Introduction
Psoriasis is a chronic inflammatory skin disease that can significantly affect patients’ quality of life [
1,
2]. In recent years, as the pathogenesis of the disease has become better understood, a variety of novel targeted treatments have become available [
1,
3]. Multiple biologic classes have demonstrated efficacy for the treatment of moderate-to-severe plaque psoriasis at set time points, including tumor necrosis factor (TNF) inhibitors, interleukin (IL)-17 inhibitors, an IL-12/23 inhibitor, and IL-23 inhibitors [
3,
4]. There is also evidence that these treatments can improve patients’ quality of life [
5,
6]. However, data on predictors and dynamics of response to biologic treatment are limited and primarily drawn from registry and retrospective studies [
7‐
10]. Therefore, generating association models from randomized controlled trial data may help with efforts to personalize healthcare in psoriasis, to maximize positive efficacy outcomes and minimize negative safety outcomes [
11].
Guselkumab, an IL-23p19 inhibitor administered by subcutaneous injection, has been extensively investigated in studies of patients with moderate-to-severe plaque psoriasis, such as the phase III clinical trials VOYAGE 1 (which included an open-label extension [OLE]) and ECLIPSE [
12‐
14]. In VOYAGE 1, higher proportions of patients achieved Investigator’s Global Assessment (IGA) 0 or 1 and Psoriasis Area and Severity Index (PASI) 90 responses at week 16 with guselkumab versus placebo (co-primary endpoint) or adalimumab [
12]. In ECLIPSE, guselkumab was superior to secukinumab on the basis of the PASI 90 primary endpoint at week 48 [
13].
In clinical practice, patient responses to biologics can be categorized into a variety of response types (RTs), such as “maintenance of clear response” and “non-acceptable response.” We hypothesize that a patient’s RT is likely to vary over time. When biologic treatment is initiated, there is usually an induction period during which initial response occurs and continues to improve (weeks 0–12/16), followed by a stabilization period when maximal efficacy is achieved and begins to level off (weeks 20–48). Beyond this point (week 52 onwards, “long term”), patients are in a maintenance phase.
The objective of these post hoc analyses of data from ECLIPSE and VOYAGE 1 was to understand the treatment response dynamics of different subgroups of patients with short-term RTs (SRTs) receiving guselkumab, adalimumab, or secukinumab in phase III trials and long-term RTs (LRTs) for those who continued guselkumab in the VOYAGE 1 OLE. We then evaluated factors associated with the most favorable treatment RT.
Methods
Source of Data
Data were derived from the ECLIPSE (NCT03090100; April 2017 through September 2018 at 142 sites) and VOYAGE 1 (NCT02207231; December 2014 through June 2020, including the OLE, at 101 sites) phase III, global, multicenter trials. Study designs have been published previously and are briefly described here [
12,
13]. Results of these post hoc analyses are presented. Data were analyzed between September 2021 and November 2022.
Participants
Eligible patients were ≥ 18 years of age and had moderate-to-severe plaque psoriasis (absolute PASI [aPASI] ≥ 12, IGA score ≥ 3, and body surface area [BSA] involvement ≥ 10%) for ≥ 6 months, and were candidates for phototherapy or systemic therapy [
12,
13]. The protocols for ECLIPSE and VOYAGE 1 were approved by relevant ethics committees and review boards. Both studies were conducted in accordance with the Declaration of Helsinki and are consistent with Good Clinical Practice. Participating patients provided written informed consent.
Randomization, Treatment, and Masking
In ECLIPSE, patients were randomized 1:1 to receive either guselkumab (100 mg at weeks 0 and 4, then every 8 weeks) or secukinumab (300 mg at weeks 0, 1, 2, 3, and 4, then every 4 weeks) through week 44 [
13]. In VOYAGE 1, patients were randomized 2:1:2 to receive either guselkumab 100 mg (weeks 0 and 4, then every 8 weeks), placebo/guselkumab (placebo at weeks 0, 4, and 12, then guselkumab at weeks 16 and 20, then every 8 weeks), or adalimumab (80 mg at week 0, 40 mg at week 1, then 40 mg every 2 weeks through week 47) [
12]. Patients in the VOYAGE 1 OLE receiving guselkumab from baseline for ≥ 156 consecutive weeks were included [
14].
Outcome Assessments
In ECLIPSE, aPASI scores were obtained weekly from baseline to week 4 and every 4 weeks thereafter until week 48; in VOYAGE 1 and the OLE, aPASI scores were obtained at baseline, week 2, every 4 weeks from week 4 to 52, then every 8 weeks to week 252.
SRTs and LRTs were author-defined at weeks 48 and 252, respectively, and were based on aPASI scores using cutoff values judged to be clinically relevant. Patients receiving treatment with guselkumab, secukinumab, or adalimumab between weeks 0 and 48 in ECLIPSE and VOYAGE 1 were categorized into six SRTs (Table
1). Patients receiving guselkumab from week 0 for ≥ 156 consecutive weeks through up to 5 years in the VOYAGE 1 OLE were categorized into four LRTs (Table
1). RTs were mutually exclusive, and each patient was assigned to the most favorable RT they achieved, based on as-observed aPASI scores reported at each time point through a defined period. Analyses were also conducted with four alternatively defined LRTs based on aPASI and Dermatology Life Quality Index (DLQI) scores reported at each time point through a defined period using clinically relevant cutoff values as determined by the authors (Table
1). DLQI scores were collected at weeks 0, 8, 16, 24, and 48 in VOYAGE 1 and weeks 76, 100, 124, 156, 172, 204, 228, and 252 in the VOYAGE 1 OLE, but were not collected in ECLIPSE.
Table 1
Response type criteria
Short term |
SRT1 | Maintenance of clear response | Patients with an aPASI score of 0 at each visit for weeks 20–48 | N/A |
SRT2 | Maintenance of optimal response | Patients who were not in SRT1 and with an aPASI score of ≤ 1 at each visit for weeks 20–48 | N/A |
SRT3 | Fluctuation around optimal response | Patients who were not in SRT1 or 2 and with an aPASI score of ≤ 1 at least once from baseline to week 16 and weeks 20–48 and with an aPASI score ≤ 3 for weeks 20–48 | N/A |
SRT4 | Gaining optimal response over time | Patients who were not in SRT1, 2, or 3 and with an aPASI score of ≤ 1 for the first time in weeks 20–48, and an aPASI score ≤ 3 for weeks 20–48 | N/A |
SRT5 | Partial response | Patients who were not in SRT1, 2, 3, or 4 and with an aPASI score ≤ 5 for weeks 20–48 | N/A |
SRT6 | Non-acceptable response | Patients who were not in SRT1, 2, 3, 4, or 5 and with an aPASI score of > 5 at any visit from week 20–48 or who discontinued the study for any reason | N/A |
Long term |
LRT1 | No disease activity | Patients with all aPASI scores of 0 during weeks 52–252 | Patients with all aPASI scores of 0 during weeks 52–252; no DLQI requirement |
LRT2 | Low disease activity | Patients who were not in LRT1 and had all aPASI scores of ≤ 2 during weeks 52–252 | Patients who were not in LRT1 and had all aPASI scores of ≤ 2 during weeks 52–252; DLQI score of ≤ 5 during weeks 20–252 |
LRT3 | Tolerable disease activity | Patients not in LRT1 or 2 who had all aPASI scores of ≤ 5 during weeks 52–252 | Patients not in LRT1 or LRT2 who had all aPASI scores of ≤ 5 during weeks 52–252; DLQI score of ≤ 10 during weeks 20–252 |
LRT4 | Variable response | Comprised all patients not in LRT1, 2, or 3 | Patients not in LRT1, LRT2, or LRT3; no DLQI requirement |
Analyses
After classification of patients into RTs and development of heatmap visualizations, baseline characteristics were compared across RTs with descriptive statistics. A Sankey diagram was developed to depict the proportion of patients in VOYAGE 1 SRTs who were recategorized into each of the respective LRTs in the VOYAGE 1 OLE.
A multivariable logistic regression model was developed to establish factors associated with RT1. For ECLIPSE and VOYAGE 1, SRT1 was defined over weeks 20–48 for patients treated with guselkumab, adalimumab, or secukinumab, and for the VOYAGE 1 OLE, LRT1 was defined over weeks 52–252 for patients treated with guselkumab, with LRTs based on aPASI or aPASI and DLQI criteria. Variables included in the model were chosen on the basis of results of comparing characteristics across RTs; variables included baseline age, body mass index (BMI), sex, smoking status, presence of psoriatic arthritis, psoriasis duration, prior exposure to systemic treatment, aPASI score, C-reactive protein level, and aPASI score at week 16. Observed data were used with no imputation. Forward and backward elimination techniques, with alpha = 0.10, were used for covariate selection. Significance was not corrected. All statistical analyses were conducted using SAS version 9.4 (SAS Institute, Inc., Cary, NC, USA) or higher using the Windows operating system.
Discussion
In these post hoc analyses of data from two multicenter randomized controlled clinical trials, patients in favorable SRTs and LRTs tended to be younger, have lower baseline BMI and aPASI scores, and have shorter duration of psoriasis than those in less favorable RTs. In multivariable logistic regression analyses, lower week 16 aPASI scores were consistently associated with a higher chance of achieving the most favorable RTs (SRT1 in ECLIPSE and VOYAGE 1, and LRT1 in the VOYAGE 1 OLE). In VOYAGE 1, but not ECLIPSE, shorter disease duration and never having smoked (versus being a former smoker) were also associated with a higher probability of achieving SRT1; no baseline demographics were associated with achieving LRT1.
As expected, on the basis of the efficacy results from the ECLIPSE and VOYAGE 1 studies, more patients receiving guselkumab were classified into favorable SRTs versus secukinumab or adalimumab. At the end of the VOYAGE 1 OLE, most patients treated with guselkumab had low disease activity (LRT2). These analyses identified differences in characteristics that help to explain these results. For example, short-term treatment with adalimumab resulted in categorization into less favorable SRTs among current smokers versus patients who had never smoked, whereas smoking status did not appear to impact SRT classification for patients receiving secukinumab or guselkumab. Higher baseline BMI appeared to result in categorization into less favorable SRTs among patients receiving secukinumab or adalimumab, but no clear pattern was observed with guselkumab. Validation studies are required to confirm the association between variables identified and the RTs defined here and to support use of these RTs for the guidance of treatment in clinical practice.
These results contribute to the ongoing effort to personalize treatment of psoriasis, which requires an understanding of the factors that may affect response to different treatments [
11,
15]. The RTs used were author-defined and based on clinical experience, but generally appear consistent with wider interpretation of aPASI scores. For example, in an analysis of data from the British Association of Dermatologists Biologics and Immunomodulators Register (BADBIR), including more than 13,000 patients with psoriasis, achieving aPASI score ≤ 4 was concordant with PASI 75 response in 88% of cases [
16]. Additionally, in a population-based cohort study including 2034 patients with moderate-to-severe psoriasis and 3 years’ sustained treatment with adalimumab, etanercept, infliximab, certolizumab pegol, ustekinumab, secukinumab, ixekizumab, or brodalumab, predictive prognostic models were used to identify optimal biologic treatment [
17]. Modeling predicted which cytokine targets were associated with a successful treatment outcome with 59.2–63.6% accuracy overall and 93.9–95.9% accuracy for the two most successful treatments. Therefore, predicted response to treatment could potentially be used to guide a holistic approach for managing patients with psoriasis, which may allow physicians to optimize the balance of treatment benefit and the risk of side effects on an individual basis [
11,
15].
The association of smoking with the onset of psoriasis and disease severity is well established [
18‐
20]. Data from real-world registries [
7‐
9], retrospective analyses [
10], and a recent meta-analysis [
21] have shown that smokers have a worse response to biologic therapy than non-smokers. However, biologics have more commonly been studied either by mechanism of action or as an overall class, but differences between individual biologics have not been established. In a recent pooled analysis evaluating the effect of baseline factors on outcomes with secukinumab, there were more non-smokers in groups with better responses to treatment; a similar, but weaker, trend was also seen in patients treated with etanercept [
22]. The value of smoking cessation for reducing psoriasis severity remains unclear, and its impact on treatment outcomes in patients with psoriasis has yet to be explored [
21]. Nonetheless, smoking cessation programs should be considered for patients with psoriasis [
20].
The less favorable outcomes observed in this study in patients with a high BMI treated with secukinumab and adalimumab are also consistent with observations in real-world and retrospective analyses [
7‐
10]. Additionally, in an analysis of pooled phase III data from patients with psoriasis (FIXTURE, ERASURE, and CLEAR trials), lower BMI was associated with a better response to treatment with secukinumab [
22]. In the prospective GUIDE study, which analyzed factors leading to complete skin clearance following guselkumab treatment, each kg/m
2 increase in BMI negatively affected the likelihood of achieving clearance [
23]. However, consistent with real-world studies [
24,
25], our analyses found that the effect of BMI on efficacy may be less with guselkumab treatment compared with other treatments. Weight loss has been shown to improve treatment outcomes in patients with psoriasis [
26,
27], and weight loss interventions should also be considered when treating patients.
Finally, prior biologic therapy may be associated with a poor short-term response to guselkumab and secukinumab. This is consistent with other studies showing that prior systemic or biologic therapy can negatively affect the efficacy of TNF inhibitors [
28,
29], ustekinumab [
28,
30], secukinumab [
22], and guselkumab [
25].
The primary strength of this analysis is that it is based on robust datasets from two randomized clinical trials that included a total of almost 2000 participants. Our analysis therefore includes comparison of three different treatments. In addition, these datasets included long-term follow-up, to a maximum of 252 weeks, allowing differentiation of short-term and long-term response. It is also a strength that the RT definitions were based on thresholds deemed clinically relevant by psoriasis experts. In contrast, a limitation of the analysis is that the RT definitions did not consider patient views. A recent consensus study, including patients with psoriasis, dermatologists, and nurses, described freedom from disease as multifaceted, with five domains identified as management of clinical symptoms, considerations beyond the skin, treatment burden, quality of care, and well-being [
31]. Hence, this analysis may have missed certain patient-relevant considerations by focusing on clinical responses, evaluating aPASI ± DLQI alone. Other limitations include the post hoc nature of these analyses, use of descriptive statistics, and the lack of statistical analyses comparing adjustment for baseline characteristics across different RTs. Long-term data were only available for guselkumab, with no comparator. Additionally, with the exception of obesity, baseline cardiometabolic parameters were not evaluated; diabetes, metabolic syndrome, and hypertension have all been shown to be associated with response to biologics in other studies [
22]. While similar results were found with RTs defined on the basis of aPASI score alone or aPASI and DLQI scores, patient-reported outcomes are important measures of treatment success [
31]. Covariates using logistic regression were derived on the basis of a relatively small number of patients, and correlation/agreement within covariates was not evaluated. Further, aPASI score was used to define RTs and was also a covariate and therefore, as expected, was associated with SRT1/LRT1. Most covariates were measured at baseline with the response variable based on a time interval; it is possible that a given characteristic could have changed (e.g., smoking status) during the time interval used to define the RT. Lastly, the performance of the models was never compared.
Declarations
Conflict of Interest
Alexander Egeberg has received research funding from AbbVie, Boehringer Ingelheim, Bristol Myers Squibb, the Danish National Psoriasis Foundation, Eli Lilly, Janssen Pharmaceuticals, the Kgl Hofbundtmager Aage Bang Foundation, Novartis, Pfizer, and the Simon Spies Foundation, and honoraria as consultant or speaker from AbbVie, Almirall, Boehringer Ingelheim, Bristol Myers Squibb, Dermavant, Eli Lilly, Galapagos NV, Galderma, Horizon Therapeutics, Janssen, LEO Pharma, Mylan, Novartis, Pfizer, Samsung Bioepis Co., Ltd., Sun Pharmaceuticals, UCB, Union Therapeutics, and Zuellig Pharma Ltd. Since the completion of this work, Alexander Egeberg has changed affiliation. His new affiliation is LEO Pharma A/S, Industriparken 55, DK-2750 Ballerup, Denmark. Curdin Conrad received research grants or served as a scientific adviser or clinical study investigator for AbbVie, Actelion, Almirall, Amgen, Boehringer Ingelheim, Bristol Myers Squibb, Celgene, Eli Lilly, Galderma, Incyte, Janssen, LEO Pharma, MSD, Novartis, Pfizer, Samsung, Sanofi, and UCB. Jozefien Buyze, Patricia Gorecki, and Sven Wegner are employees of Janssen; Sven Wegner holds stocks in Johnson & Johnson; Lorenzo Acciarri is an employee of Valos Srl, a Johnson & Johnson partner company. Diamant Thaçi has been a consultant, investigator, and speaker and participated in advisory boards for AbbVie, Almirall, Amgen, Biogen Idec, Bristol Myers Squibb, Janssen-Cilag, LEO Pharma, Lilly, Novartis, Pfizer, Regeneron, Sanofi, and UCB, and has received research/educational grants from AbbVie, LEO Pharma, Novartis, and Sanofi.