [− 113.6, 125.0]) maintained the highest NNT. The SEC 150 mg without LD lower bound indicated a lack of significance against PBO on the NNT scale (7.8 [− 113.6, 125.0]).

Cost Per Responder Over 16 Weeks

The CPR estimates among bio-naïve patients ranged from $39.5k to $102.5k. The CPR for UPA 15 mg was markedly lower (CPR [95% CrI] $39.5k [27.5k, 73.4k]) compared to most other treatments (Fig. 4a); CPR for TOF 5 mg was $44.2k ($28.3k, $88.5k). The highest CPR was observed with SEC 300 mg ($100.6k [$64.1k, $208.4k]) and SEC 150 mg ($102.5k [$68.8k, $183.8k]). CPR for the remaining treatments ranged from $54.9k (GOL 100 mg) to $65.8k (ADA 40 mg).

Fig. 4
figure 4

Cost per responder for ASAS40 over 16 weeks in bio-naïve (a) and bio-IR (b) patients with AS. aIXE 80 mg was administered every 4 weeks. bPatients received SEC 150 mg without a loading dose. ADA40 adalimumab, ARR absolute rate ratio, ASAS40 Assessment of Spondyloarthritis International Society improvement ≥ 40%, Bio biologic disease-modifying antirheumatic drug, CPR cost per responder, CrI credible interval, ETN etanercept, GOL golimumab, IR inadequate response or intolerance, IXE ixekizumab, JAK Janus kinase, PBO placebo, SEC secukinumab, TOF tofacitinib, UPA upadacitinib

Full size image

In bio-IR patients, the lowest CPR remained UPA 15 mg ($53.6k [$25.0k, $295.6k]; Fig. 4b); for remaining treatments, CPR ranged from $74.3k (TOF 5 mg) to $136.2k (IXE 80 mg Q4W).

Among TNF-IR patients, CPR was similar to that of bio-IR patients, with UPA 15 mg having the lowest CPR ($58.3k [$43.7k, $87.7k]), followed by TOF 5 mg ($66.1k [$39.5k, $186.9k]; Supplementary Material). All SEC doses and IXE 80 mg Q4W remained among the highest CPRs, with SEC 150 mg having the highest CPR of all treatments ($142.9k [$105.1k, $222.3k]).

Discussion

Among all treatments assessed in this analysis, UPA 15 mg demonstrated greater clinical and economic benefit for the treatment of AS. Indeed, UPA 15 mg had consistently numerically higher rates of response (i.e., achieving ASAS40), numerically lower NNT, and markedly lower CPR in bio-naïve, bio-IR, and TNF-IR patients with AS, followed by SEC 150 or 300 mg. Moreover, rates of AEs leading to discontinuation were lowest in patients receiving SEC 300 mg and UPA 15 mg. In contrast, IL-17 inhibitors (SEC 150 mg without LD and IXE 80 mg Q4W) had generally lower rates of response, higher NNT, and higher CPR among bio-IR and TNF-IR patients.

A 2016 study assessed comparative efficacy, NNT, and CPR among treatments approved for AS at that time (i.e., TNF and IL-17 inhibitors). In that study, patients receiving ADA, SEC, or ETN had modestly higher rates of ASAS40 achievement than patients receiving GOL or CER (49.2%, 42.4%, 41.4%, 38.6%, and 34.8%, respectively) [5]. Patients receiving ADA had lower NNTs than SEC, ETN, and GOL (ADA, 2.8; SEC, 3.5; ETN, 3.6; GOL, 4.0) [5]. However, costs for ADA, ETN, SEC without LD, and GOL were markedly lower than for patients receiving SEC with LD ($26.9k, $34.4k, $37.9k, $39.0k vs $75.7k, respectively) [5]. The JAK inhibitors, UPA and TOF, were approved after this 2016 study and were hence included in this analysis [5, 9,10,11,12].

In the current study, the lowest ASAS40 response rates and highest NNTs among bio-naïve patients were observed in those treated with ADA and GOL. Rates and NNTs in ASAS40 among SEC-treated bio-naïve patients were similar regardless of SEC dose. Among bio-IR patients, SEC 150 mg without LD had a lower rate of ASAS40 and markedly higher NNT relative to other SEC doses. To date, there are no head-to-head comparisons between therapies of different drug classes in AS. A recent study among TNF inhibitors assessed ADA versus ETN in patients with AS and found that both therapies had comparable efficacy [25]. The proportion of patients achieving ASAS40 with ADA and ETN was 22% each, which was markedly lower than that seen in this study (39.0% and 44.6%, respectively); however, small sample size in the head-to-head study (N = 18), as well as the open-label design, may be a limiting factor [25].

Cost-effectiveness studies assessing approved AS treatments are also lacking. Kim et al. [26] demonstrated that among patients with AS achieving ASAS40, those receiving ADA had a 40% higher CPR compared with those receiving SEC without a LD. The CPR for ADA-treated patients was 25% higher compared with patients receiving SEC with a LD [26]. In contrast, this study demonstrated that SEC 300 mg or 150 mg (with LD) had the highest CPR at weeks 12–16, with SEC 150 mg (without LD) and ADA having similar CPR. However, Kim et al. assessed CPR over a much longer timeframe (up to 104 weeks), which may account for the differences [26].

This analysis also reported an interesting observation wherein GOL 50 mg had a higher CPR than GOL 100 mg. While GOL 50 mg was calculated to be lower cost, its efficacy was found to be notably lower than GOL 100 mg. As CPR incorporates both efficacy and cost into the calculation, lower efficacy of GOL 50 mg leads to a higher cost per ASAS40 responder than for GOL 100 mg. Similarly, SEC 150 mg (with LD) had a higher CPR than SEC 300 mg. Per the 2022 US wholesale acquisition cost [23], SEC 150 mg (with LD) and SEC 300 mg had the same cost. However, the efficacy of SEC 150 mg (with LD) was slightly lower than the 300 mg dosing regimen. Thus, with equal cost but slightly lower efficacy, SEC 150 mg (with LD) had a higher cost per ASAS40 responder than SEC 300 mg.

This is the first study to compare efficacy among TNF, IL-17, and JAK inhibitors in a single analysis across bio-naïve, bio-IR, and TNF-IR populations; the results of this study may assist providers when making treatment decisions. In the absence of comprehensive head-to-head trials, NMA is a methodology widely described and utilized in the literature to produce indirect treatment comparisons. NMAs in this manuscript were conducted in accordance with best practices as outlined by the National Institute for Health and Care Excellence Decision Support Unit (NICE DSU), and model fit was robustly assessed following recommendations of Dias et al. [24, 27, 28]. Methodologies designed to minimize impact of potential PBO heterogeneity, such as baseline risk-adjustment or a risk difference-link, were utilized. The CPR calculation is a simple analysis that requires few assumptions; thus, results are transparent and easily described as the cost per outcome. The inclusion of data regarding treatment AEs leading to discontinuation is another strength. Previous studies did not include safety outcomes, whereas this manuscript provides insight into treatment tolerability.

The assumptions underlying NMA, namely network connectivity (i.e., that all treatments in the network are connected through a common comparator such as placebo), homogeneity (i.e., that all populations included in the network are sufficiently similar), and consistency (i.e., that effect estimates from direct and indirect evidence are aligned), must be carefully considered; if any one of the assumptions is violated, the conclusions of the NMA may be jeopardized. While these underlying assumptions have been tested as feasible in this analysis, NMA is not a substitute for head-to-head studies. Furthermore, similar to a traditional pairwise meta-analysis, conclusions from the NMA are susceptible to the methodological quality of the included studies, including reporting biases, choices of study eligibility criteria, and data that are published. Inclusion of potential future RCTs, especially head-to-head studies, may improve the generalizability and precision of NMA findings.

Data collected for this analysis corresponded to the timepoints in which primary endpoints were assessed across studies and therefore vary (i.e., 12–16 weeks). Likely owing to low event rates and low sample size, some NMA estimates are prone to notable uncertainty (e.g., SEC 150 mg without LD in the TNF-IR population). Results should be considered in this context. Furthermore, while SEC and TOF are included in the analyses for the bio-IR and TNF-IR networks, it was noted in NICE TA407 that SEC’s MEASURE 1 and MEASURE 2 were not adequately powered to detected differences in TNF inhibitor-naïve and TNF inhibitor-inadequate responder subgroups [29]. For the bio-IR and TNF-IR networks, limited trials in these populations resulted in networks that are sparsely populated, resulting in only fixed effects models producing results that passed face validity.

The CPR calculation does not consider long-term outcomes, nor does it inherently account for other potential direct or indirect medical costs other than direct treatment costs. Likewise, ongoing development, and potential future approval of biosimilar agents, especially those approved with interchangeability, may impact pricing utilized in future CPR calculations. Two commonly used therapies, infliximab and certolizumab, were not included in this analysis as there were no studies that met the inclusion criteria of the NMA and could not be included without introducing uncertainty or violating core assumptions of NMA.

Safety outcomes were not reported for bio-naïve or bio-IR specific populations but were pooled in some of the original datasets of mixed population studies. As such, safety analyses described here are for pooled populations to ensure inclusion of relevant comparators such as SEC and TOF. Splitting efficacy analysis between bio-naïve and bio-IR populations while pooling safety analysis across these populations may also be viewed through the lens of treatment effect modification. Splitting NMA networks and utilizing other adjustment techniques are common approaches to imbalances in treatment effect modifiers across studies. It may be argued that bio-naïve and bio-IR populations will respond differently to differing treatments in efficacy measures but not in treatment-associated AEs. As such, bio-naïve or bio-IR status may be considered a treatment effect modifier, and thus conducting separate bio-naïve and bio-IR efficacy analysis but pooled safety analysis is justifiable. Over time, the use of biologics has also become more common, with treaters becoming more comfortable assessing treatment outcomes and managing associated adverse events. As such, current rates of reported adverse events may differ from that reported during the earlier stages of biologic use; further research is needed to address this.

The current analysis is limited to efficacy outcomes at 12–16 weeks and safety outcomes at 12–24 weeks based on data obtained from RCTs. These timepoints are utilized to adhere to the underlying assumptions of NMA; timepoints assessed are limited to the controlled, closed-label, and randomized phases of applicable RCTs. Future RCTs may allow for formal assessment of long-term outcomes in NMA. The results remain to be validated using real-world data that reflect real-world treatment utilization and outcome patterns in the AS population in order to evaluate the generalizability of the results. Overall, given the overlapping and wide 95% CrI ranges, outcome differences should be cautiously interpreted. While ASDAS remission is a useful endpoint, such data were not included as an efficacy parameter in this systematic literature review; however, this could be addressed in future research.

Conclusions

The current analyses are the first to compare the clinical and economic benefit among approved advanced therapies for the treatment of AS. In this study, the findings suggest UPA 15 mg provides greater clinical and economic benefit as demonstrated by high response rates on ASAS40 as well as low NNT and CPR values regardless of treatment status with biologic therapies. Further analyses are warranted to confirm these findings via head-to-head or real-world comparisons of these therapies for patients with AS, as such data may be helpful to inform clinical decision-making when considering treatment options for patients with AS.