Cycle versus swap strategy after TNFi discontinuation in psoriatic arthritis and axial spondyloarthritis: a quasi-experimental study

Abstract

Objectives

To compare a bDMARD mode of action cycle vs swap treatment strategy in patients with psoriatic arthritis (PsA) or axial spondyloarthritis (axSpA) after first tumour necrosis factor inhibitor (TNFi) discontinuation.

Methods

In December 2019, our local treatment protocol for PsA and axSpA changed from a cycle strategy (first TNFi to second TNFi) to a swap strategy (first TNFi to IL-17i). We performed a retrospective comparison of the 3-year drug retention rate using multivariable Cox regression (ref: cycle group) and disease activity (DAS28-CRP for PsA, BASDAI for axSpA) in patients with a clinical diagnosis of PsA and axSpA. For subgroup analyses, Cox regression models were stratified by sex, reason of first TNFi discontinuation, and (non-)radiographic status in axSpA.

Results

In PsA patients (n=406), there was no overall significant difference in drug retention between strategies (HR: 1.17 (95% CI: 0.87 to 1.58), p=0.29), but male PsA patients had a significant higher risk for treatment discontinuation following a swap strategy. In axSpA patients (n=335), the swap strategy was overall associated with a higher risk of treatment discontinuation (HR: 1.46 (95% CI: 1.03 to 2.07), p=0.04). Patients who discontinued their first TNFi due to inefficacy and patients diagnosed with radiographic axSpA were at significant higher risk for treatment discontinuation following a swap strategy. No significant differences in disease activity were found for treatment strategies in PsA or axSpA.

Conclusion

In PsA, the cycle and swap treatment strategy performed similarly, while in axSpA, the cycle strategy was associated with a significant higher drug retention rate.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• International treatment guidelines for psoriatic arthritis (PsA) and axial spondyloarthritis (axSpA) do recommend to prescribe another biological and targeted synthetic DMARD (b/tsDMARDs) after discontinuation of a first TNF inhibitor (TNFi), but little evidence is available to specify which mode of action (MOA) performs best.

WHAT THIS STUDY ADDS

• A change in the local treatment protocol, together with strong protocol adherence, enabled the performance of the first quasi-experimental study comparing the cycle and swap treatment strategy in routine clinical practice.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• After first TNFi discontinuation in PsA or axSpA patients, no clear benefit was seen on changing the MOA, specifically when comparing an IL-17i vs a second TNFi.

• Our results support the treatment recommendations to consider cycling within the same MOA in PsA and axSpA when prescribing a second bDMARD.

Introduction

There have been substantial advances in the treatment options for psoriatic arthritis (PsA) and axial spondyloarthritis (axSpA) over the past decades. Particularly within the group of biological and targeted synthetic DMARDs (b/tsDMARDs), many drugs have entered the market in recent years. In daily clinical practice, the first bDMARD prescribed is typically a TNF inhibitor (TNFi).¹,³ However, many patients stop treatment with a TNFi either due to inefficacy or side effects,⁴ with reported 1 year TNFi discontinuation rates of approximately 30% for both PsA and axSpA.^{5 6}

Clinicians are subsequently faced with a large array of therapeutic drug options, which can broadly be categorised based on their mode of action (MOA): TNFi, IL-17 inhibitors (IL-17i), IL-12/23 inhibitors (for PsA), IL-23 inhibitors (for PsA) and Janus kinase (JAK) inhibitors. A cycle strategy refers to prescribing the same MOA again (eg, a first TNFi followed by a second TNFi); whereas, swapping refers to prescribing a new MOA (eg, a first TNFi followed by an IL-17i). The choice between a cycle or swap strategy is becoming an increasingly important question due to the increasing number of new drugs and new MOAs available, while at the same time having the option to prescribe off-patent TNFi which have both lower costs and more available data on long-term safety.⁷

As there have been no randomised controlled trials performed addressing this question, it is uncertain whether the MOA should be taken into account when prescribing a second b/tsDMARD. International guidelines rely on observational studies comparing different MOAs to formulate a recommendation.¹,³ There are several studies suggesting that continuing with a second TNFi can still be efficacious in both PsA and axSpA,^{8 9} or at least as effective as a swap strategy.¹⁰,¹⁴ Nevertheless, one could also argue that when failing to respond to a TNFi, one should alter MOA to increase the likelihood of response.¹⁵,¹⁷ Of note, in rheumatoid arthritis, although four randomised controlled studies failed to show superiority of MOA swapping over cycling to a second TNFi,¹⁸ systematic reviews including non-randomised data suggest indeed that swapping results in better outcomes.¹⁹

We performed a quasi-experimental study in which we aimed to compare the effectiveness of a second TNFi (cycle) vs an IL-17i (swap) strategy in patients with PsA or axSpA who discontinued a first TNFi. Moreover, we aimed to identify patient, disease and treatment characteristics associated with the efficacy of a cycle vs swap treatment strategy.

Methods

At the study site (Sint Maartenskliniek, the Netherlands), a local treatment protocol determines the order in which specific b/tsDMARDs are prescribed for patients with PsA and axSpA, although allowing for deviations from the protocol based on shared decision-making (eg, safety issues or extra-musculoskeletal manifestations). The first b/tsDMARD recommended in the protocol had always been a TNFi. Prior to December 2019, the protocol stated that the second b/tsDMARD prescribed should be a second TNFi. In December 2019, the protocol was changed to an IL-17i as second bDMARD. The choice of the second b/tsDMARD in the local treatment protocol did not take the cause of failure to a previous treatment (eg, inefficacy, side effects) nor the duration of the previous treatment (eg, primary failure vs secondary failure) into account. This change of drug prescription order enabled the opportunity to conduct a quasi-experimental study comparing these treatment strategies. The local ethics committee provided exemption for this study (METC Oost-Nederland; 2023–16517), as ethical approval for this type of study is not required under Dutch law (Medical Research Involving Human Subjects Act (WMO)).

Data from 2012 until March 2023 for PsA patients and May 2023 for axSpA patients were extracted from the Integral Rheumatology Information System (IRIS) in which data of the clinical outpatient visits of all rheumatology patients of the Sint Maartenskliniek are stored, except of those who object the use of clinical data for research purposes. Data included demographic and clinical characteristics: sex, age, disease duration, Body Mass Index, and (non-) radiographic status in axSpA; disease activity parameters: for PsA, Disease Activity Score 28 with CRP (DAS28-CRP), tender joint score (TJC) and swollen joint score (SJC); for axSpA, the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) including fatigue, severity and duration of morning stiffness, axial and peripheral pain, tenderness and swelling; and medication history: cs/b/tsDMARD usage (name, class, start and stop dates) and reason for treatment discontinuation.

Patients

Patients were included in the study when they: (1) had a clinical diagnosis of PsA or axSpA; (2) had discontinued their first TNFi due to inefficacy, side effects or other reasons (such as medical conditions unrelated to PsA/axSpA, discontinuation for unknown reasons) based on the clinical judgement of the rheumatologist in shared decision with the patient; (3) were subsequently prescribed a second TNFi or a first IL-17i; and (4) had at least 6 months follow-up data on the second bDMARD or information on discontinuation within 6 months available. Patients were assigned into a cycle strategy group (ie, those who received another TNFi after discontinuation of the first TNFi), and a swap strategy group (ie, those who switched to an IL-17i after discontinuation of the first TNFi). Of note, at the study site, secukinumab is prescribed at a dose of 150 mg (once per month) without loading dose as data show no clear benefit from the loading dose.²⁰,²² Baseline assessment was defined as the date patients started their second bDMARD.

Drug retention

The primary outcome was defined as the 3-year second bDMARD drug retention rate. Secondary endpoints included the second bDMARD drug retention rate at 6, 12 and 24 months. The second bDMARD treatment was considered discontinued when a stop date was registered or a third bDMARD was started. The principal reasons for treatment discontinuation were categorised as inefficacy, side effects, remission, (planned) pregnancy and other (such as medical conditions unrelated to PsA/axSpA, discontinuation for unknown reasons). Observations of patients who discontinued their second bDMARD within 3 years due to inefficacy and side effects were considered an event. Data was considered censored if (1) patients were still on therapy at the moment of data extraction (7 March 2023 for PsA; 16 May 2023 for axSpA) or (2) if the principal reason for treatment discontinuation was categorised as remission, other or (planned) pregnancy.

Disease activity

For PsA, disease activity was assessed with the DAS28-CRP.²³ For axSpA, disease activity was assessed with the BASDAI.²⁴ Disease activity was collected at baseline (−3/+3 months), 6 months (−3/+3 months), 12 months (−6/+6 months), 24 months (−6/+6 months) and 36 months (−6/+6 months).

Statistical analysis

All analyses were separately conducted for PsA and axSpA. Baseline characteristics of the cycle and swap strategy groups were reported as mean with SD, median with IQR or percentages, and subgroups were compared using the Student’s t-test, Mann-Whitney U test or χ² test, as appropriate. Drug retention of the cycle and swap strategy was visualised using Kaplan-Meier plots. Univariate Cox regression analysis was performed with all patient, disease and treatment characteristics to check for confounding in relation to drug retention. Clinically relevant covariates were selected a priori and included sex, disease duration (years), duration of first TNFi use (months) and reason of first TNFi discontinuation (inefficacy or side effects); and (non-)radiographic axSpA status in axSpA only. HRs were calculated using multivariable Cox regression, adjusted for those considered potentially meaningful in the univariable analysis (change in HR of 10% or more) and those selected a priori.²⁵ For subgroup analyses, Cox regression models were stratified by sex, reason of first TNFi discontinuation, and primary failure (defined as discontinuation of first TNFi within 9 months because of inefficacy); and (non-)radiographic axSpA status in axSpA only. The assumption of proportional hazards was evaluated by assessment of the Schoenfeld residuals over time and showed no violation. A per protocol sensitivity analysis on drug retention was performed with local treatment protocol violation being defined as following a cycle strategy after 1 December 2019 and a swap strategy before 1 December 2019. Student’s t tests were conducted to examine the differences in disease activity (DAS28-CRP/BASDAI) between the cycle and swap strategy at 6, 12, 24 and 36 months. Furthermore, the achievement of remission or low disease activity was explored. As surrogate for the achievement of remission/low disease activity, a DAS28-CRP score of ≤2.9 and a BASDAI score of <4 were used in PsA and axSpA respectively.^{24 26} Data analyses were performed using STATA 17.0.

Results

Patient characteristics

In total, 406 PsA patients and 335 axSpA patients were included. In PsA, 307 patients followed a cycle strategy and 99 a swap strategy. In axSpA, 270 patients followed a cycle strategy and 65 a swap strategy. Baseline patient and disease-related characteristics are shown in table 1 for PsA and table 2 for axSpA. The specific bDMARDs prescribed per group are listed in online supplemental table S1a and S1b. In most cases, patients initiating their second bDMARD discontinued their first bDMARD (ie, TNFi) due to inefficacy (73.4% in PsA, 66.6% in axSpA), having been treated with the first bDMARD for a median duration of 10.2 months in PsA and 12.6 months in axSpA. Overall, there were no significant differences in baseline characteristics in the cycle and swap group for PsA and axSpA, except for a difference in the reason for first TNFi discontinuation in PsA (tables1 2).

Table 1. Baseline demographic and disease-related characteristics of patients with psoriatic arthritis.

PsA	Total(n=406)	Cycle group(n=307)	Swap group(n=99)
Male, n (%)	168 (41.4)	123 (40.1)	45 (45.6)
Age (years), mean±SD	49.0±13.8	48.3±13.4	51.2±14.8
Disease duration (years), mean±SD	6.9±9.0	6.6±6.8	7.8±13.7
Concomitant csDMARD, n (%)	202 (49.8)	154 (50.2)	48 (48.5)
DAS28-CRP, mean±SD*	2.6±1.1	2.6±1.1	2.7±1.2
TJC, median (IQR)†	4 (2–8)	4 (2–9)	2 (1–6)
SJC, median (IQR)‡	2 (1–5)	2 (1–5)	3 (1–4)
CRP, median (IQR)§	3 (1–6)	2.5 (1–6)	3.5 (1–8)
Drug retention first TNFi (months), median (IQR)	10.2 (4.8–31.1)	10.9 (5.2–31.7)	8.4 (3.8–30.4)
Previous bDMARD (first TNFi), n (%)
Adalimumab	220 (54.2)	128 (41.7)	92 (92.9)
Etanercept	150 (37.0)	143 (46.6)	7 (7.1)
Infliximab	31 (7.6)	31 (10.1)	–
Certolizumab	3 (0.7)	3 (1.0)	–
Golimumab	2 (0.5)	2 (0.6)	–
Reason first TNFi discontinuation, n (%)¶
Inefficacy	298 (73.4)	217 (70.7)	81 (81.8)
Side effects	97 (23.9)	81 (26.4)	16 (16.3)
Other	11 (2.7)	9 (2.9)	2 (2.0)

^* number of missing data, N = 192, cycle group = 147; swap group = 45.

^† number of missing data, N = 206, cycle group = 158; swap group = 48.

^‡ number of missing data, N = 216, cycle group = 164; swap group = 52.

^§ number of missing data, N = 192, cycle group = 147; swap group = 45.

^¶Significantly different (p<0.05) between cycle and swap treatment strategy group.

CRP, C reactive protein; DAS28-CRP, Disease Activity Score 28 with CRP; DMARD, disease-modifying antirheumatic drugs; PsA, psoriatic arthritis; SJC, swollen joint count; TJC, tender joint count; TNFi, tumour necrosis factor α inhibitor.

Table 2. Baseline demographic and disease-related characteristics of patients with axial spondyloarthritis.

axSpA	Total(n=335)	Cycle group(n=270)	Swap group(n=65)
Male, n (%)	161 (48.1)	130 (48.2)	31 (47.7)
Age (years), mean±SD	44.1±13.6	43.5±13.9	46.9±11.9
Disease duration (years), mean±SD	7.7±8.8	8.1±8.8	6.0±9.1
Concomitant csDMARD, n (%)	34 (10.2)	30 (11.1)	4 (6.2)
Radiographic axSpA, n (%)*	218 (65.1)	176 (65.2)	42 (64.6)
BASDAI, mean±SD†	4.3±2.0	4.3±2.1	4.3±1.7
CRP, median (IQR)‡	2 (1–8)	2 (1–8.8)	2 (1–6.3)
Drug retention first-line TNFi (months), median (IQR)	12.6 (4.9–40.6)	14.2 (5.1–43.9)	8.1 (3.8–35.9)
Previous bDMARD (first TNFi), n (%)
Adalimumab	185 (55.2)	127 (47.0)	58 (89.2)
Etanercept	81 (24.2)	76 (28.2)	5 (7.7)
Infliximab	67 (20.0)	65 (24.1)	2 (3.1)
Golimumab	2 (0.6)	2 (0.7)	–
Reason first TNFi discontinuation, n (%)
Inefficacy	223 (66.6)	175 (64.8)	48 (73.9)
Side effects	92 (27.4)	79 (29.3)	13 (20.0)
Other	20 (6.0)	16 (5.9)	4 (6.1)

^* number of missing data, N = 3, cycle group = 3; swap group = 0.

^† number of missing data, N = 201, cycle group = 164; swap group = 37.

^‡ number of missing data, N = 204, cycle group = 166; swap group = 35.

axSpA, axial spondyloarthritis; BASDAI, Bath Ankylosing Spondylitis Disease Activity Index; CRP, C reactive protein; DMARD, disease-modifying antirheumatic drugs; TNFi, tumour necrosis factor α inhibitor.

Overall drug retention rates second bDMARD

In PsA, 258 discontinuation events occurred, of which 191 (74.0%) due to inefficacy and 67 (26.0%) due to side effects. In axSpA, 202 discontinuation events occurred, of which 153 (75.7%) due to inefficacy and 49 (24.3%) due to side effects. In both PsA and axSpA, no significant differences were seen in the reason for second bDMARD discontinuation between the cycle and swap group (online supplemental table S2a and S2b). In PsA, the overall drug retention was 72.1% at 6 months and decreased to 55.7% at 12 months and 40.4% at 36 months. In axSpA, the overall drug retention was 72.4% at 6 months and decreased to 60.0% at 12 months and 42.0% at 36 months. The median drug retention in PsA was 13.3 months (IQR: 5.0–42.3) and 16.9 months (IQR: 5.1–40.9) in axSpA.

Risk at second bDMARD discontinuation cycle versus swap treatment strategy

In PsA, the risk of treatment discontinuation between the cycle and swap strategy was not significantly different (HR (ref. cycle strategy): 1.17 (95% CI: 0.87 to 1.58), p=0.29) (figure 1 and online supplemental table S3a). Stratification showed that male PsA patients had a significant higher risk of treatment discontinuation when following a swap strategy (HR: 1.64 (95% CI: 1.03 to 2.60), p=0.04). Stratification by other factors (ie, reason of first TNFi discontinuation or primary failure) did not reveal significant differences in retention rate in cycle vs swap strategy.

Figure 1. Second bDMARD drug retention in psoriatic arthritis. The Kaplan-Meier survival plot shows the 3-year drug retention rate of a second TNFi in the cycle group and a first IL-17i in the swap group in patients with PsA. Ticks indicate censored data. No significant difference in drug retention was seen between the cycle and swap strategy. TNFi, tumour necrosis factor inhibitor.

In axSpA, a swap strategy was associated with a significantly higher risk of treatment discontinuation compared with a cycle strategy (HR (ref. cycle strategy): 1.46 (95% CI: 1.03 to 2.07), p=0.04) (figure 2 and online supplemental table S3b). Stratification showed similar results for patients diagnosed with radiographic axSpA (HR: 1.73 (95% CI: 1.13 to 2.66), p=0.01) and patients who experienced ineffectiveness of their first TNFi (HR: 1.77 (95% CI: 1.17 to 2.67), p<0.01). Also when inefficacy of the first TNFi occurred within 9 months (deemed primary failure), subgroup analysis showed a significant higher risk of treatment discontinuation when following a swap strategy (HR: 1.95 (95% CI: 1.11 to 3.41), p=0.02).

Figure 2. Second bDMARD drug retention in axial spondyloarthritis. The Kaplan-Meier survival plot shows the 3-year drug retention rate of a second TNFi in the cycle group and a first IL-17i in the swap group in patients with axSpA. Ticks indicate censored data. The cycle strategy was associated with significant higher drug retention rates. TNFi, tumour necrosis factor inhibitor.

In both PsA and axSpA, we did not observe an effect of year of starting the second bDMARD on the drug retention rate (online supplemental figures S1 and S2).

In total, 86% (349/406) of the PsA patients and 85% (285/335) of the axSpA patients were treated according to protocol. In both PsA and axSpA, the per protocol sensitivity showed no significant difference in drug retention between the cycle and swap strategy (online supplemental table S4a and S4b).

Disease activity

In PsA, the change in DAS28-CRP between baseline and 6 months was −0.03 (95% CI: −0.24 to 0.30) for the cycle strategy and −0.19 (95% CI: −0.20 to 0.59) for the swap strategy. Between baseline and 12 months, the change in DAS28-CRP was −0.01 (95% CI: −0.25 to 0.28) and −0.17 (95% CI: −0.25 to 0.58) for respectively the cycle and swap strategy. At 36 months, there was a reduction in DAS28-CRP of 0.00 (95% CI: −0.26 to 0.25) compared with baseline for the cycle strategy and −0.33 (95% CI: −0.31 to 0.96) for the swap strategy.

In axSpA, the change in BASDAI between baseline and 6 months was −0.28 (95% CI: −0.43 to 0.99) for the cycle strategy and −0.21 (95% CI: −0.96 to 1.37) for the swap strategy. Between baseline and 12 months, the change in BASDAI was 0.00 (95% CI: −0.70 to 0.70) and +0.23 (95% CI: −1.34 to 0.89) for respectively the cycle and swap strategy. At 36 months, there was a reduction in BASDAI of −0.27 (95% CI: −0.37 to 0.91) compared with baseline for the cycle strategy and −0.46 (95% CI: −1.23 to 2.16) for the swap strategy.

Regarding the achievement of remission or low disease activity in PsA, 66.7% of patients in the cycle group and 67.9% in the swap group achieved this state at 6 months. At 12 months, the rates were 66.0% and 69.4%, respectively, and at 36 months, they were 67.3% and 75.0%. In axSpA, 48.6% in the cycle group and 48.0% in the swap group achieved remission or low disease activity state at 6 months. At 12 months, the rates were 45.3% and 35.7%, respectively, and at 36 months, 46.3% and 57.1%.

In both PsA and axSpA, no significant differences were found in neither DAS28-CRP/BASDAI nor the achievement of remission/low disease activity between the cycle and swap strategy.

Discussion

This real-world quasi-experimental study comparing a cycle vs swap treatment strategy found that a swap strategy was not better than a cycle strategy in either PsA or axSpA patients when prescribing a second bDMARD in terms of drug retention and disease activity over a 3-year period. Our results contribute to the scientific evidence supporting recommendations to consider cycling within the same MOA in PsA and axSpA treatment.^{2 3}

Previous PsA and axSpA studies comparing a cycle and swap strategy after TNFi discontinuation are limited and often of low quality due to the nature of their design. Supporting our findings in PsA, one study found slightly better retention of TNFi vs secukinumab.²⁷ Also, data from a Nordic biological registry in PsA showed that TNFi was relatively efficacious as compared with other modes of action, even after multiple lines of treatment.²⁸ Whereas a MOA was found to impact retention rates in some PsA studies,^{17 29} many other studies found that MOA does not largely impact retention rates.^{11 12 30 31} In axSpA, many studies have investigated the efficacy of a second TNFi demonstrating its effectiveness.⁸³²,³⁵ Our results showed superiority of the cycle strategy in axSpA with respect to drug retention rate. In contrast to our findings, one study reported comparable drug retention and clinical efficacy between a second TNFi and switching to secukinumab.¹⁴ Another study also found similar drug retention rates comparing secukinumab to TNFi, although the secukinumab treatment was often given as third-line.¹³ The two treatment protocols we compared were applied to axSpA and PsA patients, regardless of whether they had experienced primary failure with the first TNFi or had the first bDMARD discontinued due to side effects or inefficacy. We found that, even in patients with primary failure or those in whom the TNFi was discontinued due to inefficacy, the swap strategy was still not superior to the cycle strategy. Overall, our results contribute to the growing evidence that MOA should only influence therapy choices for patients with clinically relevant extra-musculoskeletal manifestations.

The study findings should be interpreted in light of the strengths and limitations of our study design. A strength of this study is the use of data from clinical practice and the quasi-experimental design enabled by the local drug protocol change. Furthermore, a strong protocol adherence may have minimised the extent of allocation bias. The swap strategy from our study should be interpreted based on the standard secukinumab dose prescribed at the study site: 150 mg per month without loading dose. Although we cannot fully exclude that a higher dose might impact the drug retention rate, we do not expect that the dose or absence of a loading dose impacted our results.²⁰,²² In addition, time-related biases such as changes in clinical practice, the availability of more treatment options, and new treatment targets may have influenced our findings.³⁶ However, after comparison of the cycle vs swap strategy across different time frames, we found no time-related effect, suggesting that the drug retention rates of the cycle strategy were not influenced by the limited treatment options before 2018 or the reluctance of rheumatologists to change therapies. The retrospective nature of this study is an important limitation. Besides unequal sample sizes between the cycle and the swap strategy groups, information and confounding bias due to incomplete data may have occurred. Especially data with regard to disease activity measures, extra-musculoskeletal manifestations, and HLA-B27 status in axSpA patients were missing, such that we were unable to take these variables into account as possible confounder for drug retention.^{31 37 38} Third, the interpretation of our results with respect to disease activity outcome measures was hampered due to historical data being collected in a time period when suboptimal outcome measures (DAS28, BASDAI) for these diseases were still commonplace. We were therefore unable to describe for instance skin disease and enthesitis in PsA patients or ASDAS-CRP scores in axSpA. Finally, the study findings may not be generalisable to patients with spondyloarthropathies with clinically relevant extra-musculoskeletal manifestations, as our study focused on patients with clinical equipoise for cycle (TNFi to TNFi) or swap (TNFi to IL-17i) strategy.

Future research should focus on exploring whether other MOAs might be superior to a second TNFi (cycling). In the psoriasis field, there is robust data showing that specific MOAs are superior for treating psoriasis.³⁹ However, it remains to be seen whether new MOAs are superior to a second TNFi for treating axial disease in axSpA or treating arthritis in PsA. The ideal setting to test this would be a randomised controlled strategy trial. Given the challenges of conducting head-to-head trials with all available MOA options, we suggest that any new drug entering the spondyloarthritis market be evaluated in TNF-IR patients and that the trial always includes an active comparator arm (specifically a second TNFi, ideally a biosimilar) to formally test the added benefit of new MOAs to the spondyloarthritis treatment strategy.

In conclusion, we found no clear benefit of changing the MOA in either PsA or axSpA patients when prescribing a second bDMARD over a 3-year period. Notably, our results specifically apply to PsA and axSpA patients starting a second bDMARD, where there is clinical equipoise between choosing a second TNFi or first IL-17i. For these patients, where extra-musculoskeletal manifestations or comorbidities do not influence the decision, our results suggest that other factors such as safety profile, patient preferences, and costs, rather than MOA, should guide the choice of the second bDMARD.²

Data availability statement

Data are available upon reasonable request.