Impact of Biologic Treatment Duration on Drug-Free Remission in Moderate-to-Severe Psoriasis: A Retrospective Cohort Study Authors: Che-Chia Hsu, Yu-Huei Huang, Tsen-Fang Tsai Published: 2026 ABSTRACT IntroductionPsoriasis is a chronic inflammatory disease that often requires long-term treatment. However, treatment interruption is common. Although several predictors of drug-free remission (DFR) have been reported, the impact of treatment duration remains unclear. In Taiwan, biologics were reimbursed for up to 2.5 years, with re-treatment initiated after relapse. This study compared DFR between reimbursed patients and those enrolled in phase III pivotal trials, where treatment courses were at least 4 years.MethodsThis two-center, retrospective study included patients with moderate-to-severe plaque psoriasis treated with secukinumab, guselkumab, or risankizumab. For reimbursement, patients were required to have Psoriasis Area and Severity Index (PASI) ≥ 10, whereas pivotal trials required PASI ≥ 12 and body surface area ≥ 10%. Patients were classified into short-term (≤ 2.5 years, reimbursed) and long-term (4–6 years, clinical trials) treatment groups. Disease relapse was assessed at each visit, and DFR was evaluated using Kaplan–Meier analysis and Cox regression.ResultsAmong 164 patients, 51 received long-term therapy and 113 received short-term therapy. Long-term therapy was associated with longer DFR (median 211 vs. 111 days, log-rank p < 0.001). In multivariate analysis, long-term risankizumab reduced relapse risk by 66% (hazard ratio [HR] 0.34, 95% confidence interval [CI] 0.13–0.90, p = 0.03), with a non-significant trend for long-term guselkumab (HR 0.55, 95% CI 0.28–1.06, p = 0.07). In contrast, no significant difference was observed for secukinumab (log-rank p = 0.94). In the guselkumab subgroup, continuous long-term therapy resulted in numerically higher remission days per dose than interrupted sequential short-term therapy (9.61 ± 11.43 vs. 8.23 ± 2.78, p = 0.64).ConclusionLong-term treatment with interleukin (IL)-23 inhibitors (particularly risankizumab), but not anti-IL-17 inhibitors, may prolong DFR, suggesting a potential disease-modifying effect of prolonged IL-23 inhibition. These findings should be considered preliminary and warrant confirmation in larger prospective studies.Supplementary InformationThe online version contains supplementary material available at 10.1007/s13555-026-01785-z. Key Summary Points Why carry out this study? In real-world practice, discontinuation of biologics in psoriasis is common.Although predictors of drug-free remission (DFR), such as disease duration and mechanism of action, have been reported, the impact of treatment duration remains unclear.This study examined whether long-term biologic therapy prolongs DFR compared with short-term treatment. What was learned from the study? Long-term IL-23 inhibition, particularly with risankizumab, was associated with significantly prolonged DFR, whereas extended IL-17 inhibition did not demonstrate this benefit.These findings support consideration of longer treatment duration strategies and suggest a potential disease-modifying effect of sustained IL-23 blockade. Why carry out this study? In real-world practice, discontinuation of biologics in psoriasis is common.Although predictors of drug-free remission (DFR), such as disease duration and mechanism of action, have been reported, the impact of treatment duration remains unclear.This study examined whether long-term biologic therapy prolongs DFR compared with short-term treatment. What was learned from the study? Long-term IL-23 inhibition, particularly with risankizumab, was associated with significantly prolonged DFR, whereas extended IL-17 inhibition did not demonstrate this benefit.These findings support consideration of longer treatment duration strategies and suggest a potential disease-modifying effect of sustained IL-23 blockade. Introduction Psoriasis is a chronic inflammatory skin disease that may be associated with other comorbidities, possibly due to shared systemic manifestations [1–3]. The advent of mechanism-based therapies—including biologic agents and small-molecule inhibitors such as tyrosine kinase (TYK) inhibitors and phosphodiesterase 4 (PDE4) inhibitors—has markedly improved disease control. However, psoriasis is generally regarded as incurable and typically requires long-term continuous management to achieve sustained disease control and improved quality of life. With current novel targeted therapies, rebound is uncommon, but relapse is common and often occurs at pre-existing skin lesions, a phenomenon partly explained by the presence of tissue-resident memory T cells [3–7]. Thus, the concept of disease modification has gained increasing attention, particularly in the context of biologic therapy [1, 8, 9]. Short disease duration and early good efficacy have been associated with prolonged drug-free remission (DFR) duration in some patients and potential for dose de-escalation in some studies [3, 10]. In many other real-world settings and clinical guidelines, long-term biologic therapy is often recommended for patients with moderate-to-severe psoriasis. Treatment discontinuation in these settings more commonly results from primary or secondary loss of efficacy or adverse events rather than reimbursement restrictions [11, 12]. However, in both clinical trials and real-world practice, treatment interruption is common. This may be attributable to study design, pharmacoeconomic considerations, adverse events, or patient preference for a “treatment vacation” after achieving long-term remission [13]. In Taiwan, National Health Insurance (NHI) reimbursement policies for biologics incorporate cost considerations, requiring prior failure of conventional therapies and limiting treatment duration to approximately 2 years, with reinitiation permitted only upon relapse [14–17]. The 2-year reimbursement time frame in Taiwan provides a unique opportunity to examine the effect of treatment duration on DFR, in comparison with longer-term treatment durations of interleukin (IL) blockers in phase III pivotal studies (4–6 years). This study aimed to evaluate the impact of long-term versus short-term biologic therapy on DFR, with an exploratory analysis of the treatment efficiency of continuous therapy compared with interrupted, consecutive short-term treatment strategies. Methods Study Design and Groups This two-center, longitudinal, retrospective study included all patients with moderate-to-severe plaque psoriasis treated with three biologics (secukinumab, guselkumab, or risankizumab) at two tertiary centers by the two senior authors. Patients were categorized into short-term and long-term groups according to the planned treatment duration. The short-term group comprised patients receiving biologics reimbursed by Taiwan’s NHI, which requires a Psoriasis Area and Severity Index (PASI) score ≥ 10 after failure of two conventional therapies plus phototherapy twice or thrice weekly, each administered for at least 3 months. Eligible patients may receive up to approximately 2 years of biologic treatment, with clinical efficacy assessed every 6 months; continuation under NHI reimbursement requires achieving at least a 50% improvement from baseline PASI (PASI 50). The long-term group included patients from clinical trials who received biologics for longer planned durations (approximately 4–6 years) and met inclusion criteria of PASI ≥ 12 and body surface area (BSA) ≥ 10%. This group comprised patients from ERASURE (NCT01365455) and its extension (NCT01544595) assigned to secukinumab 300 mg, VOYAGE 1 (NCT02207231) assigned to guselkumab 100 mg, and IMMvent (NCT02694523) and its extension, LIMMitless (NCT03047395), assigned to risankizumab 150 mg. Detailed descriptions of the study designs are available in the respective publications reporting the pivotal trial results. Inclusion, Follow-up, and Outcomes Patients who completed the planned treatment—either the NHI-reimbursed period or the clinical trial protocol—and achieved clinical response at the end of therapy (PASI 50 and final PASI ≤ 10) were included. Patients who had previously received the same biologic were excluded to ensure all participants were first-time users. A flow diagram of patient recruitment, exclusion, and final analysis is provided in Supplemental Fig. 1. Patients were followed longitudinally to monitor disease relapse, defined as either a PASI score ≥ 10 or loss of PASI 50 response. The starting point of DFR was defined as the date when the drug was no longer expected to exert pharmacologic effects, estimated as the last administration date plus the dosing interval of the respective biologic agent, whereas the end was defined as the time of relapse as described above. During follow-up, phototherapies and traditional systemic treatments, including methotrexate, acitretin, and cyclosporin, were permitted. The primary outcome was time to disease relapse, while the exploratory objective was to evaluate the DFR per dose as a measure of treatment efficiency. The patients were evaluated by the same dermatologists throughout the period. Statistical Analysis Categorical variables were analyzed using the chi-square test, with Fisher’s exact test applied when ≥ 20% of expected cell counts were ≤ 5. Continuous variables were compared using Student’s t test. Survival curves for the DFR were generated using the Kaplan–Meier method, and differences between groups were assessed with the log-rank test. All analyses were performed using Python (version 3.12) with the pandas, numpy, and lifelines packages. Kaplan–Meier curves, log-rank tests, and Cox models were computed using lifelines, and figures were generated with matplotlib. All statistical tests were two-sided, with significance set at p < 0.05. Ethics Approval The study protocol was reviewed and approved by the local Institutional Review Board of Chang Gung Memorial Hospital, Linkou and Taipei branches (202000677B0); National Taiwan University Hospital (201605118RINB, 201904124RINC). Results Baseline Characteristics A total of 235 patients treated with risankizumab, guselkumab, or secukinumab were enrolled, including 73 (31.1%) patients enrolled in clinical trials and 162 (68.9%) patients reimbursed by Taiwan’s NHI. Among clinical trial patients, one of 24 patients (4.2%) treated with risankizumab withdrew because of a systemic lupus erythematosus flare-up; one of 18 patients (5.6%) in the guselkumab group experienced treatment failure; and in the secukinumab group, 14 of 31 patients (45.2%) were excluded as a result of treatment failure, while six of 31 patients (19.4%) withdrew for personal reasons. Among reimbursed patients, 10 patients (27.0%) in the risankizumab group were excluded because of treatment failure (n = 9) or adverse events (n = 1); 30 patients (34.1%) in the guselkumab group were excluded as a result of treatment failure (n = 26), adverse events (n = 2), or personal reasons (n = 2); and nine patients (24.3%) in the secukinumab subgroup were excluded because of treatment failure (Supplemental Fig. 1). Overall, 164 patients with moderate-to-severe plaque psoriasis were included in this study, with 51 (31.1%) in the long-term treatment group and 113 (68.9%) in the short-term treatment group. Patients in the long-term group were significantly younger than those in the short-term group (mean age 42.13 ± 12.16 vs. 47.27 ± 12.46 years; p = 0.01) and had a shorter psoriasis disease duration (15.60 ± 10.39 vs. 20.87 ± 9.80 years; p = 0.005). Patients in the short-term group had received more prior biologic therapies (2.34 ± 1.94 vs. 0.88 ± 0.93; p < 0.001) and exhibited a different distribution of biologic agents, with a higher proportion of guselkumab users and a lower proportion of risankizumab users compared with the long-term group (p = 0.02). No significant between-group differences were observed in sex, body weight, baseline PASI, or the presence of psoriatic arthritis (Table 1). Subgroup baseline characteristics for risankizumab, guselkumab, and secukinumab are summarized in Supplemental Tables 1–3.Table 1Baseline clinical characteristics of the study populationCharacteristicsTotalLong-term groupShort-term groupp valueNumber of patients16451 (31.1%)113 (68.9%)Age (years) Mean ± SD45.67 ± 12.5642.13 ± 12.1647.27 ± 12.460.01Body weight (kg)a Mean ± SD76.57 ± 15.7375.60 ± 16.1076.97 ± 15.630.62Psoriasis disease duration (years)b Mean ± SD19.49 ± 10.1915.60 ± 10.3920.87 ± 9.800.005Number of prior biologic therapies Mean ± SD1.88 ± 1.820.88 ± 0.932.34 ± 1.94< 0.001 Baseline PASI Mean ± SD18.95 ± 8.5818.44 ± 7.4019.18 ± 9.080.61Treatment duration (days) Mean ± SD1170.56 ± 593.092017.57 ± 231.62788.28 ± 117.69< 0.001Sex, n (%)0.14 Female347 (20.6%)27 (79.4%) Male13044 (33.8%)86 (66.2%)Psoriatic arthritis, n (%)0.16 Yes6817 (25.0%)51 (75.0%) No9634 (35.4%)62 (64.6%)Biologic agent, n (%)0.02 Guselkumab7517 (22.7%)58 (77.3%) Risankizumab5023 (46.0%)27 (54.0%) Secukinumab3911 (28.2%)28 (71.8%)PASI Psoriasis Area and Severity IndexaData were missing for 4 patientsbData were missing for 11 patients Table 1Baseline clinical characteristics of the study populationCharacteristicsTotalLong-term groupShort-term groupp valueNumber of patients16451 (31.1%)113 (68.9%)Age (years) Mean ± SD45.67 ± 12.5642.13 ± 12.1647.27 ± 12.460.01Body weight (kg)a Mean ± SD76.57 ± 15.7375.60 ± 16.1076.97 ± 15.630.62Psoriasis disease duration (years)b Mean ± SD19.49 ± 10.1915.60 ± 10.3920.87 ± 9.800.005Number of prior biologic therapies Mean ± SD1.88 ± 1.820.88 ± 0.932.34 ± 1.94< 0.001 Baseline PASI Mean ± SD18.95 ± 8.5818.44 ± 7.4019.18 ± 9.080.61Treatment duration (days) Mean ± SD1170.56 ± 593.092017.57 ± 231.62788.28 ± 117.69< 0.001Sex, n (%)0.14 Female347 (20.6%)27 (79.4%) Male13044 (33.8%)86 (66.2%)Psoriatic arthritis, n (%)0.16 Yes6817 (25.0%)51 (75.0%) No9634 (35.4%)62 (64.6%)Biologic agent, n (%)0.02 Guselkumab7517 (22.7%)58 (77.3%) Risankizumab5023 (46.0%)27 (54.0%) Secukinumab3911 (28.2%)28 (71.8%)PASI Psoriasis Area and Severity IndexaData were missing for 4 patientsbData were missing for 11 patients Baseline clinical characteristics of the study population PASI Psoriasis Area and Severity Index aData were missing for 4 patients bData were missing for 11 patients Drug-Free Remission (DFR) Overall, the DFR was significantly longer in the long-term group than in the short-term group (log-rank p < 0.001; Fig. 1), with median survival of 211 versus 111 days, respectively. Subgroup Kaplan–Meier analyses demonstrated that long-term therapy was associated with significantly longer DFR among guselkumab and risankizumab users (Fig. 2a, b), whereas no significant between-group difference was observed for secukinumab (Fig. 2c). Notably, an estimated 25% of patients in the long-term risankizumab subgroup remained disease-free for 893 days after discontinuing therapy.Fig. 1Kaplan–Meier plot showing drug-free remission duration in the long-term versus short-term treatment groupsFig. 2Kaplan–Meier plots showing drug-free remission duration for each biologic subgroup in the long-term versus short-term treatment groups. a Risankizumab. b Guselkumab. c Secukinumab Kaplan–Meier plot showing drug-free remission duration in the long-term versus short-term treatment groups Kaplan–Meier plots showing drug-free remission duration for each biologic subgroup in the long-term versus short-term treatment groups. a Risankizumab. b Guselkumab. c Secukinumab In univariate Cox regression (Table 2), long-term therapy was associated with a lower hazard of relapse for guselkumab (HR 0.48, 95% CI 0.26–0.89, p = 0.02) and risankizumab (HR 0.27, 95% CI 0.12–0.59, p = 0.001), whereas no significant effect was observed for secukinumab (HR 0.98, 95% CI 0.48–2.01, p = 0.95). Longer disease duration, a greater number of prior biologic therapies, higher baseline PASI, and the presence of psoriatic arthritis were associated with shorter DFR in select subgroups.Table 2Univariate models for drug-free remission in each subgroupVariableGuselkumabRisankizumabSecukinumabHR (95% CI)p valueHR (95% CI)p valueHR (95% CI)p valueLong-term versus short-term group0.48 (0.26–0.89)0.020.27 (0.12–0.59)0.0010.98 (0.48–2.01)0.95Body weight1.01 (0.99–1.03)0.381.01 (0.99–1.03)0.431.00 (0.98–1.02)0.86Baseline PASI1.04 (1.01–1.07)0.0101.02 (0.97–1.06)0.421.06 (1.02–1.10)0.001Age1.02 (1.00–1.04)0.101.02 (0.99–1.05)0.241.02 (0.99–1.04)0.28Sex0.77 (0.38–1.57)0.480.84 (0.38–1.87)0.670.75 (0.37–1.51)0.42Disease duration1.02 (0.99–1.04)0.171.05 (1.01–1.09)0.021.03 (0.99–1.07)0.16Number of prior biologic therapies1.35 (1.15–1.58)< 0.0011.36 (1.18–1.57)< 0.0011.15 (0.86–1.53)0.35Psoriatic arthritis0.86 (0.53–1.40)0.542.29 (1.16–4.53)0.020.98 (0.51–1.88)0.95CI confidence interval, HR hazard ratio, PASI Psoriasis Area and Severity Index Table 2Univariate models for drug-free remission in each subgroupVariableGuselkumabRisankizumabSecukinumabHR (95% CI)p valueHR (95% CI)p valueHR (95% CI)p valueLong-term versus short-term group0.48 (0.26–0.89)0.020.27 (0.12–0.59)0.0010.98 (0.48–2.01)0.95Body weight1.01 (0.99–1.03)0.381.01 (0.99–1.03)0.431.00 (0.98–1.02)0.86Baseline PASI1.04 (1.01–1.07)0.0101.02 (0.97–1.06)0.421.06 (1.02–1.10)0.001Age1.02 (1.00–1.04)0.101.02 (0.99–1.05)0.241.02 (0.99–1.04)0.28Sex0.77 (0.38–1.57)0.480.84 (0.38–1.87)0.670.75 (0.37–1.51)0.42Disease duration1.02 (0.99–1.04)0.171.05 (1.01–1.09)0.021.03 (0.99–1.07)0.16Number of prior biologic therapies1.35 (1.15–1.58)< 0.0011.36 (1.18–1.57)< 0.0011.15 (0.86–1.53)0.35Psoriatic arthritis0.86 (0.53–1.40)0.542.29 (1.16–4.53)0.020.98 (0.51–1.88)0.95CI confidence interval, HR hazard ratio, PASI Psoriasis Area and Severity Index Univariate models for drug-free remission in each subgroup CI confidence interval, HR hazard ratio, PASI Psoriasis Area and Severity Index In multivariate Cox regression adjusting for significant baseline covariates (Table 3), long-term risankizumab use was significantly associated with a reduced relapse risk compared with short-term use (HR 0.34, 95% CI 0.13–0.90, p = 0.03). Long-term guselkumab use showed a trend toward reduced relapse risk (HR 0.55, 95% CI 0.28–1.06, p = 0.07), whereas no significant difference was observed for secukinumab. A greater number of prior biologics and higher baseline PASI remained independently associated with shorter DFR duration in select subgroups.Table 3Multivariate models for drug-free remission after stepwise forward Cox proportional hazards regression in each subgroupVariableGuselkumabRisankizumabSecukinumabHR (95% CI)p valueHR (95% CI)p valueHR (95% CI)p valueLong-term versus short-term group0.55 (0.28–1.06)0.070.34 (0.13–0.90)0.03NANABody weightNANANANANANABaseline PASI1.04 (1.01–1.07)0.007NANA1.06 (1.02–1.10)0.001AgeNANANANANANASexNANANANANANADisease durationNANA1.01 (0.97–1.06)0.66NANANumber of prior biologic therapies1.25 (1.04–1.49)0.021.15 (0.97–1.38)0.11NANAPsoriatic arthritisNANA2.00 (0.88–4.55)0.10NANAVariables that were significant in the univariate analysis were included in the multivariate analysisCI confidence interval, HR hazard ratio, NA not applicable, PASI Psoriasis Area and Severity Index Table 3Multivariate models for drug-free remission after stepwise forward Cox proportional hazards regression in each subgroupVariableGuselkumabRisankizumabSecukinumabHR (95% CI)p valueHR (95% CI)p valueHR (95% CI)p valueLong-term versus short-term group0.55 (0.28–1.06)0.070.34 (0.13–0.90)0.03NANABody weightNANANANANANABaseline PASI1.04 (1.01–1.07)0.007NANA1.06 (1.02–1.10)0.001AgeNANANANANANASexNANANANANANADisease durationNANA1.01 (0.97–1.06)0.66NANANumber of prior biologic therapies1.25 (1.04–1.49)0.021.15 (0.97–1.38)0.11NANAPsoriatic arthritisNANA2.00 (0.88–4.55)0.10NANAVariables that were significant in the univariate analysis were included in the multivariate analysisCI confidence interval, HR hazard ratio, NA not applicable, PASI Psoriasis Area and Severity Index Multivariate models for drug-free remission after stepwise forward Cox proportional hazards regression in each subgroup Variables that were significant in the univariate analysis were included in the multivariate analysis CI confidence interval, HR hazard ratio, NA not applicable, PASI Psoriasis Area and Severity Index Long-Term Treatment Versus Interrupted Sequential Short Treatment Strategy Among the 58 patients who initially received short-term guselkumab therapy, nine experienced relapse and subsequently received another short-term guselkumab course. This allowed an additional analysis comparing two treatment strategies—long-term continuous therapy versus interrupted sequential short-term therapy—in the guselkumab subgroup. Long-term treatment resulted in a longer cumulative DFR compared with the interrupted sequential short-term strategy, with an effectiveness (area under curve, AUC) of 316.97 versus 232.55 days and a numerically higher DFR days per dose (9.61 ± 11.43 vs. 8.23 ± 2.78; p = 0.64) (Table 4, Fig. 3). Detailed baseline characteristics of these two groups are summarized in Supplemental Table 4.Table 4Treatment efficiency for long-term treatment versus sequential short-term treatment in guselkumab subgroupLong-term groupSequential short-term groupEffectiveness (AUC, days)316.97232.55Cost (mean dose units)33.2928.22Cost per remission day0.1050.121AUC (area under the curve) represents the cumulative drug free remission durationFig. 3Cumulative proportion of patients in drug-free remission for the guselkumab group, comparing continuous long-term treatment and sequential short-term treatment. Shaded areas under the curves indicate the cumulative drug-free remission duration Table 4Treatment efficiency for long-term treatment versus sequential short-term treatment in guselkumab subgroupLong-term groupSequential short-term groupEffectiveness (AUC, days)316.97232.55Cost (mean dose units)33.2928.22Cost per remission day0.1050.121AUC (area under the curve) represents the cumulative drug free remission duration Treatment efficiency for long-term treatment versus sequential short-term treatment in guselkumab subgroup AUC (area under the curve) represents the cumulative drug free remission duration Cumulative proportion of patients in drug-free remission for the guselkumab group, comparing continuous long-term treatment and sequential short-term treatment. Shaded areas under the curves indicate the cumulative drug-free remission duration Discussion In this study, we examined whether prolonged biologic therapy is associated with improved DFR after treatment discontinuation in patients with moderate-to-severe plaque psoriasis, both in real-world settings and following the end of clinical trials. By leveraging Taiwan’s unique NHI policy—which restricts biologic reimbursement to an approximate 2-year maximum (with an additional 6 months granted during the COVID-19 pandemic)—we were able to compare outcomes between patients who received a short, policy-mandated treatment duration and those who received substantially longer therapy through clinical trials. Our findings indicate that treatment duration has a meaningful and drug-specific influence on relapse dynamics, with long-term IL-23 inhibition demonstrating the clearest and most consistent benefits. Historically, the observation that psoriasis lesions tend to recur at the same anatomical locations has raised the possibility that memory immune cells play a role [18]. Recent studies suggest that the chronicity of psoriasis is strongly linked to the persistence of skin memory T cells (Trm) following the initial inflammatory episode [6, 7]. Psoriatic lesional skin contains significantly higher numbers of Trm compared with healthy skin, and these cells may persist even after clinical resolution, providing a cellular basis for the well-recognized tendency of psoriasis to recur at previously affected sites [19–21]. Lesion duration has also been associated with greater Trm accumulation [19, 22], suggesting that prolonged inflammatory activity may gradually expand the local memory T cell pool. Trm was also found in clinically non-lesional skin as demonstrated by another study showing that the ratio of IL-17A-producing CD8+CD103+ Trm to interferon-γ (IFNγ)-producing cells increased with disease duration [22], which provides one potential explanation for the rare occurrence of rebound phenomenon after treatment discontinuation [23]. Initial psoriasis severity may also contribute to higher Trm numbers [24], as correlations between lesion thickness and Trm counts have been reported, although some conflicting results exist [19]. Accordingly, recent research has increasingly focused on therapeutic strategies specifically aimed at modifying or depleting the cutaneous Trm pool. Emerging mechanistic studies—such as those targeting the CXCR6–CXCL16 axis—have shown that disrupting Trm maintenance can reduce relapse in preclinical psoriasis models [25]. Notably, a preliminary report from a recent knockout (KO) trial provided the first evidence suggesting a potential link between clinical remission and reduced Trm levels [8]. In the present study, long-term risankizumab therapy reduced the hazard of relapse by more than 60% compared with short-term treatment, demonstrating the most pronounced effect. Similar trends were observed for guselkumab, although the effect was less marked, potentially reflecting pharmacodynamic differences within the IL-23 inhibitor class. Consistent findings were also observed in our prior study [16]. In contrast, secukinumab did not confer longer DFR despite extended exposure. It has been frequently reported that IL-23 inhibitors show better remission duration after biologic withdrawal compared with IL-17 inhibitors [3, 6, 17, 26]. Our study further suggests that prolonged IL-17 inhibitor use may not confer additional disease-modifying effects. Exploring the potential underlying mechanisms, Th17 cells exhibit plasticity [27]; once exposed to treatment, they may undergo a transition into ex-IL-17 cells that acquire memory T cell-like phenotypes [28]. One hypothesis suggests that CD4+ dermal Trm could originate from such ex-IL-17 cells, which no longer produce IL-17 and therefore are not depleted by IL-17-targeting biologics [6]. Evidence from recent studies indicates that IL-17 inhibitors exert only limited effects on skin-resident memory T cells in psoriatic lesions [29, 30]. Prolonged IL-17 inhibition may thus have limited effects, although this hypothesis remains to be validated. By contrast, IL-23 may be critical for the maintenance of Trm, as evidenced by a mouse model showing that IL-23 is required for Trm maintenance in the skin after Candida albicans infection and during the development of imiquimod-induced dermatitis [28], as well as by a human study showing that high-dose risankizumab depletes Trm [8]. More direct evidence of the differential effects of guselkumab and secukinumab before and after treatment was also presented. Specifically, guselkumab reduced memory T cells while maintaining regulatory T cells, whereas secukinumab showed the opposite pattern [31]. The association between Trm and disease remission has not yet been fully established; nevertheless, two currently plausible strategies may help reduce the risk of relapse. The first is early initiation of biologic therapy, as growing evidence suggests that earlier treatment increases the likelihood of achieving long-term remission [3, 9, 16, 17, 32, 33]. This may be explained by the previously mentioned association between longer lesion duration and higher Trm accumulation. The second approach involves using higher biologic doses; for example, as suggested by the preliminary findings of the aforementioned KO trial [8]. There is currently limited evidence supporting the notion that simply extending treatment duration can further prolong DFR duration. Our study may provide a potential third avenue, as well as a foundation for future mechanistic research. For instance, comparing Trm levels in psoriatic skin that remains in remission after short-term versus long-term risankizumab therapy could help determine whether longer treatment further depletes the residual memory T cell pool. Further prospective studies with larger sample sizes will be required to validate this concept. Beyond the biological perspective, a small but valuable dataset comes from patients who received effective sequential short-term guselkumab therapy interrupted by NHI policy. This provided an opportunity to compare continuous versus interrupted treatment strategies. Overall, the time in remission during drug-free periods per dose was numerically higher in the long-term treatment group, and treatment efficiency may be achievable even in a budget-constrained healthcare system with longer reimbursement durations; however, the sample size was too small to detect a statistically significant difference. Previous studies suggest that continuous administration of biologic agents provides greater efficacy than intermittent use [34, 35]. Furthermore, recent studies under Taiwan’s NHI reimbursement framework indicate that therapy interruption and subsequent reinitiation may compromise treatment efficacy, both within the same drug and among patients switching between drugs with different modes of action [14, 17]. Additionally, mandatory discontinuation of therapy in patients with good disease control imposes both physical and psychological burdens [36]. Collectively, these findings highlight the need to reconsider the justification for repeated treatment interruptions versus continuous long-term therapy, particularly in the context of IL-23 inhibitors. Cost-effectiveness analyses in psoriasis have rarely been reported. Encouragingly, if confirmed in future larger studies, our hypothesis that long-term IL-23 therapy is more cost-effective than interrupted therapy would support maintaining long-term treatment even after achieving disease clearance. This study has several limitations. First, the sample size—particularly within drug-specific subgroups—was modest, which may reduce statistical power. Second, it is well known that differences in baseline demographics between clinical trials and real-world populations may affect treatment response [37], although multivariate models partially mitigated this issue. Notably, the median disease duration in both groups was relatively long (15 and 20 years, respectively), indicating that most patients did not receive early intervention. Prior biologic exposure may also influence treatment response; however, previous studies suggest that prior biologic use has limited impact on treatment outcomes with IL-23 inhibitors, particularly risankizumab [38, 39]. Taken together, differences in disease duration and prior biologic exposure were present; however, their impact on treatment outcomes in this study may be limited. It should still be noted that the inherent differences such as adherence and baseline conditions between trial and real-world populations may pose a selection bias. However, close follow-up is also required in real-world practice because of reimbursement regulations, as patients must return in person for every injection of the three drugs studied. Comorbidities have been reported to affect treatment efficacy [40, 41], but their impact on DFR remains unclear. Third, the definition of DFR is not yet standardized (in this study, defined as loss of PASI 50 with an absolute cutoff of 10 according to reimbursement criteria) and may be associated with several challenges in clinical practice. For instance, some patients may restart self-paid therapy upon perceiving mild relapse before meeting the defined criteria, making DFR difficult to interpret. Reassuringly, this scenario was likely uncommon in our cohort, as most patients under the reimbursement system could not afford the relatively high cost of biologics. In addition, patients who are doing well after treatment discontinuation may not return for follow-up, which could lead to an underestimation of DFR. We acknowledge that DFR is not yet a widely established outcome measure compared with conventional endpoints such as PASI response or drug survival. However, it reflects an important real-world clinical concern, as many patients are interested in the duration of remission after treatment discontinuation. Fourth, relapse definitions based on PASI may not fully capture patient-reported outcomes or subtle disease worsening. Fifth, using total dose as a cost proxy does not reflect true real-world pricing, including other healthcare-related fees and concomitant traditional therapies, and should be considered a coarse estimate. Additionally, the number of cases in subgroups for the exploratory analysis of DFR per dose is especially limited (n = 26), which further limits the interpretability of the results. Finally, ethnic differences have been reported among Asian and Chinese patients with psoriasis, and the generalizability of our results to other populations awaits further study [42, 43]. Conclusions In addition to disease duration, this study demonstrates that long-term anti-IL-23 therapy, particularly risankizumab, may prolong DFR, whereas prolonged anti-IL-17 therapy does not show a clear benefit. The increased cost associated with a longer treatment period may be partially offset by the extended DFR and the reduced need for reloading during re-treatment initiation. However, these findings should be interpreted with caution and considered preliminary, and they warrant confirmation in larger prospective studies because of the small sample size and differences in baseline demographics. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (PDF 314 KB) Supplementary file1 (PDF 314 KB)