In this real-world study of patients with active PsA, guselkumab was associated with significantly higher on-label persistence than SC TNFi through 24 months, regardless of prior exposure to bDMARDs. Patients receiving guselkumab were more than twice as likely to remain persistent with on-label therapy through 24 months relative to those receiving an initial SC TNFi. Similarly, biologic-naïve and biologic-experienced patients were approximately 2.4 and 1.9 times more likely, respectively, to remain persistent with on-label therapy through 24 months when initiating guselkumab versus SC TNFi. Moreover, results were robust when different therapy exposure gaps were used to define treatment discontinuation in sensitivity analyses. These real-world findings complement the phase 3, randomized, placebo-controlled trials, DISCOVER-1 and DISCOVER-2, in which guselkumab demonstrated efficacy in improving the signs and symptoms of PsA, including achievement of almost clear or clear skin, resolution of enthesitis and dactylitis, and minimal disease activity [7, 8].

The study conducted by Walsh et al. was the first to evaluate guselkumab and SC TNFi on-label persistence rates in a similar patient population through 12 months of treatment [15]. We now report findings through 24 months among patients with active PsA initiating guselkumab or SC TNFi in real-world clinical practice in the USA. The current study is consistent with the 12-month data from Walsh et al., in which on-label persistence rates were 71.5% among patients initiating guselkumab and 43.7% among those initiating SC TNFi, with the former being three times more likely to remain persistent on treatment at 12 months (P < 0.001). As such, the current study builds upon the prior analysis by confirming these first-year trends and providing evidence of sustained persistence for guselkumab over an extended follow-up of 24 months. Furthermore, the current analysis also evaluated persistence by previous biologic exposure and found that treatment persistence was observed to be consistently higher for guselkumab than SC TNFi in both patient subgroups.

Limited data are available on treatment persistence with on-label guselkumab. A previous retrospective analysis evaluated treatment effectiveness and persistence among patients with PsA in the CorEvitas Psoriatic Arthritis/Spondyloarthritis Registry who initiated guselkumab. Nearly 80% of those who initiated on-label guselkumab, 73% of whom had received at least one prior TNFi, were persistent with on-label therapy through 6 months [25]. Consistent with the previous analysis, the on-label persistence rate observed for guselkumab at 6 months in the current study was 82.1%, although patients did not have prior TNFi treatment by design. Notably, the high guselkumab persistence rates seen in the present study are also consistent with retention rates of guselkumab-randomized participants reported in the phase 3 DISCOVER-1 (94% at 48 weeks) and DISCOVER-2 trials (90% at 100 weeks) [26, 27]. With regards to TNFi, treatment persistence in PsA is well characterized in the literature, with recent studies evaluating biologic-naïve [28, 29] and biologic-experienced [30, 31] patients separately. Among biologic-naïve patients with PsA, persistence rates for TNFi (including IV agents) ranged from 37.1% (using a therapy exposure gap of 30 days) [28] to 72.3% (using a gap of 60 days) [29] at 1 year in two insurance claims-based studies, which is a range that encompasses the rate of 43.8% observed for biologic-naïve patients in the current study. Among biologic-experienced patients with PsA, the present on-label persistence rates of 40.3% at 12 months and 23.3% at 24 months are aligned with two prior observational studies estimating TNFi persistence rates of 36.8% (using a gap of 60 days) to approximately 60% (no gap defined) at 1 year and 21.3% to approximately 50% at 2 years [30, 31]. While comparability between these prior studies and the current analysis is limited by different persistence definitions, TNFi evaluated, and patient populations, the findings were generally consistent.

Despite the lack of comparative PsA literature, prior studies have shown higher persistence with guselkumab relative to TNFi among patients with psoriasis [21, 32, 33], thus reinforcing similar trends observed in the treatment of PsA in the present study. In one claims-based study including both biologic-naïve and biologic-experienced patients with psoriasis, the median time to discontinuation was 26.2 months with guselkumab and 9.9 months with adalimumab [21], which is aligned with the current study estimates of 22.0 and 9.2 months (for SC TNFi overall), respectively. Of note, IL-23p19-subunit inhibitors have demonstrated efficacy in achieving high levels of response in patients with psoriasis [34, 35]. In phase 3 studies of participants with plaque psoriasis, guselkumab was superior to adalimumab in achieving almost clear/clear skin at week 16 [36, 37]. As such, the longer persistence with guselkumab observed in the current study may have been influenced by the efficacy of guselkumab specifically in treating skin symptoms of PsA.

Given the scarcity of head-to-head comparisons between guselkumab and SC TNFi to guide PsA treatment decisions, the present analysis contributes important insights using treatment persistence, a key clinical measure that informs both effectiveness and safety of therapy. Indeed, the most common reasons for discontinuing bDMARDs among patients with psoriasis and PsA are lack of effectiveness and occurrence of adverse effects, suggesting that long-term treatment persistence may reflect positive health outcomes and optimal disease control [14, 38]. In support of this line of thought, on-label persistence with guselkumab for 6 months in the CorEvitas analysis was accompanied by significant improvements in joint, skin, and pain symptoms among patients with PsA [25]. The ability of guselkumab to improve PsA symptoms across multiple domains indicates a broad applicability for guselkumab in the treatment of PsA, based on current guidelines that recommend a domain-based approach to treatment selection [4]. Additional analysis is warranted to confirm the link between guselkumab persistence and disease management outcomes in real-world patients with PsA.

Despite high rates of on-label persistence at 6 months, persistence with both guselkumab and SC TNFi did steadily decrease by 24 months. Although reasons for treatment discontinuation were not available in the claims data, predictors of low bDMARD persistence in PsA have been identified in the literature, including female sex, smoking, comorbidities commonly associated with PsA (e.g., obesity, cardiac conditions), and prior bDMARD/systemic treatment [28, 29, 31, 39, 40]. Consistently, persistence outcomes tended to be better in biologic-naïve than biologic-experienced patients in this study, and other patient characteristics may have also influenced persistence in both cohorts. Future research may evaluate the impact of comorbidities and prior exposure to different bDMARD classes on guselkumab persistence in patients with PsA.

The strengths of the study include its large sample size and the use of real-world data, which enhance the generalizability of the findings. The study utilized rigorous statistical methods, including overlap propensity score weighting, to balance the cohorts and address potential confounding variables. Additionally, the analysis spanned 24 months, providing a robust assessment of long-term treatment persistence outcomes.

Some limitations of the study design should be noted. Imputation of days of supply is a valid approach commonly used for claims-based persistence analysis [15, 21, 41]; however, it may occasionally lead to misclassification of treatment discontinuation. For instance, a filled prescription observed in the claims data does not necessarily ensure that the treatment was taken as prescribed. Additionally, treatment effectiveness and reasons for discontinuation (e.g., lack of effectiveness, adverse effects, inconvenient dosing, worsening of related conditions) could not be assessed using claims data. While patients could initiate other treatments following the index date, combination therapy (if any) was not considered in the persistence analysis. Patients with exposure to TNFi prior to initiating guselkumab were not included in this study. Moreover, SC TNFi were evaluated collectively rather than individually, consistent with how these therapies are addressed in current treatment guidelines [4].

While the two cohorts were balanced using overlap propensity score weighting, variables that were underreported or not available in the database, such as psoriasis severity, disease activity, domain involvement, and clinical manifestations, may have led to residual confounding. As there was only limited information on provider specialty available in the database, this was not included in the analysis. Some baseline characteristics remained imbalanced after weighting (e.g., use of bDMARDs and csDMARDs); however, their inclusion as covariates in the weighted persistence analysis minimized any confounding effects. Additionally, although speculative, the exclusion criterion of a prior diagnosis code for ankylosing spondylitis may have inadvertently prevented inclusion of some patients with PsA and axial disease in the analysis population. Conversely, patients with prior or active exposure to treatments indicated only for plaque psoriasis at the time of study conduct (e.g., tildrakizumab, brodalumab) were included in the analysis, and their effects on guselkumab or SC TNFi persistence are unknown. As with all claims-based studies, there may have been billing inaccuracies or omissions in coded procedures, diagnoses, and pharmacy claims, and the extent of these factors is unknown. Lastly, as this study included patients with commercial health plans, the results may not be generalizable to patients with no or other health insurance coverage, or patients outside of the USA.