Systemic pharmacological treatments for chronic plaque psoriasis: a network meta‐analysis
Abstract
Background
Psoriasis is an immune‐mediated disease for which some people have a genetic predisposition. The condition manifests in inflammatory effects on either the skin or joints, or both, and it has a major impact on quality of life. Although there is currently no cure for psoriasis, various treatment strategies allow sustained control of disease signs and symptoms. Several randomised controlled trials (RCTs) have compared the efficacy of the different systemic treatments in psoriasis against placebo. However, the relative benefit of these treatments remains unclear due to the limited number of trials comparing them directly head to head, which is why we chose to conduct a network meta‐analysis.
Objectives
To compare the efficacy and safety of conventional systemic agents (acitretin, ciclosporin, fumaric acid esters, methotrexate), small molecules (apremilast, tofacitinib, ponesimod), anti‐TNF alpha (etanercept, infliximab, adalimumab, certolizumab), anti‐IL12/23 (ustekinumab), anti‐IL17 (secukinumab, ixekizumab, brodalumab), anti‐IL23 (guselkumab, tildrakizumab), and other biologics (alefacept, itolizumab) for patients with moderate to severe psoriasis and to provide a ranking of these treatments according to their efficacy and safety.
Search methods
We searched the following databases to December 2016: the Cochrane Skin Specialised Register, the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, Embase, and LILACS. We also searched five trials registers and the U.S. Food and Drug Administration (FDA) and European Medicines Agency (EMA) reports. We checked the reference lists of included and excluded studies for further references to relevant RCTs. We searched the trial results databases of a number of pharmaceutical companies and handsearched the conference proceedings of a number of dermatology meetings.
Selection criteria
Randomised controlled trials (RCTs) of systemic and biological treatments in adults (over 18 years of age) with moderate to severe plaque psoriasis or psoriatic arthritis whose skin had been clinically diagnosed with moderate to severe psoriasis, at any stage of treatment, in comparison to placebo or another active agent.
Data collection and analysis
Three groups of two review authors independently undertook study selection, data extraction, 'Risk of bias' assessment, and analyses. We synthesised the data using pair‐wise and network meta‐analysis (NMA) to compare the treatments of interest and rank them according to their effectiveness (as measured by the Psoriasis Area and Severity Index score (PASI) 90) and acceptability (the inverse of serious adverse effects). We assessed the certainty of the body of evidence from the NMA for the two primary outcomes, according to GRADE; we evaluated evidence as either very low, low, moderate, or high. We contacted study authors when data were unclear or missing.
Main results
We included 109 studies in our review (39,882 randomised participants, 68% men, all recruited from a hospital). The overall average age was 44 years; the overall mean PASI score at baseline was 20 (range: 9.5 to 39). Most of these studies were placebo controlled (67%), 23% were head‐to‐head studies, and 10% were multi‐armed studies with both an active comparator and placebo. We have assessed all treatments listed in the objectives (19 in total). In all, 86 trials were multicentric trials (two to 231 centres). All of the trials included in this review were limited to the induction phase (assessment at less than 24 weeks after randomisation); in fact, all trials included in the network meta‐analysis were measured between 12 and 16 weeks after randomisation. We assessed the majority of studies (48/109) as being at high risk of bias; 38 were assessed as at an unclear risk, and 23, low risk.
Network meta‐analysis at class level showed that all of the interventions (conventional systemic agents, small molecules, and biological treatments) were significantly more effective than placebo in terms of reaching PASI 90.
In terms of reaching PASI 90, the biologic treatments anti‐IL17, anti‐IL12/23, anti‐IL23, and anti‐TNF alpha were significantly more effective than the small molecules and the conventional systemic agents. Small molecules were associated with a higher chance of reaching PASI 90 compared to conventional systemic agents.
At drug level, in terms of reaching PASI 90, all of the anti‐IL17 agents and guselkumab (an anti‐IL23 drug) were significantly more effective than the anti‐TNF alpha agents infliximab, adalimumab, and etanercept, but not certolizumab. Ustekinumab was superior to etanercept. No clear difference was shown between infliximab, adalimumab, and etanercept. Only one trial assessed the efficacy of infliximab in this network; thus, these results have to be interpreted with caution. Tofacitinib was significantly superior to methotrexate, and no clear difference was shown between any of the other small molecules versus conventional treatments.
Network meta‐analysis also showed that ixekizumab, secukinumab, brodalumab, guselkumab, certolizumab, and ustekinumab outperformed other drugs when compared to placebo in terms of reaching PASI 90: the most effective drug was ixekizumab (risk ratio (RR) 32.45, 95% confidence interval (CI) 23.61 to 44.60; Surface Under the Cumulative Ranking (SUCRA) = 94.3; high‐certainty evidence), followed by secukinumab (RR 26.55, 95% CI 20.32 to 34.69; SUCRA = 86.5; high‐certainty evidence), brodalumab (RR 25.45, 95% CI 18.74 to 34.57; SUCRA = 84.3; moderate‐certainty evidence), guselkumab (RR 21.03, 95% CI 14.56 to 30.38; SUCRA = 77; moderate‐certainty evidence), certolizumab (RR 24.58, 95% CI 3.46 to 174.73; SUCRA = 75.7; moderate‐certainty evidence), and ustekinumab (RR 19.91, 95% CI 15.11 to 26.23; SUCRA = 72.6; high‐certainty evidence).
We found no significant difference between all of the interventions and the placebo regarding the risk of serious adverse effects (SAEs): the relative ranking strongly suggested that methotrexate was associated with the best safety profile regarding all of the SAEs (RR 0.23, 95% CI 0.05 to 0.99; SUCRA = 90.7; moderate‐certainty evidence), followed by ciclosporin (RR 0.23, 95% CI 0.01 to 5.10; SUCRA = 78.2; very low‐certainty evidence), certolizumab (RR 0.49, 95% CI 0.10 to 2.36; SUCRA = 70.9; moderate‐certainty evidence), infliximab (RR 0.56, 95% CI 0.10 to 3.00; SUCRA = 64.4; very low‐certainty evidence), alefacept (RR 0.72, 95% CI 0.34 to 1.55; SUCRA = 62.6; low‐certainty evidence), and fumaric acid esters (RR 0.77, 95% CI 0.30 to 1.99; SUCRA = 57.7; very low‐certainty evidence). Major adverse cardiac events, serious infections, or malignancies were reported in both the placebo and intervention groups. Nevertheless, the SAEs analyses were based on a very low number of events with low to very low certainty for just over half of the treatment estimates in total, moderate for the others. Thus, the results have to be considered with caution.
Considering both efficacy (PASI 90 outcome) and acceptability (SAEs outcome), highly effective treatments also had more SAEs compared to the other treatments, and ustekinumab, infliximab, and certolizumab appeared to have the better trade‐off between efficacy and acceptability.
Regarding the other efficacy outcomes, PASI 75 and Physician Global Assessment (PGA) 0/1, the results were very similar to the results for PASI 90.
Information on quality of life was often poorly reported and was absent for a third of the interventions.
Authors' conclusions
Our review shows that compared to placebo, the biologics ixekizumab, secukinumab, brodalumab, guselkumab, certolizumab, and ustekinumab are the best choices for achieving PASI 90 in people with moderate to severe psoriasis on the basis of moderate‐ to high‐certainty evidence. At class level, the biologic treatments anti‐IL17, anti‐IL12/23, anti‐IL23, and anti‐TNF alpha were significantly more effective than the small molecules and the conventional systemic agents, too. This NMA evidence is limited to induction therapy (outcomes were measured between 12 to 16 weeks after randomisation) and is not sufficiently relevant for a chronic disease. Moreover, low numbers of studies were found for some of the interventions, and the young age (mean age of 44 years) and high level of disease severity (PASI 20 at baseline) may not be typical of patients seen in daily clinical practice.
Another major concern is that short‐term trials provide scanty and sometimes poorly reported safety data and thus do not provide useful evidence to create a reliable risk profile of treatments. Indeed, we found no significant difference in the assessed interventions and placebo in terms of SAEs. Methotrexate appeared to have the best safety profile, but as the evidence was of very low to moderate quality, we cannot be sure of the ranking. In order to provide long‐term information on the safety of the treatments included in this review, it will be necessary to evaluate non‐randomised studies and postmarketing reports released from regulatory agencies as well.
In terms of future research, randomised trials comparing directly active agents are necessary once high‐quality evidence of benefit against placebo is established, including head‐to‐head trials amongst and between conventional systemic and small molecules, and between biological agents (anti‐IL17 versus anti‐IL23, anti‐IL23 versus anti‐IL12/23, anti‐TNF alpha versus anti‐IL12/23). Future trials should also undertake systematic subgroup analyses (e.g. assessing biological‐naïve patients, baseline psoriasis severity, presence of psoriatic arthritis, etc.). Finally, outcome measure harmonisation is needed in psoriasis trials, and researchers should look at the medium‐ and long‐term benefit and safety of the interventions and the comparative safety of different agents.
PICOs
Plain language summary
Systemic (oral or injected) medicines for psoriasis
What is the aim of this review?
The aim of this review was to compare different systemic medicines (oral or injected medicines that work throughout the entire body) used to treat chronic plaque psoriasis in adults (over 18 years of age), to find out which are the safest and most effective at clearing psoriasis. We wanted to rank the medicines in order of their safety and how well they work, to help the development of a treatment pathway for people with chronic plaque psoriasis. We collected and analysed all relevant studies to answer this question and found 109 studies.
Key messages
The results showed that a selection of treatments from the class of biological medicines appear to be the most effective systemic medicines for achieving a chronic plaque psoriasis score of PASI (Psoriasis Area and Severity Index) 90, which translates into a 90% improvement in psoriasis from the beginning of the study. We found no significant difference in serious adverse effects (SAEs) (i.e. serious side effects) when comparing any of the assessed treatments with placebo. However, as the evidence was of very low to moderate quality, we cannot be sure of these results.
For some of the interventions, we found low numbers of studies, so more research needs to be conducted to directly compare the systemic medicines with each other, rather than comparing them with placebo (an inactive substance) (once effect against placebo has been established by high‐quality evidence). In addition, longer‐term studies are needed to provide more evidence about the benefit and safety of systemic medicines and to compare their safety profiles. Indeed, the results of this review are limited to the induction treatment (i.e. outcomes were measured up to 24 weeks after participants were allocated to their treatment group), which is not an appropriate treatment option for a chronic disease.
We rated the certainty of the evidence as ranging from very low (mainly conventional medicines) to high (mainly biological medicines). We downgraded the certainty of the evidence due to risk of bias (concerns with the study methods) and then for either inconsistent results or imprecision (inaccuracy).
What was studied in the review?
Psoriasis is characterised by patches of red, flaky skin covered with scales (known as plaques) or other inflammatory effects that are seen on the skin or joints, or both. Psoriasis is caused by an abnormal response within the immune system in people who may have a genetic predisposition towards the condition.
Approximately 2% of the population have psoriasis, and 90% of those people have plaque psoriasis. Around 10% to 20% of people with chronic plaque psoriasis will need to have systemic treatments. Psoriasis impacts on quality of life, including a person's psychosocial life.
We compared 19 systemic medicines by identifying studies that compared one or more of these medicines with either placebo or with another medicine to treat moderate to severe forms of plaque psoriasis in adults who were at any stage of treatment. The medicines we assessed were conventional systemic treatments (a varied group of treatments that are the oldest treatments given to clear psoriasis), biologics (treatments that use substances made from living organisms, or synthetic versions, to target the immune system), and small molecules (which affect molecules inside immune cells). We included studies whose participants may also have had psoriatic arthritis. The main outcomes we were interested in were achievement of PASI 90 and any serious side effects that were thought to be associated with the medicines.
We combined all of the studies to allow indirect analysis of the treatments, so we could compare them with each other (network meta‐analysis).
What are the main results of the review?
The 109 studies enrolled 39,882 people (all recruited from a hospital) with moderate to severe psoriasis: 26,902 men and 12,384 women; the overall average age was 44 years, the overall mean PASI score at the start of the study was 20 (range: 9.5 to 39), indicating a high level of disease severity. Most studies (n = 73) compared the systemic medicine with a placebo treatment, a total of 25 trials compared systemic treatments with other systemic treatments, and 11 trials compared systemic treatments with systemic treatments and placebo. Most studies were short‐term, and in all, 86 trials were multicentric trials (two to 231 centres).
The outcomes presented here were measured 12 to 16 weeks after the study participants were randomised.
The results showed that compared with placebo, all treatments (assessed in the following groupings: anti‐IL17, anti‐IL12/23, anti‐IL23, and anti‐TNF alpha (i.e. the treatments known as the biologics); small molecule treatments; other biologics; and conventional systemic agents) were more effective in treating psoriasis when assessed using an index that required 90% improvement (PASI 90).
In relation to the same outcome (PASI 90), the biologic treatments anti‐IL17, anti‐IL12/23, anti‐IL23, and anti‐TNF alpha appeared to work better than the small molecules and the conventional systemic agents; and small molecules were associated with a better outcome compared to conventional systemic agents. (IL is an abbreviation of interleukin; TNF is an abbreviation of tumour necrosis factor ‐ both are types of cytokine. A cytokine affects the behaviour of a cell.)
In terms of individual drugs, again when assessing the ability to reach PASI 90, all of the anti‐IL17 drugs and guselkumab (an anti‐IL23 drug) were more effective than the anti‐TNF alpha drugs infliximab, adalimumab, and etanercept, but not certolizumab. Ustekinumab (an IL‐12/‐23 drug) was better than etanercept. No clear difference was shown between infliximab, adalimumab, and etanercept. Tofacitinib (a small molecule) was superior to methotrexate (a conventional systemic agent), and no difference was shown between the other small molecules and the conventional drugs.
Judged against placebo, six biological medicines worked best at clearing psoriasis lesions. These medicines were ranked as follows (most effective first): ixekizumab, secukinumab (both based on high‐certainty evidence), brodalumab, guselkumab, certolizumab (all based on moderate‐certainty evidence), and ustekinumab (high‐certainty evidence). Regarding the outcomes PASI 75 and Physician Global Assessment (PGA) 0/1 (i.e. achieving 75% improvement and achieving a PGA score of 0 or 1), the results were very similar to the results for PASI 90.
For the risk of serious side effects, there were no clear differences between all of the systemic medicines compared with placebo treatment. Methotrexate had the best safety profile (based on moderate‐certainty evidence), followed by ciclosporin (very low‐certainty evidence), certolizumab (moderate‐certainty evidence), infliximab (very low‐certainty evidence), alefacept (low‐certainty evidence), and fumaric acid esters (very low‐certainty evidence) (all of these are conventional treatments except for certolizumab, infliximab (anti‐TNF alpha drugs), and alefacept (classed under 'other biologics'). Major adverse cardiac events, serious infections, or malignancies were reported in both placebo and intervention groups. However, the number of serious side effects was very low, and our conclusions are based on low to very low‐ (for just over half of the results) or moderate‐certainty evidence, so they should be interpreted with caution. The most effective treatments (in terms of reaching PASI 90) had the highest numbers of reported side effects; ustekinumab, infliximab, and certolizumab appeared to have the best compromise between effectiveness and side effects.
For all studies, little information was recorded about quality of life; one third of the medicines studied had no quality of life data.
How up‐to‐date is this review?
We searched for studies that had been published up to December 2016.
Authors' conclusions
Summary of findings
| Any systemic treatment compared to placebo for chronic plaque psoriasis (network meta‐analysis) | |||||||
| Patient or population: people with chronic plaque psoriasis Setting: all the participants were recruited from a hospital setting Timescale: 12 to 16 weeks after randomisation | |||||||
| Intervention | Anticipated absolute effects* (95% CI) | Relative effect | SUCRA | № of participants | Certainty of the evidence | Comments | |
| Risk with placeboa | Risk with any systemic treatment | ||||||
| PASI 90 | |||||||
| Ixekizumab | Moderate | RR 32.45 | 94.3 | 3268 | ⊕⊕⊕⊕ | ‐ | |
| 15 per 1000 | 487 per 1000 | ||||||
| Secukinumab | Moderate | RR 26.55 | 86.5 | 2707 | ⊕⊕⊕⊕ | ‐ | |
| 15 per 1000 | 398 per 1000 | ||||||
| Brodalumab | Moderate | RR 25.45 | 84.3 | 4109 | ⊕⊕⊕⊝ | Reasons for downgrading by one level: three studies contributing to this estimate at high risk of bias in selective reporting domain | |
| 15 per 1000 | 382 per 1000 | ||||||
| Guselkumab | Moderate | RR 21.03 | 77 | 1502 | ⊕⊕⊕⊝ | Reasons for downgrading by one level: one study contributing to this estimate at high risk of bias in selective reporting domain | |
| 15 per 1000 | 315 per 1000 | ||||||
| Certolizumab | Moderate | RR 24.58 | 75.7 | 176 | ⊕⊕⊕⊝ | Downgraded one level due to imprecision: wide CIs | |
| 15 per 1000 | 369 per 1000 | ||||||
| Ustekinumab | Moderate | RR 19.91 | 72.6 | 3832 | ⊕⊕⊕⊕ | ‐ | |
| 15 per 1000 | 299 per 1000 | ||||||
| Tildrakizumab | Moderate | RR 15.63 | 63.6 | 355 | ⊕⊕⊝⊝ | Downgraded one level due to risk of bias and one level due to imprecision. The single study contributing to this estimate at unclear risk of bias in both blinding domains; wide CIs | |
| 15 per 1000 | 234 per 1000 | ||||||
| Adalimumab | Moderate | RR 14.87 | 63.1 | 3199 | ⊕⊕⊕⊝ | Downgraded one level due to inconsistency ‐ inconsistent loops of evidence | |
| 15 per 1000 | 223 per 1000 | ||||||
| Itolizumab | Moderate | RR 12.26 | 56 | 225 | ⊕⊕⊝⊝ | Downgraded one level due to imprecision (wide CIs) and one level due to risk of bias (moderate risk using credibility of evidence) | |
| 15 per 1000 | 184 per 1000 | ||||||
| Infliximab | Moderate | RR 11.18 | 53.2 | (0 RCTs) | ⊕⊝⊝⊝ | Downgraded one level due to risk of bias (credibility of risk), one level due to imprecision (wide CIs) and one level due to inconsistency (inconsistent loop of evidence) | |
| 15 per 1000 | 168 per 1000 | ||||||
| Etanercept | Moderate | RR 10.79 | 52.6 | 4954 | ⊕⊕⊕⊝ | Downgraded one level due to inconsistency (global inconsistency ‐ side‐splitting approach) | |
| 15 per 1000 | 162 per 1000 | ||||||
| Tofacitinib | Moderate | RR 8.50 | 42.5 | 2826 | ⊕⊕⊝⊝ | Downgraded one level due to risk of bias: two studies at high risk of bias in incomplete outcome data domain; and downgraded one level due to inconsistency (global approach) | |
| 15 per 1000 | 128 per 1 000 | ||||||
| Apremilast | Moderate | RR 7.66 | 39.7 | 1775 | ⊕⊕⊕⊝ | Downgraded one level due to risk of bias: one study had a slight risk of bias in selective reporting domain | |
| 15 per 1000 | 115 per 1000 | ||||||
| Ponesimod | Moderate | RR 6.60 | 37.3 | 326 | ⊕⊕⊕⊝ | Downgraded one level due to imprecision: wide CIs | |
| 15 per 1000 | 99 per 1000 | ||||||
| Alefacept | Moderate | RR 4.39 | 25.3 | (0 RCTs) | ⊕⊝⊝⊝ | Downgraded two levels due to risk of bias and a further one level due to imprecision ‐ study indirectly contributing to the estimates at high risk of bias in selective reporting domain; wide CIs | |
| 15 per 1000 | 66 per 1000 | ||||||
| Fumaric acid esters (FAEs) | Moderate | RR 4.09 | 21.9 | 704 | ⊕⊝⊝⊝ | Downgraded two levels due to risk of bias, and one level due to imprecision ‐ the studies indirectly contributing to this estimate at high risk of bias in blinding domain; wide CIs | |
| 15 per 1000 | 61 per 1000 | ||||||
| Ciclosporin | Moderate | RR 3.99 | 21.3 | (0 RCTs) | ⊕⊝⊝⊝ | Downgraded two levels due to risk of bias, and a further one level due to imprecision ‐ the single study indirectly contributing to this estimate at high risk of bias in blinding; wide CIs | |
| 15 per 1000 | 60 per 1000 | ||||||
| Methotrexate | Moderate | RR 3.61 | 20.2 | 282 | ⊕⊕⊕⊝ | Downgraded one level due to inconsistency (inconsistent loop of evidence) | |
| 15 per 1000 | 59 per 1000 | ||||||
| Acitretin | Moderate | RR 0.98 | 9.9 | (0 RCTs) | ⊕⊝⊝⊝ | Downgraded two levels due to risk of bias and a further one level due to imprecision. The single study contributing to this estimate at high risk of bias in incomplete outcome data and blinding domains; wide CIs | |
| 15 per 1000 | 15 per 1000 | ||||||
| Serious adverse events | |||||||
| Methotrexate | Moderate | RR 0.23 | 90.7 | 282 | ⊕⊕⊕⊝ | Downgraded one level due to imprecision (wide CIs) | |
| 17 per 1000 | 4 per 1000 | ||||||
| Ciclosporin | Moderate | RR 0.23 | 78.2 | (0 RCTs) | ⊕⊝⊝⊝ | Downgraded two levels due to risk of bias (credibility of evidence), and one level due to imprecision (wide CIs) | |
| 17 per 1000 | 4 per 1000 | ||||||
| Certolizumab | Moderate | RR 0.49 | 70.9 | 176 | ⊕⊕⊕⊝ | Downgraded one level due to imprecision (wide CIs) | |
| 17 per 1000 | 8 per 1000 | ||||||
| Infliximab | Moderate | RR 0.56 | 64.4 | (0 RCTs) | ⊕⊝⊝⊝ | Downgraded two levels due to risk of bias, and one level due to imprecision: credibility of evidence; wide CIs | |
| 17 per 1000 | 10 per 1000 | ||||||
| Alefacept | Moderate | RR 0.72 | 62.6 | 736 | ⊕⊕⊝⊝ | Downgraded one level due to risk of bias (credibility of evidence), and one level due to imprecision (wide CIs) | |
| 17 per 1000 | 12 per 1000 | ||||||
| Fumaric acid esters (FAEs) | Moderate | RR 0.77 | 57.7 | 704 | ⊕⊝⊝⊝ | Downgraded by one level due to risk of bias and one level due to imprecision: credibility of evidence; wide CIs | |
| 17 per 1000 | 13 per 1000 | ||||||
| Apremilast | Moderate | RR 0.84 | 54.7 | 2036 | ⊕⊕⊝⊝ | Downgraded one level due to risk of bias and one level due to imprecision: credibility of evidence and wide CIs | |
| 17 per 1000 | 14 per 1000 | ||||||
| Ustekinumab | Moderate | RR 0.89 | 52 | 4154 | ⊕⊕⊝⊝ | Downgraded one level due to risk of bias and one level due to imprecision ‐ credibility of evidence; wide CIs | |
| 17 per 1000 | 15 per 1000 | ||||||
| Acitretin | Moderate | RR 0.99 | 46.9 | (0 RCTs) | ⊕⊝⊝⊝ | Downgraded by two levels due to risk of bias and one level due to imprecision: credibility of evidence; wide CIs | |
| 17 per 1000 | 17 per 1000 | ||||||
| Tofacitinib | Moderate | RR 0.98 | 44 | 2838 | ⊕⊕⊕⊝ | Downgraded one level due to imprecision (wide CIs) | |
| 17 per 1000 | 17 per 1000 | ||||||
| Etanercept | Moderate | RR 0.99 | 43.6 | 3783 | ⊕⊕⊕⊝ | Downgraded one level due to imprecision (CIs including one) | |
| 17 per 1000 | 17 per 1000 | ||||||
| Guselkumab | Moderate | RR 1.00 | 42.6 | 1502 | ⊕⊕⊝⊝ | Downgraded one level due to risk of bias (credibility of evidence), and one level due to imprecision (CIs including one) | |
| 15 per 1000 | 15 per 1000 | ||||||
| Adalimumab | Moderate | RR 1.02 | 40.4 | 3199 | ⊕⊕⊕⊝ | Downgraded one level due to imprecision (CIs including one) | |
| 18 per 1000 | 19 per 1000 | ||||||
| Brodalumab | Moderate | RR 1.04 | 39.8 | 4109 | ⊕⊕⊝⊝ | Downgraded one level due to risk of bias (credibility of evidence) and one level due to imprecision (CIs including 1) | |
| 17 per 1000 | 18 per 1000 | ||||||
| Tildrakizumab | Moderate | RR 1.36 | 37.8 | 355 | ⊕⊕⊝⊝ | Downgraded one level due to risk of bias (credibility of evidence) and one level due to imprecision (CIs including 1) | |
| 0 per 1000 | 0 per 1000 | ||||||
| Ixekizumab | Moderate | RR 1.12 | 33.7 | 3268 | ⊕⊕⊕⊝ | Downgraded one level due to imprecision (CIs including one) | |
| 15 per 1000 | 16 per 1000 | ||||||
| Secukinumab | Moderate | RR 1.19 | 29.9 | 2707 | ⊕⊕⊕⊝ | Downgraded one level due to imprecision (CIs including one) | |
| 10 per 1000 | 12 per 1000 | ||||||
| Ponesimod | Moderate | RR 2.59 | 18.1 | 326 | ⊕⊕⊕⊝ | Downgraded one level due to imprecision (CIs including one) | |
| 15 per 1000 | 39 per 1000 | ||||||
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | |||||||
| GRADE Working Group grades of evidence | |||||||
| a 'Risk with placebo' is the median placebo‐group risk value in the included studies for the assumed risk with placebo. c The Psoriasis Area and Severity Index combines the assessment of the severity of lesions and the area affected into a single score in the range of 0 (no disease) to 72 (maximal disease); PASI 90: 90% improvement in the PASI. d SUCRA was expressed as a percentage between 0 (when a treatment is certain to be the worst) to 100% (when a treatment is certain to be the best). | |||||||
Background
Please refer to our glossary (see Table 1).
| Term | Definition |
| Antagonist | A substance that interferes with or inhibits the physiological action of another. |
| Antigen | A molecule capable of inducing an immune respons |
| Anti‐TNF alpha | A pharmaceutical drug that suppresses the physiologic response to tumor necorsis factor (TNF) |
| Biological agent | Therapeutic agents consisting of immune molecules such as soluble receptors, recombinant cytokines, and monoclonal antibodies that target effector molecules or cells of the immune system |
| CD6 | Cluster of differentiation (CD) 6 is a protein encoded by the CD6 gene |
| Cheilitis | An inflammation of the lips |
| Chimeric protein | A chimeric protein can be made by combining two different genes |
| Complex cyclophilin‐ciclosporin | Cyclophilins are a family of proteins that bind to ciclosporin, an immunosuppressant agent |
| Creatinine | A compound that is produced by metabolism of creatine and excreted in the urine |
| Cyclic adenosine monophosphate | It is a second messenger important in many biological processes |
| Cytokines | Small proteins produced by a broad range of cells that are important in cell signaling; they are immunomodulating agents |
| Dendritic cells | Antigen‐presenting cells of the immune system |
| Dermis | It is a layer of the skin |
| Epitope | It is a part of an antigen |
| Erythematous | Redness of the skin |
| Folic acid | B vitamin |
| Humanised antibody | Antibodies from non‐human species whose protein sequences have been modified to increase their similarity to antibody variants produced naturally in humans |
| IL‐17A | A pro‐inflammatory cytokine |
| IL‐23R | A cytokine receptor |
| Immune‐mediated | A group of diseases that are characterised by common inflammatory pathways leading to inflammation, and which may result from a dysregulation of the normal immune response |
| Immunogenicity | This is the ability of a particular substance, such as an antigen or epitope, to provoke an immune response in the body of a human or animal |
| Immunoglobulin 1 Fc | An antibody |
| Interferon (IFN)‐c | A protein released by cells, usually in response to a pathogen |
| Interleukin | A kind of cytokine |
| Janus kinase (JAK) inhibitors | A pharmaceutical drug that inhibits the activity of one or more of the Janus kinase family of enzymes |
| Keratinocytes | Epidermal cells that constitute 95% of the epidermis |
| Lymphocyte | A subtype of a white blood cell |
| Lymphoid organ | Part of the body that defends the body against invading pathogens that cause infections or the spread of tumours |
| Metalloproteinases | A protease enzyme |
| Monoclonal antibodies | Antibodies that are made by identical immune cells that are all clones of a unique parent cell |
| Murine sequence | Mouse genomic sequencing |
| Neutrophils | Type of white blood cell involved in the innate immune system |
| p40 | Subunit beta of interleukin 12 and 23 |
| Periumbilical | Around the navel |
| Pharmacological treatments | Drugs |
| Phase I | First‐in‐man studies |
| Phase II | Studies to assess how well the drug works, as well as to continue phase I safety assessments in a larger group of volunteers and participants |
| Phase III | Randomised controlled multicenter trials on large patient groups and are aimed at being the definitive assessment of how effective the drug is |
| Phase IV | Post‐marketing trials involve the safety surveillance |
| Phosphodiesterase 4 inhibitors | A pharmaceutical drug used to block the degradative action of phosphodiesterase 4 |
| Progressive multifocal leukoencephalopathy | A rare viral neurological disease characterised by progressive damage of the white matter of the brain at multiple locations |
| Receptor | A protein molecule that receives chemical signals from outside a cell |
| Small molecules | Chemically manufactured molecules (or SMOLs for short) |
| Sphingosine 1‐phosphate receptor agonists | A class of protein‐coupled receptors that are targets of the lipid signalling molecule Sphingosine‐1‐phosphate |
| T cells/CD4 T cells | A type of white blood cell that is of key importance to the immune system |
| Th1 and Tc1 cells | A type of T cell |
| Th17 and Tc17 cells | A type of T cell |
| TNF‐alpha | A protein that is part of the inflammatory response |
| Tumour necrosis factor antagonists | Class of biological agents |
| Umbilic | Navel |
| Xerosis | Dry skin |
Description of the condition
Psoriasis is an immune‐mediated disease for which a person can have genetic susceptibility, manifesting in chronic inflammatory effects on either the skin or joints, or both, with a prevalence ranging from 0.91% (United States) to 8.5% (Norway) (Boehncke 2015; Parisi 2013). The causes of psoriasis are not fully understood. There appears to be interaction between environmental factors and genetic susceptibility. Genome‐wide (or whole genome) association trials found several candidate genes relating to psoriasis (Elder 2010). Various environmental factors, including stress, injury, and infections, are suspected to trigger or aggravate the evolution of psoriasis. An inflammatory immune response involving dendritic cells, T cells, keratinocytes, neutrophils, and the cytokines released from immune cells initiates the pathophysiological process (Jariwala 2007; Lowes 2008; Wilson 2007; Zheng 2007).
Diagnosis is made based on clinical findings; skin biopsy is rarely used to diagnose the disease (Boehncke 2015). Several clinical types of psoriasis exist: plaque, pustular, inverse, and erythrodermic. Plaque psoriasis is the most common form, affecting 90% of people with psoriasis (Griffiths 2007). Plaque psoriasis typically appears as raised erythematous and well‐demarcated areas of inflamed skin covered with silvery white, scaly skin (Griffiths 2007). The location of the plaques is usually symmetrical on the elbows, knees, scalp, lower back, and the periumbilical region. For 5% to 25% of people with psoriatic rheumatic disease, their skin is also involved (Helliwell 2005; Zachariae 2003).
Severity
Chronicity characterises the natural history of plaque psoriasis; this means that severity varies over time, from minor localised patches to complete body coverage. The severity of the disease usually fluctuates around the same level for a particular person (Nijsten 2007), but for each person with this disease, the evolution and duration of remission is unpredictable. The psoriasis is declared clear when remission is complete.
More than a dozen outcome instruments are used to assess the severity of psoriasis and the efficacy of different treatments for psoriasis (Naldi 2010; Spuls 2010); the Psoriasis Area and Severity Index (PASI) score is one of these instruments (Schmitt 2005). The Psoriasis Area and Severity Index combines the assessment of the severity of lesions and the area affected into a single score in the range of 0 (no disease) to 72 (maximal disease). Recent clinical trials evaluating biological therapies that have received secondary marketing authorisation by the U.S. Food and Drug Administration (FDA) and the European Medicines Agency (EMA) used PASI 75, i.e. 75% improvement in the PASI score, as the primary end point. However, the PASI has substantial limitations, such as low‐response distribution, no consensus on interpretability, and low responsiveness in mild disease (Spuls 2010).
Impact and quality of life
Disease severity alone does not determine the burden of psoriasis. Multiple studies have described an impairment of the quality of life (QoL); others have focused on an evaluation of the stigma people experience; and others have studied the impact on psychosocial life (Kimball 2005).
Impairment of QoL in people with psoriasis, when measured with the 36‐item Short Form Health Survey (SF‐36) questionnaire has been found to be higher than that of people with hypertension, diabetes, or depression (Rapp 1999).
Many tools exist to measure the QoL of people with psoriasis and other skin disorders. These measures may be categorised as psoriasis‐specific (Psoriasis Index of Quality of Life (PSORIQoL), Psoriasis Disability Index (PDI)); skin‐specific (Dermatology Life Quality Index (DLQI), Skindex (a quality‐of‐life measure for patients with skin disease)); and generic QoL measures (SF‐36). However, methodological weaknesses exist in the use of QoL questionnaires, and there is poor reporting of QoL outcomes in randomised clinical trials (Le Cleach 2008). Several case‐control studies reported a higher risk of metabolic syndrome and cardiovascular comorbidities (Kremers 2007; Naldi 2005).
Description of the intervention
There is currently no cure for psoriasis, but various treatments can help to control the symptoms; thus, long‐term treatment is usually needed. In daily practice, a treatment strategy needs to be defined, and this usually involves an induction therapy, e.g. the remission of the psoriasis flare, and a maintenance therapy, e.g. increasing the period of remission.
The therapeutic approach to psoriasis includes topical treatments as a single strategy and a first‐line therapy in the management of minor forms (Mason 2013). Nevertheless, about 20% to 30% of people with psoriasis have a moderate to severe form requiring a second‐line therapy including phototherapy and conventional systemic agents, such as ciclosporin, methotrexate, or acitretin. Among the systemic agents, the choice of drug is not clear. The NICE 2012 clinical guidelines in the UK had proposed methotrexate as the first choice of systemic agent. Other countries, such as France, do not have any available guidelines. Systemic biological agents, such as the tumour necrosis factor (TNF) antagonists (infliximab, etanercept, adalimumab), the monoclonal antibody ustekinumab that targets interleukin‐12 and ‐23 (IL‐12/‐23), anti‐IL17 drugs (secukinumab or ixekizumab), and more recently new small molecules (apremilast) are "third‐line" therapies (Boehncke 2015). Indeed, there are mandatory reimbursement criteria that patients must meet before being considered for these treatments due to their high costs: moderate to severe psoriasis after failure, intolerance or contraindication to at least two conventional systemic agents (Nast 2015b).
We used the European S3 guidelines terminology to categorise the treatments (Nast 2015b).
Oral systemic treatments
Conventional systemic agents
Conventional systemic agents are a heterogeneous group of treatments that are the oldest interventions given to clear psoriasis.
The existing oral systemic pharmacological treatments available for psoriasis are ciclosporin, methotrexate, acitretin (which is the retinoid of choice for psoriasis), and fumaric acid esters (FAEs) which are licensed for psoriasis in Germany and used off‐licence in other countries (Atwan 2015).
Randomised controlled trials against placebo for both induction and maintenance therapies have demonstrated the efficacy of ciclosporin for psoriasis (Bigby 2004; Christophers 1992; Ellis 1991; Flytstrom 2008; Koo 1998; Heydendael 2003; Ho 1999; Mahrle 1995; Meffert 1997; Mrowietz 1995; Shupack 1997). In 2008, Saurat et al conducted the only randomised trial comparing the efficacy of methotrexate with placebo (Saurat CHAMPION, 2008). Randomised trials against placebo have demonstrated the efficacy of derivatives of vitamin A, the retinoids, in the treatment of plaque psoriasis (Pettit 1979). Fumaric acid esters are an alternative therapy for people with psoriasis, even though the mechanisms of action are not completely understood (Ormerod 2004). A Cochrane Review on FAEs for psoriasis was published in 2015 (Atwan 2015).
Small molecules
Small molecules affect molecules inside immune cells. Recently, small molecule drugs have been developed and show potential to treat psoriasis patients not responding to conventional treatments. These small molecule drugs include apremilast (Papp 2012b), tofacitinib (Bachelez 2015), and ponesimod (Vaclavkova 2014). Tofacitinib and ponesimod had not been approved for psoriasis at the time our analyses were done.
Biological therapies
Biological therapies use substances made from living organisms, or synthetic versions, to target the immune system. In the twentieth century, the development of biological treatments expanded the therapeutic spectrum of systemic treatments for psoriasis. All of the biologics have to be given by infusion or subcutaneous injection, and all have had at least one evaluation of their effectiveness against placebo: alefacept (Krueger 2002; Lebwohl 2003), etanercept (Leonardi 2003), infliximab (Chaudhari 2001), adalimumab (Menter REVEAL, 2008), certolizumab (Reich 2012), ustekinumab (Lebwohl 2010), secukinumab (Reich 2015), ixekizumab (Leonardi 2012), brodalumab (Papp 2012), guselkumab (Gordon X‐PLORE, 2015), tildrakizumab (Papp 2015a), and itolizumab (Krupashankar 2014). Certolizumab, tildrakizumab, and itolizumab had not been approved for psoriasis at the time our analyses were done.
How the intervention might work
Dysregulation of the immune system is a critical event in psoriasis, and the evolving knowledge of the role of the immune system in the disease has had a significant impact on treatment development.
Indeed, psoriatic plaque shows marked infiltration by activated T cells, especially CD4+ cells in the dermis. The activated T cells produce several important cytokines, namely, interferon (IFN)‐c, TNF alpha (by Th1 and Tc1 cells), IL‐17A, and IL‐23R (by Th17 and Tc17 cells) (Boehncke 2015).
Oral systemic treatments
Conventional systemic agents
Ciclosporin
Ciclosporin is an immunosuppressive agent (a drug that reduces the efficacy of the immune system); it acts by inhibiting the initial phase of the activation of CD4+ T cells, which leads to a block on the synthesis of interleukin 2 by the complex cyclophilin‐ciclosporin, thus, preventing T cell proliferation that is key to the pathogenesis of psoriasis (see above) (Ho 1996). This immunosuppression is rapid and reversible. Ciclosporin rapidly reduces the severity of the lesions (over one to three months), but the continuation of treatment is difficult after two years because of the development of adverse effects, such as elevated creatinine levels (Maza 2011). A dose of 5.0 mg/kg/day ciclosporin was significantly more effective than 2.5 mg/kg/day ciclosporin for induction of the remission of psoriasis; however, elevated creatinine was significantly more likely with 5.0 mg/kg/day ciclosporin than with 2.5 mg/kg/day ciclosporin (Christophers 1992).
Methotrexate
Methotrexate is an antimetabolite (an inhibitor of a chemical that is part of normal metabolism), which acts as an antagonist of folic acid (Montaudie 2011). Low doses of methotrexate exert anti‐inflammatory and immunomodulatory activities (Montaudie 2011). The efficacy of methotrexate cannot be assessed earlier than three months; its long‐term safety profile is good. In clinical practice, methotrexate is administered orally at 15 to 25 mg/week (Montaudie 2011).
Retinoids
Retinoids, including acitretin, are involved in the growth and differentiation of skin tissue; they bind to nuclear receptors that belong to the large family of steroid hormone receptors (Sbidian 2011). Retinoids modulate many types of proteins, including epidermal structural proteins, metalloproteinases, and cytokines (Sbidian 2011). The efficacy of retinoids is evaluated after two to three months of treatment, but skin side effects (e.g. xerosis, cheilitis) may limit the ability to increase the dose. Treatment with retinoids is best avoided in women of childbearing age because of risks to a developing foetus and the necessity of using contraception two years after discontinuation of treatment (Sbidian 2011). People receiving 50 mg/day to 75 mg/day acitretin have significantly improved psoriasis compared with those receiving 10 mg/day to 25 mg/day acitretin (Goldfarb 1988).
FAEs
FAEs are chemical compounds derived from the unsaturated dicarboxylic acid (Atwan 2015). Oral preparations of FAEs in psoriasis were developed containing dimethyl fumarate (DMF) and salts of monoethyl fumarate (MEF) as main compounds (Atwan 2015). FAEs produce anti‐inflammatory effects by preventing the proliferation of T cells (Atwan 2015).
FAEs are an effective therapy in people with psoriasis (50% to 70% achieve PASI 75 improvement within four months of treatment). Tolerance is limited by gastrointestinal side effects and flushing of the skin (Atwan 2015). Several case‐series described rare adverse events, such as progressive multifocal leukoencephalopathy (Balak 2016). In clinical practice, FAEs are administered orally. People receive this after a gradual dose incrementation the equivalent of 720 mg of DMF per day.
Small molecules
Small molecule drugs modulate proinflammatory cytokines and selectively inhibit signalling pathways: phosphodiesterase 4 inhibitors (apremilast), Janus kinase (JAK) inhibitors (tofacitinib), or sphingosine 1‐phosphate receptor agonists (ponesimod) (Torres 2015).
Apremilast
Apremilast belongs to the phosphodiesterase 4 (PDE4) inhibitors family (Torres 2015). By increasing cyclic adenosine monophosphate (cAMP) levels, PDE4 inhibitors reduce production of pro‐inflammatory TNF alpha and IFNγ in patients with psoriasis. Apremilast has recently been approved for psoriasis; its efficacy seems to be higher than conventional systemic therapy; however, no randomised controlled trials (RCTs) have assessed apremilast versus methotrexate or ciclosporin. The safety of the drug should be detailed in the near future with phase 4 studies. In clinical practice, apremilast is administered orally at 30 mg twice a day (Torres 2015).
Tofacitinib
Tofacitinib is a Janus kinase (JAK) inhibitor (Torres 2015). JAK inhibitors targets the Janus kinase (JAK)/signal transducer and activator of transcription (STAT) pathway, which is pivotal for the downstream signaling of inflammatory cytokines involved in psoriasis. Tofacitinib had not been approved for psoriasis at the time our analyses were done (Torres 2015).
Ponesimod
Ponesimod is a sphingosine 1‐phosphate receptor agonist that causes dose‐dependent sequestration of lymphocytes in lymphoid organs, thus, preventing T cell proliferation, which is key to the pathogenesis of psoriasis. Ponesimod had not been approved for psoriasis at the time our analyses were done (Torres 2015).
Biological therapies
Biological therapies have been developed in recent years and target and prevent T cell proliferation (e.g. alefacept and itolizumab, a humanised IgG1 (immunoglobulin G1) monoclonal antibody, which selectively targets CD6) or target cytokines involved in psoriasis physiopathology (e.g. anti‐TNF alpha, anti‐IL12/23, anti‐IL23, anti‐IL17).
Alefacept
Alefacept is an immunosuppressive agent (a fusion protein that blocks the growth of some types of T cells). Alefacept (either 7.5 mg intravenously (IV) or 15 mg intramuscularly (IM) once a week) is used to control inflammation in moderate to severe psoriasis with plaque formation, where it interferes with lymphocyte activation. This drug was never approved for the European drug market. It was sold in North America, Switzerland, Israel, and Australia. In 2011, the manufacturers made a decision to cease sales of alefacept. This decision was not related to any specific safety concern nor the result of any FDA‐mandated or voluntary product recall (Heffernan 2010).
Anti‐TNF alpha
Two monoclonal antibodies against tumour necrosis factor alpha (TNF‐α) (infliximab, adalimumab) and one recombinant TNF‐α receptor (etanercept) have been developed to inhibit TNF‐α signalling, thus, preventing its inflammatory effects and are approved in psoriasis (Gisondi 2004). A third, certolizumab, is being assessed for psoriasis in phase 3 trials.
-
Etanercept is a recombinant TNF‐α receptor and weakly immunogenic (provokes only a mild immune response). Its efficacy is assessed at three months. A 50 mg dose of etanercept is administered subcutaneously twice weekly for three months during the induction phase (remission of the psoriasis flare) with 50 mg administered weekly as maintenance therapy (Gisondi 2004).
-
Infliximab is a chimeric antibody that neutralises the action of TNF‐α. Its efficacy is evaluated after six to eight weeks of treatment. A dose of 5.0 mg/kg infliximab is given as an intravenous (IV) induction regimen at 0, 2, and 6 weeks followed by a maintenance regimen of 5.0 mg/kg every 8 weeks. The presence of a murine sequence at recognition sites can lead to the development of anti‐infliximab antibodies that may impair the therapeutic effect (Gisondi 2004).
-
Adalimumab is a fully humanised antibody with very low immunogenicity. Its efficacy is estimated after eight and 12 weeks of treatment. One dose of 80 mg is administered subcutaneously, followed one week later by a 40 mg subcutaneous dose, which is administered every two weeks (Mossner 2009). Those receiving TNF‐α blockers are potentially exposed to a greater risk of infection and require regular monitoring (Tubach 2009).
-
Certolizumab is an anti‐TNF alpha with a unique structure that does not contain an Fc (fragment crystallisable) portion as adalimumab or infliximab does based on the human immunoglobulin G1 Fc. Therefore, certolizumab does not display Fc‐mediated effects (improving solubility, increasing drug stability, and decreasing immunogenicity). Certolizumab had not been approved for psoriasis at the time our analyses were done (Campanati 2017).
Anti‐IL12/23, Anti‐IL23, Anti‐IL17
Additional monoclonal antibodies have been developed against pro‐inflammatory cytokines: IL‐12, IL‐23, and IL‐17 inhibit the inflammatory pathway at a different point to the anti‐TNF alpha antibodies (Dong 2017).
-
Interleukin‐12 and IL‐23 share a common domain, p40, which is the target of ustekinumab (which the FDA has recently approved) (Savage 2015). A 45 mg subcutaneous dose is administered initially (90 mg if body weight is over 100 kg), then 45 mg (or 90 mg) subcutaneously four weeks later, and thereafter 45 mg (or 90 mg) subcutaneously every 12 weeks (Savage 2015). Interleukin‐23 plays an essential role in skin inflammation in psoriasis leading to the development of agents that selectively target the IL‐23p19 subunit (Dong 2017). Drugs targeting the p19 subunit of IL‐23 are guselkumab (a fully human IgG1k monoclonal IL‐23 antagonist), tildrakizumab (a humanised IgG1k monoclonal antibody), and risankizumab (high affinity humanised IgG1 monoclonal antibody) (Dong 2017). In July 2017, the FDA approved guselkumab for psoriasis. Guselkumab is given as a 100 mg subcutaneous injection every 8 weeks, following two starter doses at week 0 and week 4. Risankizumab was assessed after we began the systematic review and will be added in the next update.
-
Interleukin‐17 inhibitors include secukinumab (a recombinant fully human anti‐IL17A IgG1k monoclonal antibody), ixekizumab (a humanised anti‐IL17 immunoglobulin G4 monoclonal antibody), and brodalumab (a human IgG2 monoclonal antibody that decreases the downstream effect of IL‐17 by antagonisng the IL‐17RA receptor) (Dong 2017). The recommended dosage for secukinumab is 300 mg administered subcutaneously at weeks 0, 1, 2, 3, and 4, and then every 4 weeks thereafter. Ixekizumab is administered at 160 mg (2 x 80 mg injections) at weeks 0, 2, 4, 6, 8, 10, and 12, and then every 4 weeks thereafter (Dong 2017).
Why it is important to do this review
To determine the treatment pathway in psoriasis, the efficacy and safety of each systemic treatment must be determined relative to other therapies. Several randomised controlled trials (RCTs) have compared against placebo the efficacy of the different systemic treatments for psoriasis. However, there are few trials comparing conventional systemic therapies head‐to‐head, systemic therapies against biological therapies, or biological therapies head‐to‐head. Several previous meta‐analyses or indirect comparison meta‐analyses have been published (Bansback 2009; Brimhall 2008; Gomez‐Garcia 2017; Gospodarevskaya 2009; Lin 2012; Loveman 2009; Nast 2015; Nelson 2008; Reich 2008; Reich 2012a; Schmitt 2008; Signorovitch 2010; Signorovitch 2015; Spuls 1997; Strober 2006; Tan 2011; Turner 2009; Woolacott 2006). However, the number of studies included in these publications was low, the searches were not exhaustive, and several trials have been published since their search dates. Also, the publications did not evaluate some systemic and biological treatments.
A network meta‐analysis enables the best use of the direct and indirect information available to determine the relative efficacy of treatments. In other words, a network meta‐analysis will help to highlight the missing key comparisons that are needed to inform clinical practice.
The plans for this review were published as a protocol 'Systemic pharmacological treatments for chronic plaque psoriasis' (Sbidian 2015).
Objectives
To compare the efficacy and safety of conventional systemic agents (acitretin, ciclosporin, fumaric acid esters, methotrexate), small molecules (apremilast, tofacitinib, ponesimod), anti‐TNF alpha (etanercept, infliximab, adalimumab, certolizumab), anti‐IL12/23 (ustekinumab), anti‐IL17 (secukinumab, ixekizumab, brodalumab), anti‐IL23 (guselkumab, tildrakizumab), and other biologics (alefacept, itolizumab) for patients with moderate to severe psoriasis and to provide a ranking of these treatments according to their efficacy and safety.
Methods
Criteria for considering studies for this review
Types of studies
We included randomised controlled trials (RCTs).
Phase I trials were not eligible because participants, outcomes, dosages, and schema of administration of interventions are too different from phase II, III, and IV studies. Cross‐over trials were not eligible (because of the unpredictable evolution of psoriasis and risk of carry‐over bias). Non‐randomised studies, including follow‐up studies, were not eligible.
Types of participants
We considered trials that included adults (over 18 years of age) with moderate to severe plaque psoriasis (i.e. needed systemic treatment) or psoriatic arthritis whose skin had been clinically diagnosed with moderate to severe psoriasis and who were at any stage of treatment.
Types of interventions
We considered trials that assessed systemic and biological treatments, irrespective of the dose and duration of treatment, compared with placebo or with each other.
Systemic and biological treatments included the following:
-
Systemic conventional treatments
-
FAEs
-
Acitretin
-
Ciclosporin
-
Methotrexate
-
-
Small molecules
-
Apremilast
-
Tofacitinib
-
Ponesimod
-
-
Anti‐TNF alpha
-
Infliximab
-
Etanercept
-
Adalimumab
-
Certolizumab
-
-
Anti‐IL12/23
-
Ustekinumab
-
-
Anti‐IL17
-
Secukinumab
-
Brodalumab
-
Ixekizumab
-
-
Anti‐IL23
-
Tildrakizumab
-
Guselkumab
-
-
Other biologic treatments
-
Itolizumab
-
Alefacept
-
We were interested to compare both the different drugs (n = 19) and the different classes of drugs (n = 7).
A new anti‐IL23 molecule (BI 655066, risankizumab) appeared after we began this review and was not included in this systematic review. However, the ongoing studies of risankizumab have been reported in this review.
Active comparators included the following:
-
any of the aforementioned systemic and biological treatments; or
-
additional treatment not of primary interest but used for the network synthesis, such as topical treatment or phototherapy.
In multi‐arm trials, study groups assessing drugs other than those mentioned above were not eligible. In cases of multi‐dose trials, we grouped together all of the different dose groups as a single arm and performed sensitivity analysis at dose level.
In our Background section, we have referred to ongoing Cochrane Reviews that address some of the systemic treatments administered to adults with plaque psoriasis. We considered these treatments in our review, and we have liaised with each of these teams to harmonise our protocols. However, the Cochrane Review on FAEs, published in 2015, included people with all types of psoriasis and not only plaque‐type psoriasis (Atwan 2015).
Types of outcome measures
Psoriasis is a chronic disease; treatments are symptomatic often with a return to baseline after discontinuation. In the absence of an existing defined core outcome set (Spuls 2016), we chose the most relevant outcomes for patients (COMET). The Psoriasis Area and Severity Index score (PASI) 75 is the most common outcome measure used. However, confronted with a debilitating and a socially and psychologically highly visible disease, a completely "clear or almost clear" skin is a more stringent test in the induction phase (remission of the psoriasis flare).
Primary outcomes
-
The proportion of participants who achieved clear or almost clear skin, that is, at least PASI 90.
-
The proportion of participants with serious adverse effects (SAE). We used the definition of severe adverse effects from the International Conference of Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use, which includes death, life‐threatening events, initial or prolonged hospitalisation, and adverse events requiring intervention to prevent permanent impairment or damage.
Secondary outcomes
-
Proportion of participants who achieve PASI 75 at induction phase.
-
Proportion of participants who achieve a Physician Global Assessment (PGA) value of 0 or 1.
-
Quality of life measured by a specific scale. Available validated scales are the Dermatology Life Quality Index (DLQI), Skindex, Psoriasis Disability Index (PDI), or Psoriasis Symptom Inventory (PSI).
-
The proportions of participants with adverse effects (AE).
-
Proportion of participants with at least one relapse in the maintenance phase (between 52 to 104 weeks).
Timings
Where possible, we evaluated the outcomes at two different timings:
-
induction therapy (short‐term remission) (evaluation less than 24 weeks after the randomisation); and
-
maintenance therapy (long‐term remission) (evaluation between 52 and 104 weeks after the randomisation).
We did not include studies that had timings outside of these time ranges in our review. All of the outcomes except the proportion of participants with at least one relapse in the maintenance phase were recorded during the randomisation phase.
Search methods for identification of studies
We aimed to identify all relevant RCTs regardless of language or publication status (published, unpublished, in press, or in progress).
Electronic searches
We searched the following databases up to 15 December 2016:
-
the Cochrane Skin Specialised Register using the search strategy in Appendix 1;
-
the Cochrane Central Register of Controlled Trials (CENTRAL; 2016, Issue 11) in the Cochrane Library using the strategy in Appendix 2;
-
MEDLINE Ovid (from 1946) using the strategy in Appendix 3;
-
Embase Ovid (from 1974) using the strategy in Appendix 4; and
-
LILACS (Latin American and Caribbean Health Science Information database, from 1982) using the strategy in Appendix 5.
Trials registers
We searched the following trials registers up to 22 December 2016 with the following search terms: psoriasis AND one by one each drug name listed in Types of interventions:
-
World Health Organization International Clinical Trials Registry Platform (ICTRP) (apps.who.int/trialsearch/);
-
ISRCTN registry (www.isrctn.com);
-
US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov);
-
Australian New Zealand Clinical Trials Registry (www.anzctr.org.au); and
-
EU Clinical Trials Register (www.clinicaltrialsregister.eu).
Searching other resources
Previous meta‐analyses and systematic reviews
We looked at the search strategies of previous meta‐analyses to improve our search strategies.
References from other studies
We checked the bibliographies of included and excluded studies for further references to relevant trials.
Unpublished literature
We searched the trial results databases of various pharmaceutical companies to identify ongoing and unpublished trials. We made attempts to locate unpublished and ongoing trials through correspondence with authors and pharmaceutical companies (see Table 2).
| Contact | Requested Information | Contacted | Reply (last check 1/03/2017) | |
| Missing data | ||||
| Prof. Akcali | Outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 8 and 21 November 2016 | No response | |
| Prof. Al‐Hamamy | Outcomes: PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 8 and 21 November 2016 | No response | |
| Prof. Asahina | Outcome: PASI 90 | 8 November 2016 | Asahina 2010 detailed report | |
| Prof. Asahina Pfizer | Outcomes: AEs & SAEs | 3 and 12 January 2017 | Additional data to the publication not provided | |
| Prof. Asawanonda | Outcomes: PASI 75, PGA 0/1, AEs & SAEs | 21 November 2016 15 December 2016 | Asawanonda 2006 sent detailed report for PASI 75 and AEs. PGA was not collected during this study. | |
| Prof. Bisonnette Innovaderm Recherches Inc. | Outcomes: PASI 90, PGA 0/1, AEs | 8 and 21 November 2016 | Additional data to the publication not provided | |
| Dr Blauvelt Novartis | Outcome: QoL scale | 8 and 21 November 2016 | Additional data to the publication not provided | |
| Prof. Fabri | Outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 8 and 21 November 2016 | Caproni 2009 sent detailed report for PASI 90 and SAEs. Other outcomes (PGA, QoL and AEs) not collected during this study. | |
| Prof. Dogra | Outcomes: PGA 0/1, QoL scale, AEs & SAEs | 8 and 21 November 2016 | No response | |
| Prof. Dogra | Outcomes: PGA 0/1, QoL scale, AEs & SAEs | 8 November 2016 | PGA & QoL scale not collected during this study. AEs & SAEs not provided per arm | |
| Dr Fallah Arani | Outcomes: PASI 90, PGA 0/1 and QoL scale | 8 and 21 November 2016 | Outcomes not collected during this study | |
| Prof. Flytstrom | Outcomes: PGA 0/1 | 12 and 19 January 2017 | Additional data to the publication not provided | |
| Prof. Gisondi | Outcomes: PASI 90, PGA 0/1, QoL scale, AEs & SAEs | 8 November 2016 | Gisondi 2008 sent detailed report for the requested outcomes except for QoL (not assessed during the study) | |
| Prof. Gordon | Outcomes: PGA0/1, AEs | 3 and 12 January 2017 | No response | |
| Prof. Gottlieb Abbvie | Outcomes: PASI 90 & QoL scale | 8 November 2016 | Gottlieb 2012 sent detailed report for the requested outcomes | |
| Prof. Gottlieb Amgen | Outcomes: PASI 90, PGA 0/1, QoL scale, AEs & SAEs | 8 November 2016 | Gottlieb 2011 sent detailed report for the requested outcomes | |
| Prof. Griffiths Janssen | Outcome: QoL scale | 16 December 2016 | QoL was not collected during this study | |
| Prof. Jacobe | Outcomes: PASI 90, PGA 0/1, QoL scale, AEs & SAEs | 8 and 20 November 2016 | No response | |
| Pfizer | Outcomes: PASI 90, QoL scale | 3 and 12 January 2017 | No response | |
| Prof. Ganapathi R&D, Biocon Research Limited | Outcomes: QoL scale, AEs & SAEs | 8 and 21 November 2016 | Krupashandar sent detailed report for the requested outcomes, however AEs and SAEs were only available for the entire trial and not at the time of the major outcome assessment | |
| Prof. Lebwohl Valeant Pharmaceuticals NA LLC | Outcomes: PASI 90 & QoL scale | 8 and 21 November 2016 | Lebwohl AMAGINE‐2, 2015 sent detailed report for PASI 90, individual scores and median difference from baseline of QoL were not available | |
| Prof. Lebwohl Valeant Pharmaceuticals NA LLC | Outcomes: PASI 90 & QoL scale | 8 and 21 November 2016 | Lebwohl AMAGINE‐3, 2015 sent detailed report for PASI 90, individual scores and median difference from baseline of QoL were not available | |
| Prof. Leonardi | Outcomes: QoL scale & AEs | 8 and 21 November 2016 | No response | |
| Prof. Kaur | Outcomes: PASI 90, PGA 0/1, QoL scale, AEs & SAEs | 8 and 21 November 2016 | No response | |
| Prof. Menter | Outcome: PGA 0/1 | 8 and 21 November 2016 | No response | |
| Prof. Menter | Outcome: PGA 0/1 | 8 and 21 November 2016 | No response | |
| Prof. Mrowietz | Outcome: QoL scale | 3 and 12 January 2017 | Additional data to the publication not provided | |
| Prof. Paul Novartis | Outcome: PASI 90 | 3 January 2017 | Additional data to the publication not provided | |
| Prof. Papp | Outcome: QoL scale | 22 November 2016 13 December 2016 | Additional data to the publication not provided | |
| Prof. Papp | Outcome: QoL scale | 22 November 2016 13 December 2016 | Additional data to the publication not provided | |
| Prof. Papp | Outcome: QoL scale, AEs & SAEs | 22 November 2016 13 December 2016 | Additional data to the publication not provided | |
| Prof. Papp | Outcome: QoL scale | 22 November 2016 13 December 2016 | Additional data to the publication not provided | |
| Prof. Papp | Outcome: PASI 90, PGA0/1, QoL scale | 3 January 2017 | Additional data to the publication not provided | |
| Prof. Paul Novartis | Outcome: QoL scale | 15 December 2016, 2 January 2017 | Additional data to the publication not provided | |
| Prof. Reich Novartis | Outcomes: PGA 0/1 & QoL scale | 8 November 2016, 16 December 2016 | Additional data to the publication not provided | |
| Prof. Reich PelotonAdvantage | Outcome: QoL scale | 4 January 2017 | Additional data to the publication not provided | |
| Prof. Rich | Outcome: QoL scale | 22 November 2016, 13 December 2016 | No response | |
| Prof. Sterry | Outcomes: PASI 90 & QoL scale | 8 and 21 November 2016 | No response | |
| Prof. Strober Abbvie | Outcome: QoL scale | 8 November 2016 | Strober sent detailed report for the requested outcomes | |
| Prof. Thaçi Novartis | Outcome: QoL scale | 8 and 21 November 2016 | Additional data to the publication not provided | |
| Prof. Torii | Outcomes: PASI 90 & PGA0/1 | 21 November 2016 | Torii sent detailed report for the requested outcomes | |
| Prof. Tyring | Outcomes: PGA 0/1 & QoL scale | 8 and 21 November 2016 | No response | |
| Dr van Bezooijen | Outcomes: PASI 90, adverse effects | 4 and 12 January 2017 | Additional data to the publication not provided | |
| Prof. van der Kherkhof Pfizer | Outcome: AEs | 8 and 21 November 2016 | Additional data to the publication not provided | |
| No contact | Outcomes: AEs and SAEs | No | Authors' email not found | |
| No contact | Outcome: PASI 90 | No | Authors' email not found | |
| Awaiting classification studies | ||||
| Prof. Elewski Abbvie | Study's protocol and outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 3 and 12 January 2017 | Will be included when published | |
| Prof. Khattri | Study's protocol and outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 3 and 12 January 2017 | No response | |
| Prof. Lee | Study's protocol and outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 3 and 12 January 2017 | No response | |
| Prof. Reich | Study's protocol and outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 3 January 2017 | Will be included when published | |
| Prof. Chow | outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 8 November 2016, 16 December 2016 | No response | |
| Prof. Gurel | Study's protocol and outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 17 and 24 January 2017 | Gurel 2015 sent detailed report for the requested outcomes. Finally Gurel study was classified in the included studies section. | |
| No contact | Outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | No | Authors' email not found | |
| Dr Nogarales, MD Celgene Corporation | Asking for study protocol and efficacy/safety results | 12 and 19 January 2017 | Email response: "Thank you very much for your email and your interest in our study in Japanese subjects. May I please enquire as to the planned timing for publication for your meta‐analysis as we have just recently submitted our primary manuscript?" Will be included when published | |
| Prof. Krishna | Asking for study protocol and efficacy/safety results | 5 and 12 January 2017 | No response | |
| Prof. Jacobi | Asking for study protocol and efficacy/safety results | 12 and 19 January 2017 | No response | |
| Prof. Shah | Asking for study protocol and efficacy/safety results | 12 and 19 January 2017 | ||
| Abstracts | ||||
| Prof. Yilmaz | Outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 16 December 2016 | Yilmaz 2002 sent detailed report for the requested outcomes. Finally Yilmaz 2002 study was classified in the included studies section. | |
| Prof. Mrowietz | Study's protocol and outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 16 December 2016, 3 January 2017 | Additional data to the publication not provided. Finally Mrowietz study was classified in the awaiting classification section. | |
| Prof. Reich | Study's protocol and outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 16 December 2016 | Additional data to the publication not provided. Finally Reich 2004 study was classified in the awaiting classification section. | |
| Ongoing studies | ||||
| Dr Yamauchi Dr Patnaik, Director, Clinical Science Institute | Asking for study protocol and efficacy/safety results | 5 January 2017 | Email response: Dear Dr Sbidian, Will be included when published | |
| Prof. Spuls | Asking for study protocol and efficacy/safety results | 5 January 2017 | Email response "The study is currently ongoing and has not yet been analysed. Therefore, we are not able to provide data on efficacy or safety. Will be included when published | |
AE: adverse events; PASI: Psoriasis Area and Severity Index; PGA: Physician Global Assessment; QoL: quality of life; SAE: serious adverse events
We searched reviews submitted to the U.S. Food and Drug Administration and the European Medicines Agency (EMA) for drug registration (using www.accessdata.fda.gov/scripts/cder/drugsatfda and www.ema.europa.eu/ema).
Conference proceedings
We handsearched the proceedings of the following conferences during the periods not included in the Cochrane Skin Specialised Register:
-
The American Academy of Dermatology (AAD) from 2008 to 2009 and from 2012 to 2013;
-
The Society for Investigative Dermatology (SID) from 2008 to 2009 and from 2012 to 2013; and
-
The European Academy of Dermatology and Venereology (EADV) from 2008 to 2013.
Adverse effects
We did not perform a separate search for rare or delayed adverse effects of the target interventions. However, we examined data on adverse effects from the included studies we identified.
Data collection and analysis
Selection of studies
Two groups of two review authors (LLC/ES or IGD/GD) independently examined each title and abstract to exclude irrelevant reports. These authors independently examined full‐text articles to determine eligibility. We contacted study authors for clarification when necessary and discussed disagreements to reach consensus. We list excluded studies and document the primary reason for exclusion.
Data extraction and management
Three groups of two review authors (LLC, GD, CH, IGD, CM, or ES) each extracted the data from published and unpublished reports independently using a standardised form. We pilot‐tested this form (Data Extraction Form) on a set of included trials. We extracted the data to populate the 'Characteristics of included studies' tables in RevMan Manager 5.3 (Revman 2014).
We extracted the data from the reports of the U.S. Food and Drug Administration (FDA) when available, if not from the US National Institutes of Health Ongoing Trials Register ClinicalTrials.gov (www.clinicaltrials.gov), and finally from the published reports.
Outcome data
We extracted (arm‐level data) from each included trial; hence, the total number of participants randomised to each intervention. For binary outcomes, we also extracted the number of participants (if available) who:
-
reached PASI 90, PASI 75, or PGA 0/1 during the induction phase;
-
had at least one relapse in the maintenance phase; and
-
had at least one SAE/one AE during the induction phase.
For quality of life, we extracted from each included trial the mean change score of the study specific scale from baseline to follow‐up.
When PASI 90 and PASI 75 outcomes were not reported and when the information was available, we extracted the PASI score at baseline and at the evaluation point (or the percentage reduction in PASI from baseline to follow up) to calculate the number of participants who reached PASI 75 and 90.
Regarding the assessment of quality of life, we recorded all specific quality of life (QoL) scales (Dermatology Life Quality Index (DLQI), Skindex, Psoriasis Disability Index (PDI), and Psoriasis Symptom Inventory (PSI)).
Data on potential effect modifiers
We extracted baseline demographic and clinical characteristics of participants that may have acted as effect modifiers (age, sex, body weight, duration of psoriasis, severity of psoriasis at baseline, previous psoriasis treatment). One review author (ES) checked and entered the data into the RevMan computer software. We contacted the authors of the trials to request missing data (see Table 2).
Assessment of risk of bias in included studies
We used Cochrane's 'Risk of bias' (RoB) tool to assess the risk of bias. Three groups of two review authors each (LLC, GD, CH, IGD, CM, or ES) independently assessed the risk of bias, and one author (LLC) resolved any disagreements. For each of the following domains and according to the general principles in section 8.4 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011), we graded the following 'Risk of bias' domains as 'low', 'high', or 'unclear'.
-
Selection bias
-
Was the allocation sequence adequately generated? We considered randomisation adequate (low risk of bias) if the allocation sequence was generated from a table of random numbers or was computer‐generated. We considered randomisation inadequate (high risk of bias) if sequences could be related to prognosis. We considered randomisation unclear if the paper stated that the trial was randomised, but did not describe the method.
-
Was allocation adequately concealed? We deemed allocation concealment as adequate if the report stated that it was undertaken by means of sequentially pre‐numbered sealed opaque envelopes or by a centralised system. We considered a double‐blind double‐dummy process as at low risk of bias even if the paper did not describe the method of allocation concealment.
-
-
Performance and detection bias
-
Was knowledge of the allocated intervention adequately prevented during the study? We evaluated the risk of bias separately for personnel and participants, outcomes assessors, and each outcome.
-
-
Attrition bias
-
Were incomplete outcome data adequately addressed? We examined if there was imbalance across intervention groups in numbers or reasons for missing data, type of measure undertaken to handle missing data, and whether the analysis was carried out on an intention‐to‐treat (ITT) basis. We assessed the use of strategies to handle missing data.
-
-
Reporting bias
-
Were reports of the study free of suggestion of selective outcome reporting? We evaluated if each outcome was measured, analysed, and reported. We compared outcomes specified in protocols (if available on the FDA website or ClinicalTrials.gov) and in material and methods with outcomes presented in the results section. We considered reporting bias inadequate if one specified outcome in protocols was lacking in the main report.
-
-
Other risk of bias
-
We did not fulfil the 'other risk of bias' item as we did not highlight particular circumstances leading to other risk of bias from particular trial designs, contamination between the experimental and control groups, and particular clinical settings.
-
Overall risk of bias
To summarise the quality of evidence and to interpret the network results, we used these six RoB criteria (random sequence generation, allocation concealment, blinding of participants, blinding of outcome assessor, incomplete outcome data, and selective outcome reporting) in order to classify each trial.
We would classify the trial as having low risk of bias if we rated none of the domains above as high risk of bias and two or less as unclear risk.
We would classify the trial as having moderate risk of bias if we rated one domain as high risk of bias, one or less domains as unclear risk, or no domains as high risk of bias but three or less were rated as unclear risk.
All other cases were assumed to pertain to high risk of bias.
Measures of treatment effect
Relative treatment effects
For each pair‐wise comparison and each dichotomous outcome at each time point, we used risk ratios (RR) with 95% confidence intervals (CI) as a measure of treatment effect. For continuous variables (e.g. quality of life scale), we used the standardised mean difference (SMD) with 95% CI.
Relative treatment ranking ‐ network meta‐analysis
For every treatment, we estimated the ranking probabilities of being at each possible rank for all outcomes. We inferred on treatment hierarchy using the surface under the cumulative ranking curve (SUCRA) (Salanti 2011). SUCRA was expressed as a percentage between 0 (when it is certain a treatment is the worst) to 100% (when it is certain a treatment is the best).
Unit of analysis issues
The primary unit of analysis was the participant. We did not consider studies with non‐standard design features that would lead to clustering (e.g. cross‐over trials).
We treated comparisons from trials with multiple intervention groups as independent two‐arm studies in the pair‐wise meta‐analyses. At the network meta‐analysis stage, we properly accounted for the within‐trial correlation.
Dealing with missing data
We extracted, when possible, both the number of randomised and analysed participants in each study arm. We contacted trial authors or sponsors by email to request missing outcome data (numbers of events and numbers of participants for important dichotomous clinical outcomes) when these were not available in study reports that were less than 10 years old (See Table 2). For the main analysis, we assumed that any participant with missing outcome data did not experience clearance, whatever the group. In a sensitivity analysis, we also synthesised the data ignoring the missing participants (complete case analysis) assuming that they were missing at random (Mavridis 2014).
Assessment of heterogeneity
We undertook meta‐analyses only if we judged participants, interventions, comparisons, and outcomes to be sufficiently similar (section 9.5.2 of the Cochrane Handbook for Systematic Reviews of Interventions) (Higgins 2011). Potential sources of heterogeneity included participants' baseline characteristics (weight, the duration of previous treatment, treatment doses, co‐interventions, and duration of treatment). When enough data were available, we investigated the distributions of these characteristics across studies and treatment comparisons. The latter allows assessing transitivity, i.e. whether there were important differences between the trials evaluating different comparisons other than the treatments being compared (Salanti 2014). To further reassure the plausibility of the transitivity assumption, we only included in our analyses trials not involving co‐interventions and with a timing of outcome assessment from 12 to 16 weeks.
In the classical meta‐analyses, we assessed statistical heterogeneity by visual inspection of the forest plots and using the Q‐test and the I² statistic. We interpreted the I² statistic according to the following thresholds (section 9.5.2 of the Cochrane Handbook for Systematic Reviews of Interventions; Higgins 2011): 0% to 40% might not be important; 30% to 60% may represent moderate heterogeneity; 50% to 90% may represent substantial heterogeneity; 75% to 100% represents considerable heterogeneity.
In the network meta‐analysis, the assessment of statistical heterogeneity in the entire network was based on the estimated heterogeneity standard deviation parameter (τ) estimated from the network meta‐analysis models (Jackson 2014). We inferred on the presence or absence of important heterogeneity by comparing the magnitude of τ with the empirical distributions provided in Turner et al and Rhodes et al (Rhodes 2015; Turner 2012). We also estimated the prediction intervals to assess how much the estimated heterogeneity affects the relative effects with respect to the additional uncertainly anticipated in future studies (Riley 2011). Where feasible, we would have investigated the possible sources of heterogeneity in subgroup analyses and meta‐regression.
Although we restricted the risk of important heterogeneity in our data by considering eligible only studies with a follow‐up period between 12 and 16 weeks and without co‐interventions, we investigated differences in heterogeneity across the different analyses. Specifically, we observed whether splitting the nodes of the network and analysing each drug separately reduced the heterogeneity estimate. We also ran a series of sensitivity analyses (see Sensitivity analysis), and we monitored whether heterogeneity became smaller or larger compared to the primary analysis.
Assessment of reporting biases
To assess reporting biases, we used an adaptation of the funnel plot by subtracting from each study‐specific effect size the mean of meta‐analysis of the study‐specific comparison, which we plotted against the study standard error (Chaimani 2013). We employed this 'comparison‐adjusted funnel plot' for all comparisons of an active treatment against placebo. When we detected funnel plot asymmetry for the two primary outcomes, we investigated the presence of small‐study effects in the network meta‐regression (Chaimani 2012).
Data synthesis
We conducted pair‐wise meta‐analyses to synthesise trials comparing one of the treatments against placebo or two treatments against each other. We performed pair‐wise meta‐analyses for all outcomes and comparisons, provided that at least two studies were available, using a random‐effects model.
We then employed network meta‐analysis to estimate the relative effects for all possible comparisons between any pair of treatments. We provided a graphical depiction of the evidence network for all outcomes to illustrate the network geometry (Chaimani 2017). We ran network meta‐analysis using the approach of multivariate meta‐analysis, which treats the different comparisons that appear in studies as different outcomes (White 2012).
We interpreted a statistically non‐significant P value (e.g. larger than 0.05) as a finding of uncertainty unless confidence intervals were sufficiently narrow to rule out an important magnitude of effect.
We assessed inconsistency (i.e. the possible disagreement between the different pieces of evidence) locally and globally. Specifically, we used the loop‐specific approach (Bucher 1997) and the side‐splitting method (Dias 2010). We also fit the design by treatment interaction model to evaluate the presence of inconsistency in the entire network (Higgins 2012).
We conducted pair‐wise meta‐analyses using Review Manager 5 (RevMan 5) (Revman 2014), and we performed all other analyses in Stata 14 using the 'network' (www.stata‐journal.com/article.html?article=st0410) and 'network graphs' packages (www.stata‐journal.com/article.html?article=st0411).
Subgroup analysis and investigation of heterogeneity
We considered running subgroup analyses and meta‐regressions to investigate potential sources of heterogeneity or inconsistency (such as weight of participants, duration of psoriasis, baseline severity, previous systemic treatments), but no sufficient data on these characteristics were available to perform these additional analyses.
Sensitivity analysis
To assess the robustness of our results, we performed the following sensitivity analyses for the two primary outcomes: (1) running the analysis at dose‐level considering that each different drug dose is a different intervention; (2) excluding trials at high risk of bias; (3) excluding trials with a total sample size smaller than 50 randomised participants; and (4) analysing only the observed participants assuming that missing participants are missing at random.
'Summary of findings' table
We included a 'Summary of findings' table in our review. We downgraded evidence based on the five Grading of Recommendations, Assessment, Development and Evaluation (GRADE) considerations (study limitations, consistency of effect, imprecision, indirectness, and publication bias) (Schunemann 2011). We assessed the confidence of the evidence estimates from network meta‐analysis, based on an extension of the standard GRADE approach which is based on the contributions of the direct comparisons to the estimation in the network meta‐analysis (Salanti 2014).
We included an overall grading of the evidence for the two main outcomes:
• PASI 90 during the induction phase
• Serious adverse effects during the induction phase
We assessed the study limitations by first evaluating the risk of bias of each direct estimate and then integrating these judgements with the contribution of each direct estimate to the network estimates.
We assessed inconsistency by considering the networks' heterogeneity (network meta‐analysis estimate of between‐study variance and prediction intervals) and using both local and global inconsistency in the networks.
We assessed imprecision by focusing on the CIs of the network meta‐analysis treatment effect estimates and by examining ranking probabilities (rankograms).
We assessed indirectness by evaluating the distribution of the potential effect modifiers (baseline demographic and clinical characteristics of participants).
We assessed publication bias by considering the comprehensive search strategy that we performed and the risk of publication bias in the specific field. The comparison‐adjusted funnel plots that test the presence of small‐study effects in the network assisted our judgement.
For each outcome, we chose the median placebo‐group risk value in the included studies for the assumed risk with placebo. According to the software GRADEpro 2008 (www.gradepro.org), we assigned four levels of certainty of evidence: high, moderate, low, or very low. We used this assessment, which two authors (LLC and ES) conducted, to inform the main text of the discussion section.
Results
Description of studies
Results of the search
The Electronic searches retrieved 4798 records after deduplication. The searches of other sources identified 622 records from trials registers and three further records from other sources. We had a total of 5422 records after removal of duplicates.
After reviewing the titles and abstracts, we discarded 4738 citations. We examined the full text of the remaining 684 citations: 410 did not meet the inclusion criteria. Within this group, 203 did not include participants with moderate to severe psoriasis and so did not meet our inclusion criteria. We have not created 'Characteristics of excluded studies' tables for this group. We had a further 207 excluded studies (see Characteristics of excluded studies). We identified 14 trials as studies awaiting classification (reported in 18 references) (see Characteristics of studies awaiting classification). We identified 34 studies as ongoing (see Characteristics of ongoing studies).
We included 109 studies, reported in 222 references. For a further description of our screening process, see the study flow diagram (Figure 1).
Study flow diagram
Included studies
Trial design
All trials used a parallel‐group design. The mean sample size was 366 (range: 10 to 1881). In all, 88 trials were multicentric trials (2 to 231 centres) and 15 were single‐centre trials (Akcali 2014; Al‐Hamamy 2014; Asawanonda 2006; Chaudhari 2001; Chladek 2005; Dogra 2013; Dogra 2012; Dubertret 1989; Ellis 1991; Gisondi 2008; Gurel 2015; Hunter 1963; Mahajan 2010; Shehzad 2004; Van Bezooijen 2016); for six trials, single‐centre or multicentric status was not clear (Caproni 2009; Engst 1994; Goldfarb 1988; Jacobe 2008; Olsen 1989; Yilmaz 2002). All of the trials recruited participants from a hospital setting. The trials took place worldwide (n = 37, 33.9%), in Europe (n = 28, 25.7%), in North America (n = 21, 19.3%), in Asia (n = 17, 15.6%), or in the Middle East (n = 1, 0.9%). The location was not stated for five trials (Caproni 2009; Engst 1994; Goldfarb 1988; Jacobe 2008; Olsen 1989).
In total, 55 trials out of 109 were multiarm; 40 multiarm trials assessed the same experimental drug at multiple dose levels; seven assessed at least two different drugs; eight assessed both the same experimental drug at multiple dose levels and different drugs.
In total, 15 trials (Al‐Hamamy 2014; Asawanonda 2006; Bissonnette 2013; Gottlieb 2012; Gurel 2015; Jacobe 2008; Lowe 1991; Mahajan 2010; Ruzicka 1990; Saurat 1988; Shehzad 2004; Sommerburg 1993; Tanew 1991; Van Bezooijen 2016; Yilmaz 2002) had a co‐intervention mainly with phototherapy. Only 14 studies were carried out before the year 2000 (Dubertret 1989; Ellis 1991; Engst 1994; Goldfarb 1988; Hunter 1963; Laburte 1994; Lowe 1991; Meffert 1997; Nugteren‐Huying 1990; Olsen 1989; Ruzicka 1990; Saurat 1988; Sommerburg 1993; Tanew 1991).
Characteristics of the participants
This review included 109 trials, with a total of 39,882 randomised participants. We summarise the characteristics of the participants in the Characteristics of included studies. The participants were reported to be between 27 and 56.5 years old, with an overall mean age of 44; there were more men (26,902) than women (12,384). Age and gender were unreported for, respectively, 743 and 596 participants (eight and nine studies). The overall mean weight was 85.6 (range: 64 to 97), and the overall mean Psoriasis Area and Severity Index (PASI) score at baseline was 20 (range: 9.5 to 39).
Characteristics of the comparisons
Trials with two parallel arms (the different dose groups were grouped together in one "arm")
Intervention versus placebo: 73 trials compared systemic treatments with placebo
-
Twenty‐one trials compared systemic conventional treatments versus placebo
-
-
Acitretin (n = 9) (Goldfarb 1988; Gurel 2015; Lowe 1991; Olsen 1989; Ruzicka 1990; Saurat 1988; Sommerburg 1993; Tanew 1991; Yilmaz 2002)
-
Fumaric acid esters (FAEs) (n = 3) (Nugteren‐Huying 1990; Mrowietz BRIDGE, 2016; Van Bezooijen 2016)
-
Ciclosporin (n = 2) (Ellis 1991; Meffert 1997)
-
Methotrexate (n = 7) (Al‐Hamamy 2014; Asawanonda 2006; Hunter 1963; Gottlieb 2012; Mahajan 2010; Shehzad 2004; Warren METOP, 2017)
-
-
Nine trials compared small molecule treatments versus placebo
-
-
Apremilast (n = 4) (Papp 2012b; Papp 2013b; Papp ESTEEM‐1, 2015; Paul ESTEEM‐2, 2015)
-
Tofacitinib (n = 4) (Krueger 2016; Papp 2012a; Papp OPT Pivotal‐1, 2015; Papp OPT Pivotal‐2, 2015)
-
Ponesimod (n = 1) (Vaclavkova 2014)
-
-
Forty‐three trials compared biological treatments versus placebo
-
-
Anti‐TNF alpha
-
-
-
-
Etanercept (n = 8) (Bagel 2012; Gottlieb 2003; Gottlieb 2011; Leonardi 2003; Papp 2005; Strober 2011; Tyring 2006; Van de Kerkhof 2008)
-
Adalimumab (n = 5) (Asahina 2010; Bissonnette 2013; Gordon 2006; Menter REVEAL, 2008; Cai 2016)
-
Infliximab (n = 6) (Chaudhari 2001; Gottlieb 2004; Reich EXPRESS, 2005; Torii 2010; Yang 2012; Menter EXPRESS‐II, 2007)
-
Certolizumab (n = 1) (Reich 2012)
-
-
-
-
Anti‐IL12/23
-
Ustekinumab (n = 6) (Igarashi 2012; Krueger 2007; Leonardi PHOENIX‐1, 2008; Papp PHOENIX‐2, 2008; Tsai PEARL, 2011; Zhu LOTUS, 2013)
-
-
-
-
Anti‐IL17
-
Secukinumab (n = 6) (Blauvelt FEATURE, 2015; Langley ERASURE, 2014; Papp 2013a; Paul JUNCTURE, 2015; Reich 2015; Rich 2013)
-
Ixekizumab (n = 2) (Gordon UNCOVER‐1, 2016; Leonardi 2012)
-
Brodalumab (n = 3) (Papp AMAGINE‐1, 2016; Papp 2012; Nakagawa 2016)
-
-
-
-
Anti‐IL23
-
Guselkumab (n = 0)
-
Tildrakizumab (n = 1) (Papp 2015a)
-
-
-
-
Other biologics
-
itolizumab (n = 1) (Krupashankar 2014)
-
Alefacept (n = 4) (Ellis 2001; Jacobe 2008; Krueger 2002; Lebwohl 2003)
-
-
Intervention versus active comparators: 25 trials compared systemic treatments with systemic treatments
-
Acitretin versus acitretin (n = 1) (Dogra 2013)
-
Acitretin versus ciclosporin (n = 1) (Akcali 2014)
-
Ciclosporin versus methotrexate (n = 4) (Flytström 2008; Heydendael 2003; Piskin 2003, Sandhu 2003)
-
Ciclosporin versus ciclosporin (n = 3) (Dubertret 1989; Engst 1994; Laburte 1994)
-
Methotrexate versus methotrexate (n = 2) (Chladek 2005; Dogra 2012)
-
Methotrexate versus FAEs (n = 1) (Fallah Arani 2011)
-
Methotrexate versus alefacept (n = 1) (Yan 2011)
-
Methotrexate versus infliximab (n = 1) (Barker RESTORE‐1, 2011)
-
Acitretine versus etanercept (n = 2) (Caproni 2009; Gisondi 2008)
-
Etanercept versus etanercept (n = 3) (Ortonne 2013; Sterry PRESTA, 2010; Strohal PRISTINE, 2013)
-
Etanercept versus infliximab (n = 1) (de Vries PIECE, 2016)
-
Etanercept versus ustekinumab (n = 1) (Griffiths ACCEPT, 2010)
-
Tofacitinib versus tofacitinib (n = 2) (Asahina 2016; Bissonnette 2015)
-
Secukinumab versus secukinumab (n = 1) (Mrowietz SCULPTURE, 2015)
-
Secukinumab versus ustekinumab (n = 1) (Thaci CLEAR, 2015)
Trials with three parallel arms (the different dose groups were grouped together in one "arm")
A total of 11 trials compared systemic treatments with systemic treatments and placebo.
-
Methotrexate versus adalimumab versus placebo (n = 1) (Saurat CHAMPION, 2008)
-
Etanercept versus ixekizumab versus placebo (n = 2) (Griffiths UNCOVER‐2, 2015; Griffiths UNCOVER‐3, 2015)
-
Etanercept versus secukinumab versus placebo (n = 1) (Langley FIXTURE, 2014)
-
Etanercept versus apremilast versus placebo (n = 1) (Reich LIBERATE, 2017)
-
Guselkumab versus adalimumab versus placebo (n = 3) (Blauvelt VOYAGE‐1, 2016; Gordon X‐PLORE, 2015; Reich VOYAGE‐2, 2017)
-
Brodalumab versus ustekinumab versus placebo (n = 2) (Lebwohl AMAGINE‐2, 2015; Lebwohl AMAGINE‐3, 2015)
-
Tofacitinib versus etanercept versus placebo (n = 1) (Bachelez 2015)
In total, the dataset consisted of 109 studies, which provide information on 204, 159, and 152 comparisons between 35 different drug doses, 20 different drugs, and 8 different drug classes, respectively (both including placebo). For the sensitivity analyses, the different drug doses were divided into the following:
-
methotrexate, taken orally, ≥ 15 or < 15 mg per week;
-
ciclosporin, taken orally, ≥ 3 or < 3 mg/Kg per day;
-
acitretin, taken orally, ≥ 35 or < 35 mg per day;
-
apremilast, taken orally, 30 mg twice a day or other dosages per day;
-
ponesimod, taken orally, 40 mg per day or other dosages per day;
-
tofacitinib, taken orally, 20 mg per day or other dosages per day;
-
etanercept, subcutaneous (S/C), 25 mg twice a week or etanercept 50 mg twice a week;
-
infliximab, intravenous, 5 mg/kg at week 0, 2, and 4 then every 6 weeks or other dosages;
-
adalimumab, S/C, 80 mg at week 0, 40 mg at week 1 then 40 mg every other week or other dosages;
-
secukinumab, S/C, 300 mg at week 0, 1, 2, 3, and 4 then every 4 weeks or other dosages;
-
ixekizumab, S/C, 80 mg every two weeks or other dosages;
-
brodalumab, S/C, 210 mg every two weeks or other dosages;
-
guselkumab, S/C, 100 mg at week 0 and 4 then every 16 weeks or other dosages.
Alefacept (S/C or intravenous (IV)), FAEs (taken orally), certolizumab (S/C), itolizumab (IV), ustekinumab (S/C 45 mg or 90 mg according to the weight) and tildrakizumab (S/C) were grouped in one dosage whatever the dosages.
For each study, we provide details of the dosage in Characteristics of included studies.
Characteristics of the outcomes
Regarding the efficacy outcomes during induction therapy (eight to 24 weeks), out of 109 trials, 82 reported PASI 90, 76 reported on Physician Global Assessment (PGA) 0/1, 93 reported PASI 75, and 54 trials reported assessment of change in quality of life. Fifty‐two studies used the dermatology‐specific instrument Dermatology Life Quality Index (DLQI); two studies used other specific skin instruments (Skindex). For all of these studies, the investigators provided citations to reports indicating that the tools had been previously validated.
Out of 109 trials, 73 reported the number of participants with adverse effects (different from the number of adverse effects), and 85 reported the number of serious adverse effects.
These outcomes were evaluated between eight and 24 weeks: eight weeks (five studies), 10 weeks (seven studies), 12 weeks (56 studies), 13 weeks (two studies), 14 weeks (two studies), 15 weeks (one study), 16 weeks (22 studies), and 24 weeks (10 studies). Timing of assessment was unknown or not clearly defined for four studies (Engst 1994; Hunter 1963; Saurat 1988; Shehzad 2004).
No trial assessed the outcome 'Proportion of participants with at least one relapse in the maintenance phase (between 52 to 104 weeks)'.
Funding
In all, 82 studies declared a source of funding, 79 studies declared a pharmaceutical company funding, four studies declared a unique institutional funding (Chladek 2005; de Vries PIECE, 2016; Flytström 2008; Heydendael 2003), five studies had no funding source (Akcali 2014; Asawanonda 2006; Fallah Arani 2011; Gurel 2015; Yan 2011), and 21 studies did not report the source of funding (Al‐Hamamy 2014; Caproni 2009; Dogra 2012; Dogra 2013; Dubertret 1989; Engst 1994; Gisondi 2008; Hunter 1963; Laburte 1994; Mahajan 2010; Meffert 1997; Nugteren‐Huying 1990; Piskin 2003; Ruzicka 1990; Sandhu 2003; Saurat 1988; Shehzad 2004; Sommerburg 1993; Torii 2010; Yang 2012; Yilmaz 2002).
Excluded studies
We excluded 410 full‐text reports. The main reason for exclusion was that the participants did not present with moderate to severe psoriasis (n = 203): these psoriasis participants were included in trials assessing the efficacy of our treatments of interest for psoriatic arthritis or had cutaneous lesions of psoriasis but not moderate to severe psoriasis. We detail the reason for exclusion of the 207 full‐text reports in Characteristics of excluded studies: we excluded 99 because they assessed another intervention, 45 were not a trial, three did not include plaque‐type psoriasis, and we excluded 60 for other reasons.
For six studies with three arms, one arm was not included as the intervention was not included in our search:
-
Saurat 1988: acitretin versus placebo versus etretinate (etretinate arm was not included);
-
Shehzad 2004: PUVA (psoralen and ultraviolet A) therapy versus methotrexate (methotrexate only was included);
-
Gottlieb 2011; Strober 2011: briakinumab versus etanercept versus placebo (briakinumab arm was not included);
-
Gisondi 2008: etanercept versus acitretin versus etanercept plus acitretin (etanercept plus acitretin arm was not included);
-
Al‐Hamamy 2014: narrowband ultraviolet B phototherapy plus methotrexate versus narrowband ultraviolet B alone and methotrexate alone (arm with methotrexate alone was not included).
Thaçi 2002 compared two different dosages of ciclosporin (a fixed dosage of 200 mg/day and a dosage corresponding to 2.5 mg/kg/day), and we were unable to classify the fixed dosage group either in the ciclosporin ≥ 3 mg/kg/day group nor in the ciclosporin < 3 mg/day group for the subgroup meta‐analysis.
Studies awaiting classification
We classified 14 trials reported in 18 references as studies awaiting classification. More details regarding the studies awaiting classification are available in Studies awaiting classification and Table 2.
Ongoing studies
We classified 34 trials as ongoing studies. More details are available in Characteristics of ongoing studies and Table 2. Most of the ongoing studies compare a biological treatments versus another biological treatment or versus placebo (n = 13 and n = 14, respectively). Three ongoing studies assess apremilast versus placebo, and four assess conventional systemic treatments versus conventional systemic treatments (n = 2) or placebo (n = 2).
Risk of bias in included studies
Figure 2 and Figure 3 summarise 'Risk of bias' assessments. Regarding the overall risk of bias across studies, 23 trials were at low risk of bias (Asahina 2016; Bachelez 2015; Blauvelt FEATURE, 2015; Blauvelt VOYAGE‐1, 2016; Cai 2016; Gordon UNCOVER‐1, 2016; Griffiths UNCOVER‐2, 2015; Griffiths UNCOVER‐3, 2015; Langley ERASURE, 2014; Langley FIXTURE, 2014; Leonardi 2012; Papp PHOENIX‐2, 2008; Papp 2012; Papp 2012a; Papp 2012b;Reich 2015;Reich 2012; Reich VOYAGE‐2, 2017; Rich 2013; Saurat CHAMPION, 2008; Thaci CLEAR, 2015; Vaclavkova 2014;Warren METOP, 2017). We categorised almost half of the studies (48/109) as at high risk of bias. Among the high‐risk group, five studies had only one high risk of bias domain with all the other dimensions at low risk (Bissonnette 2015; Lebwohl 2003; Papp 2013a; Papp OPT Pivotal‐1, 2015; Reich LIBERATE, 2017). We categorised the remaining 38 studies as unclear risk of bias because we assessed one or more criteria as unclear. Among the unclear 'Risk of bias' group, 11 studies had only one unclear risk of bias with all the other dimensions at low risk (Bagel 2012; Krueger 2016; Leonardi 2003; Leonardi PHOENIX‐1, 2008; Menter EXPRESS‐II, 2007; Menter REVEAL, 2008; Papp AMAGINE‐1, 2016; Paul JUNCTURE, 2015; Paul ESTEEM‐2, 2015; Reich EXPRESS, 2005; Tyring 2006). Further details of these assessments are available in the 'Risk of bias' table corresponding to each trial in the Characteristics of included studies.
'Risk of bias' summary: review authors' judgements about each 'Risk of bias' item for each included study
'Risk of bias' graph: review authors' judgements about each 'Risk of bias' item presented as percentages across all included studies
Allocation
The method of sequence generation was not described at all, or was at best unclear, in 48 trials. The remaining studies (n = 61) described the method used to generate the allocation sequence in sufficient detail; therefore, we judged this domain as low risk of bias for these studies. For allocation concealment, the majority of studies (n = 56) received a judgement of unclear risk of bias for this domain because of the absence of reporting the method used to guarantee concealment. We considered the risk low for the 53 remaining trials.
Blinding
Blinding of participants and personnel was achieved in 74 studies, whereas 30 studies were at high risk of performance bias. The remaining five studies were at unclear risk of performance bias. Blinding of outcome assessment was reported clearly in only 74 of the 109 included studies, whereas 21 studies were at high risk of detection bias. The risk of detection bias was unclear in the remaining 14 studies.
Incomplete outcome data
In more than half of the trials (69/109), incomplete outcome data appeared to have been adequately addressed, and any missing outcome data were reasonably well balanced across intervention groups, with similar reasons for missing data across the groups. However, in 13 studies the reporting of missing outcome data was largely inadequate because of one or more of the following reasons: the high number of withdrawn participants, an imbalance between groups in the number of withdrawn participants, an imbalance in reasons for missing outcomes, or no intention‐to‐treat (ITT) analysis provided. In 27 studies, this domain was as at unclear risk of bias due to one or more of the following reasons: the numbers of participants, reasons, or missing data methods were not reported.
Selective reporting
We considered 14 trials at high risk of selective outcome reporting because results for outcomes detailed in the methods section were not reported in the results section (Akcali 2014; Engst 1994; Hunter 1963; Jacobe 2008; Lebwohl 2003; Lebwohl AMAGINE‐2, 2015; Lebwohl AMAGINE‐3, 2015; Mrowietz BRIDGE, 2016; Nakagawa 2016;Papp 2013b; Papp 2005; Reich LIBERATE, 2017; Shehzad 2004; Yan 2011). In all, we considered 49 studies to be at low risk of bias for this domain as outcome details in the trial register and in the methods section were reported in the results section. For other trials (n = 46), we considered the risk of bias as unclear because we did not find these trials in any register.
Other potential sources of bias
As detailed in the Methods section, we did not fulfil the 'other risk of bias' item as we did not highlight particular circumstances leading to other risk of bias from particular trial designs, contamination between the experimental and control groups, and particular clinical settings.
Effects of interventions
See: summary of findings Table for the main comparison The summary of findings for the main comparison provides overall estimates of treatment effects compared with placebo and the certainty of the available evidence for the two primary outcomes (PASI 90 and serious adverse effects during the induction phase), obtained through network meta‐analysis.
Seven trials provided no usable or retrievable data and did not contribute further to the results of this review (Akcali 2014; Chladek 2005; Engst 1994; Lowe 1991; Piskin 2003; Olsen 1989; Shehzad 2004; see Table 2). The main reason we could not use their data was that these studies addressed none of our outcomes. Fifteen studies, involving 1113 participants (2.8% of the participants in this review), had a co‐intervention and did not contribute further to the results of this review as we could not assess the specific intervention effect (Al‐Hamamy 2014; Asawanonda 2006; Bissonnette 2013; Gottlieb 2012; Gurel 2015; Jacobe 2008; Lowe 1991; Mahajan 2010; Ruzicka 1990; Saurat 1988; Shehzad 2004; Sommerburg 1993; Tanew 1991; Van Bezooijen 2016; Yilmaz 2002). Twenty‐six studies had an outcome assessment before 12 weeks (Akcali 2014; Chaudhari 2001; Goldfarb 1988; Gottlieb 2004; Hunter 1963; Menter EXPRESS‐II, 2007; Meffert 1997; Olsen 1989; Reich EXPRESS, 2005; Ruzicka 1990; Sommerburg 1993; Saurat 1988; Torii 2010; Yang 2012), or later than 16 weeks (Al‐Hamamy 2014; Asahina 2016; Asawanonda 2006; Bissonnette 2013; Bissonnette 2015; de Vries PIECE, 2016; Engst 1994; Gisondi 2008; Gottlieb 2012; Ortonne 2013; Strohal PRISTINE, 2013; Van Bezooijen 2016).
In total, 35 studies, involving 4433 participants, were not included in the classical or network meta‐analysis. The interventions of the 35 studies particularly concerned the following:
-
infliximab (n = 7) (Chaudhari 2001; de Vries PIECE, 2016; Gottlieb 2004; Menter EXPRESS‐II, 2007; Reich EXPRESS, 2005; Torii 2010; Yang 2012)
-
acitretin (n = 8) (Akcali 2014; Goldfarb 1988; Gisondi 2008; Gurel 2015; Lowe 1991; Ruzicka 1990; Saurat 1988; Sommerburg 1993)
-
methotrexate (n = 5) (Asawanonda 2006; Al‐Hamamy 2014; Gottlieb 2012; Mahajan 2010; Shehzad 2004)
-
ciclosporin (n = 1) (Meffert 1997)
We included a total of 74 studies, involving 35,454 participants (88.9% participants of this review), in the network meta‐analysis for at least one of the outcomes.
Figure 4 and Figure 5 show the network diagrams for all of the outcomes included in the review. The size of the nodes is proportional to the total number of participants allocated to each class level (Figure 4)/drug level (Figure 5) intervention, and the thickness of the lines is proportional to the number of trials evaluating each direct comparison.
Network plot for all the outcomes at class‐level
The size of the nodes is proportional to the total number of participants allocated to each intervention and the thickness of the lines proportional to the number of studies evaluating each direct comparison.
AE: adverse events; PASI: Psoriasis Area and Severity Index; PGA: Physician Global Assessment; QoL: quality of life; SAE: serious adverse events
Network plot for all the outcomes at drug‐level
The size of the nodes is proportional to the total number of participants allocated to each intervention and the thickness of the lines proportional to the number of studies evaluating each direct comparison.
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
AE: adverse events; PASI: Psoriasis Area and Severity Index; PGA: Physician Global Assessment; QoL: quality of life; SAE: serious adverse events
Figure 6 shows the network meta‐analysis estimates of all of the outcomes for each comparisons at class level.
Relative effects of the class‐level intervention as estimated from the network meta‐analysis model
Drugs are reported in order of primary benefit ranking. Each cell contains the risk ratio (RR) (for dichotomous outcomes: PASI 90, serious adverse events, PASI 75, PGA 0/1, adverse events) or the standardised mean difference (SMD) (for the quality‐of‐life outcome), plus the 95% confidence interval, of the class level in the respective column versus the class level in the respective row. RRs larger than 1 for the lower triangle and smaller than 1 (or SMDs smaller than zero) for the upper triangle favour the treatment on the left. Significant results are bolded and underscored.
AE: adverse events; PASI: Psoriasis Area and Severity Index; PGA: Physician's Global Assessment; QoL: quality of life; SAE: serious adverse events
Figure 7, Figure 8, and Figure 9 show the network meta‐analysis estimates of all the outcomes for each comparison at drug level.
Relative effects of the intervention as estimated from the network meta‐analysis model for Psoriasis Area and Severity Index (PASI) 90 and serious adverse events (SAEs)
Drugs are reported in order of primary benefit ranking. Each cell contains the risk ratio (RR) and 95% confidence interval for the two primary outcomes (PASI 90 and SAEs) of the intervention in the respective column versus the class level in the respective row. RRs larger than 1 for the lower triangle and smaller than 1 for the upper triangle favour the treatment on the left. Significant results are highlighted in grey.
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; τ (Tau): estimated heterogeneity standard deviation parameter; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
Relative effects of the intervention as estimated from the network meta‐analysis model for Psoriasis Area and Severity Index (PASI 75) and adverse events (AEs)
Drugs are reported in order of primary benefit ranking. Each cell contains the Risk Ratio (RR) and 95% confidence interval for the two secondary outcomes (PASI 75 and adverse events) of the intervention in the respective column versus the comparator in the respective row. RRs larger than 1 for the lower triangle and smaller than 1 for the upper triangle favour the treatment on the left. Significant results are are highlighted in grey.
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; τ (Tau): estimated heterogeneity standard deviation parameter; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
Relative effects of the intervention as estimated from the network meta‐analysis model for Physician's Global Assessment (PGA 0/1) and quality of life (QoL)
Drugs are reported in order of primary benefit ranking. Each cell contains the risk ratio (RR) and 95% confidence interval (PGA 0/1) or standardized mean difference (quality of life) of the intervention in the respective column versus the comparator in the respective row. RRs larger than 1 for the lower triangle and smaller than 1 (or SMDs smaller than zero) for the upper triangle favour the treatment on the left. Significant results are are highlighted in grey.
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; τ (Tau): estimated heterogeneity standard deviation parameter; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
Figure 10 and Figure 11 show all of the relative effects from the network meta‐analyses against placebo with their 95% confidence and prediction intervals at class and drug level.
Interval plot. Network meta‐analysis estimates of class‐level versus placebo for all the outcomes
AE: adverse events; CI: confidence interval; PGA: Physician Global Assessment; PrI: predictive interval; QoL: Specific quality of life scale; RR: risk ratio; SAE: serious adverse events; SMD: standardised mean difference
Interval plot. Network meta‐analysis estimates of the interventions versus placebo for all the outcomes
AE: adverse events; CI: confidence interval; PGA: Physician Global Assessment; PrI: predictive interval; QoL: Specific quality of life scale; RR: risk ratio; SAE: serious adverse events; SMD: standardised mean difference
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
Figure 12 shows a two‐dimensional ranking plot based on surface under the cumulative ranking curve (SUCRA) values for benefit (PASI 90) and acceptability (serious adverse events) at class and drug level. The different colours represent different groups of interventions considering their performance on both outcomes simultaneously. Interventions belonging to the same group were assumed to have a similar performance when the two primary outcomes were considered jointly (Chaimani 2013).
Ranking plot. Ranking plot representing simultaneously the efficacy (x axis, PASI 90) and the acceptability (y axis, serious adverse events) of all the interventions (class and drug levels) for patients with moderate‐to‐severe psoriasis. Optimal treatment should be characterised by both high efficacy and acceptability and should be in the right upper corner of this graph.
The different colours represent different groups of interventions considering their performance on both outcomes simultaneously. Interventions belonging to the same group are assumed having a similar performance when the two primary outcomes are considered jointly
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
PASI: Psoriasis Area and Severity Index; SAE: serious adverse events; SUCRA: surface under the cumulative ranking curve
Figure 13 and Figure 14 show the ranking for all the outcomes at class and drug level, respectively.
Ranking for all the outcomes at class level
AE: adverse events; PASI: Psoriasis Area and Severity Index; PGA: Physician Global Assessment; QoL: quality of life; SAE: serious adverse events
Ranking for all the outcomes at drug level
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
AE: adverse events; PASI: Psoriasis Area and Severity Index; PGA: Physician Global Assessment; QoL: quality of life; SAE: serious adverse events
1. Primary outcomes
1.1 The proportion of participants who achieved clear or almost clear skin, e.g. PASI 90
DIRECT EVIDENCE
We report treatment estimates for pair‐wise meta‐analyses at class (see Figure 15) and drug level in Analysis 1.1; Analysis 1.2; Analysis 1.3; Analysis 1.4; Analysis 1.5; Analysis 1.6; Analysis 1.7; Analysis 1.8; Analysis 1.9; Analysis 1.10; and Analysis 1.11, respectively.
PASI 90: direct summary effects for comparisons including at least two studies at class level
AIL12/23: anti‐IL12/23; AIL17: anti‐IL17; AIL23: anti‐IL23, ATA: anti‐TNF alpha; CSA: conventional systemic agents; OB: other biologics; PBO: placebo; SM: small molecules
CI: confidence interval; PASI: Psoriasis Area and Severity Index; RR: risk ratio
In terms of reaching PASI 90, anti‐IL17 treatments (secukinumab, ixekizumab, and brodalumab) were more effective than placebo (risk ratio at class level (RR) 30.02, 95% confidence interval (CI) 21.14 to 42.64). These findings were also confirmed for anti‐IL23 (guselkumab and tildrakizumab) (class‐level RR 25.36, 95% CI 14.80 to 43.43); ustekinumab (RR 22.00, 95% CI 14.90 to 32.48); anti‐TNF alpha (etanercept, adalimumab, and certolizumab) (class‐level RR 12.97, 95% CI 9.89 to 17.02); and small molecules (apremilast, tofacitinib, and ponesimod) (class‐level RR 6.40, 95% CI 4.48 to 9.13). Both infliximab and adalimumab were more effective than methotrexate (respectively: RR 2.86, 95% CI 2.15 to 3.80; and RR 3.73, 95% CI 2.25 to 6.19). Ustekinumab, secukinumab, and ixekizumab were more effective than etanercept; secukinumab and brodalumab were more effective than ustekinumab; and guselkumab was more effective than adalimumab. No significant difference was observed between etanercept and tofacitinib or apremilast in terms of this outcome (reaching PASI 90).
NETWORK META‐ANALYSES
The PASI 90 outcome was available in 58 trials, involving 31,176 participants (87.9% of the participants in the meta‐analysis). This outcome was reported in two other trials (Nugteren‐Huying 1990; Sandhu 2003); however, the number of randomised participants was not available. These trials were added in the complete case analyses. This outcome was also reported in three other trials (Dogra 2012; Dogra 2013; Mrowietz SCULPTURE, 2015), comparing different dosages of the same drug in each case. These trials were added to the sensitivity analysis at dose level. PASI 90 was not reported for the remaining nine trials, and we were not able to obtain missing information from the trial authors (Table 2). Thirty‐nine trials, involving 16,888 participants, were placebo‐controlled trials; seven studies, involving 2048 participants, were head‐to‐head comparisons; and 12 studies, involving 12,240 participants, had both a placebo and at least two active treatments arms.
See Figure 4; Figure 5; Figure 6; Figure 7; Figure 10; Figure 11; Figure 13; and Figure 14.
Table 3 summarises the main results of both the direct and indirect evidence and the network meta‐analysis for PASI 90 at 12 to 16 weeks. The summary relative effects from the network meta‐analysis are presented in league tables for both class‐level (Figure 6) and drug‐level (Figure 7) analyses.
| Network meta‐analysis | Direct evidence | Indirect evidence | |||||||
| Comparisons* | RR | LCI | UCI | RR | LCI | UCI | RR | LCI | UCI |
| FAEs vs placebo | 4.09 | 1.88 | 8.88 | 4.47 | 1.97 | 10.14 | 1.86 | 0.16 | 21.16 |
| Methotrexate vs placebo | 3.91 | 2.16 | 7.08 | 1.53 | 0.66 | 3.53 | 17.16 | 5.69 | 51.75 |
| Adalimumab vs placebo | 14.87 | 10.45 | 21.14 | 14.42 | 10.08 | 20.64 | 108.8 | 2.24 | 5287.86 |
| Etanercept vs placebo | 10.79 | 8.47 | 13.73 | 10.62 | 7.52 | 15.01 | 11.21 | 7.26 | 17.32 |
| Ustekinumab vs placebo | 19.91 | 15.11 | 26.23 | 22.7 | 15.46 | 33.34 | 17.91 | 12.71 | 25.24 |
| Secukinumab vs placebo | 26.55 | 20.32 | 34.69 | 24.53 | 14.93 | 40.32 | 28.25 | 19.1 | 41.78 |
| Ixekizumab vs placebo | 32.45 | 23.61 | 44.60 | 39.46 | 20.64 | 75.44 | 24.51 | 10.05 | 59.77 |
| Brodalumab vs placebo | 25.45 | 18.74 | 34.57 | 26.58 | 16.65 | 42.41 | 23.74 | 10.09 | 55.86 |
| Apremilast vs placebo | 7.66 | 4.30 | 13.66 | 6.72 | 3.07 | 14.69 | 10.83 | 2.43 | 48.31 |
| Tofacitinib vs placebo | 8.50 | 6.23 | 11.60 | 6.3 | 4.14 | 9.56 | 17.91 | 8.3 | 38.62 |
| Guselkumab vs placebo | 21.03 | 14.56 | 30.38 | 26.1 | 14.71 | 46.3 | 12.7 | 4.28 | 37.69 |
| Methotrexate vs FAEs | 0.96 | 0.38 | 2.44 | 2 | 0.19 | 21.03 | 0.83 | 0.3 | 2.32 |
| Alefacept vs methotrexate | 1.12 | 0.42 | 3.02 | 1.12 | 0.42 | 3.02 | |||
| Ciclosporin vs methotrexate | 1.02 | 0.60 | 1.73 | 1.02 | 0.6 | 1.73 | |||
| Infliximab vs methotrexate | 2.86 | 2.06 | 3.97 | 2.86 | 2.06 | 3.97 | |||
| Adalimumab vs methotrexate | 3.80 | 2.26 | 6.39 | 3.35 | 2.02 | 5.57 | 13.2 | 3.4 | 51.32 |
| Etanercept vs acitretin | 11.00 | 0.63 | 191.47 | 11 | 0.63 | 191.47 | |||
| Guselkumab vs adalimumab | 1.41 | 1.21 | 1.65 | 1.4 | 1.18 | 1.66 | 2.88 | 0.68 | 12.21 |
| Ustekinumab vs etanercept | 1.85 | 1.50 | 2.27 | 1.8 | 1.27 | 2.55 | 1.95 | 1.37 | 2.77 |
| Secukinumab vs etanercept | 2.46 | 2.01 | 3.02 | 2.33 | 1.66 | 3.28 | 2.62 | 1.82 | 3.77 |
| Ixekizumab vs etanercept | 3.01 | 2.46 | 3.68 | 2.93 | 2.44 | 3.53 | 5.73 | 2.07 | 15.85 |
| Apremilast vs etanercept | 0.71 | 0.40 | 1.25 | 0.72 | 0.36 | 1.45 | 0.69 | 0.26 | 1.81 |
| Tofacitinib vs etanercept | 0.79 | 0.59 | 1.06 | 0.88 | 0.73 | 1.08 | 0.49 | 0.3 | 0.81 |
| Secukinumab vs ustekinumab | 1.33 | 1.11 | 1.61 | 1.38 | 1.03 | 1.84 | 1.19 | 0.79 | 1.81 |
| Brodalumab vs ustekinumab | 1.28 | 1.10 | 1.48 | 1.27 | 1.1 | 1.46 | 1.64 | 0.69 | 3.89 |
FAES: fumaric acid esters; LCI: low confidence interval; RR: risk ratio; UCI: upper confidence interval; vs: versus,
*The comparisons listed in this table were included in at least one direct‐evidence analysis.
All of the interventions appeared superior to placebo in terms of reaching PASI 90. Anti‐IL17 treatment was associated with a higher chance of reaching PASI 90 compared to all of the interventions: versus anti‐IL12/23 (risk ratio (RR) 1.33, 95% confidence interval (CI) 1.19 to 1.49); versus anti‐IL23 (RR 1.86, 95% CI 1.54 to 2.26); versus anti‐TNF alpha (RR 2.66, 95% CI 2.34 to 3.03); versus small molecules (RR 3.52, 95% CI 2.65 to 4.66); versus other biologics (RR 6.44, 95% CI 2.44 to 17.04); versus conventional systemic agents (RR 8.15, 95% CI 6.07 to 10.93) (Figure 6). In terms of reaching PASI 90, all of the biologic interventions (anti‐IL17, anti‐IL12/23, anti‐IL23, anti‐TNF alpha) appeared significantly superior to the small molecule class of treatments and the conventional systemic class of treatments. Small molecules were associated with a higher chance of reaching PASI 90 compared to conventional systemic agents (RR 2.31, 95% CI 1.63 to 3.28).
Results of comparisons between each of the drugs are available in Figure 7. There was no significant difference between the three anti‐IL17 (brodalumab, ixekizumab, and secukinumab) and the two anti‐IL23 (tildrakizumab and guselkumab) monoclonal antibodies in terms of reaching PASI 90. All of the anti‐IL17 drugs (brodalumab, ixekizumab, and secukinumab) and guselkumab (an anti‐IL23) were significantly more effective than three anti‐TNF alpha agents: infliximab, adalimumab, and etanercept. The direct comparison regarding certolizumab and tildrakizumab only included one trial each, so the interpretation of the results regarding certolizumab and tildrakizumab was difficult (related to wide CIs). Ustekinumab was superior to etanercept (RR 1.85, 95% CI 1.50 to 2.27). No significant difference was shown between the anti‐TNF alpha drugs. Tofacitinib was significantly superior to methotrexate (RR 2.17, 95% CI 1.13 to 4.20), and no significant difference was shown between apremilast and the conventional drugs (versus acitretin: RR 7.81, 95% CI 0.42 to 143.83; versus fumaric acid: RR 1.87, 95% CI 0.71 to 4.93; versus ciclosporin: RR 1.92, 95% CI 0.72 to 5.12; versus methotrexate: RR 1.96, 95% CI 0.85 to 4.50).
Ranking class‐level analysis (Figure 10; Figure 13; Table 4)
| Class‐level | SUCRA | Rank | SUCRA | Rank | SUCRA | Rank | SUCRA | Rank | SUCRA | Rank | SUCRA | Rank |
| Anti‐IL12/23 | 85.7 | 2 | 53.9 | 3 | 85.0 | 2 | 57.0 | 3 | 83.8 | 2 | 75.7 | 3 |
| Anti‐IL17 | 100.0 | 1 | 21.0 | 8 | 99.6 | 1 | 14.1 | 6 | 99.9 | 1 | 95.4 | 1 |
| Anti‐IL23 | 71.3 | 3 | 39.6 | 5 | 72.2 | 3 | 78.7 | 2 | 73.1 | 3 | 83.4 | 2 |
| Anti‐TNF alpha | 56.4 | 4 | 39.2 | 6 | 57.4 | 4 | 47.5 | 5 | 57.5 | 4 | 58.4 | 4 |
| Other biologics | 26.3 | 6 | 68.2 | 2 | 17.0 | 7 | _ | _ | 16.6 | 7 | 15.5 | 7 |
| Small molecules | 41.5 | 5 | 45.4 | 4 | 42.7 | 5 | 7.9 | 7 | 42.0 | 5 | 40.4 | 5 |
| Conventional systemic treatments | 18.7 | 7 | 94.8 | 1 | 26.0 | 6 | 50.8 | 4 | 27.1 | 6 | 30.8 | 6 |
| Placebo | 0 | 8 | 38.0 | 7 | 0 | 8 | 94.0 | 1 | 0 | 8 | 0.4 | 8 |
AE: adverse events; FAEs: fumaric acid esters; PGA: Physician Global Assessment; QoL: Specific quality of life scale; SAE: serious adverse events
Ranking analysis for PASI 90 performed with SUCRA strongly suggested that anti‐IL17 was the best treatment at class level (versus placebo: RR 30.81, 95% confidence interval (CI) 25.10 to 37.82; SUCRA = 100; high‐certainty evidence), followed by anti‐IL12/23 (versus placebo: RR 23.16, 95% CI 18.70 to 28.68; SUCRA = 85.7; high‐certainty of evidence), anti‐IL23 (versus placebo: RR 16.53, 95% CI 13.16 to 20.75; SUCRA = 71.3; moderate‐certainty evidence), then anti‐TNF alpha (versus placebo: RR 11.58, 95% CI 9.55 to 14.03; SUCRA = 56.4; moderate‐certainty evidence). The heterogeneity τ for this network overall was 0.09, which we considered low heterogeneity.
Ranking drug‐level analysis (Figure 11; Figure 14; Table 5)
| Drug | SUCRA | Rank | SUCRA | Rank | SUCRA | Rank | SUCRA | Rank | SUCRA | Rank | SUCRA | Rank |
| Acitretin | 9.9 | 19 | 46.9 | 9 | 26.0 | 15 | ‐ | ‐ | ‐ | ‐ | ‐ | ‐ |
| Adalimumab | 63.1 | 8 | 40.4 | 14 | 60.2 | 9 | 70.1 | 5 | 56.9 | 8 | 57.6 | 7 |
| Alefacept | 25.3 | 15 | 62.6 | 5 | 12.6 | 18 | ‐ | ‐ | 13.1 | 18 | 15.9 | 13 |
| Apremilast | 39.7 | 13 | 54.7 | 7 | 33.2 | 14 | 14.3 | 16 | 27.9 | 14 | 28.6 | 10 |
| Brodalumab | 84.3 | 3 | 39.8 | 15 | 82.1 | 3 | 46.4 | 9 | 84.0 | 5 | 52.3 | 8 |
| Certolizumab | 75.7 | 5 | 70.9 | 3 | 71.6 | 6 | 78.0 | 4 | 90.1 | 1 | ‐ | ‐ |
| Ciclosporin | 21.3 | 17 | 78.2 | 2 | 33.2 | 13 | 36.8 | 12 | 24.0 | 16 | ‐ | ‐ |
| Etanercept | 52.6 | 11 | 43.6 | 11 | 57.7 | 10 | 45.9 | 10 | 51.7 | 10 | 67.6 | 5 |
| FAEs | 21.9 | 16 | 57.7 | 6 | 11.1 | 19 | 17.8 | 15 | 15.4 | 17 | ‐ | ‐ |
| Guselkumab | 77.0 | 4 | 42.6 | 12 | 71.6 | 7 | 78.2 | 3 | 67.5 | 7 | 84.3 | 2 |
| Infliximab | 53.2 | 10 | 64.4 | 4 | 48.0 | 11 | 40.1 | 11 | 52.4 | 9 | ‐ | ‐ |
| Itolizumab | 56.0 | 9 | ‐ | ‐ | 71.6 | 8 | ‐ | ‐ | 29.4 | 13 | 16.0 | 12 |
| Ixekizumab | 94.3 | 1 | 33.7 | 17 | 91.8 | 1 | 18.1 | 14 | 85.9 | 3 | 99.2 | 1 |
| Methotrexate | 20.2 | 18 | 90.7 | 1 | 21.3 | 16 | 68.4 | 6 | 24.9 | 15 | 31.5 | 9 |
| Placebo | 2.9 | 20 | 42.0 | 13 | 0.0 | 20 | 88.0 | 1 | 0.3 | 19 | 1.2 | 14 |
| Ponesimod | 37.3 | 14 | 18.1 | 19 | 21.3 | 17 | 14.0 | 17 | 48.7 | 11 | 28.1 | 11 |
| Secukinumab | 86.5 | 2 | 29.9 | 18 | 86.7 | 2 | 36.3 | 13 | 84.4 | 4 | ‐ | ‐ |
| Tildrakizumab | 63.6 | 7 | 37.8 | 16 | 78.3 | 4 | 86.1 | 2 | 86.3 | 2 | 74.9 | 4 |
| Tofacitinib | 42.5 | 12 | 44.0 | 10 | 46.2 | 12 | 47.3 | 8 | 36.6 | 12 | 65.1 | 6 |
| Ustekinumab | 72.6 | 6 | 52.0 | 8 | 75.2 | 5 | 64.3 | 7 | 70.4 | 6 | 77.4 | 3 |
AE: adverse events; FAEs: fumaric acid esters; PASI: Psoriasis Area and Severity Index; PGA: Physician Global Assessment; QoL: specific quality of life scale; SAE: serious adverse events; SUCRA: Surface Under the Cumulative Ranking
Ranking analysis for PASI 90 performed with SUCRA strongly suggested that ixekizumab was the best treatment at drug level (versus placebo: RR 32.45, 95% CI 23.61 to 44.60; SUCRA = 94.3; high‐certainty evidence), followed by secukinumab (versus placebo: RR 26.55, 95% CI 20.32 to 34.69; SUCRA = 86.5; high certainty of evidence), brodalumab (versus placebo: RR 25.45, 95% CI 18.74 to 34.57; SUCRA = 84.3; moderate‐certainty evidence), guselkumab (versus placebo: RR 21.03, 95% CI 14.56 to 30.38; SUCRA = 77; moderate‐certainty evidence), certolizumab (versus placebo: RR 24.58, 95% CI 3.46 to 174.73; SUCRA = 75.7; moderate‐certainty evidence), then ustekinumab (versus placebo: RR 19.91, 95% CI 15.11 to 26.23; SUCRA = 72.6; high‐certainty evidence). The heterogeneity τ for this network overall was 0.09, which we considered low heterogeneity.
1.2 The proportion of participants with serious adverse effects
DIRECT EVIDENCE
We report treatment estimates for pair‐wise meta‐analyses at class (Figure 16) and drug level in Analysis 2.1; Analysis 2.2; Analysis 2.3; Analysis 2.4; Analysis 2.5; Analysis 2.6; Analysis 2.7; Analysis 2.8; Analysis 2.9; and Analysis 2.10, respectively. We provide details of the serious adverse effects in Table 6 (number of serious infections, number of malignancies, number of major adverse cardiac events per arm at class level).
Serious adverse effects: direct summary effects for comparisons including at least two studies at class level
AIL12/23: anti‐IL12/23; AIL17: anti‐IL17; AIL23: anti‐IL23, ATA: anti‐TNF alpha; CSA: conventional systemic agents; OB: other biologics; PBO: placebo; SM: small molecules
CI: confidence interval; RR: risk ratio; SAE: serious adverse events
| Number of randomised participants | Number of serious adverse events | Number of serious infections | Number of malignancies | Number of MACE | ||||||
| Drug | Placebo | Drug | Placebo | Drug | Placebo | Drug | Placebo | Drug | Placebo | |
| Conventional systemic agents | 767 | 220 | 19 | 10 | 0 | 0 | 0 | 0 | 0 | 0 |
| Anti‐TNF | 4508 | 2640 | 85 | 44 | 21 | 9 | 20 | 7 | 6 | 2 |
| Anti‐IL12/23 | 2547 | 1607 | 38 | 23 | 7 | 5 | 4 | 1 | 4 | 3 |
| Anti‐IL17 | 7551 | 2533 | 149 | 36 | 47 | 7 | 21 | 2 | 19 | 3 |
| Anti‐IL23 | 1347 | 510 | 23 | 7 | 4 | 1 | 0 | 0 | 1 | 0 |
| Other biologics | 509 | 227 | 15 | 10 | ‐ | ‐ | 2 | 1 | ‐ | ‐ |
| Small molecules | 3920 | 1280 | 89 | 28 | 15 | 5 | 14 | 0 | 5 | 1 |
MACE: Major adverse cardiac events
No significant differences were observed between methotrexate, FAEs, etanercept, adalimumab, certolizumab, ustekinumab, secukinumab, ixekizumab, brodalumab, guselkumab, tildrakizumab, alefacept, apremilast, tofacitinib, ponesimod, and placebo in terms of the number of paticipants with serious adverse effects (SAEs). The risk of SAEs was significantly higher for participants on infliximab compared to methotrexate (RR 2.41, 95% CI 1.04 to 5.59).
There were zero SAEs in the following trials: Fallah Arani 2011 (comparing methotrexate with FAEs); Flytström 2008 (comparing ciclosporin with methotrexate); and Heydendael 2003 (comparing ciclosporin with methotrexate).
NETWORK META‐ANALYSES
The SAE outcome was available in 60 trials, involving 30,898 participants (87.1% of the participants in the meta‐analysis). This outcome was reported in one other trial (Sterry PRESTA, 2010); however, the number of randomised participants was not available. This trial was added to the complete‐cases analyses. This outcome was also reported in two other trials (Laburte 1994; Mrowietz SCULPTURE, 2015), comparing different dosages of the same drug in each case. These studies were added to the sensitivity analysis at dose level. SAEs were not reported for the 11 remaining trials, and we were not able to obtain missing information from the trial authors (Table 2). Forty‐two trials, involving 16,822 participants, were placebo‐controlled trials; six, involving 1836 participants, were head‐to‐head comparisons, and 12, involving 12,240 participants, had both a placebo and at least two active treatments arms.
See Figure 4; Figure 5; Figure 6; Figure 7; Figure 10; Figure 11; Figure 13; and Figure 14.
Table 7 summarised the main results of both direct and indirect evidences and the network meta‐analysis for SAEs at 12 to 16 weeks. We present the summary relative effects from the network meta‐analysis in league tables for both class‐level (Figure 6) and drug‐level (Figure 7) analyses. No significant difference was found between all of the interventions and the placebo regarding the risk of SAE. Two significant associations were found: anti‐IL17 agents and anti‐TNF alpha agents had a higher risk of SAE compared with conventional systemic agents (RR 2.31, 95%CI 1.20 to 4.48; RR 2.06, 95% CI 1.13 to 3.75, respectively). The results are available in Figure 7 for comparison between each drug. Ixekizumab, secukinumab, and infliximab were at higher risk of SAE than methotrexate (RR 4.86, 95%CI 1.03 to 22.88; RR 5.14, 95% CI 1.09 to 24.29; RR 2.41, 95% CI 1.04 to 5.59, respectively).
| Network meta‐analysis | Direct evidence | Indirect evidence | |||||||
| Comparisons* | RR | LCI | UCI | RR | LCI | UCI | RR | LCI | UCI |
| FAEs vs placebo | 0.77 | 0.30 | 1.99 | 0.83 | 0.31 | 2.21 | 0.19 | 0 | 12.57 |
| Methotrexate vs placebo | 0.23 | 0.05 | 0.99 | 0.16 | 0.03 | 0.86 | 0.68 | 0.04 | 11.67 |
| Adalimumab vs placebo | 1.02 | 0.61 | 1.73 | 1.05 | 0.62 | 1.78 | 0.07 | 0 | 26.92 |
| Etanercept vs placebo | 0.99 | 0.65 | 1.51 | 1.09 | 0.65 | 1.84 | 0.76 | 0.31 | 1.89 |
| Ustekinumab vs placebo | 0.89 | 0.57 | 1.39 | 0.74 | 0.44 | 1.26 | 1.36 | 0.61 | 2.99 |
| Secukinumab vs placebo | 1.19 | 0.69 | 2.03 | 1.61 | 0.78 | 3.33 | 0.75 | 0.3 | 1.87 |
| Ixekizumab vs placebo | 1.12 | 0.66 | 1.90 | 1.16 | 0.62 | 2.16 | 0.97 | 0.18 | 5.12 |
| Brodalumab vs placebo | 1.04 | 0.62 | 1.73 | 0.92 | 0.53 | 1.62 | 2.77 | 0.38 | 20.28 |
| Apremilast vs placebo | 0.84 | 0.47 | 1.52 | 0.78 | 0.42 | 1.44 | 4.33 | 0.09 | 201.27 |
| Tofacitinib vs placebo | 0.98 | 0.55 | 1.76 | 1.05 | 0.53 | 2.06 | 0.67 | 0.08 | 5.35 |
| Guselkumab vs placebo | 1.00 | 0.49 | 2.04 | 1.21 | 0.51 | 2.85 | 0.52 | 0.08 | 3.41 |
| Methotrexate vs FAEs | 0.30 | 0.06 | 1.59 | 1 | 0.02 | 48.83 | 0.23 | 0.04 | 1.45 |
| Ciclosporin vs methotrexate | 0.98 | 0.06 | 15.38 | 0.98 | 0.06 | 15.38 | ‐ | ‐ | ‐ |
| Infliximab vs methotrexate | 2.41 | 1.04 | 5.59 | 2.41 | 1.04 | 5.59 | ‐ | ‐ | ‐ |
| Adalimumab vs methotrexate | 4.43 | 0.99 | 19.81 | 2.24 | 0.21 | 23.56 | 6.68 | 1.04 | 42.76 |
| Etanercept vs acitretin | 1.00 | 0.02 | 48.82 | 1 | 0.02 | 48.83 | ‐ | ‐ | ‐ |
| Guselkumab vs adalimumab | 0.98 | 0.51 | 1.88 | 0.89 | 0.44 | 1.79 | 2.07 | 0.26 | 16.45 |
| Ustekinumab vs etanercept | 0.90 | 0.52 | 1.57 | 1.25 | 0.38 | 4.11 | 0.83 | 0.44 | 1.54 |
| Secukinumab vs etanercept | 1.20 | 0.66 | 2.19 | 1.17 | 0.45 | 3.04 | 1.22 | 0.56 | 2.65 |
| Ixekizumab vs etanercept | 1.14 | 0.66 | 1.94 | 1.02 | 0.53 | 1.95 | 1.47 | 0.53 | 4.09 |
| Apremilast vs etanercept | 0.85 | 0.42 | 1.72 | 2.69 | 0.41 | 17.5 | 0.7 | 0.33 | 1.5 |
| Tofacitinib vs etanercept | 0.99 | 0.53 | 1.87 | 0.87 | 0.35 | 2.19 | 1.12 | 0.47 | 2.7 |
| Secukinumab vs ustekinumab | 1.33 | 0.74 | 2.38 | 1.01 | 0.42 | 2.39 | 1.68 | 0.77 | 3.68 |
| Brodalumab vs ustekinumab | 1.16 | 0.64 | 2.11 | 1.32 | 0.59 | 2.98 | 0.95 | 0.33 | 2.71 |
FAES: fumaric acid esters; LCI: low confidence interval; RR: risk ratio; UCI: upper confidence interval
*The comparisons listed in this table were included in at least one direct‐evidence analysis.
Ranking class‐level analysis (Figure 10; Figure 13; Table 4)
Ranking analysis for SAE performed with SUCRA strongly suggested that conventional systemic treatment was associated with the best safety profile at class level in terms of serious adverse events (versus placebo: RR 0.48, 95% CI 0.26 to 0.88; SUCRA = 94.8), followed by other biologics (versus placebo: RR 0.72, 95% CI 0.34 to 1.55; SUCRA = 68.2), anti‐IL12/23 (versus placebo: RR 0.89, 95% CI 0.58 to 1.37; SUCRA = 53.9), and then small molecules (versus placebo: RR 0.95, 95% CI 0.63 to 1.42; SUCRA = 45.4). The heterogeneity τ for this network overall was 0, which we considered low heterogeneity.
Ranking drug‐level analysis (Figure 11; Figure 14; Table 5)
Ranking analysis for SAE performed with SUCRA strongly suggested that methotrexate was associated with the best safety profile at drug level in terms of serious adverse events (versus placebo: RR 0.23, 95% CI 0.05 to 0.99; SUCRA = 90.7; moderate‐certainty evidence), followed by ciclosporin (versus placebo: RR 0.23, 95% CI 0.01 to 5.10; SUCRA = 78.2; very low‐certainty evidence), certolizumab (versus placebo: RR 0.49, 95% CI 0.10 to 2.36; SUCRA = 70.9; moderate‐certainty evidence), infliximab (versus placebo: RR 0.56, 95% CI 0.10 to 3.00; SUCRA = 64.4; very low‐certainty evidence), alefacept (versus placebo: RR 0.72, 95% CI 0.34 to 1.55; SUCRA = 62.6; low‐certainty evidence), and then the FAEs (versus placebo: RR 0.77, 95% CI 0.30 to 1.99; SUCRA = 57.7; very low‐certainty evidence). The heterogeneity τ for this network overall was 0, which we considered low heterogeneity.
Placebo had a worse ranking for SAE than conventional systemic agents, other biologics, anti‐IL12/23, and small molecules (see Table 5). Nevertheless, analyses on serious adverse events were based on a very low number of events and were reduced to the short time frame of the trials. Table 6 gives details of the types of SAE; major adverse cardiac events, serious infections, or malignancies were reported in both placebo and intervention groups.
1.3 Relationship between PASI 90 and serious adverse events
See Figure 12.
These findings for both efficacy (PASI 90) and acceptability (serious adverse events) were combined together in a bivariate ranking plot, where serious adverse events was transformed into acceptability by using the inverse values of the corresponding RRs so that higher values indicate higher acceptability (due to lower SAE): accordingly, the ideal treatment (highest performance = best efficacy + best acceptability) should appear in the upper right corner of the plot.
At class level, the highly effective treatments had serious adverse events. However, the anti‐IL12/23 treatment group was the class with the better compromise between efficacy and acceptability.
At drug level, ustekinumab, certolizumab, and infliximab might be the overall best treatments considering both outcomes jointly. This result has to be considered with cautioun for certolizumab and infliximab as only one trial was available for this drug.
2. Secondary outcomes
2.1 Mean difference of quality of life measured by a specific scale
DIRECT EVIDENCE
We report treatment estimates for pair‐wise meta‐analyses at class (Figure 17) and drug level in Analysis 3.1; Analysis 3.2; Analysis 3.3; Analysis 3.4; Analysis 3.5; Analysis 3.6; Analysis 3.7; Analysis 3.8; Analysis 3.9; Analysis 3.10; and Analysis 3.11 respectively.
Specific quality of life scale: direct summary effects for comparisons including at least two studies at class level
AIL12/23: anti‐IL12/23; AIL17: anti‐IL17; AIL23: anti‐IL23, ATA: anti‐TNF alpha; CSA: conventional systemic agents; OB: other biologics; PBO: placebo; SM: small molecules
CI: confidence interval; QoL: specific quality of life scale; SMD: standardised mean difference
NETWORK META‐ANALYSES
The quality of life outcome was available in 39 trials, involving 21,745 participants (61.3% of the participants in this review). This outcome was reported in one other trial (Krueger 2002); however, the number of randomised participants was not available. This trial were added to the complete case analyses. This outcome was also reported in another trial (Mrowietz SCULPTURE, 2015), comparing different dosages of the same drug. This trial, Mrowietz SCULPTURE, 2015, was added in the sensitivity analyses at dose level. The quality of life outcome was not reported for the 35 remaining trials, and we were not able to obtain missing information from the trial authors (Table 2). Twenty‐eight trials, involving 13,040 participants, were placebo‐controlled trials; two, involving 1080 participants, were head‐to‐head comparisons; and nine, involving 7625 participants, had both a placebo and at least two active treatments arms.
See Figure 4; Figure 5; Figure 6; Figure 9; Figure 10; Figure 11; Figure 13; and Figure 14.
We present the summary relative effects from the network meta‐analysis in league tables for both class‐level (Figure 6) and drug‐level (Figure 9) analyses. All of the interventions appeared superior to placebo in terms of showing significant improvement on a quality of life scale. Anti‐IL17, anti‐IL23, and anti‐IL12/23 were associated with a higher chance of improving quality of life compared to small molecules and conventional systemic agents (Figure 6). These differences were statistically significant for all of the classes. No significant difference was shown between the different biological agents except for anti‐IL17 and anti‐TNF alpha (anti‐IL17 was more favourable than anti‐TNF alpha). No significant differences were shown between the small molecules and the conventional agents. Results of comparisons between each of the drugs are available in Figure 9.
Ranking class‐level analysis (Figure 10; Figure 13; Table 4)
Ranking analysis for quality of life performed with SUCRA strongly suggested that anti‐IL17 was the best treatment at class level (versus placebo: standardised mean difference (SMD) ‐1.44, 95% confidence interval (CI) ‐1.68 to ‐1.19; SUCRA = 95.4), followed by anti‐IL23 (versus placebo: SMD ‐1.30 95% CI ‐1.60 to ‐0.99; SUCRA = 83.4), anti‐IL12/23 (versus placebo: SMD ‐1.21 95% CI ‐1.45 to ‐0.96; SUCRA = 75.7), then anti‐TNF alpha (versus placebo: SMD ‐1.03 95% CI ‐1.18 to ‐0.88 SUCRA = 58.4). The heterogeneity τ for this network overall was 0.27, which we considered moderate heterogeneity.
Ranking drug‐level analysis (Figure 11; Figure 14Table 5)
Ranking analysis for quality of life performed with SUCRA strongly suggested that ixekizumab was the best treatment at drug level (versus placebo: SMD ‐1.68 95% CI ‐1.93 to ‐1.43; SUCRA = 99.2), followed by guselkumab (versus placebo: SMD ‐1.31 95% CI ‐1.61 to ‐1.01; SUCRA = 84.3), ustekinumab (versus placebo: SMD ‐1.21 95% CI ‐1.42 to ‐1.00; SUCRA = 77.4), tildrakizumab (versus placebo: SMD ‐1.23 95% CI ‐1.77 to ‐0.68; SUCRA = 74.9), then etanercept (versus placebo: SMD ‐1.11 95% CI ‐1.29 to ‐0.93; SUCRA = 67.6). The heterogeneity τ for this network overall was 0.22, which we considered low to moderate heterogeneity. Moreover, six interventions (acitretin, certolizumab, ciclosporin, fumaric acid, infliximab, secukinumab) were not included in the ranking at drug level, due to no available data.
In total, available information on quality of life was poorly reported and lacking for a third of the interventions, so has to be considered with cautious.
2.2 Proportion of participants who achieve a Physician Global Assessment (PGA) value at 0 or 1
DIRECT EVIDENCE
We report treatment estimates for pair‐wise meta‐analyses at class (Figure 18) and drug level in Analysis 4.1; Analysis 4.2; Analysis 4.3; Analysis 4.4; Analysis 4.5; Analysis 4.6; Analysis 4.7; Analysis 4.8; Analysis 4.9; Analysis 4.10; and Analysis 4.11, respectively.
Physician Global Assessment 0/1: direct summary effects for comparisons including at least two studies at class‐level
AIL12/23: anti‐IL12/23; AIL17: anti‐IL17; AIL23: anti‐IL23, ATA: anti‐TNF alpha; CSA: conventional systemic agents; OB: other biologics; PBO: placebo; SM: small molecules
AE: adverse events; CI: confidence interval; PGA: Physician Global Assessment; RR: risk ratio
NETWORK META‐ANALYSES
The PGA 0/1 outcome was available in 56 trials, involving 31,030 participants (87.5% of the participants in this review). This outcome was reported in four other studies (Krueger 2002; Nugteren‐Huying 1990; Sandhu 2003; Sterry PRESTA, 2010); however, the number of randomised participants was not available. These trials were added to the complete case analyses. This outcome was also reported in another trial (Mrowietz SCULPTURE, 2015), comparing different dosages of the same drug. These trials were added in the sensitivity analysis at dose level. PGA 0/1 was not reported for the 13 remaining trials, and we were not able to obtain missing information from the trial authors (Table 2). Forty trials, involving 16,946 participants, were placebo‐controlled trials; four, involving 1844 participants, were head‐to‐head comparisons; and 12, involving 12,240 participants, had both a placebo and at least two active treatments arms.
See Figure 4; Figure 5; Figure 6; Figure 9Figure 10; Figure 11; Figure 13; and Figure 14.
We presented the summary relative effects as estimated from the network meta‐analysis in league tables at class level (Figure 6) and drug level (Figure 9). All of the interventions appeared superior to placebo in terms of reaching PGA 0/1, and anti‐IL17 monoclonal antibodies were associated with a better chance in terms of this outcome compared to the other drug classes (Figure 6). These differences were statistically significant. All of the interventions (anti‐IL17, anti‐IL23, anti‐IL12/23, anti‐TNF alpha) appeared significantly superior to the small molecule class of treatments and the conventional systemic class of treatments. No significant difference was found between small molecule and conventional systemic agents. Results of comparisons between each of the drugs are available in Figure 9.
Ranking class‐level analysis (Figure 10; Figure 13; Table 4)
Ranking analysis for PGA 0/1 performed with SUCRA strongly suggested that anti‐IL17 was the best treatment at class level (versus placebo: RR 15.85, 95% confidence interval (CI) 13.08 to 19.20; SUCRA = 99.9), followed by anti‐IL12/23 (versus placebo: RR 11.80, 95% CI 9.67 to 14.39; SUCRA = 83.8), anti‐IL23 (versus placebo: RR 9.93, 95% CI 7.58 to 13.02; SUCRA = 73.1), then anti‐TNF alpha (versus placebo: RR 7.82, 95% CI 6.66 to 9.17; SUCRA = 57.5). The heterogeneity τ for this network overall was 0.21, which we considered low to moderate heterogeneity.
Ranking drug‐level analysis (Figure 11; Figure 14; Table 5)
Ranking analysis for PGA 0/1 performed with SUCRA strongly suggested that certolizumab was the best treatment at drug level (versus placebo: RR 35.88, 95% CI 4.86 to 265.07; SUCRA = 90.1), followed by tildrakizumab (versus placebo: RR 27.54, 95% CI 3.76 to 201.98; SUCRA = 86.3), ixekizumab (versus placebo: RR 16.11, 95% CI 11.72 to 22.17; SUCRA = 85.9), secukinumab (versus placebo: RR 15.46, 95% CI 11.19 to 21.37; SUCRA = 84.4), brodalumab (versus placebo: RR 15.31, 95% CI 10.84 to 21.63; SUCRA = 84), then ustekinumab (versus placebo: RR 11.52, 95% CI 9.17 to 14.4; SUCRA = 70.4). The heterogeneity τ for this network overall was 0.23, which we considered low to moderate heterogeneity.
2.3 Proportion of participants who achieve PASI 75
DIRECT EVIDENCE
We report treatment estimates for pair‐wise meta‐analyses at class (Figure 19) and drug level in Analysis 3.1; Analysis 3.2; Analysis 3.3; Analysis 3.4; Analysis 3.5; Analysis 3.6; Analysis 3.7; Analysis 3.8; Analysis 3.9; Analysis 3.10; and Analysis 3.11, respectively.
PASI 75: direct summary effects for comparisons including at least two studies at class level
AIL12/23: anti‐IL12/23; AIL17: anti‐IL17; AIL23: anti‐IL23, ATA: anti‐TNF alpha; CSA: conventional systemic agents; OB: other biologics; PBO: placebo; SM: small molecules
CI: confidence interval; PASI: Psoriasis Area and Severity Index; RR: risk ratio
NETWORK META‐ANALYSES
PASI 75 outcome was available in 64 trials, involving 32,518 participants (91.7% of the participants in this review). This outcome was reported in two other trials (Krueger 2002; Sterry PRESTA, 2010); however, the number of randomised participants was not available. These trials were added to the complete case analyses. This outcome was also reported in five other trials (Dogra 2012; Dogra 2013; Dubertret 1989; Laburte 1994; Mrowietz SCULPTURE, 2015), comparing different dosages of the same drug in each case. These trials were added in the sensitivity analysis at dose level. PASI 75 was not reported for the three remaining trials, and we were not able to obtain missing information from the trial authors (Table 2). Forty‐five trials, involving 18,330 participants, were placebo‐controlled trials; seven, involving 1948 participants, were head‐to‐head comparisons; and 12, involving 12,240 participants, had both a placebo and at least two active treatments arms.
See Figure 4; Figure 5; Figure 6; Figure 8; Figure 10; Figure 11; Figure 13; and Figure 14.
We present the summary relative effects from the network meta‐analysis in league tables for both class‐level (Figure 6) and drug‐level (Figure 8) analyses. All of the interventions appeared superior to placebo in terms of reaching PASI 75. The anti‐IL17 class of drugs was associated with a higher chance of reaching PASI 75 compared to the other classes (Figure 6). These differences were statistically significant for all of the classes. All of the interventions (anti‐IL17, anti‐IL23, anti‐IL12/23, anti‐TNF alpha) appeared significantly superior to the small molecule class and the conventional systemic class, and the small molecules appeared significantly superior to the conventional systemic agents. Results of comparisons between each of the drugs are available in Figure 8.
Ranking class‐level analysis (Figure 10; Figure 13; Table 4)
Ranking analysis for PASI 75 performed with SUCRA strongly suggested that anti‐IL17 was the best treatment at class level (versus placebo: RR 14.32, 95% CI 12.20 to 16.81; SUCRA = 99.6), followed by anti‐IL12/23 (versus placebo: RR 12.21, 95% CI 10.23 to 14.57; SUCRA = 85.0), anti‐IL23 (versus placebo: RR 10.07, 95% CI 8.03 to 12.63; SUCRA = 72.2), then anti‐TNF alpha (versus placebo: RR 8.23 95% CI 7.20 to 9.42; SUCRA = 57.4). The heterogeneity τ for this network overall was 0.16, which we considered low heterogeneity.
Ranking drug‐level analysis (Figure 11; Figure 14; Table 5)
Ranking analysis for PASI 75 performed with SUCRA strongly suggested that ixekizumab was the best treatment at drug level (versus placebo: RR 15.81, 95% CI 12.35 to 20.23; SUCRA = 91.8), followed by secukinumab (versus placebo: RR 14.16, 95% CI 11.12 to 18.03; SUCRA = 86.7), brodalumab (versus placebo: RR 13.04 95% CI 10.17 to 16.71; SUCRA = 82.1), tildrakizumab (versus placebo: RR 14.51, 95% CI 3.60 to 58.45; SUCRA = 78.3), then ustekinumab (versus placebo: RR 11.84, 95% CI 9.79 to 14.33; SUCRA = 75.2). The heterogeneity τ for this network overall was 0.16, which we considered low heterogeneity.
Focusing on efficacy outcomes (PASI 90, PASI 75, and PGA 0/1), the results were identical at class level (Figure 10) and very close at drug level (Figure 11).
2.4 The proportions of participants with adverse effects
DIRECT EVIDENCE
We report treatment estimates for pair‐wise meta‐analyses at class (Figure 20) and drug level in Analysis 6.1; Analysis 6.2; Analysis 6.3; Analysis 6.4; Analysis 6.5; Analysis 6.6; Analysis 6.7; Analysis 6.8; Analysis 6.9; and Analysis 6.10, respectively.
Adverse effects : direct summary effects for comparisons including at least two studies at class‐level
AIL12/23: anti‐IL12/23; AIL17: anti‐IL17; AIL23: anti‐IL23, ATA: anti‐TNF alpha; CSA: conventional systemic agents; OB: other biologics; PBO: placebo; SM: small molecules
AE: adverse events; CI: confidence interval; RR: risk ratio
NETWORK META‐ANALYSES
Adverse events (AEs) outcome was available in 54 trials, involving 29,699 participants (83.8% of the participants in this review). AEs were not reported for the 36 remaining trials, and we were not able to obtain missing information from the trial authors (Table 2). This outcome was also reported in another trial (Mrowietz SCULPTURE, 2015), comparing different dosages of the same drug. Mrowietz SCULPTURE, 2015 was added to the sensitivity analyse at dose level. Thirty‐seven trials, involving 15,683 participants, were placebo‐controlled trials; five, involving 1,776 participants, were head‐to‐head comparisons; and 12, involving 12,240 participants, had both a placebo and at least two active treatments arms.
See Figure 4; Figure 5; Figure 6; Figure 8; Figure 10; Figure 11; Figure 13; and Figure 14.
We present the summary relative effects from the network meta‐analysis in league tables for both class‐level (Figure 6) and drug‐level (Figure 8) analyses. All of the interventions had a more significant risk of AEs compared to placebo. Significant associations were found: anti‐IL17 had a higher risk of AE compared with all the other interventions. Results of comparisons between each of the drugs are available in Figure 8.
Ranking class‐level analysis (Figure 10; Figure 13; Table 4)
Ranking analysis for AEs performed with SUCRA strongly suggested that placebo was associated with the best safety profile regarding all the adverse events (SUCRA 94.0). Anti‐IL23 was the best treatment at class level (versus placebo: RR 1.03, 95% CI 0.93 to 1.13; SUCRA = 78.7), followed by anti‐IL12/23 (versus placebo: RR 1.07, 95% CI 1.01 to 1.14; SUCRA = 57.0), then conventional systemic treatment (versus placebo: RR 1.08, 95% CI 0.99 to 1.17; SUCRA = 50.8). The heterogeneity τ for this network overall was 0.05, which we considered low heterogeneity.
Ranking drug‐level analysis (Figure 11; Figure 14; Table 5)
Ranking analysis for AE performed with SUCRA strongly suggested that placebo was associated with the best safety profile regarding all the adverse events (SUCRA = 88), then tildrakizumab (versus placebo: RR 0.95, 95% CI 0.76 to 1.19; SUCRA = 86.1), followed by guselkumab (versus placebo: RR 1.02, 95% CI 0.80 to 1.13; SUCRA = 78.2) and certolizumab (versus placebo: RR 1.00 95% CI 0.80 to 1.23; SUCRA = 78). The heterogeneity τ for this network overall was 0.04, which we considered low heterogeneity.
2.5. Participants with at least 1 relapse in the maintenance phase (between 52 to 104 weeks)
There were no data available for the maintenance phase.
3. Assessment of heterogeneity and inconsistency
We did not identify important heterogeneity neither in direct meta‐analyses nor in network meta‐analysis. The common outcome‐specified network heterogeneity and the prediction intervals suggested the presence of low heterogeneity for all outcomes except for quality of life, which appeared to have moderate heterogeneity. We investigated differences in heterogeneity between class‐ and drug‐level analysis, and we also investigated differences in heterogeneity between primary and sensitivity analyses for the primary outcomes (see 4. subgroup and sensitivity analyses). The results were very closed.
The distribution of some participant characteristics (age, sex ratio, weight, severity of psoriasis) did not give an indication of important differences in these characteristics across comparisons (see Figure 21; Figure 22).
Distributions of age and sex ratio of participants across comparisons
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
Distributions of weight of participants and PASI score at baseline across comparisons
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
At class‐level analysis, the global test for inconsistency was not significant for all of the outcomes except for PASI 75 (data not shown). At drug‐level analysis, the global test for inconsistency was not significant for all of the outcomes but only marginally non‐significant for PASI 90. Results of a global test for inconsistency, at drug level, are detailed in Figure 23 and Figure 24 for PASI 90 and SAEs, respectively. The loop‐specific and side‐splitting approaches indicated a handful of loops and comparisons with statistically significant inconsistency (Figure 25; Figure 26). This apparent inconsistency does not generally exceed however the expected level of inconsistency that has been suggested by empirical evidence (Veroniki 2013), which is about 10% of the total number of loops.
Side‐splitting approach and design‐by‐treatment interaction model for inconsistency for Psoriasis Area and Severity Index (PASI) 90
Treatment codes: A = PBO, B = FUM, C = MTX, D = ACI, E = ALEFACEPT, F = CICLO, G = IFX, H = ADA, I = ETA, J = USK, K = SECU, L = IXE, M = BRODA, N = CERTO, O = APRE, P = TOFA, Q = GUSEL, R = TILDRA, S = PONE, T = ITO
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
Side‐splitting approach and design‐by‐treatment interaction model for inconsistency for serious adverse events (SAEs)
Treatment codes: A = PBO, B = FUM, C = MTX, D = ACI, E = ALEFACEPT, F = CICLO, G = IFX, H = ADA, I = ETA, J = USK, K = SECU, L = IXE, M = BRODA, N = CERTO, O = APRE, P = TOFA, Q = GUSEL, R = TILDRA, S = PONE, T = ITO
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
Inconsistency plots for all the outcomes at class‐level
Inconsistency factor (IF) is calculated as the risk ratio (RR)/standardised mean difference (SMD) for direct evidence over the RR/SMD for indirect evidence in the loop with its 95% confidence interval (CI). IF value close to 0 indicates the absence of evidence for disagreement between direct and indirect evidence.
AIL12/23: anti‐IL12/23; AIL17: anti‐IL17; AIL23: anti‐IL23, ATA: anti‐TNF alpha; CSA: conventional systemic agents; OB: other biologics; PBO: placebo; SM: small molecules
Inconsistency plots for all the outcomes at drug level
Inconsistency factor (IF) is calculated as the risk ratio (RR)/standardised mean difference (SMD) for direct evidence over the RR/SMD for indirect evidence in the loop with its 95% confidence interval (CI). IF value close to 0 indicates the absence of evidence for disagreement between direct and indirect evidence.
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
Possible explanation of this apparent inconsistency could be the differences between the previous treatment allowed across these trials: for example, participants enrolled in the Saurat CHAMPION, 2008 trial (adalimumab versus methotrexate versus placebo) were naïve to methotrexate and TNF alpha antagonists whereas participants enrolled in the Menter REVEAL, 2008 trial (adalimumab versus placebo) could have received previous systemic treatment including methotrexate.
4. Subgroup and sensitivity analyses
We had enough data for none of the aforementioned characteristics that may act as effect modifiers and therefore we were not able to run subgroup analyses and meta‐regressions to investigate their potential effect on the results.
Results of the sensitivity analyses involving the following were similar to those of the main analysis for the two primary outcomes:
-
excluding studies with less than 50 participants (Figure 27) (the heterogeneity τ for this subgroup network was 0.08 for PASI 90 and 0 for SAEs, which we considered low heterogeneity);
Figure 27
Sensitivity analyses ‐ Interval plot. Network meta‐analysis results for primary outcomes (PASI 90 and serious adverse events) for trials with at least 50 participants.
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
CI: confidence interval; PASI: Psoriasis Area and Severity Index; RR: risk ratio; SAE: serious adverse events
-
completers (Figure 28) (the heterogeneity τ for this subgroup network was 0.09 for PASI 90 and 0 for SAEs, which we considered low heterogeneity);
Figure 28
Sensitivity analyses ‐ Interval plot. Network meta‐analysis results for primary outcomes (PASI 90 and serious adverse events) for the completers.
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
CI: confidence interval; PASI: Psoriasis Area and Severity Index; RR: risk ratio; SAE: serious adverse events
-
analyses at dose level (Figure 29) (the heterogeneity τ for this subgroup network was 0.10 for PASI 90 and 0 for SAEs, which we considered low heterogeneity); and
Figure 29
Sensitivity analyses ‐ Interval plot. Network meta‐analysis results for primary outcomes (PASI 90 and serious adverse events) for all the interventions depending on the doses
MTX ≥ 15/MTX other: methotrexate ≥ 15 mg per week/methotrexate < 15 mg per week; ALEFACEPT: alefacept all dosages; CICLO High: ciclosporin ≥ 3 mg/kg/day; ACI ≥ 35: acitretin ≥ 35 mg per day; FUM: fumaric acid esters all dosages; APRE 30: apremilast 30 mg twice daily; PONE 40: ponesimod 40 mg per day; TOFA 20: tofacitinib 20 mg per day; ETA 25/ETA 50: etanercept 25 mg twice a week/etanercept 50 mg twice a week; IFX: infliximab 5 mg/kg week O, 2, 4 every 6 weeks; ADA: adalimumab 80 mg Week 0, 40 mg Week 1 then 40 mg every other week; CERTO 200/400: certolizumab all dosages; USK 45: ustekinumab 45 mg; SECU 300/SECU other: secukinumab 300 mg every injection/secukinumab other dosages; IXE 200/IXE other: ixekizumab 200 mg per injection/ixekizumab other dosages; TILDRA 100/200: tildrakizumab all dosages; GUSEL 100: guselkumab 100 mg per injection; BRODA 210: brodalumab 210 mg per injection
CI: confidence interval; PASI: Psoriasis Area and Severity Index; RR: risk ratio; SAE: serious adverse events
-
excluding studies at high risk of bias (Figure 30) (the heterogeneity τ for this subgroup network was 0.12 for PASI 90 and 0 for SAEs, which we considered low heterogeneity).
Figure 30
Sensitivity analyses ‐ Interval plot. Network meta‐analysis results for primary outcomes (PASI 90 and serious adverse events) for all the interventions excluding studies at high risk of bias.
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
CI: confidence interval; PASI: Psoriasis Area and Severity Index; RR: risk ratio; SAE: serious adverse events
5. Reporting bias
The comparison‐adjusted funnel plots generally appeared symmetrical, and only the graph for quality of life presented some evidence of small‐study effects which might be caused by selective outcome reporting (Figure 31).
Funnel plot for network meta‐analysis of all the outcomes
AE: adverse event; lnRR: Mean effect size; PASI: Psoriasis Area and Severity Index; QoL: Specific quality of life scale; RR: Risk ratio; SAE: serious adverse events; SMD: standardised mean difference
6. Grading of the evidence
We graded the evidence for the two primary outcomes, PASI 90 and serious adverse events, for all of the network intervention estimates according to the approach proposed by Salanti 2014. We considered five domains: study limitations (by first evaluating the risk of bias of each direct estimate (Figure 2) and then by integrating these judgements with the contribution of each direct estimate to the network estimates (Figure 32)), consistency of effect, imprecision, indirectness, and publication bias.
Study bias distribution for each primary outcome (PASI 90 and serious adverse events)
The following graphs show how much information (i.e. the percentage contribution of each direct comparison in the network estimates) comes from low (green), unclear/moderate (yellow) and high (red) risk of bias studies. Here we have all drugs versus placebo as it is difficult to have all comparisons due to space limitations. To evaluate the direct comparisons we used the mean level of bias of the included studies in each comparison.
We used the web application CINeMA (CINeMA 2017).
The codes of the treatments are A = Placebo, B = Fumaric acid esters, C = Methotrexate, D = Acitretin, E = Alefacept, F = Ciclosporin, G = Infliximab, H = Adalimumab, I = Etanercept, J = Ustekinumab, K = Secukinumab, L = Ixekizumab, M = Brodalumab, N = Certolizumab, O = Apremilast, P = Tofacitinib, Q = Guselkumab, R = Tildrakizumab, S = Ponesimod, T = Itolizumab
For PASI 90, we judged the confidence in the treatment estimate to be high for ixekizumab, secukinumab, and ustekinumab; moderate for brodalumab, guselkumab (reasons for downgrading: studies limitations), certolizumab (imprecision), adalimumab (inconsistency), etanercept (inconsistency), apremilast (study limitations), ponesimod (imprecision), and methotrexate (inconsistency); and low or very low for all of the other treatments (tildrakizumab, itolizumab, tofacitinib, infliximab, acitretin, ciclosporin, fumaric acid esters, alefacept). More detail on the reasons for downgrading are available in summary of findings Table for the main comparison.
For serious adverse events, we judged the confidence in the treatment estimate to be low to very low for almost all of the treatment, except methotrexate, certolizumab, tofacitinib, etanercept, adalimumab, ixekizumab, secukinumab, and ponesimod, which we assessed as moderate certainty (downgrading linked to imprecision for all "moderate certainty" drugs). More detail on the reasons for downgrading are available in summary of findings Table for the main comparison.
Discussion
Summary of main results
Our systematic review and meta‐analysis compared all drugs and drugs undergoing phase II/III trials used for moderate to severe psoriasis in 2017 except a new anti‐IL23 molecule (BI 655066, risankizumab).
In total, this review included 109 studies, involving 39,882 randomised adult participants, which assessed outcomes during the induction phase (less than 24 weeks after randomisation). In total, 55 trials were multiarm. Seventy‐three trials compared systematic treatment against placebo, 25 were head to head trials, and 11 had both active comparator and placebo. Fifteen trials had a co‐intervention mainly phototherapy. Finally, 79 studies declared pharmaceutical company funding, and 21 studies did not report the source of funding.
We included 74 studies (without co‐intervention and with a timing of outcome assessment from 12 to 16 weeks after randomisation (classed as induction therapy)), involving 35,454 participants (88.9% participants of this review), in the network meta‐analysis. Conventional systemic treatments, the oldest class‐level treatment (acitretin, ciclosporin, fumaric acid esters, methotrexate); anti‐TNF alpha treatments (etanercept, infliximab, adalimumab, certolizumab); an anti‐IL12/23 treatment (ustekinumab); and anti‐IL17 treatments (secukinumab, ixekizumab, brodalumab) have all been approved for psoriasis except certolizumab. And except for apremilast and alefacept, small molecule drugs (tofacitinib, ponesimod), anti‐IL23 treatments (guselkumab and tildrakizumab), and other biologics (itolizumab) had not been approved for psoriasis at the time we conducted our analyses.
All of the assessed interventions appeared superior to placebo in terms of reaching Psoriasis Area and Severity Index (PASI) 90.
At class level, network meta‐analysis showed that the biologics anti‐IL17, followed by anti‐IL12/23, anti‐IL23, and anti‐TNF alpha outperformed the small molecules and the conventional systemic agents in terms of reaching PASI 90 measured at the twelfth to the sixteenth week of treatment after randomisation, with small molecules producing a better outcome than conventional systemic agents.
The most effective drug for reaching PASI 90 when compared to placebo was ixekizumab (high‐certainty evidence), followed by secukinumab (high‐certainty evidence), brodalumab (moderate‐certainty evidence), guselkumab (moderate‐certainty evidence), certolizumab (moderate‐certainty evidence), then ustekinumab (high‐certainty evidence) (see summary of findings Table for the main comparison).
At drug‐level, all of the anti‐IL17 agents and guselkumab (an anti‐IL23 drug) were significantly more effective in reaching PASI 90 than three anti‐TNF alpha agents (infliximab, adalimumab, and etanercept, but not certolizumab), and ustekinumab was superior to etanercept. No statistically significant difference was shown between infliximab, adalimumab, and etanercept. Only one trial assessed the efficacy of infliximab in this network; thus, the results involving infliximab have to be interpreted with caution. Tofacitinib was significantly superior to methotrexate, and no clear difference was shown between any of the other small molecules versus conventional treatments. The results were almost the same for the other efficacy outcome PASI 75.
No significant difference was found between all of the interventions and the placebo regarding the risk of serious adverse effects (SAEs). The relative ranking for SAEs strongly suggested that methotrexate was associated with the best safety profile regarding all the SAEs (moderate‐certainty evidence), followed by ciclosporin (very low‐certainty evidence), certolizumab (moderate‐certainty evidence), infliximab (very low‐certainty evidence), alefacept (low‐certainty evidence), then fumaric acid esters (FAEs) (very low‐certainty evidence). Major adverse cardiac events, serious infections, or malignancies (see Table 6) were reported in both placebo and intervention groups.
Information on quality of life was often poorly reported and was absent for a third of the interventions.
Finally, considering both efficacy (PASI 90 outcome) and acceptability (SAE outcome), highly effective treatments had also more SAE than the other treatments, and ustekinumab, infliximab, and certolizumab appeared to be the better compromise between efficacy and acceptability (bearing in mind the limitations that affect interpretation of the SAE results, such as the very low number of events on which the results were based, with just over half of the treatment estimates being based on low to very low certainty evidence (the rest moderate)).
Overall completeness and applicability of evidence
We were able to draw some conclusions on the effectiveness (and ranking) of the systemic treatment options for moderate to severe chronic plaque psoriasis during the induction phase. Long‐term efficacy and safety data are lacking. Specific details are listed below.
Participants
Participants in the included studies had a mean age of 44 years and had moderate to severe psoriasis with an overall mean PASI score at baseline of 20 (range: 9.5 to 39). This young age and the high level of disease severity may not be typical of patients seen in daily clinical practice especially for patients who need a first‐line systemic treatment. In addition, patients selected for randomised controlled trials (RCTs) generally have few major comorbidities. Almost all studies including one biological arm excluded patients with history of infectious diseases or malignancies and signs of severe renal, cardiac, hepatic, demyelinating, or other disorders. This may impact the generalisibility of these results for clinical practice. However, some participants characteristics (such as being overweight, imbalanced sex ratio in favour of males, presence of metabolic syndrome) were reflective of a moderate to severe psoriasis population, comparable to literature data (Wolkenstein 2009).
Interventions
Evidence on 19 treatments included in this review was derived from 74 trials (searched for up to December 2016). We included all interventions irrespective of the dose. Thus, we increased the number of available RCTs per intervention and had more power to assess SAEs and adverse events (AEs). The number of studies was still low for the following interventions: certolizumab, tildrakizumab, itolizumab, infliximab, ponesimod, acitretin, ciclosporin, alefacept, fumaric acid, and methotrexate, meaning we must be cautious of the conclusions drawn for these drugs. In terms of efficacy, the results from the subgroup analysis using a standard dose for each intervention was similar for PASI 90 (and SAE) compared to the main analyses, making us confident with the results of the main analysis.
For drugs just approved or not yet approved for psoriasis, ongoing studies are still investigating guselkumab, tildrakizumab, a third anti‐IL23 (risankizumab, which will be included in the next update of this review), certolizumab, tofacitinib, and itolizumab (Characteristics of ongoing studies). Ponesimod development in psoriasis is most uncertain and should be excluded from the next update of this review.
Comparisons
The majority of the studies included in the review were placebo‐controlled (around 70%) as were the identified ongoing studies. Once the benefit of a treatment has been established against placebo using high‐quality evidence, only head‐to‐head trials would be helpful to provide physicians with efficacy estimates between the different biologics based on stronger evidence than indirect comparisons.
Outcomes
Many of the trials included in this review provided evidence for the proportion of participants who reached PASI 90, PASI 75, or Physician Global Assessment (PGA) 0/1 or who experienced SAE or AE. On the other hand, patient‐reported outcome (PRO) data were scanty and poorly reported. Moreover, the heterogeneity of the scales used for PRO in psoriasis trials required using the standardised mean difference in the network. So, from a clinical point of view, the interpretation of the results was difficult: a significant result for PRO between two drugs did not mean that the result was clinically useful for the patients.
Timing
All of the included trials assessed the efficacy of the different treatments during the induction treatment phase (less than 24 weeks, with evidence in the network meta‐analysis (NMA) assessed 12 to 16 weeks after randomisation). This is an unwelcome finding for a chronic disease. The trials were designed to detect differences in the severity of psoriasis in response to therapy over short periods of treatment and are often underpowered and of insufficient duration to detect rare or long‐term adverse events. Therefore, it is of interest to conduct studies taking into account the induction of remission but also the long‐term management (long‐term remission) and the long‐term safety of the drug. In order to provide long‐term information on the safety of the treatments included in this review, it will be necessary to also evaluate non‐randomised studies and postmarketing reports released from regulatory agencies.
Quality of the evidence
Overall, we judged the confidence in the treatment estimate for PASI 90 to be high or moderate for anti‐IL17 agents, anti‐IL12/23 agents, anti‐IL23 agents, anti‐TNF alpha agents (except infliximab), methotrexate, and apremilast. We judged the confidence in treatment estimate for PASI 90 as low or very low certainty for most of the comparisons involving conventional systemic agents (except for methotrexate), infliximab, other biologics, and tofacitinib; we downgraded the certainty of the evidence due to risk of bias and then either for inconsistency or imprecision. We judged the confidence in the treatment estimate for SAEs to be low to very low certainty for half of the treatment estimates, moderate for the others; we downgraded the certainty of the evidence due to imprecision and risk of bias.
Risk of bias
The risk of bias in included studies appeared to be globally moderate to low (Figure 2; Figure 3). However, some limitations should be discussed.
-
There was variation in how well the studies took measures to blind investigators and participants: a third of trials in this systematic review were considered at high or unclear risk of performance bias (35 out of 109). This is an important point to highlight as the outcomes used for assessing efficacy were subjective. However, the proportion of trials at high risk of blinding used in the network meta‐analyses decreased to 15% (13 out of 74).
-
The reporting of missing outcome data was largely inadequate in a few studies. Since we chose a likely scenario that any participant with missing outcome data did not experience clearance for the overall analyses, the risk of overestimating efficacy due to how we reported missing data was minimised.
-
Finally, a few trials were considered at high risk of selective outcome reporting. However, we chose a stringent definition of studies at high risk of selective outcome reporting: we considered reporting bias inadequate if one specified outcome in protocols was lacking in the main report. A large proportion of included trials did not report the PRO outcomes in the main report but only in slicing publications (see Included studies). We extracted outcomes of interest both in main and slicing publications, but this disadvantaged trials that did not report all of the specified outcomes in the main report.
Indirect comparison and network meta‐analyses as standard pair‐wise meta‐analyses provide "observational" evidence since the treatments being compared have not been randomised across studies. However, we considered carefully the assumption underpinning the validity of indirect comparisons to reassure a sufficiently coherent evidence base (Cipriani 2013). The limitations of this review are reflected by the GRADE evaluations.
Heterogeneity (i.e. variation in effect modifiers within comparisons) and inconsistency (imbalance in effect modifiers between comparisons)
No evidence of the presence of heterogeneity either in direct comparisons or in the entire networks was found except in relation to the quality of life outcome (poorly reported, few studies per comparisons). There was no global inconsistency for the two primary outcomes, and the global test for inconsistency was significant only for PASI 75 at the class‐level analysis. According to the local tests, for each outcome, a handful of loops and comparisons, which does not exceed the expected level of inconsistency from empirical evidence (Veroniki 2013), appeared to have important inconsistency. Thus, we downgraded the strength of evidence for inconsistency for methotrexate, adalimumab, etanercept, infliximab, and tofacitinib.
Imprecision
The number of studies was low for the following interventions (one to two studies per interventions): certolizumab, tildrakizumab, itolizumab, infliximab, ponesimod, acitretin, ciclosporin, alefacept, fumaric acid, and methotrexate. We downgraded the strength of evidence for imprecision for all of these interventions for the two primary outcomes.
Indirectness or transitivity assumption
We did not find any evidence that important variables, such as age, sex, weight, and duration and severity of psoriasis, varied across comparisons (see Characteristics of included studies and Figure 31 and Figure 32). However, several comparisons had only one or two studies, and the lack of data did not allow us to check the distributions of previous treatments across comparisons; thus, transitivity cannot be assessed statistically properly.
Several participant characteristics have changed in newer trials, such as participants' exclusion criteria. However, most of the included trials were conducted after 2000, minimising the variability across trial participant characteristics. The location of the trial could also create some differences between participants as the response of treatment could be related to genetic background (Chiu 2014). To further reassure the plausibility of the transitivity assumption, we only included in our analyses trials not involving co‐interventions. Moreover, the trials were also fairly similar in terms of outcome assessment (less than 24 weeks). As a consequence, we excluded from the meta‐analyses most of the trials assessing infliximab efficacy. Indeed, timing of efficacy outcome assessments was from 8 to 10 weeks for infliximab trials. However, as differences in response rates of biologics for the treatment of psoriasis have been reported in several meta‐analyses published to date mainly related to the primary endpoint times, we assumed the importance of a similar timing of outcome assessment between trials (Puig 2014). Thus, the possibility of intransitivity seems to be unlikely even if it could not be totally excluded.
Publication bias
We assessed publication bias considering the comprehensive search strategy we performed and the risk for publication bias in the specific field. The comparison‐adjusted funnel plot for all placebo‐controlled trials for all the outcomes did not indicate any evident risk of publication bias for the two primary outcomes.
Potential biases in the review process
We performed a wide search for trials, including five trials registers and databases of each company when available, and we searched the U.S. Food and Drug Administration and the European Medicines Agency (EMA) databases and abstract proceedings of seven congresses up to a maximum of 10 years. We did not approach pharmaceutical companies for additional data when their databases were not open access, and it is possible that additional data from this source could contribute to this review. The probability that we missed a trial is low considering our wide search and is supported by the absence of small‐study effects (testing by the comparison‐adjusted funnel plots). However, the fact that 14 studies have not yet been incorporated may be a source of potential bias.
Study selection, data extraction, and 'Risk of bias' assessments were done in duplicate and independently, and we reached consensus by discussing any discrepancies. Some published trial reports did not provide enough details to extract outcomes and adequately assess risk of bias, especially studies performed before 2000 (e.g. before the International Committee of Medical Journal Editors issued the requirement of trial registration for publication). However, we contacted the authors of the trials to request missing data, but we cannot avoid some biased assessment in the review process due to incomplete reporting of trial details, results, or both.
We had some departures from the protocol plans (see Differences between protocol and review). After protocol publication, we added five biological drugs either approved for psoriasis or for which there were ongoing phase 3 trials. We chose to keep only PASI 90 as the primary efficacy outcome and not a composite outcome including PASI and Physician Global Assessment (PGA). Indeed, PASI 90 and PGA do not reflect the same measures (see Agreements and disagreements with other studies or reviews). To minimise inconsistency, we assessed the primary outcome between the twelth and the sixteenth week rather than less than 24 weeks.
Agreements and disagreements with other studies or reviews
We searched in MEDLINE Ovid (from 1946) using the strategy "Psoriasis" AND "Meta‐analysis" for already published network meta‐analyses. Seven network meta‐analyses were systematically reviewed and have compared the short‐term efficacy of treatments for moderate to severe psoriasis (Gomez‐Garcia 2017; Gupta 2014; Jabbar‐Lopez 2017; Lin 2012; Reich 2012a; Schmitt 2014;Signorovitch 2015).
We compared our findings with the four most recent network meta‐analyses (Gomez‐Garcia 2017; Jabbar‐Lopez 2017; Schmitt 2014;Signorovitch 2015). Schmitt 2014 included 48 trials, involving 16,696 participants, assessing both conventional systemic (ciclosporin, methotrexate, acitretin, FAEs) and biologic treatments (infliximab, adalimumab, etanercept, alefacept, and ustekinumab). Signorovitch 2015 included 15 trials, involving 7388 participants, assessing only anti‐TNF alpha agents (infliximab, adalimumab, etanercept) and anti‐IL12/23 drugs (ustekinumab). Gomez‐Garcia 2017 included 27 trials, involving 10,629 participants, assessing three anti‐TNF alpha agents (infliximab, etanercept, and adalimumab), one anti‐IL12/23 agent (ustekinumab), and one anti‐IL17 agent (secukinumab). Jabbar‐Lopez 2017 included 41 trials, involving 20,561 participants, assessing the same drugs as Gomez‐Garcia 2017, plus ixekizumab (another anti‐IL17 agent) and methotrexate.
Thus, compared to previous reviews, we included more interventions and consequently more trials (n = 109) and participants (n = 39,882). Regarding the overlapping period, we also included more trials than the other meta‐analyses. Indeed, we performed a larger search in terms of the number of databases used, including trials registers and other resources (unpublished literature), irrespective of the date or language limitations.
Schmitt 2014 and Signorovitch 2015 chose PASI 75 as primary outcome during the induction phase (less than 16 weeks); however, data on PASI 90 were also available. Gomez‐Garcia 2017 presented both PASI 75 and PASI 90 results. Finally, Jabbar‐Lopez 2017 chose a composite outcome: PASI 90 or Physician Global Assessment (PGA) 1. We chose PASI 90 as our primary efficacy outcome because complete clearance seems the less subjective outcome and the most relevant regarding patient expectation in short‐term assessment (induction phase). The composite outcome used by Jabbar‐Lopez 2017 did not reflect complete or almost complete clearance. Indeed, PGA 1 is highly correlated to PASI 75 and not PASI 90, which could lead to a classification bias (Robinson 2012).
Jabbar‐Lopez 2017 presented their results using number needed to treat (NNT). Although NNT is an easily understandable and very useful measure for patients and clinicians, it can be misleading in a network meta‐analysis since it requires the assumption of a common average control group risk applying to all studies. This is a rather strong assumption particularly in networks involving also head‐to‐head studies without a control group as here.
Infliximab was the best drug in terms of reaching PASI 75 in the network meta‐analyses of Schmitt 2014 and Signorovitch 2015. Adalimumab and ustekinumab were more likely to reach PASI 75 than etanercept. These last results are partly confirmed by our review: ustekinumab was more effective at reaching both PASI 75 and 90 than etanercept; however, no significant difference was shown between adalimumab and etanercept, as in the most recent network meta‐analyses from Gomez‐Garcia et al (Gomez‐Garcia 2017) and Jabbar‐Lopez et al (Jabbar‐Lopez 2017). Infliximab was also the most effective drug in Gomez‐Garcia 2017, without significant difference between infliximab and secukinumab. Infliximab was ranked in third place after ixekizumab and secukinumab in Jabbar‐Lopez 2017, without significant difference between infliximab and secukinumab. Our findings did not find such efficacy for infliximab. To prevent inconsistency in our network meta‐analaysis, we chose to include trials assessing outcomes between 12 to 16 weeks. Thus, only one trial, Barker RESTORE‐1, 2011, which compared infliximab versus methotrexate, was taken into account for this intervention. Regarding the four previous network meta‐analyses, two did not assess inconsistency (Schmitt 2014 ; Signorovitch 2015), and two reported significant global and local inconsistency for PASI 75, which preclude interpretation of their results (Gomez‐Garcia 2017; Jabbar‐Lopez 2017).
'Risk of bias' summary: review authors' judgements about each 'Risk of bias' item for each included study
'Risk of bias' graph: review authors' judgements about each 'Risk of bias' item presented as percentages across all included studies
Network plot for all the outcomes at class‐level
The size of the nodes is proportional to the total number of participants allocated to each intervention and the thickness of the lines proportional to the number of studies evaluating each direct comparison.
AE: adverse events; PASI: Psoriasis Area and Severity Index; PGA: Physician Global Assessment; QoL: quality of life; SAE: serious adverse events
Network plot for all the outcomes at drug‐level
The size of the nodes is proportional to the total number of participants allocated to each intervention and the thickness of the lines proportional to the number of studies evaluating each direct comparison.
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
AE: adverse events; PASI: Psoriasis Area and Severity Index; PGA: Physician Global Assessment; QoL: quality of life; SAE: serious adverse events
Relative effects of the class‐level intervention as estimated from the network meta‐analysis model
Drugs are reported in order of primary benefit ranking. Each cell contains the risk ratio (RR) (for dichotomous outcomes: PASI 90, serious adverse events, PASI 75, PGA 0/1, adverse events) or the standardised mean difference (SMD) (for the quality‐of‐life outcome), plus the 95% confidence interval, of the class level in the respective column versus the class level in the respective row. RRs larger than 1 for the lower triangle and smaller than 1 (or SMDs smaller than zero) for the upper triangle favour the treatment on the left. Significant results are bolded and underscored.
AE: adverse events; PASI: Psoriasis Area and Severity Index; PGA: Physician's Global Assessment; QoL: quality of life; SAE: serious adverse events
Relative effects of the intervention as estimated from the network meta‐analysis model for Psoriasis Area and Severity Index (PASI) 90 and serious adverse events (SAEs)
Drugs are reported in order of primary benefit ranking. Each cell contains the risk ratio (RR) and 95% confidence interval for the two primary outcomes (PASI 90 and SAEs) of the intervention in the respective column versus the class level in the respective row. RRs larger than 1 for the lower triangle and smaller than 1 for the upper triangle favour the treatment on the left. Significant results are highlighted in grey.
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; τ (Tau): estimated heterogeneity standard deviation parameter; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
Relative effects of the intervention as estimated from the network meta‐analysis model for Psoriasis Area and Severity Index (PASI 75) and adverse events (AEs)
Drugs are reported in order of primary benefit ranking. Each cell contains the Risk Ratio (RR) and 95% confidence interval for the two secondary outcomes (PASI 75 and adverse events) of the intervention in the respective column versus the comparator in the respective row. RRs larger than 1 for the lower triangle and smaller than 1 for the upper triangle favour the treatment on the left. Significant results are are highlighted in grey.
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; τ (Tau): estimated heterogeneity standard deviation parameter; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
Relative effects of the intervention as estimated from the network meta‐analysis model for Physician's Global Assessment (PGA 0/1) and quality of life (QoL)
Drugs are reported in order of primary benefit ranking. Each cell contains the risk ratio (RR) and 95% confidence interval (PGA 0/1) or standardized mean difference (quality of life) of the intervention in the respective column versus the comparator in the respective row. RRs larger than 1 for the lower triangle and smaller than 1 (or SMDs smaller than zero) for the upper triangle favour the treatment on the left. Significant results are are highlighted in grey.
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; τ (Tau): estimated heterogeneity standard deviation parameter; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
Interval plot. Network meta‐analysis estimates of class‐level versus placebo for all the outcomes
AE: adverse events; CI: confidence interval; PGA: Physician Global Assessment; PrI: predictive interval; QoL: Specific quality of life scale; RR: risk ratio; SAE: serious adverse events; SMD: standardised mean difference
Interval plot. Network meta‐analysis estimates of the interventions versus placebo for all the outcomes
AE: adverse events; CI: confidence interval; PGA: Physician Global Assessment; PrI: predictive interval; QoL: Specific quality of life scale; RR: risk ratio; SAE: serious adverse events; SMD: standardised mean difference
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
Ranking plot. Ranking plot representing simultaneously the efficacy (x axis, PASI 90) and the acceptability (y axis, serious adverse events) of all the interventions (class and drug levels) for patients with moderate‐to‐severe psoriasis. Optimal treatment should be characterised by both high efficacy and acceptability and should be in the right upper corner of this graph.
The different colours represent different groups of interventions considering their performance on both outcomes simultaneously. Interventions belonging to the same group are assumed having a similar performance when the two primary outcomes are considered jointly
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
PASI: Psoriasis Area and Severity Index; SAE: serious adverse events; SUCRA: surface under the cumulative ranking curve
Ranking for all the outcomes at class level
AE: adverse events; PASI: Psoriasis Area and Severity Index; PGA: Physician Global Assessment; QoL: quality of life; SAE: serious adverse events
Ranking for all the outcomes at drug level
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
AE: adverse events; PASI: Psoriasis Area and Severity Index; PGA: Physician Global Assessment; QoL: quality of life; SAE: serious adverse events
PASI 90: direct summary effects for comparisons including at least two studies at class level
AIL12/23: anti‐IL12/23; AIL17: anti‐IL17; AIL23: anti‐IL23, ATA: anti‐TNF alpha; CSA: conventional systemic agents; OB: other biologics; PBO: placebo; SM: small molecules
CI: confidence interval; PASI: Psoriasis Area and Severity Index; RR: risk ratio
Serious adverse effects: direct summary effects for comparisons including at least two studies at class level
AIL12/23: anti‐IL12/23; AIL17: anti‐IL17; AIL23: anti‐IL23, ATA: anti‐TNF alpha; CSA: conventional systemic agents; OB: other biologics; PBO: placebo; SM: small molecules
CI: confidence interval; RR: risk ratio; SAE: serious adverse events
Specific quality of life scale: direct summary effects for comparisons including at least two studies at class level
AIL12/23: anti‐IL12/23; AIL17: anti‐IL17; AIL23: anti‐IL23, ATA: anti‐TNF alpha; CSA: conventional systemic agents; OB: other biologics; PBO: placebo; SM: small molecules
CI: confidence interval; QoL: specific quality of life scale; SMD: standardised mean difference
Physician Global Assessment 0/1: direct summary effects for comparisons including at least two studies at class‐level
AIL12/23: anti‐IL12/23; AIL17: anti‐IL17; AIL23: anti‐IL23, ATA: anti‐TNF alpha; CSA: conventional systemic agents; OB: other biologics; PBO: placebo; SM: small molecules
AE: adverse events; CI: confidence interval; PGA: Physician Global Assessment; RR: risk ratio
PASI 75: direct summary effects for comparisons including at least two studies at class level
AIL12/23: anti‐IL12/23; AIL17: anti‐IL17; AIL23: anti‐IL23, ATA: anti‐TNF alpha; CSA: conventional systemic agents; OB: other biologics; PBO: placebo; SM: small molecules
CI: confidence interval; PASI: Psoriasis Area and Severity Index; RR: risk ratio
Adverse effects : direct summary effects for comparisons including at least two studies at class‐level
AIL12/23: anti‐IL12/23; AIL17: anti‐IL17; AIL23: anti‐IL23, ATA: anti‐TNF alpha; CSA: conventional systemic agents; OB: other biologics; PBO: placebo; SM: small molecules
AE: adverse events; CI: confidence interval; RR: risk ratio
Distributions of age and sex ratio of participants across comparisons
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
Distributions of weight of participants and PASI score at baseline across comparisons
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
Side‐splitting approach and design‐by‐treatment interaction model for inconsistency for Psoriasis Area and Severity Index (PASI) 90
Treatment codes: A = PBO, B = FUM, C = MTX, D = ACI, E = ALEFACEPT, F = CICLO, G = IFX, H = ADA, I = ETA, J = USK, K = SECU, L = IXE, M = BRODA, N = CERTO, O = APRE, P = TOFA, Q = GUSEL, R = TILDRA, S = PONE, T = ITO
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
Side‐splitting approach and design‐by‐treatment interaction model for inconsistency for serious adverse events (SAEs)
Treatment codes: A = PBO, B = FUM, C = MTX, D = ACI, E = ALEFACEPT, F = CICLO, G = IFX, H = ADA, I = ETA, J = USK, K = SECU, L = IXE, M = BRODA, N = CERTO, O = APRE, P = TOFA, Q = GUSEL, R = TILDRA, S = PONE, T = ITO
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
Inconsistency plots for all the outcomes at class‐level
Inconsistency factor (IF) is calculated as the risk ratio (RR)/standardised mean difference (SMD) for direct evidence over the RR/SMD for indirect evidence in the loop with its 95% confidence interval (CI). IF value close to 0 indicates the absence of evidence for disagreement between direct and indirect evidence.
AIL12/23: anti‐IL12/23; AIL17: anti‐IL17; AIL23: anti‐IL23, ATA: anti‐TNF alpha; CSA: conventional systemic agents; OB: other biologics; PBO: placebo; SM: small molecules
Inconsistency plots for all the outcomes at drug level
Inconsistency factor (IF) is calculated as the risk ratio (RR)/standardised mean difference (SMD) for direct evidence over the RR/SMD for indirect evidence in the loop with its 95% confidence interval (CI). IF value close to 0 indicates the absence of evidence for disagreement between direct and indirect evidence.
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
Sensitivity analyses ‐ Interval plot. Network meta‐analysis results for primary outcomes (PASI 90 and serious adverse events) for trials with at least 50 participants.
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
CI: confidence interval; PASI: Psoriasis Area and Severity Index; RR: risk ratio; SAE: serious adverse events
Sensitivity analyses ‐ Interval plot. Network meta‐analysis results for primary outcomes (PASI 90 and serious adverse events) for the completers.
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
CI: confidence interval; PASI: Psoriasis Area and Severity Index; RR: risk ratio; SAE: serious adverse events
Sensitivity analyses ‐ Interval plot. Network meta‐analysis results for primary outcomes (PASI 90 and serious adverse events) for all the interventions depending on the doses
MTX ≥ 15/MTX other: methotrexate ≥ 15 mg per week/methotrexate < 15 mg per week; ALEFACEPT: alefacept all dosages; CICLO High: ciclosporin ≥ 3 mg/kg/day; ACI ≥ 35: acitretin ≥ 35 mg per day; FUM: fumaric acid esters all dosages; APRE 30: apremilast 30 mg twice daily; PONE 40: ponesimod 40 mg per day; TOFA 20: tofacitinib 20 mg per day; ETA 25/ETA 50: etanercept 25 mg twice a week/etanercept 50 mg twice a week; IFX: infliximab 5 mg/kg week O, 2, 4 every 6 weeks; ADA: adalimumab 80 mg Week 0, 40 mg Week 1 then 40 mg every other week; CERTO 200/400: certolizumab all dosages; USK 45: ustekinumab 45 mg; SECU 300/SECU other: secukinumab 300 mg every injection/secukinumab other dosages; IXE 200/IXE other: ixekizumab 200 mg per injection/ixekizumab other dosages; TILDRA 100/200: tildrakizumab all dosages; GUSEL 100: guselkumab 100 mg per injection; BRODA 210: brodalumab 210 mg per injection
CI: confidence interval; PASI: Psoriasis Area and Severity Index; RR: risk ratio; SAE: serious adverse events
Sensitivity analyses ‐ Interval plot. Network meta‐analysis results for primary outcomes (PASI 90 and serious adverse events) for all the interventions excluding studies at high risk of bias.
ACI: acitretin; ADA: adalimumab; APRE: apremilast; BRODA: brodalumab; CERTO: certolizumab; CICLO: ciclosporin; ETA: etanercept; FUM: fumaric acid; IFX: infliximab; ITO: itolizumab; IXE: ixekizumab; GUSEL: guselkumab; MTX: methotrexate; PBO: placebo; PONE: ponesimod; SECU: secukinumab; TILDRA: tildrakizumab; TOFA: tofacitinib; USK: ustekinumab
CI: confidence interval; PASI: Psoriasis Area and Severity Index; RR: risk ratio; SAE: serious adverse events
Funnel plot for network meta‐analysis of all the outcomes
AE: adverse event; lnRR: Mean effect size; PASI: Psoriasis Area and Severity Index; QoL: Specific quality of life scale; RR: Risk ratio; SAE: serious adverse events; SMD: standardised mean difference
Study bias distribution for each primary outcome (PASI 90 and serious adverse events)
The following graphs show how much information (i.e. the percentage contribution of each direct comparison in the network estimates) comes from low (green), unclear/moderate (yellow) and high (red) risk of bias studies. Here we have all drugs versus placebo as it is difficult to have all comparisons due to space limitations. To evaluate the direct comparisons we used the mean level of bias of the included studies in each comparison.
We used the web application CINeMA (CINeMA 2017).
The codes of the treatments are A = Placebo, B = Fumaric acid esters, C = Methotrexate, D = Acitretin, E = Alefacept, F = Ciclosporin, G = Infliximab, H = Adalimumab, I = Etanercept, J = Ustekinumab, K = Secukinumab, L = Ixekizumab, M = Brodalumab, N = Certolizumab, O = Apremilast, P = Tofacitinib, Q = Guselkumab, R = Tildrakizumab, S = Ponesimod, T = Itolizumab
Comparison 1 Primary outcome ‐ PASI 90, Outcome 1 Conventional systemic agents versus placebo.
Comparison 1 Primary outcome ‐ PASI 90, Outcome 2 Conventional systemic 1 versus conventional systemic 2.
Comparison 1 Primary outcome ‐ PASI 90, Outcome 3 Anti‐TNF alpha versus placebo.
Comparison 1 Primary outcome ‐ PASI 90, Outcome 4 Ustekinumab versus placebo.
Comparison 1 Primary outcome ‐ PASI 90, Outcome 5 Anti‐IL17 versus placebo.
Comparison 1 Primary outcome ‐ PASI 90, Outcome 6 Anti‐IL23 versus placebo.
Comparison 1 Primary outcome ‐ PASI 90, Outcome 7 Other biologics.
Comparison 1 Primary outcome ‐ PASI 90, Outcome 8 Biologic versus conventional systemic treatments.
Comparison 1 Primary outcome ‐ PASI 90, Outcome 9 Biologic 1 versus biologic 2.
Comparison 1 Primary outcome ‐ PASI 90, Outcome 10 Small molecules versus placebo.
Comparison 1 Primary outcome ‐ PASI 90, Outcome 11 Biologic versus small molecules.
Comparison 2 Primary outcome ‐ serious adverse events, Outcome 1 Conventional systemic agents versus placebo.
Comparison 2 Primary outcome ‐ serious adverse events, Outcome 2 Anti‐TNF alpha versus placebo.
Comparison 2 Primary outcome ‐ serious adverse events, Outcome 3 Ustekinumab versus placebo.
Comparison 2 Primary outcome ‐ serious adverse events, Outcome 4 Anti‐IL17 versus placebo.
Comparison 2 Primary outcome ‐ serious adverse events, Outcome 5 Anti‐IL23 versus placebo.
Comparison 2 Primary outcome ‐ serious adverse events, Outcome 6 Other biologics.
Comparison 2 Primary outcome ‐ serious adverse events, Outcome 7 Biologic versus conventional systemic treatments.
Comparison 2 Primary outcome ‐ serious adverse events, Outcome 8 Biologic 1 versus biologic 2.
Comparison 2 Primary outcome ‐ serious adverse events, Outcome 9 Small molecules versus placebo.
Comparison 2 Primary outcome ‐ serious adverse events, Outcome 10 Biologic versus small molecules.
Comparison 3 Secondary outcome ‐ PASI 75, Outcome 1 Conventional systemic agents versus placebo.
Comparison 3 Secondary outcome ‐ PASI 75, Outcome 2 Conventional systemic 1 versus conventional systemic 2.
Comparison 3 Secondary outcome ‐ PASI 75, Outcome 3 Anti‐TNF alpha versus placebo.
Comparison 3 Secondary outcome ‐ PASI 75, Outcome 4 Ustekinumab versus placebo.
Comparison 3 Secondary outcome ‐ PASI 75, Outcome 5 Anti‐IL17 versus placebo.
Comparison 3 Secondary outcome ‐ PASI 75, Outcome 6 Anti‐IL23 versus placebo.
Comparison 3 Secondary outcome ‐ PASI 75, Outcome 7 Other biologics.
Comparison 3 Secondary outcome ‐ PASI 75, Outcome 8 Biologic versus conventional systemic treatments.
Comparison 3 Secondary outcome ‐ PASI 75, Outcome 9 Biologic 1 versus biologic 2.
Comparison 3 Secondary outcome ‐ PASI 75, Outcome 10 Small molecules versus placebo.
Comparison 3 Secondary outcome ‐ PASI 75, Outcome 11 Biologic versus small molecules.
Comparison 4 Secondary outcome ‐ PGA 0/1, Outcome 1 Conventional systemic agents versus placebo.
Comparison 4 Secondary outcome ‐ PGA 0/1, Outcome 2 Conventional systemic 1 versus conventional systemic 2.
Comparison 4 Secondary outcome ‐ PGA 0/1, Outcome 3 Anti‐TNF alpha versus placebo.
Comparison 4 Secondary outcome ‐ PGA 0/1, Outcome 4 Ustekinumab versus placebo.
Comparison 4 Secondary outcome ‐ PGA 0/1, Outcome 5 Anti‐IL17 versus placebo.
Comparison 4 Secondary outcome ‐ PGA 0/1, Outcome 6 Anti‐IL23 versus placebo.
Comparison 4 Secondary outcome ‐ PGA 0/1, Outcome 7 Other biologics.
Comparison 4 Secondary outcome ‐ PGA 0/1, Outcome 8 Biologic versus conventional systemic treatments.
Comparison 4 Secondary outcome ‐ PGA 0/1, Outcome 9 Biologic 1 versus biologic 2.
Comparison 4 Secondary outcome ‐ PGA 0/1, Outcome 10 Small molecules versus placebo.
Comparison 4 Secondary outcome ‐ PGA 0/1, Outcome 11 Biologic versus small molecules.
Comparison 5 Secondary outcome ‐ quality of life, Outcome 1 Conventional systemic agents versus placebo.
Comparison 5 Secondary outcome ‐ quality of life, Outcome 2 Anti‐TNF alpha versus placebo.
Comparison 5 Secondary outcome ‐ quality of life, Outcome 3 Ustekinumab versus placebo.
Comparison 5 Secondary outcome ‐ quality of life, Outcome 4 Anti‐IL17 versus placebo.
Comparison 5 Secondary outcome ‐ quality of life, Outcome 5 Anti‐IL23 versus placebo.
Comparison 5 Secondary outcome ‐ quality of life, Outcome 6 Other biologics.
Comparison 5 Secondary outcome ‐ quality of life, Outcome 7 Biologic versus conventional systemic treatments.
Comparison 5 Secondary outcome ‐ quality of life, Outcome 8 Biologic 1 versus biologic 2.
Comparison 5 Secondary outcome ‐ quality of life, Outcome 9 Small molecules versus placebo.
Comparison 5 Secondary outcome ‐ quality of life, Outcome 10 Biologic versus small molecules.
Comparison 6 Secondary outcome ‐ adverse events, Outcome 1 Conventional systemic agents versus placebo.
Comparison 6 Secondary outcome ‐ adverse events, Outcome 2 Conventional systemic 1 versus conventional systemic 2.
Comparison 6 Secondary outcome ‐ adverse events, Outcome 3 Anti‐TNF alpha versus placebo.
Comparison 6 Secondary outcome ‐ adverse events, Outcome 4 Ustekinumab versus placebo.
Comparison 6 Secondary outcome ‐ adverse events, Outcome 5 Anti‐IL17 versus placebo.
Comparison 6 Secondary outcome ‐ adverse events, Outcome 6 Anti‐IL23 versus placebo.
Comparison 6 Secondary outcome ‐ adverse events, Outcome 7 Biologic versus conventional systemic treatments.
Comparison 6 Secondary outcome ‐ adverse events, Outcome 8 Biologic 1 versus biologic 2.
Comparison 6 Secondary outcome ‐ adverse events, Outcome 9 Small molecules versus placebo.
Comparison 6 Secondary outcome ‐ adverse events, Outcome 10 Biologic versus small molecules.
| Any systemic treatment compared to placebo for chronic plaque psoriasis (network meta‐analysis) | |||||||
| Patient or population: people with chronic plaque psoriasis Setting: all the participants were recruited from a hospital setting Timescale: 12 to 16 weeks after randomisation | |||||||
| Intervention | Anticipated absolute effects* (95% CI) | Relative effect | SUCRA | № of participants | Certainty of the evidence | Comments | |
| Risk with placeboa | Risk with any systemic treatment | ||||||
| PASI 90 | |||||||
| Ixekizumab | Moderate | RR 32.45 | 94.3 | 3268 | ⊕⊕⊕⊕ | ‐ | |
| 15 per 1000 | 487 per 1000 | ||||||
| Secukinumab | Moderate | RR 26.55 | 86.5 | 2707 | ⊕⊕⊕⊕ | ‐ | |
| 15 per 1000 | 398 per 1000 | ||||||
| Brodalumab | Moderate | RR 25.45 | 84.3 | 4109 | ⊕⊕⊕⊝ | Reasons for downgrading by one level: three studies contributing to this estimate at high risk of bias in selective reporting domain | |
| 15 per 1000 | 382 per 1000 | ||||||
| Guselkumab | Moderate | RR 21.03 | 77 | 1502 | ⊕⊕⊕⊝ | Reasons for downgrading by one level: one study contributing to this estimate at high risk of bias in selective reporting domain | |
| 15 per 1000 | 315 per 1000 | ||||||
| Certolizumab | Moderate | RR 24.58 | 75.7 | 176 | ⊕⊕⊕⊝ | Downgraded one level due to imprecision: wide CIs | |
| 15 per 1000 | 369 per 1000 | ||||||
| Ustekinumab | Moderate | RR 19.91 | 72.6 | 3832 | ⊕⊕⊕⊕ | ‐ | |
| 15 per 1000 | 299 per 1000 | ||||||
| Tildrakizumab | Moderate | RR 15.63 | 63.6 | 355 | ⊕⊕⊝⊝ | Downgraded one level due to risk of bias and one level due to imprecision. The single study contributing to this estimate at unclear risk of bias in both blinding domains; wide CIs | |
| 15 per 1000 | 234 per 1000 | ||||||
| Adalimumab | Moderate | RR 14.87 | 63.1 | 3199 | ⊕⊕⊕⊝ | Downgraded one level due to inconsistency ‐ inconsistent loops of evidence | |
| 15 per 1000 | 223 per 1000 | ||||||
| Itolizumab | Moderate | RR 12.26 | 56 | 225 | ⊕⊕⊝⊝ | Downgraded one level due to imprecision (wide CIs) and one level due to risk of bias (moderate risk using credibility of evidence) | |
| 15 per 1000 | 184 per 1000 | ||||||
| Infliximab | Moderate | RR 11.18 | 53.2 | (0 RCTs) | ⊕⊝⊝⊝ | Downgraded one level due to risk of bias (credibility of risk), one level due to imprecision (wide CIs) and one level due to inconsistency (inconsistent loop of evidence) | |
| 15 per 1000 | 168 per 1000 | ||||||
| Etanercept | Moderate | RR 10.79 | 52.6 | 4954 | ⊕⊕⊕⊝ | Downgraded one level due to inconsistency (global inconsistency ‐ side‐splitting approach) | |
| 15 per 1000 | 162 per 1000 | ||||||
| Tofacitinib | Moderate | RR 8.50 | 42.5 | 2826 | ⊕⊕⊝⊝ | Downgraded one level due to risk of bias: two studies at high risk of bias in incomplete outcome data domain; and downgraded one level due to inconsistency (global approach) | |
| 15 per 1000 | 128 per 1 000 | ||||||
| Apremilast | Moderate | RR 7.66 | 39.7 | 1775 | ⊕⊕⊕⊝ | Downgraded one level due to risk of bias: one study had a slight risk of bias in selective reporting domain | |
| 15 per 1000 | 115 per 1000 | ||||||
| Ponesimod | Moderate | RR 6.60 | 37.3 | 326 | ⊕⊕⊕⊝ | Downgraded one level due to imprecision: wide CIs | |
| 15 per 1000 | 99 per 1000 | ||||||
| Alefacept | Moderate | RR 4.39 | 25.3 | (0 RCTs) | ⊕⊝⊝⊝ | Downgraded two levels due to risk of bias and a further one level due to imprecision ‐ study indirectly contributing to the estimates at high risk of bias in selective reporting domain; wide CIs | |
| 15 per 1000 | 66 per 1000 | ||||||
| Fumaric acid esters (FAEs) | Moderate | RR 4.09 | 21.9 | 704 | ⊕⊝⊝⊝ | Downgraded two levels due to risk of bias, and one level due to imprecision ‐ the studies indirectly contributing to this estimate at high risk of bias in blinding domain; wide CIs | |
| 15 per 1000 | 61 per 1000 | ||||||
| Ciclosporin | Moderate | RR 3.99 | 21.3 | (0 RCTs) | ⊕⊝⊝⊝ | Downgraded two levels due to risk of bias, and a further one level due to imprecision ‐ the single study indirectly contributing to this estimate at high risk of bias in blinding; wide CIs | |
| 15 per 1000 | 60 per 1000 | ||||||
| Methotrexate | Moderate | RR 3.61 | 20.2 | 282 | ⊕⊕⊕⊝ | Downgraded one level due to inconsistency (inconsistent loop of evidence) | |
| 15 per 1000 | 59 per 1000 | ||||||
| Acitretin | Moderate | RR 0.98 | 9.9 | (0 RCTs) | ⊕⊝⊝⊝ | Downgraded two levels due to risk of bias and a further one level due to imprecision. The single study contributing to this estimate at high risk of bias in incomplete outcome data and blinding domains; wide CIs | |
| 15 per 1000 | 15 per 1000 | ||||||
| Serious adverse events | |||||||
| Methotrexate | Moderate | RR 0.23 | 90.7 | 282 | ⊕⊕⊕⊝ | Downgraded one level due to imprecision (wide CIs) | |
| 17 per 1000 | 4 per 1000 | ||||||
| Ciclosporin | Moderate | RR 0.23 | 78.2 | (0 RCTs) | ⊕⊝⊝⊝ | Downgraded two levels due to risk of bias (credibility of evidence), and one level due to imprecision (wide CIs) | |
| 17 per 1000 | 4 per 1000 | ||||||
| Certolizumab | Moderate | RR 0.49 | 70.9 | 176 | ⊕⊕⊕⊝ | Downgraded one level due to imprecision (wide CIs) | |
| 17 per 1000 | 8 per 1000 | ||||||
| Infliximab | Moderate | RR 0.56 | 64.4 | (0 RCTs) | ⊕⊝⊝⊝ | Downgraded two levels due to risk of bias, and one level due to imprecision: credibility of evidence; wide CIs | |
| 17 per 1000 | 10 per 1000 | ||||||
| Alefacept | Moderate | RR 0.72 | 62.6 | 736 | ⊕⊕⊝⊝ | Downgraded one level due to risk of bias (credibility of evidence), and one level due to imprecision (wide CIs) | |
| 17 per 1000 | 12 per 1000 | ||||||
| Fumaric acid esters (FAEs) | Moderate | RR 0.77 | 57.7 | 704 | ⊕⊝⊝⊝ | Downgraded by one level due to risk of bias and one level due to imprecision: credibility of evidence; wide CIs | |
| 17 per 1000 | 13 per 1000 | ||||||
| Apremilast | Moderate | RR 0.84 | 54.7 | 2036 | ⊕⊕⊝⊝ | Downgraded one level due to risk of bias and one level due to imprecision: credibility of evidence and wide CIs | |
| 17 per 1000 | 14 per 1000 | ||||||
| Ustekinumab | Moderate | RR 0.89 | 52 | 4154 | ⊕⊕⊝⊝ | Downgraded one level due to risk of bias and one level due to imprecision ‐ credibility of evidence; wide CIs | |
| 17 per 1000 | 15 per 1000 | ||||||
| Acitretin | Moderate | RR 0.99 | 46.9 | (0 RCTs) | ⊕⊝⊝⊝ | Downgraded by two levels due to risk of bias and one level due to imprecision: credibility of evidence; wide CIs | |
| 17 per 1000 | 17 per 1000 | ||||||
| Tofacitinib | Moderate | RR 0.98 | 44 | 2838 | ⊕⊕⊕⊝ | Downgraded one level due to imprecision (wide CIs) | |
| 17 per 1000 | 17 per 1000 | ||||||
| Etanercept | Moderate | RR 0.99 | 43.6 | 3783 | ⊕⊕⊕⊝ | Downgraded one level due to imprecision (CIs including one) | |
| 17 per 1000 | 17 per 1000 | ||||||
| Guselkumab | Moderate | RR 1.00 | 42.6 | 1502 | ⊕⊕⊝⊝ | Downgraded one level due to risk of bias (credibility of evidence), and one level due to imprecision (CIs including one) | |
| 15 per 1000 | 15 per 1000 | ||||||
| Adalimumab | Moderate | RR 1.02 | 40.4 | 3199 | ⊕⊕⊕⊝ | Downgraded one level due to imprecision (CIs including one) | |
| 18 per 1000 | 19 per 1000 | ||||||
| Brodalumab | Moderate | RR 1.04 | 39.8 | 4109 | ⊕⊕⊝⊝ | Downgraded one level due to risk of bias (credibility of evidence) and one level due to imprecision (CIs including 1) | |
| 17 per 1000 | 18 per 1000 | ||||||
| Tildrakizumab | Moderate | RR 1.36 | 37.8 | 355 | ⊕⊕⊝⊝ | Downgraded one level due to risk of bias (credibility of evidence) and one level due to imprecision (CIs including 1) | |
| 0 per 1000 | 0 per 1000 | ||||||
| Ixekizumab | Moderate | RR 1.12 | 33.7 | 3268 | ⊕⊕⊕⊝ | Downgraded one level due to imprecision (CIs including one) | |
| 15 per 1000 | 16 per 1000 | ||||||
| Secukinumab | Moderate | RR 1.19 | 29.9 | 2707 | ⊕⊕⊕⊝ | Downgraded one level due to imprecision (CIs including one) | |
| 10 per 1000 | 12 per 1000 | ||||||
| Ponesimod | Moderate | RR 2.59 | 18.1 | 326 | ⊕⊕⊕⊝ | Downgraded one level due to imprecision (CIs including one) | |
| 15 per 1000 | 39 per 1000 | ||||||
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | |||||||
| GRADE Working Group grades of evidence | |||||||
| a 'Risk with placebo' is the median placebo‐group risk value in the included studies for the assumed risk with placebo. c The Psoriasis Area and Severity Index combines the assessment of the severity of lesions and the area affected into a single score in the range of 0 (no disease) to 72 (maximal disease); PASI 90: 90% improvement in the PASI. d SUCRA was expressed as a percentage between 0 (when a treatment is certain to be the worst) to 100% (when a treatment is certain to be the best). | |||||||
| Term | Definition |
| Antagonist | A substance that interferes with or inhibits the physiological action of another. |
| Antigen | A molecule capable of inducing an immune respons |
| Anti‐TNF alpha | A pharmaceutical drug that suppresses the physiologic response to tumor necorsis factor (TNF) |
| Biological agent | Therapeutic agents consisting of immune molecules such as soluble receptors, recombinant cytokines, and monoclonal antibodies that target effector molecules or cells of the immune system |
| CD6 | Cluster of differentiation (CD) 6 is a protein encoded by the CD6 gene |
| Cheilitis | An inflammation of the lips |
| Chimeric protein | A chimeric protein can be made by combining two different genes |
| Complex cyclophilin‐ciclosporin | Cyclophilins are a family of proteins that bind to ciclosporin, an immunosuppressant agent |
| Creatinine | A compound that is produced by metabolism of creatine and excreted in the urine |
| Cyclic adenosine monophosphate | It is a second messenger important in many biological processes |
| Cytokines | Small proteins produced by a broad range of cells that are important in cell signaling; they are immunomodulating agents |
| Dendritic cells | Antigen‐presenting cells of the immune system |
| Dermis | It is a layer of the skin |
| Epitope | It is a part of an antigen |
| Erythematous | Redness of the skin |
| Folic acid | B vitamin |
| Humanised antibody | Antibodies from non‐human species whose protein sequences have been modified to increase their similarity to antibody variants produced naturally in humans |
| IL‐17A | A pro‐inflammatory cytokine |
| IL‐23R | A cytokine receptor |
| Immune‐mediated | A group of diseases that are characterised by common inflammatory pathways leading to inflammation, and which may result from a dysregulation of the normal immune response |
| Immunogenicity | This is the ability of a particular substance, such as an antigen or epitope, to provoke an immune response in the body of a human or animal |
| Immunoglobulin 1 Fc | An antibody |
| Interferon (IFN)‐c | A protein released by cells, usually in response to a pathogen |
| Interleukin | A kind of cytokine |
| Janus kinase (JAK) inhibitors | A pharmaceutical drug that inhibits the activity of one or more of the Janus kinase family of enzymes |
| Keratinocytes | Epidermal cells that constitute 95% of the epidermis |
| Lymphocyte | A subtype of a white blood cell |
| Lymphoid organ | Part of the body that defends the body against invading pathogens that cause infections or the spread of tumours |
| Metalloproteinases | A protease enzyme |
| Monoclonal antibodies | Antibodies that are made by identical immune cells that are all clones of a unique parent cell |
| Murine sequence | Mouse genomic sequencing |
| Neutrophils | Type of white blood cell involved in the innate immune system |
| p40 | Subunit beta of interleukin 12 and 23 |
| Periumbilical | Around the navel |
| Pharmacological treatments | Drugs |
| Phase I | First‐in‐man studies |
| Phase II | Studies to assess how well the drug works, as well as to continue phase I safety assessments in a larger group of volunteers and participants |
| Phase III | Randomised controlled multicenter trials on large patient groups and are aimed at being the definitive assessment of how effective the drug is |
| Phase IV | Post‐marketing trials involve the safety surveillance |
| Phosphodiesterase 4 inhibitors | A pharmaceutical drug used to block the degradative action of phosphodiesterase 4 |
| Progressive multifocal leukoencephalopathy | A rare viral neurological disease characterised by progressive damage of the white matter of the brain at multiple locations |
| Receptor | A protein molecule that receives chemical signals from outside a cell |
| Small molecules | Chemically manufactured molecules (or SMOLs for short) |
| Sphingosine 1‐phosphate receptor agonists | A class of protein‐coupled receptors that are targets of the lipid signalling molecule Sphingosine‐1‐phosphate |
| T cells/CD4 T cells | A type of white blood cell that is of key importance to the immune system |
| Th1 and Tc1 cells | A type of T cell |
| Th17 and Tc17 cells | A type of T cell |
| TNF‐alpha | A protein that is part of the inflammatory response |
| Tumour necrosis factor antagonists | Class of biological agents |
| Umbilic | Navel |
| Xerosis | Dry skin |
| Contact | Requested Information | Contacted | Reply (last check 1/03/2017) | |
| Missing data | ||||
| Prof. Akcali | Outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 8 and 21 November 2016 | No response | |
| Prof. Al‐Hamamy | Outcomes: PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 8 and 21 November 2016 | No response | |
| Prof. Asahina | Outcome: PASI 90 | 8 November 2016 | Asahina 2010 detailed report | |
| Prof. Asahina Pfizer | Outcomes: AEs & SAEs | 3 and 12 January 2017 | Additional data to the publication not provided | |
| Prof. Asawanonda | Outcomes: PASI 75, PGA 0/1, AEs & SAEs | 21 November 2016 15 December 2016 | Asawanonda 2006 sent detailed report for PASI 75 and AEs. PGA was not collected during this study. | |
| Prof. Bisonnette Innovaderm Recherches Inc. | Outcomes: PASI 90, PGA 0/1, AEs | 8 and 21 November 2016 | Additional data to the publication not provided | |
| Dr Blauvelt Novartis | Outcome: QoL scale | 8 and 21 November 2016 | Additional data to the publication not provided | |
| Prof. Fabri | Outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 8 and 21 November 2016 | Caproni 2009 sent detailed report for PASI 90 and SAEs. Other outcomes (PGA, QoL and AEs) not collected during this study. | |
| Prof. Dogra | Outcomes: PGA 0/1, QoL scale, AEs & SAEs | 8 and 21 November 2016 | No response | |
| Prof. Dogra | Outcomes: PGA 0/1, QoL scale, AEs & SAEs | 8 November 2016 | PGA & QoL scale not collected during this study. AEs & SAEs not provided per arm | |
| Dr Fallah Arani | Outcomes: PASI 90, PGA 0/1 and QoL scale | 8 and 21 November 2016 | Outcomes not collected during this study | |
| Prof. Flytstrom | Outcomes: PGA 0/1 | 12 and 19 January 2017 | Additional data to the publication not provided | |
| Prof. Gisondi | Outcomes: PASI 90, PGA 0/1, QoL scale, AEs & SAEs | 8 November 2016 | Gisondi 2008 sent detailed report for the requested outcomes except for QoL (not assessed during the study) | |
| Prof. Gordon | Outcomes: PGA0/1, AEs | 3 and 12 January 2017 | No response | |
| Prof. Gottlieb Abbvie | Outcomes: PASI 90 & QoL scale | 8 November 2016 | Gottlieb 2012 sent detailed report for the requested outcomes | |
| Prof. Gottlieb Amgen | Outcomes: PASI 90, PGA 0/1, QoL scale, AEs & SAEs | 8 November 2016 | Gottlieb 2011 sent detailed report for the requested outcomes | |
| Prof. Griffiths Janssen | Outcome: QoL scale | 16 December 2016 | QoL was not collected during this study | |
| Prof. Jacobe | Outcomes: PASI 90, PGA 0/1, QoL scale, AEs & SAEs | 8 and 20 November 2016 | No response | |
| Pfizer | Outcomes: PASI 90, QoL scale | 3 and 12 January 2017 | No response | |
| Prof. Ganapathi R&D, Biocon Research Limited | Outcomes: QoL scale, AEs & SAEs | 8 and 21 November 2016 | Krupashandar sent detailed report for the requested outcomes, however AEs and SAEs were only available for the entire trial and not at the time of the major outcome assessment | |
| Prof. Lebwohl Valeant Pharmaceuticals NA LLC | Outcomes: PASI 90 & QoL scale | 8 and 21 November 2016 | Lebwohl AMAGINE‐2, 2015 sent detailed report for PASI 90, individual scores and median difference from baseline of QoL were not available | |
| Prof. Lebwohl Valeant Pharmaceuticals NA LLC | Outcomes: PASI 90 & QoL scale | 8 and 21 November 2016 | Lebwohl AMAGINE‐3, 2015 sent detailed report for PASI 90, individual scores and median difference from baseline of QoL were not available | |
| Prof. Leonardi | Outcomes: QoL scale & AEs | 8 and 21 November 2016 | No response | |
| Prof. Kaur | Outcomes: PASI 90, PGA 0/1, QoL scale, AEs & SAEs | 8 and 21 November 2016 | No response | |
| Prof. Menter | Outcome: PGA 0/1 | 8 and 21 November 2016 | No response | |
| Prof. Menter | Outcome: PGA 0/1 | 8 and 21 November 2016 | No response | |
| Prof. Mrowietz | Outcome: QoL scale | 3 and 12 January 2017 | Additional data to the publication not provided | |
| Prof. Paul Novartis | Outcome: PASI 90 | 3 January 2017 | Additional data to the publication not provided | |
| Prof. Papp | Outcome: QoL scale | 22 November 2016 13 December 2016 | Additional data to the publication not provided | |
| Prof. Papp | Outcome: QoL scale | 22 November 2016 13 December 2016 | Additional data to the publication not provided | |
| Prof. Papp | Outcome: QoL scale, AEs & SAEs | 22 November 2016 13 December 2016 | Additional data to the publication not provided | |
| Prof. Papp | Outcome: QoL scale | 22 November 2016 13 December 2016 | Additional data to the publication not provided | |
| Prof. Papp | Outcome: PASI 90, PGA0/1, QoL scale | 3 January 2017 | Additional data to the publication not provided | |
| Prof. Paul Novartis | Outcome: QoL scale | 15 December 2016, 2 January 2017 | Additional data to the publication not provided | |
| Prof. Reich Novartis | Outcomes: PGA 0/1 & QoL scale | 8 November 2016, 16 December 2016 | Additional data to the publication not provided | |
| Prof. Reich PelotonAdvantage | Outcome: QoL scale | 4 January 2017 | Additional data to the publication not provided | |
| Prof. Rich | Outcome: QoL scale | 22 November 2016, 13 December 2016 | No response | |
| Prof. Sterry | Outcomes: PASI 90 & QoL scale | 8 and 21 November 2016 | No response | |
| Prof. Strober Abbvie | Outcome: QoL scale | 8 November 2016 | Strober sent detailed report for the requested outcomes | |
| Prof. Thaçi Novartis | Outcome: QoL scale | 8 and 21 November 2016 | Additional data to the publication not provided | |
| Prof. Torii | Outcomes: PASI 90 & PGA0/1 | 21 November 2016 | Torii sent detailed report for the requested outcomes | |
| Prof. Tyring | Outcomes: PGA 0/1 & QoL scale | 8 and 21 November 2016 | No response | |
| Dr van Bezooijen | Outcomes: PASI 90, adverse effects | 4 and 12 January 2017 | Additional data to the publication not provided | |
| Prof. van der Kherkhof Pfizer | Outcome: AEs | 8 and 21 November 2016 | Additional data to the publication not provided | |
| No contact | Outcomes: AEs and SAEs | No | Authors' email not found | |
| No contact | Outcome: PASI 90 | No | Authors' email not found | |
| Awaiting classification studies | ||||
| Prof. Elewski Abbvie | Study's protocol and outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 3 and 12 January 2017 | Will be included when published | |
| Prof. Khattri | Study's protocol and outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 3 and 12 January 2017 | No response | |
| Prof. Lee | Study's protocol and outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 3 and 12 January 2017 | No response | |
| Prof. Reich | Study's protocol and outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 3 January 2017 | Will be included when published | |
| Prof. Chow | outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 8 November 2016, 16 December 2016 | No response | |
| Prof. Gurel | Study's protocol and outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 17 and 24 January 2017 | Gurel 2015 sent detailed report for the requested outcomes. Finally Gurel study was classified in the included studies section. | |
| No contact | Outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | No | Authors' email not found | |
| Dr Nogarales, MD Celgene Corporation | Asking for study protocol and efficacy/safety results | 12 and 19 January 2017 | Email response: "Thank you very much for your email and your interest in our study in Japanese subjects. May I please enquire as to the planned timing for publication for your meta‐analysis as we have just recently submitted our primary manuscript?" Will be included when published | |
| Prof. Krishna | Asking for study protocol and efficacy/safety results | 5 and 12 January 2017 | No response | |
| Prof. Jacobi | Asking for study protocol and efficacy/safety results | 12 and 19 January 2017 | No response | |
| Prof. Shah | Asking for study protocol and efficacy/safety results | 12 and 19 January 2017 | ||
| Abstracts | ||||
| Prof. Yilmaz | Outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 16 December 2016 | Yilmaz 2002 sent detailed report for the requested outcomes. Finally Yilmaz 2002 study was classified in the included studies section. | |
| Prof. Mrowietz | Study's protocol and outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 16 December 2016, 3 January 2017 | Additional data to the publication not provided. Finally Mrowietz study was classified in the awaiting classification section. | |
| Prof. Reich | Study's protocol and outcomes: PASI 90, PASI 75, PGA 0/1, QoL scale, AEs & SAEs | 16 December 2016 | Additional data to the publication not provided. Finally Reich 2004 study was classified in the awaiting classification section. | |
| Ongoing studies | ||||
| Dr Yamauchi Dr Patnaik, Director, Clinical Science Institute | Asking for study protocol and efficacy/safety results | 5 January 2017 | Email response: Dear Dr Sbidian, Will be included when published | |
| Prof. Spuls | Asking for study protocol and efficacy/safety results | 5 January 2017 | Email response "The study is currently ongoing and has not yet been analysed. Therefore, we are not able to provide data on efficacy or safety. Will be included when published | |
| AE: adverse events; PASI: Psoriasis Area and Severity Index; PGA: Physician Global Assessment; QoL: quality of life; SAE: serious adverse events | ||||
| Network meta‐analysis | Direct evidence | Indirect evidence | |||||||
| Comparisons* | RR | LCI | UCI | RR | LCI | UCI | RR | LCI | UCI |
| FAEs vs placebo | 4.09 | 1.88 | 8.88 | 4.47 | 1.97 | 10.14 | 1.86 | 0.16 | 21.16 |
| Methotrexate vs placebo | 3.91 | 2.16 | 7.08 | 1.53 | 0.66 | 3.53 | 17.16 | 5.69 | 51.75 |
| Adalimumab vs placebo | 14.87 | 10.45 | 21.14 | 14.42 | 10.08 | 20.64 | 108.8 | 2.24 | 5287.86 |
| Etanercept vs placebo | 10.79 | 8.47 | 13.73 | 10.62 | 7.52 | 15.01 | 11.21 | 7.26 | 17.32 |
| Ustekinumab vs placebo | 19.91 | 15.11 | 26.23 | 22.7 | 15.46 | 33.34 | 17.91 | 12.71 | 25.24 |
| Secukinumab vs placebo | 26.55 | 20.32 | 34.69 | 24.53 | 14.93 | 40.32 | 28.25 | 19.1 | 41.78 |
| Ixekizumab vs placebo | 32.45 | 23.61 | 44.60 | 39.46 | 20.64 | 75.44 | 24.51 | 10.05 | 59.77 |
| Brodalumab vs placebo | 25.45 | 18.74 | 34.57 | 26.58 | 16.65 | 42.41 | 23.74 | 10.09 | 55.86 |
| Apremilast vs placebo | 7.66 | 4.30 | 13.66 | 6.72 | 3.07 | 14.69 | 10.83 | 2.43 | 48.31 |
| Tofacitinib vs placebo | 8.50 | 6.23 | 11.60 | 6.3 | 4.14 | 9.56 | 17.91 | 8.3 | 38.62 |
| Guselkumab vs placebo | 21.03 | 14.56 | 30.38 | 26.1 | 14.71 | 46.3 | 12.7 | 4.28 | 37.69 |
| Methotrexate vs FAEs | 0.96 | 0.38 | 2.44 | 2 | 0.19 | 21.03 | 0.83 | 0.3 | 2.32 |
| Alefacept vs methotrexate | 1.12 | 0.42 | 3.02 | 1.12 | 0.42 | 3.02 | |||
| Ciclosporin vs methotrexate | 1.02 | 0.60 | 1.73 | 1.02 | 0.6 | 1.73 | |||
| Infliximab vs methotrexate | 2.86 | 2.06 | 3.97 | 2.86 | 2.06 | 3.97 | |||
| Adalimumab vs methotrexate | 3.80 | 2.26 | 6.39 | 3.35 | 2.02 | 5.57 | 13.2 | 3.4 | 51.32 |
| Etanercept vs acitretin | 11.00 | 0.63 | 191.47 | 11 | 0.63 | 191.47 | |||
| Guselkumab vs adalimumab | 1.41 | 1.21 | 1.65 | 1.4 | 1.18 | 1.66 | 2.88 | 0.68 | 12.21 |
| Ustekinumab vs etanercept | 1.85 | 1.50 | 2.27 | 1.8 | 1.27 | 2.55 | 1.95 | 1.37 | 2.77 |
| Secukinumab vs etanercept | 2.46 | 2.01 | 3.02 | 2.33 | 1.66 | 3.28 | 2.62 | 1.82 | 3.77 |
| Ixekizumab vs etanercept | 3.01 | 2.46 | 3.68 | 2.93 | 2.44 | 3.53 | 5.73 | 2.07 | 15.85 |
| Apremilast vs etanercept | 0.71 | 0.40 | 1.25 | 0.72 | 0.36 | 1.45 | 0.69 | 0.26 | 1.81 |
| Tofacitinib vs etanercept | 0.79 | 0.59 | 1.06 | 0.88 | 0.73 | 1.08 | 0.49 | 0.3 | 0.81 |
| Secukinumab vs ustekinumab | 1.33 | 1.11 | 1.61 | 1.38 | 1.03 | 1.84 | 1.19 | 0.79 | 1.81 |
| Brodalumab vs ustekinumab | 1.28 | 1.10 | 1.48 | 1.27 | 1.1 | 1.46 | 1.64 | 0.69 | 3.89 |
| FAES: fumaric acid esters; LCI: low confidence interval; RR: risk ratio; UCI: upper confidence interval; vs: versus, *The comparisons listed in this table were included in at least one direct‐evidence analysis. | |||||||||
| Class‐level | SUCRA | Rank | SUCRA | Rank | SUCRA | Rank | SUCRA | Rank | SUCRA | Rank | SUCRA | Rank |
| Anti‐IL12/23 | 85.7 | 2 | 53.9 | 3 | 85.0 | 2 | 57.0 | 3 | 83.8 | 2 | 75.7 | 3 |
| Anti‐IL17 | 100.0 | 1 | 21.0 | 8 | 99.6 | 1 | 14.1 | 6 | 99.9 | 1 | 95.4 | 1 |
| Anti‐IL23 | 71.3 | 3 | 39.6 | 5 | 72.2 | 3 | 78.7 | 2 | 73.1 | 3 | 83.4 | 2 |
| Anti‐TNF alpha | 56.4 | 4 | 39.2 | 6 | 57.4 | 4 | 47.5 | 5 | 57.5 | 4 | 58.4 | 4 |
| Other biologics | 26.3 | 6 | 68.2 | 2 | 17.0 | 7 | _ | _ | 16.6 | 7 | 15.5 | 7 |
| Small molecules | 41.5 | 5 | 45.4 | 4 | 42.7 | 5 | 7.9 | 7 | 42.0 | 5 | 40.4 | 5 |
| Conventional systemic treatments | 18.7 | 7 | 94.8 | 1 | 26.0 | 6 | 50.8 | 4 | 27.1 | 6 | 30.8 | 6 |
| Placebo | 0 | 8 | 38.0 | 7 | 0 | 8 | 94.0 | 1 | 0 | 8 | 0.4 | 8 |
| AE: adverse events; FAEs: fumaric acid esters; PGA: Physician Global Assessment; QoL: Specific quality of life scale; SAE: serious adverse events | ||||||||||||
| Drug | SUCRA | Rank | SUCRA | Rank | SUCRA | Rank | SUCRA | Rank | SUCRA | Rank | SUCRA | Rank |
| Acitretin | 9.9 | 19 | 46.9 | 9 | 26.0 | 15 | ‐ | ‐ | ‐ | ‐ | ‐ | ‐ |
| Adalimumab | 63.1 | 8 | 40.4 | 14 | 60.2 | 9 | 70.1 | 5 | 56.9 | 8 | 57.6 | 7 |
| Alefacept | 25.3 | 15 | 62.6 | 5 | 12.6 | 18 | ‐ | ‐ | 13.1 | 18 | 15.9 | 13 |
| Apremilast | 39.7 | 13 | 54.7 | 7 | 33.2 | 14 | 14.3 | 16 | 27.9 | 14 | 28.6 | 10 |
| Brodalumab | 84.3 | 3 | 39.8 | 15 | 82.1 | 3 | 46.4 | 9 | 84.0 | 5 | 52.3 | 8 |
| Certolizumab | 75.7 | 5 | 70.9 | 3 | 71.6 | 6 | 78.0 | 4 | 90.1 | 1 | ‐ | ‐ |
| Ciclosporin | 21.3 | 17 | 78.2 | 2 | 33.2 | 13 | 36.8 | 12 | 24.0 | 16 | ‐ | ‐ |
| Etanercept | 52.6 | 11 | 43.6 | 11 | 57.7 | 10 | 45.9 | 10 | 51.7 | 10 | 67.6 | 5 |
| FAEs | 21.9 | 16 | 57.7 | 6 | 11.1 | 19 | 17.8 | 15 | 15.4 | 17 | ‐ | ‐ |
| Guselkumab | 77.0 | 4 | 42.6 | 12 | 71.6 | 7 | 78.2 | 3 | 67.5 | 7 | 84.3 | 2 |
| Infliximab | 53.2 | 10 | 64.4 | 4 | 48.0 | 11 | 40.1 | 11 | 52.4 | 9 | ‐ | ‐ |
| Itolizumab | 56.0 | 9 | ‐ | ‐ | 71.6 | 8 | ‐ | ‐ | 29.4 | 13 | 16.0 | 12 |
| Ixekizumab | 94.3 | 1 | 33.7 | 17 | 91.8 | 1 | 18.1 | 14 | 85.9 | 3 | 99.2 | 1 |
| Methotrexate | 20.2 | 18 | 90.7 | 1 | 21.3 | 16 | 68.4 | 6 | 24.9 | 15 | 31.5 | 9 |
| Placebo | 2.9 | 20 | 42.0 | 13 | 0.0 | 20 | 88.0 | 1 | 0.3 | 19 | 1.2 | 14 |
| Ponesimod | 37.3 | 14 | 18.1 | 19 | 21.3 | 17 | 14.0 | 17 | 48.7 | 11 | 28.1 | 11 |
| Secukinumab | 86.5 | 2 | 29.9 | 18 | 86.7 | 2 | 36.3 | 13 | 84.4 | 4 | ‐ | ‐ |
| Tildrakizumab | 63.6 | 7 | 37.8 | 16 | 78.3 | 4 | 86.1 | 2 | 86.3 | 2 | 74.9 | 4 |
| Tofacitinib | 42.5 | 12 | 44.0 | 10 | 46.2 | 12 | 47.3 | 8 | 36.6 | 12 | 65.1 | 6 |
| Ustekinumab | 72.6 | 6 | 52.0 | 8 | 75.2 | 5 | 64.3 | 7 | 70.4 | 6 | 77.4 | 3 |
| AE: adverse events; FAEs: fumaric acid esters; PASI: Psoriasis Area and Severity Index; PGA: Physician Global Assessment; QoL: specific quality of life scale; SAE: serious adverse events; SUCRA: Surface Under the Cumulative Ranking | ||||||||||||
| Number of randomised participants | Number of serious adverse events | Number of serious infections | Number of malignancies | Number of MACE | ||||||
| Drug | Placebo | Drug | Placebo | Drug | Placebo | Drug | Placebo | Drug | Placebo | |
| Conventional systemic agents | 767 | 220 | 19 | 10 | 0 | 0 | 0 | 0 | 0 | 0 |
| Anti‐TNF | 4508 | 2640 | 85 | 44 | 21 | 9 | 20 | 7 | 6 | 2 |
| Anti‐IL12/23 | 2547 | 1607 | 38 | 23 | 7 | 5 | 4 | 1 | 4 | 3 |
| Anti‐IL17 | 7551 | 2533 | 149 | 36 | 47 | 7 | 21 | 2 | 19 | 3 |
| Anti‐IL23 | 1347 | 510 | 23 | 7 | 4 | 1 | 0 | 0 | 1 | 0 |
| Other biologics | 509 | 227 | 15 | 10 | ‐ | ‐ | 2 | 1 | ‐ | ‐ |
| Small molecules | 3920 | 1280 | 89 | 28 | 15 | 5 | 14 | 0 | 5 | 1 |
| MACE: Major adverse cardiac events | ||||||||||
| Network meta‐analysis | Direct evidence | Indirect evidence | |||||||
| Comparisons* | RR | LCI | UCI | RR | LCI | UCI | RR | LCI | UCI |
| FAEs vs placebo | 0.77 | 0.30 | 1.99 | 0.83 | 0.31 | 2.21 | 0.19 | 0 | 12.57 |
| Methotrexate vs placebo | 0.23 | 0.05 | 0.99 | 0.16 | 0.03 | 0.86 | 0.68 | 0.04 | 11.67 |
| Adalimumab vs placebo | 1.02 | 0.61 | 1.73 | 1.05 | 0.62 | 1.78 | 0.07 | 0 | 26.92 |
| Etanercept vs placebo | 0.99 | 0.65 | 1.51 | 1.09 | 0.65 | 1.84 | 0.76 | 0.31 | 1.89 |
| Ustekinumab vs placebo | 0.89 | 0.57 | 1.39 | 0.74 | 0.44 | 1.26 | 1.36 | 0.61 | 2.99 |
| Secukinumab vs placebo | 1.19 | 0.69 | 2.03 | 1.61 | 0.78 | 3.33 | 0.75 | 0.3 | 1.87 |
| Ixekizumab vs placebo | 1.12 | 0.66 | 1.90 | 1.16 | 0.62 | 2.16 | 0.97 | 0.18 | 5.12 |
| Brodalumab vs placebo | 1.04 | 0.62 | 1.73 | 0.92 | 0.53 | 1.62 | 2.77 | 0.38 | 20.28 |
| Apremilast vs placebo | 0.84 | 0.47 | 1.52 | 0.78 | 0.42 | 1.44 | 4.33 | 0.09 | 201.27 |
| Tofacitinib vs placebo | 0.98 | 0.55 | 1.76 | 1.05 | 0.53 | 2.06 | 0.67 | 0.08 | 5.35 |
| Guselkumab vs placebo | 1.00 | 0.49 | 2.04 | 1.21 | 0.51 | 2.85 | 0.52 | 0.08 | 3.41 |
| Methotrexate vs FAEs | 0.30 | 0.06 | 1.59 | 1 | 0.02 | 48.83 | 0.23 | 0.04 | 1.45 |
| Ciclosporin vs methotrexate | 0.98 | 0.06 | 15.38 | 0.98 | 0.06 | 15.38 | ‐ | ‐ | ‐ |
| Infliximab vs methotrexate | 2.41 | 1.04 | 5.59 | 2.41 | 1.04 | 5.59 | ‐ | ‐ | ‐ |
| Adalimumab vs methotrexate | 4.43 | 0.99 | 19.81 | 2.24 | 0.21 | 23.56 | 6.68 | 1.04 | 42.76 |
| Etanercept vs acitretin | 1.00 | 0.02 | 48.82 | 1 | 0.02 | 48.83 | ‐ | ‐ | ‐ |
| Guselkumab vs adalimumab | 0.98 | 0.51 | 1.88 | 0.89 | 0.44 | 1.79 | 2.07 | 0.26 | 16.45 |
| Ustekinumab vs etanercept | 0.90 | 0.52 | 1.57 | 1.25 | 0.38 | 4.11 | 0.83 | 0.44 | 1.54 |
| Secukinumab vs etanercept | 1.20 | 0.66 | 2.19 | 1.17 | 0.45 | 3.04 | 1.22 | 0.56 | 2.65 |
| Ixekizumab vs etanercept | 1.14 | 0.66 | 1.94 | 1.02 | 0.53 | 1.95 | 1.47 | 0.53 | 4.09 |
| Apremilast vs etanercept | 0.85 | 0.42 | 1.72 | 2.69 | 0.41 | 17.5 | 0.7 | 0.33 | 1.5 |
| Tofacitinib vs etanercept | 0.99 | 0.53 | 1.87 | 0.87 | 0.35 | 2.19 | 1.12 | 0.47 | 2.7 |
| Secukinumab vs ustekinumab | 1.33 | 0.74 | 2.38 | 1.01 | 0.42 | 2.39 | 1.68 | 0.77 | 3.68 |
| Brodalumab vs ustekinumab | 1.16 | 0.64 | 2.11 | 1.32 | 0.59 | 2.98 | 0.95 | 0.33 | 2.71 |
| FAES: fumaric acid esters; LCI: low confidence interval; RR: risk ratio; UCI: upper confidence interval *The comparisons listed in this table were included in at least one direct‐evidence analysis. | |||||||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Conventional systemic agents versus placebo Show forest plot | 3 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 1.1 Methotrexate | 2 | 282 | Risk Ratio (M‐H, Random, 95% CI) | 2.60 [0.26, 25.90] |
| 1.2 Fumaric acid esters | 1 | 704 | Risk Ratio (M‐H, Random, 95% CI) | 4.47 [2.01, 9.95] |
| 2 Conventional systemic 1 versus conventional systemic 2 Show forest plot | 3 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 2.1 Ciclosporin versus methotrexate | 2 | 172 | Risk Ratio (M‐H, Random, 95% CI) | 1.18 [0.47, 2.98] |
| 2.2 Methotrexate versus fumaric acid esters | 1 | 60 | Risk Ratio (M‐H, Random, 95% CI) | 2.0 [0.19, 20.90] |
| 3 Anti‐TNF alpha versus placebo Show forest plot | 21 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 3.1 Etanercept versus placebo | 12 | 4954 | Risk Ratio (M‐H, Random, 95% CI) | 11.17 [7.66, 16.28] |
| 3.2 Adalimumab versus placebo | 8 | 3199 | Risk Ratio (M‐H, Random, 95% CI) | 14.86 [8.93, 24.73] |
| 3.3 Certolizumab versus placebo | 1 | 176 | Risk Ratio (M‐H, Random, 95% CI) | 24.58 [3.48, 173.49] |
| 4 Ustekinumab versus placebo Show forest plot | 7 | 3832 | Risk Ratio (M‐H, Random, 95% CI) | 22.59 [14.74, 34.64] |
| 5 Anti‐IL17 versus placebo Show forest plot | 16 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 5.1 Secukinumab versus placebo | 7 | 2707 | Risk Ratio (M‐H, Random, 95% CI) | 26.52 [14.91, 47.17] |
| 5.2 Ixekizumab versus placebo | 4 | 3268 | Risk Ratio (M‐H, Random, 95% CI) | 53.85 [15.34, 189.07] |
| 5.3 Brodalumab versus placebo | 5 | 4109 | Risk Ratio (M‐H, Random, 95% CI) | 26.33 [16.77, 41.33] |
| 6 Anti‐IL23 versus placebo Show forest plot | 4 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 6.1 Guselkumab versus placebo | 3 | 1502 | Risk Ratio (M‐H, Random, 95% CI) | 24.87 [14.20, 43.55] |
| 6.2 Tildrakizumab versus placebo | 1 | 355 | Risk Ratio (M‐H, Random, 95% CI) | 15.63 [2.24, 109.29] |
| 7 Other biologics Show forest plot | 1 | 225 | Risk Ratio (M‐H, Random, 95% CI) | 12.26 [0.76, 197.54] |
| 7.1 Itolizumab versus placebo | 1 | 225 | Risk Ratio (M‐H, Random, 95% CI) | 12.26 [0.76, 197.54] |
| 8 Biologic versus conventional systemic treatments Show forest plot | 4 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 8.1 Etanercept versus acitretin | 1 | 60 | Risk Ratio (M‐H, Random, 95% CI) | 11.00 [0.64, 190.53] |
| 8.2 Infliximab versus methotrexate | 1 | 868 | Risk Ratio (M‐H, Random, 95% CI) | 2.86 [2.15, 3.80] |
| 8.3 Adalimumab versus methotrexate | 1 | 218 | Risk Ratio (M‐H, Random, 95% CI) | 3.73 [2.25, 6.19] |
| 8.4 Alefacept versus methotrexate | 1 | 212 | Risk Ratio (M‐H, Random, 95% CI) | 1.12 [0.42, 2.98] |
| 9 Biologic 1 versus biologic 2 Show forest plot | 10 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 9.1 Ustekinumab versus Etanercept | 1 | 903 | Risk Ratio (M‐H, Random, 95% CI) | 1.80 [1.45, 2.24] |
| 9.2 Secukinumab versus etanercept | 1 | 980 | Risk Ratio (M‐H, Random, 95% CI) | 2.32 [1.85, 2.92] |
| 9.3 Ixekizumab versus etanercept | 2 | 2209 | Risk Ratio (M‐H, Random, 95% CI) | 2.98 [2.24, 3.98] |
| 9.4 Secukinumab versus ustekinumab | 1 | 676 | Risk Ratio (M‐H, Random, 95% CI) | 1.38 [1.23, 1.53] |
| 9.5 Brodalumab versus ustekinumab | 2 | 3088 | Risk Ratio (M‐H, Random, 95% CI) | 1.27 [1.16, 1.39] |
| 9.6 Guselkumab versus adalimumab | 3 | 1658 | Risk Ratio (M‐H, Random, 95% CI) | 1.41 [1.25, 1.60] |
| 10 Small molecules versus placebo Show forest plot | 9 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 10.1 Apremilast versus placebo | 4 | 1775 | Risk Ratio (M‐H, Random, 95% CI) | 6.78 [3.03, 15.17] |
| 10.2 Tofacitinib versus placebo | 4 | 2826 | Risk Ratio (M‐H, Random, 95% CI) | 6.80 [3.86, 11.99] |
| 10.3 Ponesimod versus placebo | 1 | 326 | Risk Ratio (M‐H, Random, 95% CI) | 6.60 [1.65, 26.41] |
| 11 Biologic versus small molecules Show forest plot | 2 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 11.1 Etanercept versus Tofacitinib | 1 | 998 | Risk Ratio (M‐H, Random, 95% CI) | 1.13 [0.93, 1.38] |
| 11.2 Etanercept versus apremilast | 1 | 166 | Risk Ratio (M‐H, Random, 95% CI) | 1.42 [0.72, 2.78] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Conventional systemic agents versus placebo Show forest plot | 3 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 1.1 Methotrexate | 2 | 283 | Risk Ratio (M‐H, Random, 95% CI) | 0.16 [0.03, 0.88] |
| 1.2 Fumaric acid esters | 1 | 704 | Risk Ratio (M‐H, Random, 95% CI) | 0.83 [0.31, 2.21] |
| 2 Anti‐TNF alpha versus placebo Show forest plot | 20 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 2.1 Etanercept versus placebo | 11 | 3783 | Risk Ratio (M‐H, Random, 95% CI) | 1.06 [0.61, 1.83] |
| 2.2 Adalimumab versus placebo | 8 | 3199 | Risk Ratio (M‐H, Random, 95% CI) | 1.02 [0.60, 1.73] |
| 2.3 Certolizumab versus placebo | 1 | 176 | Risk Ratio (M‐H, Random, 95% CI) | 0.49 [0.10, 2.36] |
| 3 Ustekinumab versus placebo Show forest plot | 8 | 4154 | Risk Ratio (M‐H, Random, 95% CI) | 0.89 [0.50, 1.58] |
| 4 Anti‐IL17 versus placebo Show forest plot | 16 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 4.1 Secukinumab versus placebo | 7 | 2707 | Risk Ratio (M‐H, Random, 95% CI) | 1.67 [0.79, 3.53] |
| 4.2 Ixekizumab versus placebo | 4 | 3268 | Risk Ratio (M‐H, Random, 95% CI) | 1.23 [0.65, 2.32] |
| 4.3 Brodalumab versus placebo | 5 | 4109 | Risk Ratio (M‐H, Random, 95% CI) | 0.92 [0.52, 1.61] |
| 5 Anti‐IL23 versus placebo Show forest plot | 4 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 5.1 Guselkumab versus placebo | 3 | 1502 | Risk Ratio (M‐H, Random, 95% CI) | 1.21 [0.51, 2.85] |
| 5.2 Tildrakizumab versus placebo | 1 | 355 | Risk Ratio (M‐H, Random, 95% CI) | 1.36 [0.07, 24.94] |
| 6 Other biologics Show forest plot | 2 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 6.1 Alefacept versus placebo | 2 | 736 | Risk Ratio (M‐H, Random, 95% CI) | 0.74 [0.34, 1.62] |
| 7 Biologic versus conventional systemic treatments Show forest plot | 3 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 7.1 Etanercept versus acitretin | 1 | 60 | Risk Ratio (M‐H, Random, 95% CI) | 0.0 [0.0, 0.0] |
| 7.2 Infliximab versus methotrexate | 1 | 868 | Risk Ratio (M‐H, Random, 95% CI) | 2.41 [1.04, 5.59] |
| 7.3 Adalimumab versus methotrexate | 1 | 218 | Risk Ratio (M‐H, Random, 95% CI) | 2.04 [0.19, 22.14] |
| 8 Biologic 1 versus biologic 2 Show forest plot | 10 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 8.1 Ustekinumab versus etanercept | 1 | 903 | Risk Ratio (M‐H, Random, 95% CI) | 1.25 [0.38, 4.11] |
| 8.2 Secukinumab versus etanercept | 1 | 980 | Risk Ratio (M‐H, Random, 95% CI) | 1.08 [0.41, 2.82] |
| 8.3 Ixekizumab versus etanercept | 2 | 2209 | Risk Ratio (M‐H, Random, 95% CI) | 1.07 [0.55, 2.06] |
| 8.4 Secukinumab versus ustekinumab | 1 | 676 | Risk Ratio (M‐H, Random, 95% CI) | 1.01 [0.42, 2.39] |
| 8.5 Brodalumab versus ustekinumab | 2 | 3088 | Risk Ratio (M‐H, Random, 95% CI) | 1.51 [0.64, 3.56] |
| 8.6 Guselkumab versus adalimumab | 3 | 1658 | Risk Ratio (M‐H, Random, 95% CI) | 0.90 [0.44, 1.82] |
| 9 Small molecules versus placebo Show forest plot | 11 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 9.1 Apremilast versus placebo | 5 | 2036 | Risk Ratio (M‐H, Random, 95% CI) | 0.79 [0.41, 1.49] |
| 9.2 Tofacitinib versus placebo | 5 | 2838 | Risk Ratio (M‐H, Random, 95% CI) | 1.05 [0.53, 2.06] |
| 9.3 Ponesimod versus placebo | 1 | 326 | Risk Ratio (M‐H, Random, 95% CI) | 2.59 [0.34, 19.85] |
| 10 Biologic versus small molecules Show forest plot | 2 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 10.1 Etanercept versus tofacitinib | 1 | 998 | Risk Ratio (M‐H, Random, 95% CI) | 1.15 [0.46, 2.89] |
| 10.2 Etanercept versus apremilast | 1 | 166 | Risk Ratio (M‐H, Random, 95% CI) | 0.33 [0.04, 3.14] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Conventional systemic agents versus placebo Show forest plot | 3 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 1.1 Methotrexate | 2 | 283 | Risk Ratio (M‐H, Random, 95% CI) | 2.36 [1.19, 4.68] |
| 1.2 Fumaric acid esters | 1 | 704 | Risk Ratio (M‐H, Random, 95% CI) | 2.56 [1.68, 3.89] |
| 2 Conventional systemic 1 versus conventional systemic 2 Show forest plot | 3 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 2.1 Ciclosporin versus methotrexate | 2 | 172 | Risk Ratio (M‐H, Random, 95% CI) | 1.37 [0.84, 2.23] |
| 2.2 Methotrexate versus fumaric acid esters | 1 | 60 | Risk Ratio (M‐H, Random, 95% CI) | 1.2 [0.41, 3.51] |
| 3 Anti‐TNF alpha versus placebo Show forest plot | 22 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 3.1 Etanercept versus placebo | 13 | 5066 | Risk Ratio (M‐H, Random, 95% CI) | 8.55 [6.94, 10.52] |
| 3.2 Adalimumab versus placebo | 8 | 3199 | Risk Ratio (M‐H, Random, 95% CI) | 9.08 [6.52, 12.65] |
| 3.3 Certolizumab versus placebo | 1 | 176 | Risk Ratio (M‐H, Random, 95% CI) | 11.31 [4.37, 29.24] |
| 4 Ustekinumab versus placebo Show forest plot | 8 | 4154 | Risk Ratio (M‐H, Random, 95% CI) | 12.41 [8.69, 17.71] |
| 5 Anti‐IL17 versus placebo Show forest plot | 16 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 5.1 Secukinumab versus placebo | 7 | 2707 | Risk Ratio (M‐H, Random, 95% CI) | 15.70 [11.27, 21.87] |
| 5.2 Ixekizumab versus placebo | 4 | 3268 | Risk Ratio (M‐H, Random, 95% CI) | 17.44 [10.45, 29.10] |
| 5.3 Brodalumab versus placebo | 5 | 4109 | Risk Ratio (M‐H, Random, 95% CI) | 12.80 [8.46, 19.36] |
| 6 Anti‐IL23 versus placebo Show forest plot | 4 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 6.1 Guselkumab versus Placebo | 3 | 1502 | Risk Ratio (M‐H, Random, 95% CI) | 12.28 [8.79, 17.17] |
| 6.2 Tildrakizumab versus placebo | 1 | 355 | Risk Ratio (M‐H, Random, 95% CI) | 14.51 [3.73, 56.45] |
| 7 Other biologics Show forest plot | 3 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 7.1 Alefacept versus placebo | 2 | 736 | Risk Ratio (M‐H, Random, 95% CI) | 2.95 [1.76, 4.94] |
| 7.2 Itolizumab versus placebo | 1 | 225 | Risk Ratio (M‐H, Random, 95% CI) | 13.23 [1.88, 92.93] |
| 8 Biologic versus conventional systemic treatments Show forest plot | 4 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 8.1 Etanercept versus acitretin | 1 | 60 | Risk Ratio (M‐H, Random, 95% CI) | 2.13 [1.09, 4.16] |
| 8.2 Alefacept versus methotrexate | 1 | 212 | Risk Ratio (M‐H, Random, 95% CI) | 0.74 [0.46, 1.21] |
| 8.3 Infliximab versus methotrexate | 1 | 868 | Risk Ratio (M‐H, Random, 95% CI) | 1.86 [1.58, 2.19] |
| 8.4 Adalimumab versus methotrexate | 1 | 218 | Risk Ratio (M‐H, Random, 95% CI) | 2.25 [1.72, 2.94] |
| 9 Biologic 1 versus biologic 2 Show forest plot | 10 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 9.1 Ustekinumab versus etanercept | 1 | 903 | Risk Ratio (M‐H, Random, 95% CI) | 1.26 [1.13, 1.40] |
| 9.2 Secukinumab versus etanercept | 1 | 980 | Risk Ratio (M‐H, Random, 95% CI) | 1.64 [1.44, 1.88] |
| 9.3 Ixekizumab versus etanercept | 2 | 2209 | Risk Ratio (M‐H, Random, 95% CI) | 1.79 [1.43, 2.24] |
| 9.4 Secukinumab versus ustekinumab | 1 | 676 | Risk Ratio (M‐H, Random, 95% CI) | 1.13 [1.06, 1.20] |
| 9.5 Brodalumab versus ustekinumab | 2 | 3088 | Risk Ratio (M‐H, Random, 95% CI) | 1.10 [1.04, 1.17] |
| 9.6 Guselkumab versus adalimumab | 3 | 1658 | Risk Ratio (M‐H, Random, 95% CI) | 1.21 [1.13, 1.30] |
| 10 Small molecules versus placebo Show forest plot | 11 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 10.1 Apremilast versus placebo | 5 | 2036 | Risk Ratio (M‐H, Random, 95% CI) | 3.88 [2.42, 6.22] |
| 10.2 Tofacitinib versus placebo | 5 | 2838 | Risk Ratio (M‐H, Random, 95% CI) | 6.41 [3.84, 10.71] |
| 10.3 Ponesimod versus placebo | 1 | 326 | Risk Ratio (M‐H, Random, 95% CI) | 3.51 [1.88, 6.53] |
| 11 Biologic versus small molecules Show forest plot | 2 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 11.1 Etanercept versus Tofacitinib | 1 | 998 | Risk Ratio (M‐H, Random, 95% CI) | 1.14 [1.02, 1.28] |
| 11.2 Etanercept versus apremilast | 1 | 166 | Risk Ratio (M‐H, Random, 95% CI) | 1.21 [0.86, 1.71] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Conventional systemic agents versus placebo Show forest plot | 3 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 1.1 Methotrexate | 2 | 283 | Risk Ratio (M‐H, Random, 95% CI) | 2.94 [1.47, 5.89] |
| 1.2 Fumaric acid esters | 1 | 704 | Risk Ratio (M‐H, Random, 95% CI) | 2.73 [1.72, 4.32] |
| 2 Conventional systemic 1 versus conventional systemic 2 Show forest plot | 1 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 2.1 Ciclosporin versus methotrexate | 1 | 88 | Risk Ratio (M‐H, Random, 95% CI) | 0.82 [0.47, 1.46] |
| 3 Anti‐TNF alpha versus placebo Show forest plot | 19 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 3.1 Etanercept versus placebo | 11 | 4334 | Risk Ratio (M‐H, Random, 95% CI) | 7.77 [5.98, 10.10] |
| 3.2 Adalimumab versus placebo | 7 | 3051 | Risk Ratio (M‐H, Random, 95% CI) | 8.38 [6.28, 11.18] |
| 3.3 Certolizumab versus placebo | 1 | 176 | Risk Ratio (M‐H, Random, 95% CI) | 35.88 [5.11, 251.73] |
| 4 Ustekinumab versus placebo Show forest plot | 8 | 4154 | Risk Ratio (M‐H, Random, 95% CI) | 11.33 [7.38, 17.39] |
| 5 Anti‐IL17 versus placebo Show forest plot | 15 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 5.1 Secukinumab versus placebo | 6 | 2607 | Risk Ratio (M‐H, Random, 95% CI) | 17.16 [7.48, 39.36] |
| 5.2 Ixekizumab versus placebo | 4 | 3268 | Risk Ratio (M‐H, Random, 95% CI) | 17.46 [9.87, 30.90] |
| 5.3 Brodalumab versus placebo | 5 | 4109 | Risk Ratio (M‐H, Random, 95% CI) | 18.78 [13.29, 26.55] |
| 6 Anti‐IL23 versus placebo Show forest plot | 4 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 6.1 Guselkumab versus placebo | 3 | 1502 | Risk Ratio (M‐H, Random, 95% CI) | 10.59 [7.73, 14.51] |
| 6.2 Tildrakizumab versus placebo | 1 | 355 | Risk Ratio (M‐H, Random, 95% CI) | 27.54 [3.95, 191.78] |
| 7 Other biologics Show forest plot | 2 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 7.1 Alefacept versus placebo | 1 | 507 | Risk Ratio (M‐H, Random, 95% CI) | 2.54 [1.22, 5.29] |
| 7.2 Itolizumab versus placebo | 1 | 225 | Risk Ratio (M‐H, Random, 95% CI) | 3.78 [0.94, 15.17] |
| 8 Biologic versus conventional systemic treatments Show forest plot | 3 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 8.1 Alefacept versus methotrexate | 1 | 212 | Risk Ratio (M‐H, Random, 95% CI) | 0.69 [0.37, 1.29] |
| 8.2 Infliximab versus methotrexate | 1 | 868 | Risk Ratio (M‐H, Random, 95% CI) | 1.99 [1.67, 2.37] |
| 8.3 Adalimumab versus methotrexate | 1 | 218 | Risk Ratio (M‐H, Random, 95% CI) | 2.44 [1.79, 3.32] |
| 9 Biologic 1 versus biologic 2 Show forest plot | 10 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 9.1 Ustekinumab versus etanercept | 1 | 903 | Risk Ratio (M‐H, Random, 95% CI) | 1.40 [1.24, 1.58] |
| 9.2 Secukinumab versus etanercept | 1 | 980 | Risk Ratio (M‐H, Random, 95% CI) | 2.09 [1.73, 2.53] |
| 9.3 Ixekizumab versus etanercept | 2 | 2209 | Risk Ratio (M‐H, Random, 95% CI) | 2.01 [1.74, 2.31] |
| 9.4 Secukinumab versus ustekinumab | 1 | 676 | Risk Ratio (M‐H, Random, 95% CI) | 1.23 [1.13, 1.35] |
| 9.5 Brodalumab versus ustekinumab | 2 | 3088 | Risk Ratio (M‐H, Random, 95% CI) | 1.17 [1.07, 1.27] |
| 9.6 Guselkumab versus adalimumab | 3 | 1658 | Risk Ratio (M‐H, Random, 95% CI) | 1.24 [1.17, 1.32] |
| 10 Small molecules versus placebo Show forest plot | 10 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 10.1 Apremilast versus placebo | 4 | 1776 | Risk Ratio (M‐H, Random, 95% CI) | 3.88 [2.04, 7.38] |
| 10.2 Tofacitinib versus placebo | 5 | 2838 | Risk Ratio (M‐H, Random, 95% CI) | 4.48 [3.51, 5.71] |
| 10.3 Ponesimod versus placebo | 1 | 326 | Risk Ratio (M‐H, Random, 95% CI) | 6.73 [2.19, 20.64] |
| 11 Biologic versus small molecules Show forest plot | 2 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 11.1 Etanercept versus tofacitinib | 1 | 998 | Risk Ratio (M‐H, Random, 95% CI) | 1.15 [1.04, 1.27] |
| 11.2 Etanercept versus apremilast | 1 | 166 | Risk Ratio (M‐H, Random, 95% CI) | 1.33 [0.78, 2.27] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Conventional systemic agents versus placebo Show forest plot | 2 | Std. Mean Difference (IV, Random, 95% CI) | Subtotals only | |
| 1.1 Methotrexate | 2 | 283 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.67 [‐1.40, 0.06] |
| 2 Anti‐TNF alpha versus placebo Show forest plot | 14 | Std. Mean Difference (IV, Random, 95% CI) | Subtotals only | |
| 2.1 Etanercept versus placebo | 7 | 2779 | Std. Mean Difference (IV, Random, 95% CI) | ‐1.10 [‐1.37, ‐0.83] |
| 2.2 Adalimumab versus placebo | 7 | 2774 | Std. Mean Difference (IV, Random, 95% CI) | ‐1.02 [‐1.16, ‐0.88] |
| 3 Ustekinumab versus placebo Show forest plot | 6 | 2917 | Std. Mean Difference (IV, Random, 95% CI) | ‐1.21 [‐1.39, ‐1.03] |
| 4 Anti‐IL17 versus placebo Show forest plot | 5 | Std. Mean Difference (IV, Random, 95% CI) | Subtotals only | |
| 4.1 Ixekizumab versus placebo | 3 | 3126 | Std. Mean Difference (IV, Random, 95% CI) | ‐1.76 [‐2.09, ‐1.43] |
| 4.2 Brodalumab versus placebo | 2 | 349 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.96 [‐1.44, ‐0.47] |
| 5 Anti‐IL23 versus placebo Show forest plot | 3 | Std. Mean Difference (IV, Random, 95% CI) | Subtotals only | |
| 5.1 Guselkumab versus placebo | 2 | 1252 | Std. Mean Difference (IV, Random, 95% CI) | ‐1.39 [‐1.63, ‐1.14] |
| 5.2 Tildrakizumab versus placebo | 1 | 355 | Std. Mean Difference (IV, Random, 95% CI) | ‐1.23 [‐1.55, ‐0.91] |
| 6 Other biologics Show forest plot | 2 | Std. Mean Difference (IV, Random, 95% CI) | Subtotals only | |
| 6.1 Alefacept versus placebo | 1 | 229 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.32 [‐0.62, ‐0.02] |
| 6.2 Itolizumab versus placebo | 1 | 225 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.34 [‐0.68, ‐0.01] |
| 7 Biologic versus conventional systemic treatments Show forest plot | 2 | Mean Difference (IV, Fixed, 95% CI) | Subtotals only | |
| 7.1 Alefacept versus methotrexate | 1 | 212 | Mean Difference (IV, Fixed, 95% CI) | 1.31 [‐0.28, 2.90] |
| 7.2 Adalimumab versus methotrexate | 1 | 218 | Mean Difference (IV, Fixed, 95% CI) | ‐3.40 [‐5.75, ‐1.05] |
| 8 Biologic 1 versus biologic 2 Show forest plot | 4 | Mean Difference (IV, Fixed, 95% CI) | Subtotals only | |
| 8.1 Ixekizumab versus etanercept | 2 | 2209 | Mean Difference (IV, Fixed, 95% CI) | ‐1.99 [‐2.39, ‐1.59] |
| 8.2 Guselkumab versus adalimumab | 2 | 1407 | Mean Difference (IV, Fixed, 95% CI) | ‐1.73 [‐2.50, ‐0.97] |
| 9 Small molecules versus placebo Show forest plot | 7 | Std. Mean Difference (IV, Random, 95% CI) | Subtotals only | |
| 9.1 Apremilast versus placebo | 3 | 1609 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.62 [‐0.77, ‐0.47] |
| 9.2 Tofacitinib versus placebo | 3 | 2629 | Std. Mean Difference (IV, Random, 95% CI) | ‐1.09 [‐1.28, ‐0.89] |
| 9.3 Ponesimod versus placebo | 1 | 326 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.58 [‐0.86, ‐0.31] |
| 10 Biologic versus small molecules Show forest plot | 1 | Std. Mean Difference (IV, Random, 95% CI) | Subtotals only | |
| 10.1 Etanercept versus Tofacitinib | 1 | 998 | Std. Mean Difference (IV, Random, 95% CI) | ‐0.06 [‐0.19, 0.07] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Conventional systemic agents versus placebo Show forest plot | 3 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 1.1 Methotrexate | 2 | 283 | Risk Ratio (M‐H, Random, 95% CI) | 0.94 [0.81, 1.10] |
| 1.2 Fumaric acid esters | 1 | 704 | Risk Ratio (M‐H, Random, 95% CI) | 1.40 [1.22, 1.62] |
| 2 Conventional systemic 1 versus conventional systemic 2 Show forest plot | 3 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 2.1 Ciclosporin versus methotrexate | 2 | 172 | Risk Ratio (M‐H, Random, 95% CI) | 1.10 [0.90, 1.34] |
| 2.2 Methotrexate versus fumaric acid esters | 1 | 60 | Risk Ratio (M‐H, Random, 95% CI) | 1.13 [0.91, 1.39] |
| 3 Anti‐TNF alpha versus placebo Show forest plot | 17 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 3.1 Etanercept versus placebo | 9 | 3529 | Risk Ratio (M‐H, Random, 95% CI) | 1.12 [1.05, 1.20] |
| 3.2 Adalimumab versus placebo | 7 | 3051 | Risk Ratio (M‐H, Random, 95% CI) | 1.06 [1.00, 1.13] |
| 3.3 Certolizumab versus placebo | 1 | 176 | Risk Ratio (M‐H, Random, 95% CI) | 1.00 [0.81, 1.22] |
| 4 Ustekinumab versus placebo Show forest plot | 8 | 4154 | Risk Ratio (M‐H, Random, 95% CI) | 1.06 [1.00, 1.13] |
| 5 Anti‐IL17 versus placebo Show forest plot | 16 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 5.1 Secukinumab versus placebo | 7 | 2707 | Risk Ratio (M‐H, Random, 95% CI) | 1.15 [1.02, 1.29] |
| 5.2 Ixekizumab versus placebo | 4 | 3268 | Risk Ratio (M‐H, Random, 95% CI) | 1.24 [1.07, 1.45] |
| 5.3 Brodalumab versus placebo | 5 | 4109 | Risk Ratio (M‐H, Random, 95% CI) | 1.15 [1.00, 1.32] |
| 6 Anti‐IL23 versus placebo Show forest plot | 4 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 6.1 Guselkumab versus placebo | 3 | 1502 | Risk Ratio (M‐H, Random, 95% CI) | 1.03 [0.92, 1.16] |
| 6.2 Tildrakizumab versus placebo | 1 | 355 | Risk Ratio (M‐H, Random, 95% CI) | 0.95 [0.76, 1.18] |
| 7 Biologic versus conventional systemic treatments Show forest plot | 2 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 7.1 Infliximab versus methotrexate | 1 | 868 | Risk Ratio (M‐H, Random, 95% CI) | 1.08 [0.97, 1.20] |
| 7.2 Adalimumab versus methotrexate | 1 | 218 | Risk Ratio (M‐H, Random, 95% CI) | 0.90 [0.78, 1.05] |
| 8 Biologic 1 versus biologic 2 Show forest plot | 10 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 8.1 Ustekinumab versus etanercept | 1 | 903 | Risk Ratio (M‐H, Random, 95% CI) | 0.97 [0.89, 1.06] |
| 8.2 Secukinumab versus etanercept | 1 | 980 | Risk Ratio (M‐H, Random, 95% CI) | 1.00 [0.89, 1.12] |
| 8.3 Ixekizumab versus etanercept | 2 | 2209 | Risk Ratio (M‐H, Random, 95% CI) | 1.06 [0.97, 1.15] |
| 8.4 Secukinumab versus ustekinumab | 1 | 676 | Risk Ratio (M‐H, Random, 95% CI) | 1.10 [0.98, 1.25] |
| 8.5 Brodalumab versus ustekinumab | 2 | 3088 | Risk Ratio (M‐H, Random, 95% CI) | 1.00 [0.93, 1.09] |
| 8.6 Guselkumab versus adalimumab | 3 | 1658 | Risk Ratio (M‐H, Random, 95% CI) | 0.97 [0.88, 1.07] |
| 9 Small molecules versus placebo Show forest plot | 10 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 9.1 Apremilast versus placebo | 5 | 2036 | Risk Ratio (M‐H, Random, 95% CI) | 1.22 [1.07, 1.38] |
| 9.2 Tofacitinib versus placebo | 4 | 2641 | Risk Ratio (M‐H, Random, 95% CI) | 1.14 [1.03, 1.25] |
| 9.3 Ponesimod versus placebo | 1 | 326 | Risk Ratio (M‐H, Random, 95% CI) | 1.31 [1.02, 1.68] |
| 10 Biologic versus small molecules Show forest plot | 2 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 10.1 Etanercept versus tofacitinib | 1 | 998 | Risk Ratio (M‐H, Random, 95% CI) | 1.00 [0.89, 1.12] |
| 10.2 Etanercept versus apremilast | 1 | 166 | Risk Ratio (M‐H, Random, 95% CI) | 1.32 [1.03, 1.69] |
