Persistence and Long-Term Disease Control of Interleukin-17 Inhibitors and Interleukin-23 Inhibitors in Patients with Psoriasis: A Nationwide Cohort Study in Japan

Introduction

Psoriasis is a chronic, immune-mediated inflammatory skin disease, affecting approximately 2% of the global population and 0.34% of the Japanese population.^1,2 The interleukin (IL)-23/IL-17 immune axis plays a central role in the immunopathogenesis of psoriasis, with IL-23 serving as a pivotal upstream regulator of IL-17.^3,4 Biologics that selectively block IL-23 or IL-17A provide superior efficacy compared with the earlier class of biologics for the treatment of moderate-to-severe psoriasis.^3,5–7 IL-17 inhibitors (IL-17i) typically exhibit a rapid onset of action, whereas IL-23 inhibitors (IL-23i) tend to provide more durable clinical benefit with more favorable safety profile.^4,5,8

IL-17i and IL-23i were first approved in Japan in 2015 and 2018, respectively, and both are reimbursed nationwide under the National Health Insurance.⁹ For plaque psoriasis, biologics are generally indicated for patients with body surface area (BSA)≥10% and psoriasis area and severity index (PASI) ≥ 12, comparable with globally accepted criteria for moderate-to-severe psoriasis.^10,11 Biologics are prescribed at designated medical institutions, and ongoing follow-up with regular monitoring of efficacy and safety is required after treatment initiation.

Psoriasis requires long-term treatment and ongoing management. Biologics frequently achieve satisfactory disease control, but their efficacy may decline over time. Loss of response and safety concerns are common reasons for treatment discontinuation.^11–14 Persistence is defined as the duration of continuous exposure to a biologic from initiation to discontinuation.^15,16 In real-world studies, persistence is widely used as a proxy endpoint for effectiveness and safety, particularly in analyses of claims data that typically lack clinical measures such as PASI.^17–21 Furthermore, treatment patterns after discontinuation may provide indirect evidence of disease control achieved with the prior biologic. Immediate switching to another biologic often indicates failure of the index therapy, whereas a sustained treatment-free status is typically associated with clinical remission.^9,19,22

Previous studies have generally reported longer persistence with IL-23i compared to other biologic classes.^16,19,22,23 However, real-world evidence from Asian populations remains relatively limited, with existing studies often having short follow-up or lacking adjustment for potential confounding.^24,25 To address this evidence gap, we conducted a retrospective cohort study using nationwide claims data. The study evaluated treatment persistence and post-discontinuation treatment patterns in patients with psoriasis treated with IL-17i versus IL-23i in real-world clinical practice in Japan.

Materials and Methods

Study Design and Database

This retrospective cohort study included patients with psoriasis who newly initiated treatment with an IL-17i (secukinumab/SEC, ixekizumab/IXE, brodalumab/BRO and bimekizumab/BIM) or an IL-23i (gusulkumab/GUS, risankizumab/RIS and tildrakizumab/TIL). Data collection began at each class’s market launch (2015 for IL-17i and 2018 for IL-23i) and continued through 31 December 2022, yielding maximum observation periods of up to 8 years for IL-17i and 5 years for IL-23i. The first biologic prescribed to each patient was designated as the index biologic, and the date of its initial prescription was defined as the index date. Patients were followed from the index date until the last available records no later than 31 December 2022.

The Medical Data Vision (MDV) database, one of the largest medical databases in Japan, was the data source for this study. MDV contains hospital-based claims data from over 550 hospitals and covers approximately 54 million patients across the country.²⁶ The study was conducted in accordance with the Japanese Ethical Guidelines for Medical and Biological Research Involving Human Subjects and the Declaration of Helsinki. The requirement for informed consent was waived as all personal and identifying data were anonymized in the study database.

Participants

Patients diagnosed with plaque psoriasis (International Classification of Disease version 9 [ICD-9] code 696.1x and 696.8x, and ICD-10 codes L40.0, L40.8, and L40.9) who initiated the biologics of interest between 01 January 2015 and 31 December 2022 were included. At least one psoriasis diagnosis during the 1-year baseline period prior to the index date was required. Exclusion criteria included age <18 years at the index date and a diagnosis at baseline of other conditions that can be treated with biologics, including rheumatoid arthritis, ankylosing spondylitis, juvenile idiopathic arthritis, Crohn’s disease, ulcerative colitis, non-infectious uveitis, and hidradenitis suppurativa. A 1-year washout of biologic exposure prior to the index date was required. Each patient was considered in the study population only once. Patients were grouped into either the IL-17i cohort or IL-23i cohort based on the index biologic. Persistence of the index biologic was evaluated for all patients. Post-discontinuation treatment patterns were analyzed for subsets of patients who discontinued their index biologic.

Study Outcomes

Persistence of the Index Biologic

Persistence was defined as the duration of continuous exposure to the index biologic from initiation to discontinuation. Discontinuation was determined through three steps. First, days of supply for each index biologic were identified based on the maintenance dosing interval recommended by the label (BRO 2 weeks, SEC/IXE/BIM 4 weeks, GUS 8 weeks, RIS/TIL 12 weeks). Second, prescription dates were extracted from the claims data. For a small number of prescriptions that included multiple doses, the number of administrations and the corresponding administration dates were inferred from the total dose and the per-administration dose. Third, discontinuation was defined as a prescription gap, which is the interval between two consecutive prescriptions or the interval between the last prescription and the study end date (31 December 2022), exceeding a predefined threshold.^15,16 In our study, the maximum allowed prescription gap in the primary analyses was days of supply plus one maintenance dosing interval. The sensitivity analyses employed two alternative definitions of the maximum allowed prescription gap: (1) days of supply plus two maintenance dosing intervals and (2) days of supply plus fixed 90 days. (Figure S1a).

Bio-Switch and Treatment-Free Status

After discontinuation of the index biologic, we continued to assess the subsequent treatment patterns to provide more insights into disease control. Bio-switch was defined as initiating a subsequent biologic before the end of the index biologic’s prescription gap (Figure S1b). The duration of the treatment-free period was defined as the interval between the discontinuation date and the first occurrence of any systemic therapy (biologic or non-biologic) for psoriasis.¹⁹ Patients who were receiving concomitant systemic therapy at the discontinuation date were excluded from the treatment-free analysis.

Control of Bias

To compare the persistence of IL-17i versus IL-23i, propensity scores (PS) were generated using a logistic regression model incorporating covariates that were considered as potential confounders based on prior knowledge, including baseline demographic characteristics (age and sex), comorbid conditions (Charlson comorbidity index [CCI], psoriatic arthritis [PsA] and depression), year of biologic initiation, history of topical and nonbiologic systemic treatments for psoriasis, treating hospital size, and the duration of psoriasis disease. PS-based inverse probability of treatment weighting (IPTW) methods and PS-based matching were employed to balance potential confounders between cohorts. The comparability of groups was assessed using standardized mean differences (SMD), where a SMD of <10% indicated well-balanced characteristics.

To compare treatment-free status, new PS were established based on patient characteristics during a newly defined baseline period of 12 months prior to the discontinuation date, and both PS-IPTW and PS-matching methods were applied. An SMD of <10% indicated well-balanced characteristics.

Statistical Analysis

Categorical variables were described with counts and percentages, and continuous variables were described with means and standard deviations (SD) and medians with first and third quartiles (Q1-Q3). Persistence was assessed by Kaplan-Meier (KM) methodology from the index date to discontinuation. Patients without discontinuation events were censored at the last available record (no later than 31 December 2022). The median persistence time (when ≥50% of patients remained persistent, if reached) and the probability of persistence at each year following treatment initiation were reported. Stratified analyses of persistence were performed based on age (<50 or ≥50 years), year of biologic initiation (before or after IL-23i launch in 2018), and baseline PsA status. The proportion of patients who underwent bio-switch was reported. Among the patients without bio-switch, the duration of the treatment-free period was estimated by KM methodology from the discontinuation date to the first consumption of any systemic therapy for psoriasis. Patients without post-discontinuation systemic therapy were censored at the last available record (no later than 31 December 2022).

In the IPTW analyses, inverse-probability-of-treatment weights were estimated and applied in weighted Cox proportional hazard models to compare the risks of discontinuation (persistence analysis) and resumption of systemic therapy (treatment-free analysis). Results were presented as hazard ratios (HR) with 95% confidence intervals (CI). In the PS-matching analyses, patients were matched 1:1 without replacement using a caliper width of 0.2 times the pooled standard deviation of the propensity-score logit. Cox models were then fitted in the matched samples to estimate HR (95% CI) for discontinuation and for post-discontinuation systemic therapy. PsA was considered as a potential effect modifier for persistence.²¹ Cox models on persistence including an interaction term between the index biologic and PsA status were fitted to report stratum-specific HRs. Sensitivity analyses of treatment-free status were conducted in a subset of patients with long-term persistence (ie, persistence ≥6 months) of the index biologic.

Data management and analyses were performed using SAS 9.4 (SAS Institute Inc., NC, USA).

Results

Participants and Baseline Characteristics

This study included 1,751 patients in the IL-17i cohort (SEC n=919, IXE n=485, BRO n=319 and BIM n=28), and 1,721 patients in the IL-23i cohort (GUS n=828, RIS n=742 and TIL n=151) (Figure 1). In the IL-17i and IL-23i cohorts, 67.7% and 64.1% of patients were male, with median ages of 55 [Q1-Q3: 45–67] and 58 [Q1-Q3: 48–71] years, respectively. Approximately 45% of patients in both cohorts had received non-biologic systemic treatments for psoriasis during the baseline period, and 9.5% had been treated with phototherapy. A higher proportion of patients in the IL-17i cohort (21.1%) had comorbid PsA at baseline compared to the IL-23i cohort (11.7%). The distribution of other comorbidities was generally similar across cohorts, with both exhibiting a median CCI of 0 [Q1-Q3: 0–1] (Table 1).

Table 1 Baseline Characteristics of Patients Receiving Interleukin-17 Inhibitors and Interleukin-23 Inhibitors for the Persistence Analysis (Original Population)

Figure 1 Flowchart of participant selection. *Patients for persistence analysis. Abbreviations: IL-17i, interleukin-17 inhibitors; IL-23i, interleukin-23 inhibitors.

Persistence

Original Population

The IL-17i cohort exhibited a 1-year persistence rate of 55.7% [95% CI 53.2–58.1%], which declined to 21.5% [95% CI 18.9–24.2%] at year 4. The 1-year persistence rate in IL-23i cohort was 77.7% [95% CI 75.4–79.8%] and decreased to 47.8% [95% CI 42.2–53.2%] at year 4 (Figure 2). Median persistence durations were 1.3 years for the IL-17i cohort and 3.9 years for the IL-23i cohort. Persistence patterns for IL-17i remained similar before and after the launch of IL-23i in 2018 (Figure 3a). When stratified by age, higher persistence with IL‑23i than IL‑17i was observed in both the <50 and ≥50-year age groups (Figure 3b). When stratified by baseline PsA status, persistence trends differed between cohorts across PsA strata, indicating that PsA may act as an effect modifier of the association between biologics and persistence (Figure 3c).

Figure 2 Kaplan-Meier curves for biologic persistence in the original population. Shading indicates 95% CI. Abbreviations: IL-17i, interleukin-17 inhibitors; IL-23i, interleukin-23 inhibitors.

Figure 3 Kaplan-Meier curves for biologic persistence in the original population. (a) Stratified by launch year, (b) Stratified by age group, (c) Stratified by baseline PsA status. Shading indicates 95% CI. Abbreviations: IL-17i, interleukin-17 inhibitors; IL-23i, interleukin-23 inhibitors; PsA, psoriatic arthritis.

Balanced Populations

After PS-IPTW, data were derived for 1,750 and 1,735 individuals in the IL-17i and IL-23i cohorts, respectively, with all SMD < 10% (Table S1), indicating good covariate balance. KM persistence curves differed significantly between the weighted IL-17i and IL-23i cohorts (Figure S2a), and the adjusted hazard ratio (aHR) for discontinuation was 0.49 (95% CI 0.44–0.54) for IL-23i versus IL-17i (Figure 4). When the interaction term of baseline PsA and biologic class was included, the stratum-specific aHRs was 0.74 (95% CI 0.58–0.94) in the PsA subgroup and 0.44 (95% CI 0.39–0.50) in the non-PsA subgroup (Figure 4). In the PS-matching analysis, 1,141 individuals were matched in each cohort with all SMD <10% (Table S1), and the adjusted KM curves and aHRs were similar to the IPTW results (Figures S2b and 4). Sensitivity analyses using alternative definitions of discontinuation (days of supply plus two maintenance dosing intervals or plus 90 days) produced results that were directionally consistent with the primary analyses (Figures S3a-c and S4a-c).

Figure 4 Cox proportional hazard analysis for biologic persistence in patients with psoriasis using different models (IL-23i vs IL-17i). Abbreviations: IL-17i, interleukin-17 inhibitors; IL-23i, interleukin-23 inhibitors; PSA, psoriatic arthritis; IPTW, inverse probability of treatment weighting.

Post-Discontinuation Treatment Patterns

Original Population

Among the 1,067 patients who discontinued their index IL-17i, 265 switched to another biologic shortly after discontinuation, representing 15.1% of the baseline IL-17i cohort. In the IL-23i cohort, 479 discontinued the index IL-23i and 155 underwent a bio-switch, representing 9.0% of the baseline IL-23i cohort. The remaining patients eligible for the treatment-free analysis numbered 779 and 309 in IL-17i and IL-23i cohort, respectively (Figure 1). The IL-17i cohort was younger than the IL-23i cohort with median ages of 57 [Q1-Q3: 45–70] vs 62 [Q1-Q3: 49–73] years. The IL-17i cohort had higher proportions of males (70.2% vs 62.8%) and baseline PsA cases (22.3% vs 10.7%) (Table 2). The 1-year treatment-free rates were 19.2% [95% CI 16.1–22.4%] for IL-17i cohort and 34.5% [95% CI 27.8–41.2%] for IL-23i cohort (Figure 5).

Table 2 Baseline Characteristics of Patients Receiving Interleukin-17 Inhibitors and Interleukin-23 Inhibitors for the Treatment-Free Analysis (Original Population)

Figure 5 Kaplan-Meier curves for free from systemic treatment after biologic discontinuation in the original population. Abbreviations: IL-17i, interleukin-17 inhibitors; IL-23i, interleukin-23 inhibitors.

Balanced Populations

After PS-IPTW, data were derived for 779 IL-17i and 309 IL-23i patients, with all SMD <10% (Table S2); KM curves for treatment-free status differed significantly between the weighted cohorts (Figure S5a), and the aHRs for resumption of systemic therapy was 0.57 (95% CI, 0.49–0.67) for IL-23i versus IL-17i (Figure 6). In the PS-matching analyses, 307 patients were matched in each group with all SMDs <10% (Table S2), and the adjusted KM curves and aHRs were similar with the IPTW results (Figures S5b and 6). Sensitivity analyses restricted to subsets of patients with long‑term persistence (≥6 months) yielded directionally consistent HR estimates (Figure 6).

Figure 6 Cox proportional hazard analysis for treatment-free status after biologic discontinuation in patients with psoriasis using different models (IL-23i vs IL-17i). Abbreviations: IL-17i, interleukin-17 inhibitors; IL-23i, interleukin-23 inhibitors; IPTW, inverse probability of treatment weighting.

Discussion

This retrospective cohort study compared IL‑17i with IL‑23i in patients with psoriasis using nationwide Japanese claims data and employed PS-IPTW and PS-matching to control for potential confounding. The study covered a long period of observation and included a substantially larger sample than prior real‑world studies in Japan. Longer persistence was observed for IL-23i than IL-17i when discontinuation was defined by a prescription gap of days of supply plus one maintenance dosing interval, which was our primary analysis. Sensitivity analyses using two alternative definitions, days of supply plus two maintenance dosing intervals or plus fixed 90 days, yielded results in the same direction. IL‑23i was also associated with a longer post-discontinuation treatment‑free duration than IL‑17i, and this pattern remained in sensitivity analyses restricted to patients with long‑term persistence.

Several Japanese studies have evaluated persistence with IL-17i and IL-23i,^{14,24,25,27,28} including three analyses using the Japan Medical Data Center (JMDC), which is an employer-based claims database containing data on approximately 17 million insured individuals.^24,25,27 Across the three studies, the 1-year persistence for IL-17i ranged from 38%-73%, whereas the 1-year persistence for IL-23i remained above 70%. These results are broadly consistent with our study (1-year persistence: IL-17i 56% vs IL-23i 78%), supporting the external validity of our findings within the Japanese context. Studies using the United States claims data reported 1-year persistence of approximately 45–53% for IL-17i and 59–69% for IL-23i.^16,19,22,23 A multicenter cohort study covering 16 dermatology centers in Europe and North America reported 18-month persistence of 80–86% for IL-17i and 91–96% for IL-23i.²⁹ The heterogeneity across studies may reflect differences in healthcare settings, data sources, population characteristics, follow-up duration, and the definitions of persistence and discontinuation.

Biologic therapies for psoriasis generally require continuous treatment, and decisions about discontinuation are made based on comprehensive clinical considerations.^9,18 The most common reason for discontinuation is loss of efficacy.^14,20,21 Yanase et al reported rates of loss of efficacy ranging from 5% to 22% for IL‑17i and <5% for IL‑23i.¹⁴ Safety concerns also contribute to discontinuation.³⁰ Although both IL-17i and IL-23i are generally well tolerated in real-world settings, some studies have reported a higher occurrence of adverse events (AEs) during IL-17i than IL-23i therapy.^5,21 From another perspective, patients and physicians may electively discontinue therapy when a satisfactory response, such as PASI-90/100 (clinical clearance) or Dermatology Life Quality Index remission, has been achieved and maintained. To date, there has been no global consensus on the optimal timing of biologic discontinuation.^10,18 Japanese guidelines recommend that discontinuation can be considered after remission has been sustained for 9–12 months.⁹

Multiple factors may explain the longer persistence frequently observed with IL-23i relative to IL-17i. Mechanistically, IL-17 lies downstream of IL-23 in the pathogenic IL-23/IL-17 immune axis. Accordingly, blockade of IL-17 commonly yields a more rapid symptomatic response, whereas IL-23 inhibition, by targeting upstream drivers of Th17 differentiation, may lead to a slower onset but more sustained modification of the pathogenic pathway.^3,4,8 However, agent-specific pharmacokinetics/pharmacodynamics may modify these general patterns.³¹ IL-17i are administered at higher doses and with greater frequency than IL-23i, which may also contribute to the higher occurrence of AEs observed with IL-17i.⁵ In addition, higher administration frequency can reduce the convenience of treatment and impair adherence, thereby increasing the likelihood of treatment discontinuation.²³ Other non-clinical factors that may affect real-world persistence include accessibility of care, drug availability, cost and reimbursement, patient and clinician preferences, and possible administrative issues, among others.

In our study, PsA modified the persistence curves of IL-17i and IL-23i in opposite directions (Figure 3c) and the interaction term was statistically significant (p<0.05), indicating that PsA was an effect modifier of persistence.²¹ Baseline PsA was present in 21.2% of patients in the IL-17i cohort and 11.7% in the IL-23i cohort. In the IL-17i cohort, patients with PsA showed slightly higher persistence than those without PsA, whereas in the IL-23i cohort persistence tended to be lower among patients with PsA. Multiple studies have demonstrated the preventive effect against joint destruction for IL-17i, and the current Japanese guidelines recommend tumor necrosis factor-α inhibitors and IL-17i as the preferred options for patients with concomitant PsA.^9,32 Although beneficial effects of IL-23i on joint symptoms have been reported in clinical studies, longer follow-up is needed to confirm their effectiveness in joint prevention in real-world settings.^33,34

Guidelines recommend continued follow-up after biologic discontinuation, and post-discontinuation treatment patterns can provide real-world insights into long-term disease control.¹⁰ In our study, 15% of patients with IL-17i and 9% of patients with IL-23i switched to a subsequent biologic shortly after discontinuation, a pattern commonly interpreted as reflecting inadequate effectiveness or safety concerns with the preceding biologic.^9,35 Analyses using US claims data reported an overall 1-year bio-switch rate of 14%, with the lowest switch rates observed for IL-23i compared with other biologic classes.³⁶ A chart-review study in England reported a 1-year bio-switch rate of 12%, where all switches were due to ineffectiveness and no patient achieved PASI 75.³⁷

In our study, 19.2% of IL-17i patients and 34.5% of IL-23i patients remained free of any systemic therapies of psoriasis at 1 year after biologic discontinuation. Cox models showed that patients in the IL-23i cohort had 40–50% lower hazard of resuming systemic treatment, which has been used as a proxy for disease relapses, compared with the IL-17i cohort (aHR ranged 0.5 to 0.6), and similar findings were observed in the long-persistence subgroups (persistence > 6 months). A prior review reported that the longest time to relapse was observed with IL-23i compared with other biologics.³¹ The relatively shorter time to relapse of IL-17i may reflect its downstream position in the IL-23/IL-17 pathway and the correspondingly shorter duration of effect following IL-17 blockade.⁵ A study using US claims data reported 1-year treatment-free rates of approximately 16–18% for IL-17i and 28% for IL-23i, broadly concordant with our findings.¹⁹ Zhdanava et al, used treatment-free status as a proxy for disease remission and found that patients treated with IL-23i were about 30% more likely to achieve remission than those treated with IL-17i.²² Sustaining a treatment-free status is a desirable objective in real-world psoriasis management. In our study, we cannot exclude that non-clinical factors such as patient preferences, cost, healthcare access, as well as data completeness, may have also influenced the observed results.

Study limitations must be acknowledged. First, claims data capture prescription information which may not accurately reflect actual medication administration. In Japan, IL-23i must be administered in medical institutions, so prescription records are likely to closely reflect actual drug administration. However, IL-17i are allowed to be administered by self-injection, which may lead to over- or underestimation of persistence patterns when they are based on prescription data. Second, claims data lack clinical information such as PASI and AEs, and the underlying reasons for biologic discontinuation and treatment switching are unknown, therefore persistence and treatment-free status are considered as proxy measures and not direct indicators of disease control. Third, although the measured covariates were balanced by PS-related methods, residual confounding from unmeasured factors may bias the estimated associations. Fourth, the MDV is a hospital-based database and follow-up may be interrupted if patients switch hospitals. Even so, Japan’s strict biologic management policies typically ensure that patients consistently attend the same facility for biologic therapy, mitigating this limitation.⁹

Conclusion

In this retrospective cohort study, IL-23i was associated with longer treatment persistence and a longer post-discontinuation treatment-free period compared with IL-17i in patients with psoriasis. These findings may provide actionable insights for healthcare providers and patients as they develop treatment strategies and plan the long-term management of psoriasis. Although PS-IPTW and PS-matching were employed to adjust for covariates, the limited clinical information available in claims data may leave residual confounding. Future research integrating comprehensive clinical data (such as PASI) is warranted to compare different biologic treatment strategies and assess their impacts on disease control, thereby informing clinical decision-making.