Comparison of Biologics and Oral Treatments for Plaque Psoriasis: A Meta-analysis

Key Points

Question  What is the short-term and long-term comparative efficacy among biologics and oral agents for plaque psoriasis?

Findings  In a network meta-analysis of 60 clinical trials for short-term efficacy, brodalumab, guselkumab, ixekizumab, and risankizumab-rzaa had the highest Psoriasis Area and Severity Index response rates at 10 to 16 weeks from baseline. A meta-analysis of long-term efficacy suggested that brodalumab, guselkumab, ixekizumab, and risankizumab-rzaa had the highest response rates at 44 to 60 weeks.

Meaning  This study provides an assessment of both short-term and long-term comparative efficacy among treatments for moderate to severe plaque psoriasis which can help health care stakeholders optimize treatment regimens.

Importance  The clinical benefits of novel treatments for moderate to severe psoriasis are well established, but wide variations exist in patient response across different therapies. In the absence of head-to-head randomized trials, meta-analyses synthesizing data from multiple studies are needed to assess comparative efficacy among psoriasis treatments.

Objective  To estimate the relative short-term and long-term efficacy of biologics and oral agents for the treatment of moderate to severe psoriasis.

Data Sources  A systematic literature review was conducted on December 4, 2017, and updated on September 17, 2018. The Embase, MEDLINE, and Cochrane Central Register databases were included.

Study Selection  Phase 2, 3, or 4 randomized clinical trials of treatments licensed by the US Food and Drug Administration and the European Medicines Agency for adults with moderate to severe psoriasis with data on Psoriasis Area and Severity Index assessment of 75%, 90%, and 100% reductions (PASI 75, 90, and 100) at 10 to 16 weeks (short-term efficacy) or 44 to 60 weeks (long-term efficacy) from baseline.

Data Extraction and Synthesis  Data were extracted based on the Preferred Reporting Items for Systematic Review and Meta-analysis guidelines. A bayesian network meta-analysis was conducted to estimate short-term PASI response rates; to account for variation across trials, an ordinal model that adjusted for reference arm response was implemented. The long-term PASI rates were estimated via a traditional meta-analysis.

Main Outcomes and Measures  PASI 75, 90, and 100 response rates at 10 to 16 weeks and 44 to 60 weeks from baseline.

Results  Sixty trials meeting all inclusion criteria were included. At weeks 10 to 16, the highest PASI 90 rates were seen with risankizumab-rzaa (71.6%; 95% credible interval [CrI], 67.5%-75.4%), brodalumab (70.8%; 95% CrI, 66.8%-74.6%), ixekizumab (70.6%; 95% CrI, 66.8%-74.6%), and guselkumab (67.3%; 62.5%-71.9%). At weeks 44 to 60, the treatments with the highest PASI 90 rates were risankizumab-rzaa (79.4%, 95% CI, 75.5%-82.9%), guselkumab (76.5%; 95% CI, 72.1%-80.5%), brodalumab (74.0%; 95% CI, 69.3%-78.1%), and ixekizumab (73.9%; 95% CI, 69.9%-77.5%). Findings were consistent for short-term and long-term PASI 75 and 100 responses.

Conclusions and Relevance  This study provides an assessment of the comparative efficacy among treatments for moderate to severe plaque psoriasis. The meta-analysis suggests that brodalumab, guselkumab, ixekizumab, and risankizumab-rzaa were associated with the highest PASI response rates in both short-term and long-term therapy.

The treatment options for patients with moderate to severe psoriasis have expanded greatly over the past decade.^1-4 Among the treatments, Quiz Ref IDbiologics provide targeted inhibition of immune-mediated pathways involving specific cytokines, such as tumor necrosis factor (TNF), interleukin (IL)-17, and IL-23.^5,6 Biologics licensed by the US Food and Drug Administration and the European Medicines Agency for the treatment of moderate to severe psoriasis include the tumor necrosis factor inhibitors adalimumab, etanercept, infliximab, and certolizumab pegol; the IL-12/23 inhibitor ustekinumab; the IL-17 inhibitors secukinumab, ixekizumab, and brodalumab; and the IL-23 inhibitors tildrakizumab-asmn, guselkumab, and risankizumab-rzaa.^7-9

Although the increased options of biologics and oral treatments for moderate to severe psoriasis have provided substantial benefit to patients, it can be challenging for clinicians to determine how the medications compare with one another. Variations exist across different therapies with regard to efficacy, safety, and dosing profiles.^10,11

Although several head-to-head trials exist,^12-22 they are not available for all possible comparisons. In the absence of head-to-head trials across the entire set of comparators, studies that combine and analyze data from multiple studies are needed to determine comparative efficacy. The overall objective in this study is to evaluate the comparative efficacy of systemic treatments for psoriasis, including newly developed biologics. Specifically, the short-term, relative rates of Psoriasis Area and Severity Index (PASI) response are estimated via a network meta-analysis (NMA), and the long-term PASI response rates following maintenance therapy are estimated via a traditional meta-analysis.

A systematic literature review was performed on December 4, 2017, and updated on September 17, 2018, to identify randomized clinical trials of treatments licensed by the US Food and Drug Administration and the European Medicines Agency for adults with moderate to severe psoriasis. This systematic review was conducted based on the Preferred Reporting Items for Systematic Review and Meta-Analysis guidelines for Network Meta-Analysis (PRISMA-NMA) using Embase, MEDLINE, and the Cochrane Central Register of Controlled Trials.²³ Identified studies were independently assessed by 2 reviewers for inclusion at each stage of study selection. Any discrepancies between the inclusion/exclusion decisions were resolved by a third independent reviewer.

To be included in the analyses, studies were required to (1) be a phase 2, 3, or 4 randomized clinical trial on adults with moderate to severe psoriasis who were eligible for systemic therapies and phototherapy, (2) include European Medicines Agency–licensed treatments and dosages for moderate to severe psoriasis, and (3) report at least 1 of the efficacy outcomes of interest (PASI 75, 90, and 100, indicating 75%, 90%, or 100% reductions) at the end of the primary response period (10-16 weeks from baseline) or at the end of the maintenance period (44-60 weeks from baseline).

Comparators for the base case analyses included anti-TNF agents (etanercept, 25 mg, twice weekly or 50 mg once weekly, or etanercept, 50 mg twice weekly until week 12, then once weekly; adalimumab, 80 mg, at week 0, then 40 mg every 2 weeks starting at week 1; certolizumab pegol, 400 mg, at weeks 0, 2, and 4, then 200 mg every 2 weeks; certolizumab pegol, 400 mg, every 2 weeks; infliximab, 5 mg/kg, at weeks 0, 2, and 6, then every 8 weeks), apremilast, 30 mg, twice daily after initial titration schedule; dimethyl fumarate, ustekinumab weight-based dosage (45 mg ≤100 kg, 90 mg >100 kg at weeks 0 and 4, then every 12 weeks), anti-IL-17 agents (secukinumab, 300 mg, at weeks 0, 1, 2, and 3, then every 4 weeks starting at week 4; ixekizumab, 160 mg, at week 0, then 80 mg every 2 weeks; brodalumab, 210 mg, at weeks 0, 1, and 2, then every 2 weeks), and anti-IL-23 agents (tildrakizumab-asmn, 100 mg, at weeks 0, 4, then every 12 weeks; tildrakizumab-asmn, 200 mg, at weeks 0, 4, then every 12 weeks; guselkumab, 100 mg, at weeks 0, 4, then every 8 weeks; risankizumab-rzaa, 150 mg, at weeks 0, 4, then every 12 weeks). Different dosing schedules of etanercept with 25 mg twice weekly and 50 mg once weekly were assumed to have the same clinical efficacy, and the 2 dosages were pooled into a single etanercept, 25 mg twice weekly/50 mg once weekly, treatment arm. For the outcomes at the end of the maintenance period, a dose of 50 mg twice weekly until week 12 then once weekly (approved by the US Food and Drug Administration and also as an alternative dosage licensed by the European Medicines Agency) was used. In addition, the following conventional systemic treatments were included in a sensitivity analysis for the short-term analyses: acitretin, 0.4 mg/kg daily, cyclosporine, 2.5-3 mg/kg/d (initial dose), fumaric acid esters (licensed in Germany only), low-dose methotrexate (initial dose of 5-7.5 mg once weekly, increased as needed and as tolerated up to a maximum of 15-25 mg once weekly), and high-dose methotrexate (initial dose of 15 mg once weekly, increased as needed up to a maximum of 20-22.5 mg once weekly).

In this analysis, the outcomes of interest were the proportion of patients that achieved 75%, 90%, and 100% reductions in PASI and the associated number needed to treat (NNT) at the end of the primary response period (10-16 weeks from baseline) and at the end of the maintenance period (44-60 weeks from baseline).

An ordinal bayesian random effects NMA²⁴ was conducted to compare the relative PASI 75, 90, and 100 responses at the end of the primary response period (10-16 weeks from baseline) across treatments. To account for heterogeneity across trials, a model that adjusted for reference-arm response was implemented.^25,26 The base-case analysis included all biologics, apremilast, and dimethyl fumarate. The estimated PASI response rates were summarized using posterior medians and their associated 95% credible intervals (CrIs). The NNT for each treatment was calculated as the reciprocal of the treatment effect difference vs placebo. Pairwise comparisons of the estimated treatment effects relative to risankizumab-rzaa, ixekizumab, brodalumab, and guselkumab were summarized using odds ratios (ORs) and their associated 95% CrIs. All analyses were conducted using R, version 3.5.1 statistical software (R Foundation) and WinBUGS, version 1.4.3 with noninformative priors, 40 000 burn-in iterations, a thinning factor of 10, and 3 chains each with 100 000 posterior iterations. Three sensitivity analyses were conducted: 1 that included only global trials (country-specific trials were excluded to assess the associations of homogeneous populations with the study results), 1 that included only phase 3 trials, and 1 that focused on an expanded treatment space that also included conventional systemic treatments (acitretin, cyclosporine, fumaric acid esters, and methotrexate).

Most of the psoriasis trials feature crossover (typically, the placebo arm switching to the active treatment) after the primary short-term end point. In addition, some psoriasis trials rerandomize patients based on the level of PASI response achieved. Owing to the relative dearth of trials that did not feature crossover or rerandomization, a traditional random-effects meta-analysis²⁷ was conducted to estimate the PASI 75, 90, and 100 response rates at the end of the maintenance period (44-60 weeks from baseline). Only data from trial treatment arms in which baseline treatment assignment was maintained during the maintenance period were used. Each intervention’s response rate and its associated 95% CI were summarized. Pairwise comparisons of the estimated response rates relative to brodalumab, guselkumab, ixekizumab, and risankizumab-rzaa were summarized using ORs and their associated 95% CIs. Three sensitivity analyses were conducted: an analysis of only global trials (country-specific trials were excluded to assess the association between the homogeneous populations and the study results), 1 that included only phase 3 trials, and 1 that included only trials that reported nonresponse imputed (NRI) data.

A total of 11 476 studies were systematically identified. Following elimination of duplicate studies, 8037 studies were screened based on title and abstract and 713 full-text studies were evaluated. An updated search on September 17, 2018, identified an additional 59 randomized clinical trials of treatments and dosages licensed by the European Medicines Agency for adults with moderate to severe psoriasis. A total of 412 publications covering 160 studies met the inclusion criteria (Figure 1). The full list of clinical trials^{13,15-21,28-86} and their reported efficacy data that were analyzed are summarized in eTable 1 in the Supplement.

The evidence network for the base-case NMA of short-term efficacy 10 to 16 weeks from baseline is illustrated in Figure 2.^{13,15-21,30-35,37-47,49-69,71-74,83,84} A total of 60 trials meeting all inclusion criteria were included in the base-case NMA. The median posterior probability of each intervention achieving at least the given PASI response (PASI 75, 90, and 100) is presented in the Table. Quiz Ref IDThe estimated PASI 90 response rates at 10 to 16 weeks from baseline were 71.6% (95% CrI, 67.5%-75.4%) for risankizumab-rzaa; 70.8% (95% CrI, 66.8%-74.6%) for brodalumab; 70.6% (95% CrI, 66.8%-74.6%) for ixekizumab; 67.3% (95% CrI, 62.5%-71.9%) for guselkumab; 61.4% (95% CrI, 57.2%-65.6%) for secukinumab; 57.4% (95% CrI, 52.2%-62.8%) for infliximab; 45.6% (39.3%-52.2%) for certolizumab pegol, 400 mg; 43.9% (40.2%-47.9%) for ustekinumab; 43.7% (95% CrI, 40.0%-47.4%) for adalimumab; 40.2% (95% CrI, 33.5%-47.2%) for certolizumab pegol, 200 mg; 38.8% (95% CrI, 33.3%-44.7%) for tildrakizumab-asmn, 200 mg; 36.8% (31.4%-42.5%) for tildrakizumab-asmn, 100 mg; 17.9% (CrI, 14.9%-21.4%) for etanercept; 12.1% (95% CrI, 9.9%-14.7%) for apremilast; 11.4% (95% CrI, 7.5%-16.7%) for dimethyl fumarate; and 1.1% (95% CrI, 1.0%-1.3%) for placebo. In addition, the ORs of achieving the given PASI response (PASI 75, 90, and 100) for each intervention vs brodalumab, guselkumab, ixekizumab, and risankizumab-rzaa are illustrated in eFigure 1 in the Supplement.

Quiz Ref IDAmong all of the comparators evaluated, brodalumab, guselkumab, ixekizumab, and risankizumab-rzaa had the highest PASI 75, 90, and 100 rates at the end of the primary response period, and there were no statistically significant differences among these treatments. Brodalumab, guselkumab, ixekizumab, and risankizumab-rzaa had significantly higher PASI 75, 90, and 100 rates (with 95% probability) compared with adalimumab, apremilast, certolizumab pegol, dimethyl fumarate, etanercept, tildrakizumab-asmn, and ustekinumab. Brodalumab, ixekizumab, and risankizumab-rzaa also had significantly higher rates compared with infliximab and secukinumab across all PASI outcomes (with 95% probability).

The NNT to achieve PASI 75, 90, or 100 for each treatment relative to placebo is presented in Figure 3. To achieve a PASI 75 response, the NNT was 1.19 (95% CrI, 1.16-1.23) with risankizumab-rzaa, 1.20 (95% CrI, 1.17-1.23) with ixekizumab and brodalumab, and 1.23 (95% CrI, 1.19-1.27) with guselkumab. The NNT to achieve a PASI 90 response was 1.42 (95% CrI, 1.35-1.51) with risankizumab-rzaa, 1.43 (95% CrI, 1.36-1.52) with ixekizumab, 1.44 (95% CrI, 1.36-1.52) with brodalumab, and 1.51 (95% CrI, 1.41-1.63) with guselkumab. The NNT to achieve a PASI 100 response was 2.48 (95% CrI, 2.23-2.79) with risankizumab-rzaa, 2.54 (95% CrI, 2.28-2.85) with ixekizumab, 2.56 (95% CrI, 2.28-2.85) with brodalumab, and 2.81 (95% CrI, 2.46-3.25) with guselkumab.

Among the 60 trials included in the base-case, 10 are country-specific trials and were excluded from this sensitivity analysis. The median posterior probability of PASI 75, 90, and 100 rates for each intervention are presented in eTable 2 in the Supplement. Overall, the results from the sensitivity analysis including only global trials were consistent with the base-case analysis.

Among the 60 trials included in the base case, 47 were phase 3 trials that were included in the following sensitivity analysis. The median posterior probability of PASI 75, 90, and 100 rates for each intervention are presented in eTable 2 in the Supplement. The results from the sensitivity analysis based on phase 3 trials were consistent with those from the base-case analysis.

For the sensitivity analysis of the expanded treatment space, a total of 70 trials were included. The median posterior probability of PASI 75, 90, and 100 rates for each intervention are presented in eTable 2 in the Supplement. The results from the sensitivity analysis of the expanded treatment space were also consistent with those from the base-case analysis.

A total of 22 trials on 10 treatments meeting all inclusion criteria were included in the meta-analysis. The estimated response rates (PASI 75, 90, and 100) of each intervention are presented in Figure 4.^{13,15-18,20,21,34,41,42,47,50-52,73,74,84-86}Quiz Ref ID The estimated PASI 90 response rates at 44 to 60 weeks from baseline were 79.4% (95% CI, 75.5%-82.9%) for risankizumab-rzaa, 76.5% (95% CI, 72.1%-80.5%) for guselkumab, 74.0% (95% CI, 69.3%-78.1%) for brodalumab, 73.9% (95% CI, 69.9%-77.5%) for ixekizumab, 71.3% (95% CI, 64.2%-77.5%) for secukinumab, 52.4% (95% CI, 47.1%-57.7%) for ustekinumab, 46.2% (95% CI, 38.6%-53.9%) for adalimumab, 40.1% (95% CI, 30.0%-51.1%) for infliximab, 33.4% (95% CI, 28.5%-38.7%) for etanercept, and 16.0% (95% CI, 10.7%-23.3%) for apremilast. In addition, the ORs of achieving PASI 75, 90, and 100 for each intervention vs brodalumab, guselkumab, ixekizumab, and risankizumab-rzaa are presented in eFigure 2 in the Supplement. Quiz Ref IDAmong the comparators, the highest estimated PASI 75, 90, and 100 response rates at 44 to 60 weeks from baseline occurred following treatment with brodalumab, guselkumab, ixekizumab, risankizumab-rzaa, and secukinumab, which were significantly higher than those corresponding to adalimumab, apremilast, etanercept, infliximab, and ustekinumab. In addition, risankizumab-rzaa had significantly higher PASI 90 rates than ixekizumab and secukinumab; risankizumab-rzaa, brodalumab, and ixekizumab had significantly higher PASI 100 rates than secukinumab, and risankizumab-rzaa and ixekizumab also had significantly higher PASI 100 rates than guselkumab (eFigure 2 in the Supplement).

The estimated long-term response rates for PASI 75, 90, and 100 of each intervention in the sensitivity analyses are presented in eTable 3 in the Supplement. Among the 22 trials included in the base-case analysis, the sensitivity analysis including only global trials excluded 2 country-specific trials and produced results consistent with the base-case analysis. A total of 14 trials that reported NRI data at the end of the maintenance period (44-60 weeks from baseline) were included in the sensitivity analysis including only NRI data. The results of this sensitivity analysis were mostly consistent with the base-case analysis with the exception of secukinumab, which slightly increased because only 1 of the 5 trials that were included in the base case reported NRI values, and this trial had the highest PASI responses of the trials considered.¹³ Three phase 2 trials were excluded from the sensitivity analysis including only phase 3 trials; the results of this sensitivity analyses were consistent with the base-case analysis.

To test whether the method of meta-analysis would affect the results, we conducted an NMA using a fixed-effects bayesian multinomial likelihood model with a probit link for the long-term PASI outcomes using identified randomized clinical trials that maintained randomization through 52 weeks. Only 7 randomized clinical trials (AMAGINE 2,⁸⁷ AMAGINE 3,⁸⁸ CLEAR,¹³ FIXTURE,⁸⁹ IXORA-S,⁹⁰ UltIMMa1,¹⁵ and UltIMMa2¹⁵) on 6 treatments (brodalumab, etanercept, ixekizumab, risankizumab-rzaa, secukinumab, and ustekinumab) were able to be included into this NMA because of the crossover in the studies evaluating long-term PASI outcomes (eFigure 3 in the Supplement). In the NMA, risankizumab-rzaa (81.3%; 95% CrI, 75.7%-86.1%) had the highest estimated PASI 90 rates, followed by brodalumab (76.6%; 95% CrI, 71.3%-81.4%), ixekizumab (69.8%; 95% CrI, 60.1%-78.3%), secukinumab (65.2%; 95% CrI, 58.1%-71.6%), ustekinumab (52.5%; 95% CrI, 49.3%-55.7%), and etanercept (35.9%; 95% CrI, 27.1%-45.6%). As demonstrated by this NMA, risankizumab-rzaa had significantly higher PASI response rates compared with ixekizumab (PASI 90 OR, 1.89; 95% CrI, 1.12-1.99), secukinumab (OR, 2.33; 95% CrI, 1.54-3.56), ustekinumab (OR, 3.95; 95% CrI, 2.90-5.45), and etanercept (OR, 7.80; 95% CrI, 4.75-12.94), and brodalumab had significantly higher PASI response rates compared with secukinumab (OR, 1.76; 95% CrI, 1.22-2.54), ustekinumab (OR, 2.97; 95% CrI, 2.32-3.83), and etanercept (OR, 5.86; 95% CrI, 3.72-9.36) (eTable 4 and eTable 5 in the Supplement).

With increased therapeutic options for psoriasis, it has become increasingly challenging for clinicians to compare across the available treatments for clinical decision making. Meta-analytic approaches have important roles in integrating evidence from multiple studies to determine the relative efficacy of treatments and inform clinical decision making.⁹¹ This study provides a comprehensive assessment of the short-term and long-term comparative efficacy of biologics and oral therapies approved by the US Food and Drug Administration and the European Medicines Agency for the treatment of moderate to severe plaque psoriasis.

Results from this study suggest that brodalumab, guselkumab, ixekizumab, and risankizumab-rzaa had significantly higher PASI response rates than adalimumab, apremilast, certolizumab pegol, dimethyl fumarate, etanercept, tildrakizumab-asmn, and ustekinumab at the end of the primary response period. There were no statistically significant differences in short-term efficacy among brodalumab, guselkumab, ixekizumab, and risankizumab-rzaa. Over the long-term, brodalumab, guselkumab, ixekizumab, risankizumab-rzaa, and secukinumab had significantly higher PASI response rates than those corresponding to adalimumab, apremilast, infliximab, etanercept, and ustekinumab.

The short-term results from this study generally align with published NMAs that compared the efficacy of other biologic therapies for treatment of moderate to severe plaque psoriasis at the end of the primary response period.^28,29,92-98 For example, a recent NMA study by Sawyer et al⁹⁵ used similar methods and a similar list of trials to suggest that brodalumab, ixekizumab, risankizumab-rzaa, and guselkumab have the highest short-term PASI response rates.

The Sawyer et al⁹⁵ NMA also pointed out the lack of a comprehensive systematic evaluation of long-term efficacy of systemic treatments in patients with moderate to severe psoriasis. Work by Nast et al⁹⁹ found infliximab, secukinumab, and ustekinumab to be the most effective long-term treatment options. However, owing to clinical and methodologic differences among available studies and a lack of sufficiently long-term, head-to-head trials, the strength of the ranking of treatment efficacy was limited beyond 24 to 28 weeks.⁹⁹ A 2017 NMA of four 52-week randomized clinical trials found that brodalumab achieved significantly higher PASI 100 response rates compared with secukinumab, ustekinumab, and etanercept.¹⁰⁰ Nevertheless, only 4 trials were able to be included in the primary indirect comparison owing to study design heterogeneity and lack of long-term, active-comparator trial data. The present study overcame these limitations by using traditional meta-analysis methods.

Despite the differences in methods, the results from the long-term traditional meta-analysis are consistent with this NMA, suggesting the robustness of the meta-analysis results. However, the meta-analysis directly synthesized the PASI response rates for patients receiving each active treatment during both the induction and maintenance periods, making it possible to include additional treatments beyond the connected network.

The present work analyzes the short-term and long-term efficacy that included all licensed medications to date (eg, brodalumab, dimethyl fumarate, guselkumab, risankizumab-rzaa, and tildrakizumab-asmn). This study also has several other strengths. First, the analytic methods used in the study are well-established, rigorous, treatment comparison techniques. Meta-analyses are a statistical tool used to synthesize direct evidence from multiple studies to compare 2 interventions at a time. Network meta-analyses extend traditional meta-analyses to allow for simultaneous comparison of several different treatments that are regularly used in clinical practice but have not necessarily been compared head-to-head in a randomized trial. These methods provide useful evidence about the comparative effectiveness of competing interventions—which would otherwise be lacking—in a cost-effective and timely manner. The NMA approach used in the analysis of short-term efficacy enabled an estimation of comparative effects among treatments that have not yet been investigated in head-to-head randomized clinical trials for moderate to severe plaque psoriasis. Adjusting for the reference arm response can reduce the outcomes associated with cross-trial heterogeneity, as the reference arm response rate integrates the results of other observed and unobserved trial-level factors likely to affect treatment arm outcomes. The traditional meta-analysis used to assess the long-term outcomes of treatments allowed for the amalgamation of scientific evidence and offered an objective appraisal of the available evidence. By combining data from multiple independent trials of the same treatment, the meta-analysis increased the sample size and improved statistical precision for the estimation of long-term PASI responses. In addition, the results from the sensitivity analyses were consistent with the base-case analyses for both short-term and long-term treatment, which supports the robustness of the primary findings. The comparative analyses of recently available biologic treatments in the present study fills an important gap in the current body of literature. Additional analyses comparing the short-term and long-term safety outcomes are warranted to provide a complete benefit-risk assessment of novel psoriasis therapies for health care stakeholders.^101,102

The results of this analysis should be interpreted within the context of certain limitations. First, as with all meta-analyses, the presence of cross-trial differences and patient characteristics that may modify the treatment effect can introduce bias in the comparisons. Although adjusting for the reference arm response in this NMA can reduce the effects of cross-trial heterogeneity, there is no guarantee that adjustments will eliminate these effects. In addition, subgroup analyses (eg, by relevant patient characteristics) are generally not feasible because most published trials do not report detailed PASI outcomes for subgroups. Second, the findings of this study are based on clinical trial data, which may not be generalizable to real-world settings. Future research may study the comparative efficacy among treatments in real-world settings. Moreover, a holistic comparison of psoriasis treatments should include other key aspects, such as safety profile, effect on quality of life (eg, the Dermatology Life Quality Index), as well as administration and dosing frequency. Third, the short-term efficacy analysis only included efficacy data up to week 16, and any agent’s full efficacy may not have been achieved at this time point; however, the long-term analyses serve as a complement to this short-term assessment. Fourth, the long-term meta-analysis focused on conducting random-effects meta-analysis within treatments because an anchor-based approach (eg, NMA) would be limited for all treatments owing to crossover and rerandomization. Accordingly, the current meta-analysis using the DerSimonian and Laird²⁷ method can only account for heterogeneity among trials of the same treatment but not for different treatments. Hence, the long-term meta-analysis results should be interpreted with caution beyond the relationships examined in the long-term NMA.

In the absence of head-to-head randomized clinical trials of treatments for moderate to severe plaque psoriasis, this study provides what we believe to be a comprehensive assessment of the comparative short-term and long-term efficacy among several novel treatments. Over the primary response period and long-term maintenance period, brodalumab, guselkumab, ixekizumab, and risankizumab-rzaa were associated with the highest estimated PASI 75, 90, and 100 response rates.

Accepted for Publication: October 25, 2019.

Corresponding Author: April W. Armstrong, MD, MPH, Department of Dermatology, Keck School of Medicine, University of Southern California, HC4 2000, 1450 San Pablo, Health Sciences Campus, Los Angeles, CA 90033 (aprilarmstrong@post.harvard.edu).

Published Online: February 5, 2020. doi:10.1001/jamadermatol.2019.4029

Author Contributions: Drs Li and Betts had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Armstrong, Joshi, Skup, Li, Betts, Augustin.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Armstrong, Joshi, Skup, Williams, Li, Betts.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Armstrong, Puig, Joshi, Li, Betts.

Administrative, technical, or material support: Joshi, Williams.

Supervision: Armstrong, Puig, Joshi, Skup, Williams, Betts, Augustin.

Conflict of Interest Disclosures: Dr Armstrong reported receiving grants and personal fees from AbbVie, Eli Lilly, Leo Pharma, and Novartis Pharmaceuticals Corp; personal fees from Boehringer Ingelheim/Parexel, Bristol-Myers Squibb, Celgene, Dermavant, Janssen Pharmaceuticals Inc, Merck, Modernizing Medicine, Ortho Dermatologics, Pfizer Inc, Regeneron Pharmaceuticals, Sanofi Genzyme, Science 37 Inc, Genentech, GlaxoSmithKline, and Valeant; and grants from Dermira, Janssen-Ortho Inc, Kyowa Hakko Kirin, and UCB Pharma outside the submitted work. Dr Puig reported receiving grants and personal fees from AbbVie during the conduct of the study; grants and personal fees from Almirall, Amgen, Boehringer Ingelheim, Celgene, Janssen, Leo-Pharma, Lilly, Novartis, Pfizer, Regeneron, Sanofi, Roche, and UCB outside the submitted work; and personal fees from Sandoz, Merck-Serono, MSD, Mylan, and Samsung-Bioepis. Dr Joshi is an employee AbbVie during the conduct of the study. Dr Skup is an employee AbbVie. Dr Williams is an of AbbVie. Dr Li reported receiving grants from AbbVie during the conduct of the study outside the submitted work. Dr Betts reported receiving consultancy fees from AbbVie Inc. during the conduct of the study. No other disclosures were reported.

Funding/Support: Design, study conduct, and financial support for the study were provided by AbbVie.

Role of the Funder/Sponsor: AbbVie participated in the interpretation of data, review, and approval of the abstract; all authors contributed to the development of the publication and maintained control over the final content.

Additional Contributions: Yan Wang, ScD, and Viviana Garcia-Horton, PhD, provided analytical support and Gloria DeWalt, PhD (Analysis Group Inc), provided editorial assistance. Analysis Group Inc received payment from AbbVie for participation in this research. Palvi Gupta, MPharm, Jatin Gupta, MPharm, and Rashi Patel, MPharm (DRG Abacus) provided support with literature search and study selection.