Relative sensitivity and preference of indicators in pharmaceutical trials of psoriasis and psoriatic arthritis

Abstract

Objectives

Numerous indicators have been proposed to evaluate the efficacy for randomized clinical trials (RCTs) of psoriasis (Pso) and psoriatic arthritis (PsA), but their comparability and correlation remain unknown. We aim to evaluate the preference and relative sensitivity of the most widely used indicators that report response rate, and to offer guidance for the primary endpoint selection for Pso and PsA trials.

Methods

We conducted a systematic search, including five databases and four registries, to identify all pharmacological intervention-controlled RCTs. A Bayesian hierarchical linear mixed model was employed to assess relative discriminations and provide a ranking of these indicators. This model, considered the gold standard for sparse and heterogeneous data, was applied to estimate differences between control and intervention groups and assess the preference and relative sensitivity of outcome indicators in Pso and PsA.

Results

Altogether, 386 RCTs met our inclusion criteria. We included 9 and 8 commonly used response rate indicators for Pso and PsA trials, respectively, all of which were treated as primary endpoints. We found evidence of significant differences among indicators. PASI 50, PASI 75 and IGA 0,1 proved to be robust indicators for assessing pharmacological efficacy in the majority of RCTs of Pso. Conversely, PASI 125, DIQI 0,1 and NRS-4 were not preferred under different circumstances. Furthermore, PASI 50, PASI 75 and PASI 90 appeared to be highly effective in almost all categories of pharmacological RCTs of PsA. However, due to their extreme sensitivity, it was advisable to use ACR 20 to prevent an overestimation of the therapeutic benefits of interventions. ACR 50, ACR 70 and MDA were less sensitive, but they were supposed to be more cautious in evaluating disease changing. The choice of indicators was slightly influenced by disease severity, intervention type and administration method.

Conclusions

The notable efficacy discrimination ability of indicators underscores the importance of flexibility and comprehensiveness in selecting primary outcome(s). Our findings provide practical implications for optimizing indicator selection in future trial design, ensuring better alignment with trial objectives and disease characteristics.

PROSPERO number: CRD42022337725.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40001-025-02995-5.

Key summary

• The comparability of indicators in RCTs for Pso and PsA remains unclear.

• A Bayesian hierarchical linear mixed model is applied for primary endpoint selection.

• •Indicator preference is influenced by severity, drug type and trial outcomes

• PASI 50 and PASI 75 are robust for assessing efficacy in most RCTs of Pso and PsA.

• Our findings emphasize a multidimensional criterion for evaluating trial validity.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40001-025-02995-5.

Introduction

Psoriasis (Pso) is a common chronic papulosquamous disease with both cutaneous and systemic manifestations, affecting approximately 2–5% of the general population [1]. Approximately one-third of patients will develop psoriatic arthritis (PsA) eventually, which is the main comorbidity associated with Pso, characterized by heterogeneous clinical features and widespread musculoskeletal inflammation [2]. The introduction of biologic therapies has significantly improved both short- and long-term outcomes by controlling symptoms, enhancing quality of life, and preventing structural joint damage and disability [3]. This therapeutic progress has also prompted increasing attention to the choice of outcome indicators in both Pso and PsA clinical trials, as accurate assessment is essential for evaluating treatment benefit and guiding decision-making.

A wide range of outcome measures are currently used in daily practice and clinical research for Pso, yet many lack consistent validation and comparability, and no universally accepted standard has been established [4]. Since psoriatic skin lesions are readily visible, they are relatively easy to quantify. However, lesion extent and severity alone do not fully reflect disease burden, as the psychosocial impact varies among individuals [5]. Common tools used to score Pso include the Psoriasis Area and Severity Index (PASI) that evaluates lesion characteristics along with the affected area, and the Investigator’s Global Assessment (IGA) focusing on the overall severity of lesion. In spite of the wide acceptance of PASI as the currently optimum measure in research, it has been criticized for lacking consensus on interpretability, having a low response distribution and non-linear scale [4, 6]. Complementary instruments, such as the Patient’s Global Assessment (PatGA) and Dermatology Life Quality Index (DLQI), have been introduced to better capture patient perspectives [4]. In this context, it is important to distinguish between two categories of measures: severity measures, which assess baseline disease status or stratify severity levels, and efficacy indicators, which are used to detect changes in disease activity in response to interventions. In this study, two analytical dimensions, preference and relative sensitivity, were introduced to compare outcome indicators. Preference refers to the extent to which an indicator is favored in specific trial contexts based on its ability to reflect perceived treatment efficacy. Relative sensitivity refers to the capacity of an indicator to detect treatment-induced changes in disease status, especially when compared against other indicators. More sensitive indicators can detect subtle differences and may be particularly useful in trials comparing active therapies.

While Pso primarily manifests in the skin, PsA is distinguished by musculoskeletal involvement such as peripheral arthritis, enthesitis, dactylitis, and axial disease. The clinical expression of skin and joint involvement can vary independently, and their therapeutic responses may not always align. Therefore, clear differentiation between skin and joint involvement is essential for accurate disease characterization, outcome assessment, and treatment evaluation [7]. Recognizing this heterogeneity, contemporary clinical research increasingly emphasizes the use of domain-specific and composite outcome measures that reflect the multifaceted nature of psoriatic disease. However, the evaluation of PsA remains complex. Patients are often misdiagnosed due to its broad clinical spectrum and the lack of specific diagnostic tests. In response to these challenges, the Group for Research and Assessment of Psoriasis and Psoriatic Arthritis (GRAPPA) has promoted the use of PsA-specific composite measures. While early tools such as the American College of Rheumatology (ACR) criteria and Disease Activity Score (DAS) were adapted from rheumatoid arthritis, they have limited validation in PsA. Subsequently, Psoriatic Arthritis Response Criteria (PsARC) and Minimal Disease Activity (MDA) were developed specifically for PsA, but their comparability and sensitivity remain incompletely assessed [8].

The increasing availability of highly effective therapies has expanded the evidence base on Pso and PsA treatments. Accordingly, comparisons across outcome indicators within the same patient population are needed to assess their relative performance in randomized controlled trials (RCTs). In this study, we systematically evaluated the preference and relative sensitivity of various efficacy indicators in different trial contexts. This comparison helps identify context-appropriate tools for accurate efficacy evaluation and prevents underestimation of treatment effects. We emphasize the importance of adopting a multidimensional evaluation framework in clinical trials, tailored to the clinical features and therapeutic goals of psoriatic diseases.

Materials and methods

This systematic review and network meta-analysis was performed in line with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [9]. The detailed protocol of our study was prospectively registered in the PROSPERO database, and the registration number was CRD42022337725.

Search strategy and bias risk evaluation

Two researchers (JR.T and DY.Z) independently screened titles and abstracts, assessed full texts and collected relevant data from qualifying studies. The search for RCTs included published articles from peer-reviewed English-language journals and registered trials in clinical trials registries, both from inception until December 31, 2023. Five electronic databases were systematically searched (namely, PubMed, EMBASE, Web of Science, Cochrane Library Central Register of Controlled Trials (CENTRAL), and China National Knowledge Infrastructure (CNKI)). Bibliographies cited in eligible reviews were also manually searched to identify additional studies. Records of registered RCTs were collected from four publicly available web-based clinical trial registries, including the ClinicalTrials.gov of the US National Library of Medicine, the International Standard Randomised Controlled Trial Number Register (ISRCTN), the Australian and New Zealand Clinical Trials Registry (ANZCTR) and the Chinese Clinical Trial Register. Studies were included only if they utilized two or more specific outcome measures reporting response rate. In addition, all studies were assessed for risk of bias using the Cochrane Risk of Bias 2.0 tool for RCTs by JR.T and ST.K independently [10]. All results were cross-checked and any disagreement at each stage was resolved by mutual negotiation or a third reviewer. Details of the search procedure and algorithm, study selection and data extraction, and the results of risk of bias were provided in the Appendices 1–5.

Indicators reporting response rate

We selected the most commonly used efficacy evaluation indicators reporting response rate in RCTs for Pso and PsA separately. These indicators were selected based on their frequency of use in eligible RCTs identified during the systematic review phase. Although there were several terms and variant nomenclatures for the same type of measures, the issue was addressed by standardizing the naming of certain similar measures into a unified format [4]. Outcome measures for Pso trials included “PASI 50”, “PASI 75”, “PASI 90” (≥ 50%, ≥ 75%, and ≥ 90% reductions in the PASI scores from baseline), “PASI 125”(worsening over baseline value with more than 125%), “IGA 0,1” (clear or almost clear (grade 0 or 1)), “IGA 0,1, Reduction ≥ 2” (clear or almost clear (grade 0 or 1) and at least a two-grade improvement from baseline), “PatGA 0,1” (clear or almost clear (grade 0 or 1)), “DLQI 0,1” (no effect at all on the participant's life (grade 0 or 1)) and “NRS-4” (≥ 4-point improvement from baseline using an itch Numerical Rating Scale). Meanwhile, outcome measures for PsA trials included “ACR 20”, “ACR 50”, “ACR 70” (≥ 20%, ≥ 50%, and ≥ 70% improvement in ACR components from baseline), “MDA”, “PsARC”, “PASI 50”, “PASI 75” and “PASI 90”. Detailed description of these indicators was shown in Appendix 6. The outcome of interest was the percentage change between intervention and control groups.

Data analysis

Conventional methods only permitted limited pairwise comparisons of indicators directly compared within the same trial. To overcome this, we adopted a network meta-analysis approach, which integrated both direct and indirect comparisons across studies. This method allowed for a more complete and reliable comparison of indicator performance, even when head-to-head trials were lacking. It also allowed for the quantification of inconsistencies between direct and indirect evidence, offering a measure of overall reliability. For the model construction method, please refer to Appendix 7 or our previous publications [11, 12]. To minimize the influence of confounding factors and improve statistical power when dealing with sparse and heterogeneous data, the gold standard model for sparse and heterogeneous data [13–15] — a Bayesian hierarchical linear mixed model was applied to estimate the difference between control group and intervention group to assess preference and relative sensitivity of outcome indicators in Pso and PsA (Table 1).

Table 1.

Definition of preference and relative sensitivity

Concept	Definition	Application in RCT Context
Preference	The extent to which an outcome indicator is favored across specific clinical contexts	Reflects which indicator is more likely to show larger treatment differences across trials
Relative sensitivity	The ability of an indicator to detect subtle changes in disease status or treatment effect	Indicates which indicator is more reactive/responsive to treatment changes within the same trial

We defined the “intervention group possibility” as the proportion of positive outcomes among individuals in the intervention group, and the “control group possibility” as the corresponding proportion in the control group. Essentially, these values represented the response rates for each indicator within the respective groups. Then, we calculated the percentage change between two groups (intervention group possibility – control group possibility) for each indicator and employed the Bayesian hierarchical linear mixed model to estimate differences of the percentage change between different indicators. The statistical analysis was conducted using the brms package in R (version 4.0.5), which implemented Bayesian inference via the Hamiltonian Markov Chain Monte Carlo method. We run 8000 iterations across 4 chains to ensure convergence and model stability. In the model, we included three covariates with fixed effects: application method (topical or systemic), age group, and disease severity. The model also accounted for the nested relationship between intervention type and the specific intervention used, reflecting their hierarchical structure. Although there was variation in the variables, the posterior distributions for indicator differences remained stable across chains, supporting the reliability of our findings (Appendix 8). To simplify interpretation for clinicians and researchers, we highlighted the top three ranked indicators with statistically significant differences. If two indicators were closely ranked in third place, we extend our recommendation to the top four. In scenarios where no significant differences were found, we suggested relying on the top one or two indicators. This consistent approach was also applied to identify less recommended indicators, ensuring a uniform judgment methodology.

In brief, we applied a Bayesian hierarchical linear mixed model to estimate the effectiveness of each indicator by calculating the difference in response rates between intervention and control groups. This approach enabled both direct and indirect comparisons while adjusting for factors such as intervention type, administration route, and disease severity, allowing for a robust and comprehensive ranking of indicators based on sensitivity and clinical relevance.

Results

Overview of indicators reporting response rate in pharmacological intervention controlled RCTs for Pso and PsA

The systematic search yielded a total of 70,942 articles from the five electronic databases and 1770 trials from the four clinical trial registries. Altogether, 386 RCTs met the criteria and were selected for inclusion in the analysis, with 342 records on Pso and 74 on PsA, respectively. The characteristics of included trials were summarized in the Appendices 9–10. In terms of study quality, 90.2% of all included studies were rated as low-medium risk of bias, demonstrating the overall credibility and high quality of the studies included in this research. Network plots of indicator comparisons were displayed in Fig. 1A and B, where node size corresponded to the number of trials, and line width was proportional to the number of trials for each direct comparison. The most commonly studied indicators reporting response rate were PASI 75 (86.5%, 296 trials) in Pso trials and ACR 20 (95.9%, 71 trials) in PsA trials. RCTs of both Pso and PsA were divided into three subgroups, based on topical or systemic applications, disease severity and type of intervention.

Fig. 1.

Network plots of eligible comparisons between different pharmacological treatments for efficacy evaluation indicators. The size of each node is proportionate to the number of trials. The width of lines is proportional to the number of trials in which each direct comparison is made. A Network of the comparisons for psoriasis outcome indicators. B Network of the comparisons for psoriatic arthritis outcome indicators

Preference of indicators reporting response rate in pharmacological intervention controlled RCTs for Pso

The general preference of indicators was evaluated by Bayesian model considering the influence of age, disease severity and topical or systemic application (Appendix 11). We reported the effectiveness estimate with 95% confidence intervals (CI) and a statistically significant difference was detected if the null value was not included in the 95% CI. Since each indicator was estimated by its intervention group possibility minus its control group possibility, a greater positive estimated value indicated better distinguishing ability, while a greater negative value suggested the opposite. In the ranking of indicators reporting response rates, a higher cumulative probability value (closer to 1) indicated a greater ability to distinguish treatment efficacy. However, while higher rankings generally suggested stronger discriminatory performance, they should be interpreted within the broader context of indicator sensitivity, clinical relevance, and the phase of trial design. In light of this, both PASI 50 and PASI 75 demonstrated comparability and significantly higher response rates in intervention groups compared to control groups, in contrast to PASI 125, DLQI 0,1, and PASI 90. This suggested that PASI 50 and PASI 75 were more likely to reveal the effectiveness of pharmacological interventions than other indicators within the same participants. Moreover, IGA 0,1 was the third preferred indicator and preceded PASI 125, DLQI 0,1 and PASI 90. Conversely, PASI 125 was identified as the least favorable indicator, exhibiting a significant difference in response rate compared to the other indices. This suggested that PASI 125 might inadequately reflect the distinctions between pharmacological interventions and controls. DLQI 0,1 also showed to be the less effective indicator compared to other indicators except NRS-4 (Fig. 2). The preference of indicators was altered when evaluating groups with different disease severity, intervention types or application methods.

Fig. 2.

Preference of indicators reporting response rate in pharmacological intervention-controlled RCTs for psoriasis. A Bayesian hierarchical linear mixed model estimated effectiveness with 95% confidence intervals on indicators reporting response rate in pharmacological intervention-controlled RCTs for psoriasis. B The rank of indicators reporting response rate. The sooner an indicator reaches 1, the stronger ability to discriminate treatment efficacy

Different disease severity

In participants with all severities, although mean difference of percentage change had relative less discrepancy, PASI 50 tended to be a more robust indicator than others compared to PASI 90 and PASI 75. Instead, PASI 90 was considered the least favorable (Figure S5). Though no significant difference was observed when assessing moderate patients, PASI 75 still showed an advantage compared to PASI 90 (Figure S6). When evaluating the efficacy in severe, mild-to-moderate and moderate-to-severe patients, PASI 50 and PASI 75 demonstrated significantly stronger efficacy revealing ability. However, PASI 125 had significantly limited discrimination ability compared with all indicators (Figure S7-S9). Additionally, in RCTs where subject characteristics were not mentioned, PASI 50 and PASI 75 were significantly better than PASI 125 and PASI 90, which remained the leading indicators, while PASI 125 was still significantly worse than all other indicators (Figure S10).

Different intervention types

When assessing antibodies, PASI 50, PASI 75 and IGA 0,1, Reduction ≥ 2 were significantly superb than PASI 125, DLQI 0,1 and PASI 90, while PASI 125, DLQI 0,1 and NRS-4 lagged behind (Figure S11). Of note, PASI 50, PASI 75, NRS-4 and PatGA 0,1 sequentially showed significant benefit to efficacy evaluation in small molecule interventions, whereas PASI 125, DLQI 0,1 and PASI 90 were the weakest (Figure S12). Besides, PASI 75, IGA 0,1 and PASI 90 presented similar effects with significantly higher relative percentage change than PASI 125 in the evaluation of non-biologics interventions (Figure S13).

Different application methods

For participants with topical application IGA 0,1, PASI 50, and PASI 75 were comparable and had significant tendencies to uncover the intervention effectiveness than PASI 125 and PASI 90 (Figure S14). Additionally, PASI 50 and PASI 75 demonstrated comparable efficacy discrimination ability in systemically administered patients, surpassing PASI 125, DLQI 0,1, PASI 90 and IGA 0,1 significantly (Figure S15). In all, both PASI 50 and PASI 75 were reliable as outcome assessment measures for different application types.

Preference of indicators reporting response rate in pharmacological intervention controlled RCTs for PsA

PASI-derived indicators were not attenuated in terms of evaluating and comparing the treatment efficacy for participants with PsA, although they were used on average only one-third as often as measures related to ACR (Appendix 12). In the overall PsA trials, PASI 75 and PASI 50 were comparable, and followed by PASI 90, they were the best indicators reporting response rate and demonstrated significantly satisfied response rate compared to the rest of the metrics. However, ACR 70 was the least effective indicator with significant difference from others except MDA, which was also not recommended (Fig. 3).

Fig. 3.

Preference of indicators reporting response rate in pharmacological intervention-controlled RCTs for psoriatic arthritis. A Bayesian hierarchical linear mixed model estimated effectiveness with 95% confidence intervals on indicators reporting response rate in pharmacological intervention-controlled RCTs for psoriatic arthritis. B The rank of indicators reporting response rate. The sooner an indicator reaches 1, the stronger ability to discriminate treatment efficacy

Similar to the RCTs of Pso, the preference of indicators reporting response rate was also influenced by different disease severity and intervention type of involved patients. When evaluating all-severity and moderate-to-severe psoriasis patients, preference for indicators was consistent with overall estimation, it was easier to obtain significant results by selecting the PASI endpoints instead of the ACR endpoints, MDA and PsARC. In contrast, ACR 70, MDA and PsARC were shown to be the least effective indicators compared to the PASI endpoints (Figure S16-S17). Still, in participants with unmentioned severity, PASI 75 was slightly superior to PASI 50 and PASI 90, all showing significant advantages compared to ACR 70, ACR 50 and MDA (Figure S18). Similarly, among participants treated with antibodies, the comparisons of indicators aligned with the overall results (Figure S19). It was noteworthy that ACR 20 and PsARC showed similar effects in small molecule pharmacological intervention, significantly outperforming PASI 90 (Figure S20). Within non-biologics interventions, there were no clear superiority and inferiority among these indicators. Nevertheless, according to the rank of sensitivity, PsARC was more relatively sensitive and ACR 50 was less recommended (Figure S21). Due to the absence of topical application in the included PsA studies, the discriminative effect of outcome measures in systemic therapy was consistent with the overall assessment as well (Figure S22).

Discussion

This article conducted a Bayesian hierarchical linear mixed model designed to identify the most appropriate indicators for assessing the outcome of pharmacological intervention controlled RCTs. PASI 50 and PASI 75 were considered the most valid and sensitive indicators in nearly all types of pharmacological RCTs involving patients with and those with PsA exhibiting cutaneous involvement. In contrast, PASI 125 and ACR 70 performed worst in discriminating efficacy under different circumstances. Indicator recommendations varied slightly with disease severity, intervention type and application type.

In clinical trials, the primary outcome played a dominant role in determining intervention efficacy statistically [16, 17]. Well-defined efficacy measures were helpful to guide investigators and physicians in their evaluation and care of patients, thereby obviating poor outcomes. Within the period 1977–2000, Naldi et al. reviewed 171 RCTs of Pso therapies, and found PASI was the most popular among 44 different scoring systems [18]. Later, Puzenat et al. analyzed all clinical studies grading the severity of Pso patients between 1980 and 2009 with methodological validation criteria, and concluded PASI was the most extensively studied score and thoroughly validated despite limitations [19]. Currently, PASI 75 was considered the benchmark of primary endpoints in evaluating therapies for Pso, which stemmed from the deliberations between the Food and Drug Administration (FDA) and the Dermatology Advisory Council in 1998 [20–22]. A previous study also suggested that PASI 50 represented a clinically meaningful improvement and was suitable for distinguishing treatment efficacy [22]. Our results supported these findings, demonstrating that both PASI 50 and PASI 75 offered strong discriminatory capacity across RCT types. Additionally, PASI 90 was suggested as another standard endpoint to measure the efficacy of pharmaceuticals for psoriasis, but no significant advantage of PASI 90 was detected for Pso RCTs in our analysis [23]. Of note, it was observed that a few drugs (anti-TNFα drugs, anti-IL17 drugs, methotrexate etc.) achieved PASI 90 or better response correlating with significant improvements in DLQI in phase II or III clinical trials [24, 25]. Since those were the minority of all drugs, PASI 90 couldn’t replace PASI 75 response to meet therapeutic expectations in most situations as it was too stringent. Meanwhile, PASI 90 wasn’t excluded in corresponding trials and might become an important secondary endpoint.

IGA was an average evaluation of psoriatic lesions that didn’t include information such as body surface area and individual lesion locations, and was still recommended in combination with PASI by the European Medicines Agency [26]. Controversy about IGA response as the primary outcome had persisted due to the variation in nomenclature, scale size, definitions and outcome interpretation. Some defined efficacy as clear or almost clear lesions (IGA 0,1), whereas others added a 2-point reduction in the score (IGA 0,1, Reduction ≥ 2). Robinson et al. reported that IGA and PASI correlated quite tightly and yielded very similar results in relatively efficacious therapies for moderate-to-severe patients, which were redundant [26]. They recommended PASI alone because it was statistically superior and incorporated body surface area. Particularly, since the high sensitivity represented lower specificity, PASI 50, PASI 75 and IGA uncovered therapeutic efficacy easily but perhaps overestimated the impact of placebo. PASI 125 consistently underperformed due to its inherent nature of reflecting disease worsening. Specifically, it captured post-treatment exacerbation rather than clinical improvement, serving more as an indicator of safety and tolerability. As a result, it did not align with the standard efficacy objectives of most RCTs and was therefore considered the least effective indicator for assessing treatment efficacy. What’s more, DLQI 0,1, PatGA 0,1 and NRS-4 were scored directly by patients and might be regarded as subjective and less rigorous, as also indicated in our results. Nevertheless, patient-oriented outcomes truly reflected what was important to the patient and might be preferred outside clinical trials since they could better tell the patient’s feelings on the therapy [19]. In addition, most international guidelines recommended the combination of objective measurements and subjective assessments from the patient’s perspective [27].

Assessment of treatment efficacy in PsA had historically relied on the standard criteria developed for rheumatoid arthritis. However, given the heterogeneous nature of PsA, disease-specific indicators were essential to comprehensively evaluate its multiple domains, including peripheral joint activity, skin activity, patient global assessment, pain assessment, physical function and health-related quality of life [28]. Although PASI scores were well-established and widely accepted as sensitive measures of skin improvement in psoriasis, they were used infrequently in PsA trials [29]. This might partly reflect an outdated emphasis on articular symptoms, despite that more than 80% of PsA patients presented with both joint and active cutaneous involvement [30, 31]. Overlooking the skin lesions not only underestimated the full burden of disease but might also compromise therapeutic assessment. Recent studies have demonstrated a potential association between cutaneous and articular disease activity in PsA, suggesting that although PASI was originally developed to assess skin involvement, it might also provide indirect insights into overall disease activity in PsA patients, particularly in patients who present with both skin and joint symptoms [32–35].

Our findings emphasized the distinct roles of PASI endpoints and ACR endpoints indicators in PsA. In practice, PASI endpoints reflected skin response while ACR endpoints captured joint improvement; a composite view was needed for holistic PsA assessment. ACR 20 was favored for its sensitivity in early-phase trials but might overestimate minor improvements; conversely, ACR 50, ACR 70, and MDA were stricter but risk underestimating drug benefit [29, 36]. In our analysis, PsARC showed inconsistent discriminatory ability, echoing concerns about its limited validation [8, 37]. To better inform clinical trial design, we proposed that early-phase studies might benefit from more sensitive indicators like PASI 50 and ACR 20 to detect early signals, whereas confirmatory late-phase trials should emphasize more stringent or composite endpoints such as ACR 50, or MDA.

To guide practical implementation, suggestions for primary outcome indicator(s) selection in future RCTs of Pso and PsA were listed in Table 2, outlining preferred indicators based on disease type, intervention class and trial phase. This table might assist investigators in selecting context-appropriate primary outcomes that balanced sensitivity with specificity. Due to the limited number of trials, detailed subgroup analysis was not possible. A balanced indicator selection was always required since the more sensitive indicator was associated with more false positives. What’s more, the combination of biomarkers might provide a more comprehensive picture of the disease status in patients with Pso and PsA [38]. A further exploration of the evaluation ability of outcome measures reporting score change was also warranted.

Table 2.

Recommendations for the selection of primary outcome of RCTs for psoriasis and psoriatic arthritis

Items	Suggested Indicators		Not suggested Indicators
	Psoriasis	Psoriatic arthritis	Psoriasis	Psoriatic arthritis
Overall	PASI 50 PASI 75 IGA 0,1	PASI 75 PASI 50 PASI 90	PASI 125 DLQI 0,1 NRS-4	ACR 70 MDA ACR 50
Baseline severity of participants
All	PASI 50 IGA 0,1	PASI 75 PASI 90 ACR 20	PASI 90 PatGA 0,1	ACR 70 MDA ACR 50
Moderate	PASI 75 PatGA 0,1	–	PASI 90	–
Severe	PASI 75 PASI 50	–	DLQI 0,1	–
Mild to moderate	PASI 50 PASI 75 IGA 0,1	–	PASI 125 PASI 90 IGA 0,1, Reduction ≥ 2	–
Moderate to severe	PASI 50 PASI 75 IGA 0,1, Reduction ≥ 2	PASI 75 PASI 50 PASI 90 ACR 20	PASI 125 DLQI 0,1 NRS-4	MDA ACR 70 ACR 50
Not mentioned	PASI 50 PASI 75 IGA 0,1	PASI 75 PASI 50 PASI 90 PsARC	PASI 125 DLQI 0,1 IGA 0,1, Reduction ≥ 2	ACR 70 ACR 50 MDA
Type of intervention
Antibodies	PASI 50 PASI 75 IGA 0,1, Reduction ≥ 2	PASI 75 PASI 50 PASI 90 ACR 20	PASI 125 DLQI 0,1 NRS-4	ACR 70 MDA ACR 50
Small molecules	PASI 50 PASI 75 NRS-4	ACR 20 PsARC ACR 50	PASI 125 DLQI 0,1 PASI 90	PASI 90 ACR 70 MDA
Non-biologics	PASI 75 IGA 0,1 PASI 90	PsARC	PASI 125 DLQI 0,1	ACR 50
Application method
Topical	IGA 0,1 PASI 50 PASI 75	–	PASI 125 PASI 90 IGA 0,1, Reduction ≥ 2	–
Systemic	PASI 50 PASI 75 IGA 0,1, Reduction ≥ 2	PASI 75 PASI 50 PASI 90 PsARC	PASI 125 DLQI 0,1 NRS-4	ACR 70 MDA ACR 50

PASI 50: ≥ 50% reduction in the Psoriasis Area and Severity Index (PASI) score from baseline; PASI 75: ≥ 75% reduction in the PASI score from baseline; PASI 90: ≥ 90% reduction in the PASI score from baseline; PASI 125: ≥ 125% of the baseline PASI score; IGA 0,1: Investigator’s Global Assessment (IGA) score of 0 (clear) or 1 (almost clear) at endpoint; IGA 0,1, Reduction ≥ 2: IGA score of 0 or 1 with at least a 2-grade improvement from baseline; PatGA 0,1: Patient’s Global Assessment (PatGA) score of 0 (clear) or 1 (almost clear) at endpoint; DLQI 0,1: Dermatology Life Quality Index (DLQI) score of 0–1, indicating no impact on quality of life; NRS-4: ≥ 4-point reduction in Itch Numeric Rating Scale (NRS) from baseline; ACR 20: American College of Rheumatology (ACR) 20% improvement criteria; ACR 50: ACR 50% improvement criteria; ACR 70: ACR 70% improvement criteria; MDA: Minimal Disease Activity; PsARC: Psoriatic Arthritis Response Criteria

This study has several limitations. First, the number of RCTs included in the analysis, particularly those involving patients with PsA, remains limited. This scarcity reflects the lack of high-quality trials reporting response rates for various outcome indicators. Second, among the included studies, the representation of PsA patients evaluated using the ACR 20. Although it's widely accepted for assessing treatment response in inflammatory arthritis, the small sample sizes and clinical heterogeneity of PsA patients may reduce the robustness and generalizability of conclusions drawn from ACR 20 assessments alone. Third, our study relies on retrospective data extracted from registered clinical trials, which may introduce heterogeneity and selection bias and limit the applicability of the findings to the broader populations of Pso and PsA patients. Despite these limitations, our findings provide a valuable foundation for the selection and standardization of outcome measures in clinical trials involving Pso and PsA. This can improve comparability across studies and support more informed, evidence-based treatment decisions. Future research should focus on prospective validation of these results in larger and more diverse cohorts. Additionally, exploring novel outcome indicators and employing advanced statistical or modeling approaches will be crucial to further optimize efficacy evaluation in these complex conditions. Ultimately, this work may inform the design of more efficient and patient-centered clinical trials in psoriatic disease.

Conclusion

In summary, our findings presented evidence to determine the preference of indicators reporting response rate as primary outcome(s). This comparative analysis of outcome indicators offered practical guidance for designing more sensitive, valid, and patient-centered clinical trials in Pso and PsA. To better evaluate and reveal the therapeutic advantages of pharmacological interventions, PASI 50 and PASI 75 were recommended as the primary outcome of Pso RCTs, and the PASI endpoints along with ACR 20 were recommended for PsA RCTs. Comprehensive assessments together with other types of indicators were also essential. As for trials that only used indicators with extremely sensitivity, attention should be paid to interpret the outcomes to avoid exaggerating the efficacy of treatment.

Abbreviations

ACR

American College of Rheumatology

ANZCTR

Australian and New Zealand Clinical Trials Registry

CNKI

China National Knowledge Infrastructure

CENTRAL

Cochrane Library Central Register of Controlled Trials

CI

Confidence Intervals

DLQI

Dermatology Life Quality Index

DAS

Disease Activity Score

FDA

Food and Drug Administration

GRAPPA

Group for Research and Assessment of Psoriasis and Psoriatic Arthritis

ISRCTN

International Standard Randomised Controlled Trial Number Register

IGA

Investigator’s Global Assessment

MDA

Minimal Disease Activity

PatGA

Patient’s Global Assessment

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta-Analyses

Pso

Psoriasis

PASI

Psoriasis Area and Severity Index

PsA

Psoriatic Arthritis

PsARC

Psoriatic Arthritis Response Criteria

RCTs

Randomized Clinical Trials

Author contributions

QL and JT conceived of the study. JT and QL developed the protocol. JT, DZ and SK did the literature search. SK and JT appraised study quality, and extracted and analyzed the data. DZ was in charge of computation and coding. JT, SK, DZ, MZ, XY and QL interpreted the data. SK and JT wrote the first draft of the article. XY and YH revised the first draft of the article. QL reviewed and critically evaluated the draft paper. QL is responsible for the overall content as the guarantors. All authors reviewed the manuscript.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Ethics approval was not applicable, since this research was based exclusively on previously published studies.

IRB approval status

Not applicable.

Patient consent

Not applicable.

Competing interests

The authors declare no competing interests.