Systemic non-biological therapy

National Clinical Guideline Centre (UK)

See 2019 Partial Update

Update information September 2017: The guideline has been revised throughout to link to MHRA advice and NICE technology appraisals that have been completed since original publication. Minor updates since publication August 2019: Links to the MHRA safety advice on the risk of using retinoids in pregnancy have been updated to the June 2019 version.

10Systemic non-biological therapy

Publication Details

Systemic therapy³⁰⁷ is invariably indicated in patients with life-threatening forms of unstable psoriasis such as generalised pustular psoriasis and erythroderma; these are rare. Systemic therapy is more commonly used in people with extensive stable plaque psoriasis where topical therapy would be impractical and potentially unsafe and where phototherapy is not appropriate or has failed (see chapter 6). People with localised plaque psoriasis associated with significant functional impairment and/or psychological distress (for example severe nail disease, hand and foot involvement), palmoplantar pustulosis and extensive ‘guttate type’ psoriasis may also benefit from systemic therapy. The presence of psoriatic arthritis can have a major influence on when systemic therapy is considered in the treatment pathway for skin psoriasis and the choice of agent is also critical since acitretin and fumaric acid esters have no benefit in psoriatic arthritis, in contrast to, for example methotrexate. Accurate UK data on the proportion of people with psoriasis who are treated with systemic therapy is not available. In one US based study, the proportion of people with BSA >10% was 5.25% of all people with psoriasis²⁰⁴ and could be used as a crude surrogate indicator of those potentially suitable for systemic therapy but is likely to be inaccurate.

Ciclosporin (CSA), methotrexate (MTX), acitretin and fumaric acid esters are the most commonly used systemic therapies to treat psoriasis and will be referred to as systemic non-biological therapies for clarity. In other inflammatory diseases, induction of remission and maintenance therapy are often considered separately. Recent European guidelines for the treatment of psoriasis have adopted this approach in considering achievement of PASI 75 over 12–16 weeks³⁰⁷. In practice, once satisfactory control is achieved, the same treatment is continued at the minimal effective dose in order to maintain disease control and quality of life. Ciclosporin is the exception to this given the predictable nephrotoxic effects of the drug with continuous use, and is not generally considered suitable for long-term disease management. All the interventions can be complicated by poor tolerability, short and long-term toxicity and poor or inadequate efficacy. Supplementary treatment with topicals is commonly required.

Which agent to choose is influenced by multiple factors and must be tailored to the needs of the individual. The type and pattern of psoriasis, extent of involvement and whether or not rapid control is necessary are important. For example, stable chronic plaque psoriasis requires a very different treatment strategy to generalised pustular psoriasis. The presence of psoriatic arthritis, comorbidities, age, conception plans, preferences of patient and clinician, logistical issues around safe drug administration and monitoring as well as many other factors also need to be taken into account. Nevertheless, it is useful to review the evidence on the relative efficacy and safety of the available agents to inform the decision-making process.

The evidence review excluded data on fumaric acid esters as this is not licensed for any indication in the UK and therefore falls outside the agreed standard operating procedures for NICE guidelines.

The GDG agreed to ask the following question: in people with psoriasis (all types), what are the clinical effectiveness, safety, tolerability and cost effectiveness of systemic methotrexate, ciclosporin and acitretin compared with each other or with placebo?

10.1. Methodological introduction

A literature search was conducted for randomised controlled trials or systematic reviews that compared the efficacy and safety of methotrexate, ciclosporin and acitretin with each other or with placebo/no treatment for the induction or maintenance of remission in people with psoriasis. Comparisons of different doses of a particular treatment and of different maintenance schedules were also sought. Additionally, long-term safety data was sought from cohort or case control studies.

No time limit was placed on the literature search and there were no limitations on duration of follow-up. Indirect populations were excluded as were studies with a sample size of less than 10.

The outcomes considered were:

PASI75
PASI50
Change in PASI (mean improvement) or final PASI as a surrogate outcome
Clear or nearly clear (minimal residual activity[MRA]/PASI>90/0 or 1 on PGA)
Improved (for PPP population only)
Time-to-relapse (loss of PASI50)
Time-to-remission/max response
Change in DLQI
Severe adverse events
Specific adverse events were assessed for each intervention (methotrexate: hepatotoxicity, marrow suppression and pneumonitis; acitretin: hyperlipidaemia, hepatotoxicity, skeletal AEs and cheilitis; ciclosporin: renal impairment, hypertension, gout and hyperuricaemia)
Withdrawal due to toxicity

Twenty eight RCTs were found that addressed the question and were included in the review. There was no suitable long-term observational data and no studies were available that assessed systemic non-biological therapy in an exclusively paediatric population. The studies differed in terms of the disease severity stated as an inclusion criterion (Table 137).

Table 137

Disease severity inclusion and dosing schedules of included studies.

The systematic review protocol specified clear or nearly clear disease as an outcome and this was defined as either: i) minimal residual activity; ii) PASI90; or III) 0 or 1 on PGA. The data from the studies identified for this section showed that PASI90 and 0 or 1 on PGA were not equivalent outcomes. PASI90 was found to be a more stringent criterion of response. For this reason both outcomes are reported separately.

10.2. Methotrexate vs placebo for induction of remission

10.2.1. Evidence profile

Table 138

Evidence profile comparing methotrexate vs placebo for induction of remission.

10.2.2. Evidence statements

In people with psoriasis, incrementally dosed methotrexate was statistically significantly better than placebo for:

Clear/nearly clear (PGA) at 16 weeks [1 study; 156 participants; moderate quality evidence]³⁵³
PASI75 at 4–6 months [2 studies; 192 participants; moderate quality evidence]¹⁵⁰^,³⁵³
PASI50 at 4–6 months [2 studies; 192 participants; moderate quality evidence]¹⁵⁰^,³⁵³
PASI change/final score at 4–6 months [2 studies; 192 participants; moderate quality evidence]¹⁵⁰^,³⁵³

In people with psoriasis, there was no statistically significant difference between incrementally dosed methotrexate and placebo for:

PASI90 at 16 weeks [1 study; 156 participants; very low quality evidence]³⁵³
Severe adverse events at 26 weeks [1 study; 163 participants; very low quality evidence]³⁵³
Withdrawal due to toxicity at 26 weeks [1 study; 159 participants; very low quality evidence]³⁵³
Raised liver enzymes at 26 weeks [1 study; 163 participants; very low quality evidence]³⁵³

10.3. Methotrexate vs ciclosporin for induction of remission

10.3.1. Evidence profile

Table

Only ITT data were available for the Flytstrom and Heydendael studies, and the assumptions were not stated so it was not possible to use an available case analysis.

The dosing schedules were considered clinically similar enough to pool in the Flytstrom and Heydendael studies, but the Sandhu study was considered to be different. Therefore, data from Flytstrom and Heydendael were pooled unless there was significant heterogeneity.

10.3.2. Evidence statements

In people with psoriasis, ciclosporin was statistically significantly better than methotrexate for:

PASI75 at 12 weeks (incremental MTX dose; 7.5 up to 15 mg/wk) [1 study; 68 participants; moderate quality evidence]¹⁰⁴
PASI50 at 12 weeks (incremental MTX dose; 7.5 up to 15 mg/wk) [1 study; 68 participants; moderate quality evidence]¹⁰⁴
Final PASI at 12–16 weeks (incremental dose MTX within licensed range; maximum 22.5 mg/wk) [2 studies; 153 participants; low quality evidence]¹⁰⁴^,¹⁴⁷
Elevated liver enzymes at 12–24 weeks (MTX dose within licensed range; maximum 22.5 mg/wk) [3 studies; 190 participants; moderate quality evidence]¹⁰⁴^,¹³⁴^,¹⁴⁷
Withdrawal due to toxicity at 16 weeks (incremental dose MTX; 15 up to 22.5 mg/wk) [1 study; 85 participants; high quality evidence]¹⁴⁷

In people with psoriasis, methotrexate was statistically significantly better than ciclosporin for:

Final PASI at 12 weeks (high dose MTX; 0.5 mg/kg/wk) [1 study; 30 participants; very low quality evidence]³⁵⁰
Clearance at 10 weeks (high dose MTX; 0.5 mg/kg/wk) [1 study; 30 participants; very low quality evidence]³⁵⁰
Elevated creatinine at 12–24 weeks (standard MTX dose range; maximum 15 mg/wk) [2 studies; 105 participants; moderate quality evidence]¹⁰⁴^,¹³⁴

In people with psoriasis, there was no statistically significant difference between ciclosporin and methotrexate for:

Clear/nearly clear (PASI90) at 12 weeks (incremental MTX dose; 7.5 up to 15 mg/wk) [1 study; 68 participants; low quality evidence]¹⁰⁴
Clear/nearly clear (PASI90) at 16 weeks (incremental dose MTX; 15 up to 22.5 mg/wk) [1 study; 85 participants; very low quality evidence]¹⁴⁷
Time-to-PASI75 (incremental dose MTX; 15 up to 22.5 mg/wk) after follow-up for a maximum of 16 weeks [1 study; 85 participants; low quality evidence]¹⁴⁷
Time-to-PASI90 (incremental dose MTX; 15 up to 22.5 mg/wk) after follow-up for a maximum of 16 weeks [1 study; 85 participants; very low quality evidence]¹⁴⁷
PASI75 at 16 weeks (incremental dose MTX; 15 up to 22.5 mg/wk) [1 study; 85 participants; low quality evidence]¹⁴⁷
Remaining clear at 12 weeks (after tapering) [1 study; 19 participants; very low quality evidence]³⁵⁰
Change in NAPSI (decreasing MTX dose; 15 mg/wk reduced to 10 mg/wk) at 6 months [1 study; 37 participants; low quality evidence]¹³⁴
Hypertension at 16 weeks (incremental dose MTX; 15 up to 22.5 mg/wk) [1 study; 85 participants; very low quality evidence]¹⁴⁷
Hypertension at 12 weeks (high dose MTX; 0.5 mg/kg/wk) [1 study; 30 participants; very low quality evidence]³⁵⁰
Withdrawal due to toxicity at 12–16 weeks (standard MTX dose range; maximum 15 mg/wk) [2 studies; 105 participants; low quality evidence]¹⁰⁴^,¹³⁴

Evidence statements for individual studies where insufficient data were available to perform original statistical analysis comparing ciclosporin and methotrexate in people with psoriasis:

Percentage change in DLQI from baseline to 12 weeks was statistically significantly better with ciclosporin than methotrexate (incremental dose; 7.5 up to 15 mg/wk) at 8 weeks [1 study; 68 participants; low quality evidence]¹⁰⁴
There was no significant difference between ciclosporin and methotrexate (incremental dose; 7.5 up to 15 mg/wk) for change in DLQI from baseline to 12 weeks [1 study; 68 participants; very low quality evidence]¹⁰⁴
There was no significant difference between ciclosporin and methotrexate (incremental dose; 15 up to 22.5 mg/wk) in median time to relapse after a maximum follow-up of 8 weeks post-treatment [1 study; 85 participants; low quality evidence]¹⁴⁷

10.3.3. Subgroups and heterogeneity

Heterogeneity was present for the outcomes of clear or nearly clear, PASI75, final PASI and withdrawal due to toxicity between three studies¹⁰⁴^,¹⁴⁷^,³⁵⁰. This was thought to be due to the different dosing regimens of methotrexate used in the included studies, as the estimate of efficacy moved towards favouring methotrexate compared with ciclosporin as the dose of methotrexate used increased (while the dose of ciclosporin was similar among the studies). Conversely, there were relatively more withdrawals due to toxicity with higher dose methotrexate compared with ciclosporin. However, it is also possible that the differences were caused or contributed to by the differences in the use of folic acid. The Flytstrom study¹⁰⁴, which also used the lowest dosing schedule, was the only one to have administered folic acid which may have reduced the efficacy of methotrexate while also making it more tolerable.

It was unclear why there was no heterogeneity between the Heydendael and Flytstrom studies for the outcome of final PASI in contrast to the outcome of PASI75. However, the final scores do mask a slightly greater difference in the change in PASI between the two studies owing to baseline differences, with the difference in change scores between the methotrexate and ciclosporin groups being greater in the Flytstrom study in which methotrexate showed lower efficacy than in the Heydendael study (the percentage change in PASI was greater in the ciclosporin group by 16.5% in the Flytstrom study but 10.2% in the Heydendael study).

10.4. Acitretin vs placebo for induction of remission

10.4.1. Evidence profile

Table

10.4.2. Evidence statements

In people with psoriasis, acitretin was statistically significantly better than placebo for:

PASI75 (50 mg acitretin) at 8 weeks [2 studies; 63 participants; very low quality evidence]¹²³^,²¹³

In people with psoriasis, acitretin was statistically significantly more likely than placebo to result in:

Cheilitis at 8 weeks (10, 25 and 50 mg acitretin) [2 studies; 54, 53 and 60 participants, respectively; very low quality evidence]¹²³^,²¹³
Cheilitis at 8 weeks (75 mg acitretin) [1 study; 17 participants; very low quality evidence]¹²³
Cheilitis at 6 months (10, 25 and 50 mg acitretin) [1 study; 40 participants; very low quality evidence]²¹³
Hair loss at 8 weeks (50 mg acitretin) [2 studies; 60 participants; very low quality evidence]¹²³^,²¹³
Hair loss at 6 months (50 mg acitretin) [1 study; 40 participants; very low quality evidence]²¹³

In people with psoriasis, there was no statistically significant difference between acitretin and placebo for:

PASI75 at 8 weeks (10 and 25 mg acitretin) [2 studies; 57 participants; very low quality evidence]¹²³^,²¹³
PASI75 at 8 weeks (75 mg acitretin) [1 study; 17 participants; very low quality evidence]¹²³
Withdrawal due to toxicity at 8 weeks [1 study; 76 participants; very low quality evidence]²¹³
Hair loss at 8 weeks (10 and 25 mg acitretin) [2 studies; 54 and 53 participants, respectively; very low quality evidence] ¹²³^,²¹³
Hair loss at 8 weeks (75 mg acitretin) [1 study; 17 participants; very low quality evidence]¹²³
Hair loss at 6 months (10 and 25 mg acitretin) [1 study; 40 participants; very low quality evidence]²¹³
Increased triglycerides at 8 weeks (10, 25 and 50 mg acitretin) [1 study; 37, 36 and 37 participants, respectively; very low quality evidence]²¹³
Increased triglycerides at 6 months (10, 25 and 50 mg acitretin) [1 study; 35, 34 and 34 participants, respectively; very low quality evidence]²¹³
Increased liver enzymes at 8 weeks (10 mg acitretin) [1 study; 37 participants; very low quality evidence]²¹³
Increased liver enzymes at 6 months (10, 25 and 50 mg acitretin) [1 study; 35, 34 and 34 participants, respectively; very low quality evidence]²¹³
Increased cholesterol at 8 weeks (10, 25 and 50 mg acitretin) [1 study; 37, 36 and 37 participants, respectively; very low quality evidence]²¹³
Increased cholesterol at 6 months (10 and 25 mg acitretin) [1 study; 35 participants; very low quality evidence]²¹³

In people with psoriasis there were no events with either acitretin or placebo for:

Increased liver enzymes at 8 weeks (25 and 50 mg acitretin) [1 study; 37 participants; very low quality evidence]²¹³

Evidence statements for individual studies where insufficient data were available to perform original statistical analysis comparing acitretin and placebo in people with psoriasis:

Acitretin 50 or 75 mg was better than placebo or 10 mg acitretin for improvement in scaling, erythema, thickness and pustulation at 8 weeks [1 study; 21 participants; very low quality evidence]¹⁸⁵
Reduction in PASI at 8 weeks was significantly greater in the groups receiving 25 mg/day and 50 mg/day compared with placebo, but there was no significant difference between the 25 and 50 mg groups. Additionally, the mean percentage decrease in PASI score in the 10 mg group was greater than in the placebo group, but did not differ significantly from 25 or 50 mg groups [1 study; 80 participants; very low quality evidence]²¹³
There was no significant difference in PASI score at 6 months between the placebo, 10, 25 and 50 mg groups at 6 months [1 study; 80 participants; very low quality evidence]²¹³
There were more side effects at higher doses of acitretin at 6 months [1 study; 21 participants; very low quality evidence]¹⁸⁵

10.4.3. Subgroups and heterogeneity

For the outcomes of PASI75, hair loss and cheilitis from two studies¹²³^,²¹³ there was no statistically significant difference between the dose subgroups, suggesting that the increase in efficacy and toxicity is negligible. However, the small size of the studies and wide confidence intervals may mean that the true difference in effect has not been detected, although the point estimates did increase in favour of acitretin for efficacy and in favour of placebo for toxicity as the dose increased.

10.5. Increasing vs decreasing acitretin dosing schedule for induction of remission

10.5.1. Evidence profile

Table

10.5.2. Evidence statements

In people with psoriasis, there was no statistically significant difference between acitretin increasing and decreasing doses for:

Cheilitis at 6 weeks [1 study; 42 participants; low quality evidence]²⁶
Hair loss at 6 weeks [1 study; 42 participants; very low quality evidence]²⁶
Withdrawal due to toxicity at 6 weeks [1 study; 41 participants; very low quality evidence]²⁶

Table 139

Summary of non-analysed data for increasing vs decreasing acitretin dosing.

Evidence statements for individual studies where insufficient data were available to perform original statistical analysis comparing increasing and decreasing acitretin dosing in people with psoriasis:

Decreasing acitretin was slightly better than increasing doses for percentage change in PASI at 6 weeks [1 study; 40 participants; very low quality evidence]²⁶. However, there was no statistically significant difference between the three treatment groups (increasing, decreasing and constant dosing) for percentage improvement in PASI (p=0.42).
The severe adverse reactions at 6 weeks were dose dependent: their frequency and intensity increased progressively with increasing dose and decreased with decreasing dose.
- There were statistically significantly more adverse events for patients using 50 vs 10 mg acitretin [1 study; 42 participants; very low quality evidence] ²⁶

10.6. Increasing vs constant acitretin dosing schedule for induction of remission

10.6.1. Evidence profile

Table

10.6.2. Evidence statements

In people with psoriasis, there was no statistically significant difference between acitretin increasing and constant doses for:

Cheilitis at 6 weeks [1 study; 44 participants; low quality evidence]²⁶
Hair loss at 6 weeks [1 study; 44 participants; very low quality evidence]²⁶
Withdrawal due to toxicity at 6 weeks [1 study; 42 participants; very low quality evidence]²⁶

Evidence statements for individual studies where insufficient data were available to perform original statistical analysis comparing increasing and constant acitretin dosing in people with psoriasis:

Increasing acitretin was slightly better than constant dosing for percentage change in PASI at 6 weeks [1 study; 40 participants; very low quality evidence]²⁶. However, there was no statistically significant difference between the three treatment groups (increasing, decreasing and constant dosing) for percentage improvement in PASI (p=0.42).

10.7. Ciclosporin vs placebo for induction of remission

10.7.1. Evidence profile

Table

10.7.2. Evidence statements

In people with psoriasis, ciclosporin administered for induction of remission was statistically significantly better than placebo for:

Clear/nearly clear on PGA at 8 weeks (3, 5 or 7.5 mg/kg/day) [1 study; 50, 45 and 40 participants, respectively; moderate quality evidence]⁸⁵
PASI75 at 8–10 weeks (2.5–3.0 or 5 mg/kg) [2 studies; 157 participants; moderate quality evidence]⁸⁵^,²⁴⁷
PASI50 at 4–10 weeks [2 studies; 43 participants; low quality evidence]¹³³^,⁴¹³
Mean % change in PASI (1.25 and 2.5 mg/kg/day CSA) [1 study; 79 and 80 participants; moderate quality evidence]²⁴⁷

In people with psoriasis, there was no statistically significant difference between ciclosporin and placebo for:

Clearance at 4 weeks (14 mg/kg/day) [1 study; 21 participants; very low quality evidence]⁸⁶
PASI75 at 10 weeks (1.25 mg/kg) [1 study; 84 participants; very low quality evidence]²⁴⁷
Hypertension at 8–10 weeks [2 studies; 44 participants; very low quality evidence]⁸⁶^,¹³³
Decreased glomerular filtration rate at 8 weeks (3, 5 and 7.5 mg/kg/day) [1 study; 21, 19 and 21 participants, respectively; low to very low quality evidence]⁸⁵

There were no events with either ciclosporin or placebo for:

Withdrawal due to toxicity at 4 weeks [2 studies; 41 participants; moderate quality evidence]⁸⁶^,⁴¹³

Evidence statements for individual studies where no numerical analyses could be performed due to insufficient information comparing ciclosporin and placebo in people with psoriasis:

Ciclosporin (3.0, 5.0 or 7.5 mg/kg/day) administered for induction of remission was statistically significantly better than placebo for improvement in PASI at 8 weeks [1 study; 85 participants; low quality evidence]⁸⁵
Ciclosporin (5.0 or 7.5 mg/kg/day) administered for induction of remission is statistically significantly better than ciclosporin (3.0 mg/kg/day) for improvement in PASI at 8 weeks, but there was no significant difference between 5 and 7.5 mg/kg/day [1 study; 85 participants; low quality evidence]⁸⁵

10.7.3. Subgroups and heterogeneity

For the outcomes of clear/nearly clear on PGA, PASI75 and decrease in glomerular filtration rate from two studies⁸⁵^,²⁴⁷ there was no statistically significant subgroup differences between the ciclosporin doses (3, 5 and 7.5 mg/kg/day in one study⁸⁵ and 1.25 or 2.5 mg/kg/day in the other²⁴⁷), suggesting that the increase in efficacy and toxicity is negligible. However, the small size of the studies and wide confidence intervals may mean that the true difference in effect has not been detected, although the point estimates did increase in favour of ciclosporin for efficacy and in favour of placebo for toxicity as the dose increased.

For the outcome of percentage change in PASI there was a statistically significant difference between the 1.25 and 2.5 mg/kg/day dose subgroups from one study²⁴⁷. The percentage change was significantly greater compared with placebo in the higher dose group.

10.8. Ciclosporin dosage comparisons for induction of remission

10.8.1. Evidence profile

Table

10.8.2. Evidence statements

In people with psoriasis, 5.0 mg/kg ciclosporin was statistically significantly better than 2.5 mg/kg ciclosporin administered for induction of remission for:

PASI75 at 12 weeks [1 study; 251 participants; very low quality evidence]²⁰⁶

In people with psoriasis, 5.0 mg/kg ciclosporin was statistically significantly more likely than 2.5 mg/kg ciclosporin administered for induction of remission to result in:

Elevated creatinine at 12–36 weeks [1 study; 243 participants; very low quality evidence]⁵⁹
Elevated uric acid at 12–36 weeks [1 study; 243 participants; very low quality evidence]⁵⁹

In people with psoriasis, 2.5 mg/kg ciclosporin was statistically significantly more likely than 1.25 mg/kg ciclosporin administered for induction of remission to result in:

Hypertension at 12–36 weeks [1 study; 292 participants; very low quality evidence]⁵⁹

In people with psoriasis, there was no statistically significant difference between an initial dose of 1.25 and 2.5 mg/kg ciclosporin administered for induction of remission for:

PASI75 at 12–36 weeks [1 study; 217 participants; very low quality evidence]⁵⁹
Elevated creatinine at 12–36 weeks [1 study; 292 participants; very low quality evidence]⁵⁹
Elevated uric acid at 12–36 weeks [1 study; 292 participants; very low quality evidence]⁵⁹

In people with psoriasis, there was no statistically significant difference between 2.5 and 5.0 mg/kg ciclosporin administered for induction of remission for:

Hypertension at 12–36 weeks [1 study; 243 participants; very low quality evidence]⁵⁹

Table 140

Summary of non-analysed data for ciclosporin dosing increments for induction.

Evidence statements for non-randomised data comparing ciclosporin doses for induction of remission:

In people with psoriasis, increasing the dose of ciclosporin allowed the achievement of PASI75 when lower doses were ineffective after 12–36 weeks [1 study; 109 participants; very low quality evidence]⁵⁹

10.9. Ciclosporin vs placebo for maintenance of remission

There were four studies⁶³^,⁸⁴^,³⁷¹^,³⁹⁹ that addressed the use of ciclosporin for the maintenance of remission in psoriasis; therefore, all had an initial induction period and only those who responded were randomised to the maintenance phase. The Ellis study⁸⁴ defined remission as achieving clear or nearly clear status on ciclosporin induction therapy and followed up for a further 4 months with low-dose ciclosporin (1.5 or 3 mg/kg/day) or placebo for 4 months. The Shupack study³⁷¹ defined remission as 70% improvement in BSA maintained for 2 weeks during a 16-week induction phase with 5.0 mg/kg/day ciclosporin, and the maintenance treatments were placebo or ciclosporin 3.0 mg/kg/day for 24 weeks. The Colombo study⁶³ defined remission as PASI75 during an 8–16-week induction period with any dose of ciclosporin and the maintenance dose was 5 mg/kg/day ciclosporin or placebo just on two consecutive days per week. The Thaci study³⁹⁹ had an induction period where participants received either 200 mg/day or 2.5 mg/kg/day increased stepwise by 50 mg if response was insufficient and only those who achieved PASI75 by week 12 were randomised to the maintenance phase to receive either the last effective dose of ciclosporin 3-times a week or placebo for a further 12 weeks. The dosing regimens in the latter two studies were not considered similar enough to the former two studies for pooling to be appropriate.

10.9.1. Evidence profile

Table

10.9.2. Evidence statements

In people with psoriasis, continuous ciclosporin administered for maintenance of remission was statistically significantly better than placebo for:

Mean time to relapse and relapse rate after a maximum follow-up of 4 months (3 mg/kg/day CSA) [1 study; 41 participants; moderate to low quality evidence]⁸⁴
Time-to-relapse (CSA three-times a week or 3 mg/kg/day) after a maximum follow-up of 12 or 24 weeks [2 studies; 224 participants; moderate quality evidence]³⁷¹^,³⁹⁹

In people with psoriasis, there was no statistically significant difference between ciclosporin administered for maintenance of remission and placebo for:

PASI75 at 24 weeks (CSA 5 mg/kg/day at weekends only) [1 study; 189 participants; very low quality evidence]⁶³
Mean final PASI at 24 weeks (CSA 5 mg/kg/day at weekends only) [1 study; 189 participants; very low quality evidence]⁶³
Maintaining at least mild psoriasis following PASI75 at 12 weeks (3-times weekly dosing) [1 study; 53 participants; low quality evidence]³⁹⁹
Mean time to relapse and relapse rate after a maximum follow-up of 4 months (1.5 mg/kg/day CSA) [1 study; 40 participants; low to very low quality evidence]⁸⁴
Relapse rate after a maximum follow-up of 24 weeks (CSA 5 mg/kg/day at weekends only) [1 study; 189 participants; very low quality evidence]⁶³
Withdrawal due to toxicity at 24 weeks (CSA 5 mg/kg/day at weekends only) [1 study; 239 participants; very low quality evidence]⁶³
Severe adverse events at 24 weeks (CSA 5 mg/kg/day at weekends only) [1 study; 239 participants; very low quality evidence]⁶³
Elevated creatinine at 24 weeks (CSA 5 mg/kg/day at weekends only) [1 study; 239 participants; very low quality evidence]⁶³

In people with psoriasis, there were no events with either ciclosporin administered for maintenance of remission or placebo for:

Elevated creatinine (at two consecutive visits) at 12 weeks (3-times weekly dosing) [1 study; 93 participants; moderate quality evidence]³⁹⁹

Evidence statements for individual studies where no original statistical analysis could be performed comparing ciclosporin and placebo administered for maintenance of remission:

Time to relapse was longer with two- or three-times weekly ciclosporin than placebo after a maximum follow-up of 12 or 24 weeks [2 studies; 332 participants; low to very low quality evidence]⁶³^,³⁹⁹
There was a greater increase in PASI at 12 weeks during maintenance with placebo than three-times weekly ciclosporin [1 study; 93 participants; low quality evidence]³⁹⁹

10.9.3. Subgroups and heterogeneity

For the outcomes of mean time to relapse and relapse rate from one study⁸⁴ there was a statistically significant difference between the dose subgroups. The time to relapse was significantly shorter and the relapse rate significant lower compared with placebo in the 3 mg/kg/day dose group compared with 1.5 mg/kg/day.

10.10. Intermittent (abrupt cessation) vs continuous ciclosporin for maintenance of remission

One study⁵³ defined intermittent dosing as ciclosporin being abruptly stopped after induction followed by an 12-week course of ciclosporin if relapse occurred, and continuous dosing as a tapering of the dose by 0.5mg/kg/day bi-monthly down to a maintenance level (the lowest marginally effective dose).

Two studies¹⁵¹^,¹⁵² defined intermittent ciclosporin as abruptly stopped ciclosporin being abruptly stopped after induction followed by an additional course of ciclosporin if relapse occurred, and continuous ciclosporin dosing as a tapering of the dose by 1 mg/kg/day until the treatment was stopped completely within 4 weeks, then an additional course was administered on relapse.

10.10.1. Evidence profile

Table

10.10.2. Evidence statements

In people with psoriasis, continuous ciclosporin was statistically significantly better than intermittent ciclosporin administered for maintenance of remission for:

Clear/nearly clear (PASI90) at 9 months [1 study; 45 participants; low quality evidence]⁵³
PASI75 at 9 months [1 study; 45 participants; very low quality evidence]⁵³
Time-to-relapse after a maximum follow-up of 1 year [1 study; 365 participants; very low quality evidence] ¹⁵¹

In people with psoriasis, there was no statistically significant difference between continuous and intermittent ciclosporin for maintenance of remission for:

PASI50 at 9 months [1 study; 45 participants; low quality evidence]⁵³
Increased creatinine at 9 months [1 study; 45 participants; very low quality evidence]⁵³
Hypertension at 9 months [1 study; 45 participants; very low quality evidence]⁵³

Evidence statements for individual studies where no statistical analysis could be performed comparing intermittent (abrupt cessation) and continuous ciclosporin administered for maintenance of remission in people with psoriasis:

Median time-to-relapse after a maximum follow-up of 2 years was longer with continuous than intermittent ciclosporin [1 study; 76 participants; very low quality evidence]¹⁵²

10.11. Intermittent (taper to withdraw) vs continuous (taper to minimum dose) ciclosporin for the maintenance of remission

Two studies induced remission using 3–5 mg/kg/day ciclosporin and defined the maintenance schedules as follows. ‘Continuous’ ciclosporin entailed dose reduction by 0.5–1.0 mg/kg/day each week and being continued at the lowest effective dose (in the range 0.5–3 mg/kg/day). If relapse occurred, the dose was increased to 3–5 mg/kg/day until remission was achieved, and the same procedure was repeated. ‘Intermittent’ ciclosporin entailed dose reduction by 0.5–1.0 mg/kg/day every other week followed by withdrawal. During withdrawal, topical steroids (10 g/day or less) of strong or medium potency were applied and if relapse occurred, the dose was increased to 3–5 mg/kg/day until remission was achieved. Treatment was withdrawn on remission and topical steroids were again applied.

10.11.1. Evidence profile

Table

10.11.2. Evidence statements

In people with psoriasis, an intermittent (taper to withdrawal) schedule was statistically significantly better than a continuous schedule of ciclosporin administered for maintenance of remission for:

Percentage change in PASI at 48 months [1 study; 37 participants; very low quality evidence]²⁹⁹

In people with psoriasis, a continuous schedule was statistically significantly better than an intermittent (taper to withdrawal) schedule of ciclosporin administered for maintenance of remission for:

Final PASI at 48 months [1 study; 31 participants; very low quality evidence]²⁹⁰

In people with psoriasis, there was no statistically significant difference between intermittent (taper to withdrawal) vs continuous ciclosporin administered for maintenance of remission for:

Withdrawal due to toxicity at 48 months [1 study; 68 participants; very low quality evidence]²⁹⁹
Hypertension at 1 year [1 study; 122 participants; very low quality evidence]²⁹⁰
Increased creatinine at 1 year [1 study; 122 participants; very low quality evidence]²⁹⁰
Hyperuricaemia at 1 year [1 study; 122 participants; very low quality evidence]²⁹⁰
Increased liver enzymes at 1 year [1 study; 122 participants; very low quality evidence]²⁹⁰

10.11.3. Subgroups and heterogeneity

For the outcomes of percentage change in PASI and final PASI the two studies²⁹⁰^,²⁹⁹ were not pooled as heterogeneity was present. This was not explained by any of the pre-defined subgroups; however, both studies were at high risk of bias owing to differences in baseline PASI score, which was higher in the intermittent group in both studies by 5.2–6.4 points, which was greater than the mean difference at the end point of the study in both cases. Additionally, both had a high drop-out rate in both the continuous and intermittent groups (32% and 29.5% for Ozawa²⁹⁹ and 75.4% and 73.8% for Ohtsuki²⁹⁰).

10.12. Ciclosporin dosage comparisons for maintenance

One study induced remission using 2.5 vs 5.0 mg/kg/day ciclosporin and patients achieving remission entered a maintenance phase, receiving 2.5 or 5.0 mg/kg/day. The 5 mg/kg/day dose was tapered to 2.5 over 3 months and the dose was tapered in all participants from months 9–12.

10.12.1. Evidence profile

Table

10.12.2. Evidence statements

In people with psoriasis, 2.5 mg/kg/day ciclosporin was statistically significantly better than 5.0 mg/kg/day ciclosporin administered for maintenance of remission for:

Severe adverse events at 18 months [1 study; 251 participants; low quality evidence]²⁰⁶
Elevated creatinine at 18 months [1 study; 251 participants; low quality evidence]²⁰⁶

In people with psoriasis, there was no statistically significant difference between 2.5 and 5.0 mg/kg/day ciclosporin administered for maintenance of remission for:

Hypertension at 18 months [1 study; 251 participants; very low quality evidence]²⁰⁶
Elevated uric acid at 18 months [1 study; 251 participants; very low quality evidence]²⁰⁶

Table 141

Summary of non-analysed data for ciclosporin in the maintenance of remission.

Evidence statements for individual studies where no statistical analysis could be performed comparing different doses of ciclosporin administered for maintenance of remission:

In people with psoriasis, there was no clinically relevant difference between 2.5 and 5.0 mg/kg/day ciclosporin for maintenance for change in PASI at 18 months [1 study; 251 participants; very low quality evidence]²⁰⁶.

10.13. Ciclosporin vs placebo for induction of remission in palmoplantar pustulosis

Note that the Reitamo study³²⁸ included data from both a double-blind placebo-controlled phase and an open dose-finding phase in which non-responders from the placebo group were given 1.25mg/kg/day ciclosporin at week 4 and further dose increases at monthly intervals in steps of 1.25mg/kg/day up to maximum of 3.75mg/kg/day until week 16 if still unresponsive. Responders in the ciclosporin group continued previous treatment, while non-responders in ciclosporin group had the dose increased to 3.75mg/kg/day

10.13.1. Evidence profile

Table

10.13.2. Evidence statements

In people with palmoplantar pustulosis, ciclosporin was statistically significantly better than placebo for:

Improvement at 4 weeks [2 studies; 96 participants; moderate quality evidence]⁸⁸^,³²⁸.

In people with palmoplantar pustulosis, placebo was statistically significantly better than ciclosporin for:

Hypertension at 12 months [1 study; 58 participants; very low quality evidence]⁸⁸.

In people with palmoplantar pustulosis, there was no statistically significant difference between ciclosporin and placebo for:

Hypertension at 4 weeks [1 study; 58 participants; very low quality evidence]⁸⁸
Increased serum creatinine at 12 months [1 study; 58 participants; very low quality evidence]⁸⁸
Improvement at 4 months during open phase [1 study; 28 participants; very low quality evidence]³²⁸
Relapse rate during open (4 months) and withdrawal (6 months) phases [1 study; 32 and 22 participants, respectively; very low quality evidence]³²⁸.

In people with palmoplantar pustulosis, there were no events with either ciclosporin or placebo for:

Increased serum creatinine at 4 weeks [1 study; 38 participants; moderate quality evidence]³²⁸.

10.14. Time to maximum effect

10.14.1. Evidence profiles

10.14.1.1. Ciclosporin

Table

10.14.1.2. Methotrexate

Table

10.14.1.3. Acitretin

Table

10.14.2. Data summary table

Table 142

Absolute data on time to maximum effect or time to remission.

10.14.3. Evidence statements

Evidence statements for individual studies that provide data regarding the time to remission or time to maximum response for systemic non-biological therapies (no statistical analysis could be performed).

10.14.3.1. Ciclosporin

In people with psoriasis, the time to remission when taking ciclosporin varied between studies:

Median time to 70–90% reduction in BSA ranged from 8–12 weeks [2 studies; 546 participants; very low quality evidence]¹⁵¹^,³⁷¹
Mean time to PASI80: 15.4 weeks [1 study; 37 participants; very low quality evidence]²⁹⁹

In people with psoriasis, the time to maximum response when taking ciclosporin varied between studies:

Mean PASI score still decreasing gradually at 12 weeks (although most rapid improvement was seen over the first 0–8 weeks) [1 study; 31 participants; very low quality evidence]¹⁰⁴
Mean PASI score reached maximal response at 12 weeks [1 study; 42 participants; very low quality evidence]¹⁴⁷
Mean PASI score reached maximal response at 16 weeks [1 study; 17 participants; very low quality evidence]¹³⁴
Mean percentage change in PASI reaching a maximum between 8 and 12 weeks [1 study; 94 participants; very low quality evidence]⁵⁹

Summary

The majority of the evidence suggests that 2.5–5.0mg/kg/day ciclosporin leads to remission or maximum response after between 9 and 12 weeks of treatment

10.14.3.2. Methotrexate

In people with psoriasis, the time to maximum response when taking methotrexate varied between studies:

Mean percentage improvement in PASI score still increasing gradually at 16 weeks [1 study; 110 participants; very low quality evidence]³⁵³
Mean PASI score reached a maximum response between 4 and 6 months [1 study; 20 participants; very low quality evidence]¹⁵⁰
Mean PASI score still decreasing gradually at 12 weeks [1 study; 37 participants; very low quality evidence]¹⁰⁴
Mean PASI score reached maximal response at 12 weeks [1 study; 43 participants; very low quality evidence]¹⁴⁷
Mean PASI score reached maximal response at 8 weeks [1 study; 17 participants; very low quality evidence]¹³⁴

Summary

The majority of the evidence suggests that methotrexate leads to remission or maximum response after between 16 and 24 weeks of treatment, although the higher initial dose of 15 mg/wk in two studies¹³⁴^,¹⁴⁷ appeared to achieve maximal response after 8–12 weeks of treatment

10.14.3.3. Acitretin

In people with psoriasis, the time to maximum response when taking acitretin varied between studies:

Mean improvement in global score and percentage coverage of body surface area (pooled data for all doses of acitretin) were maximal at 20 weeks [1 study; 37 participants; very low quality evidence]¹²³
Mean percentage improvement in PASI score was still increasing at 2 months on 10, 25 and 50 mg/day acitretin [1 study; 60 participants; very low quality evidence]²¹³
Percentage improvement in PASI had not reached a maximum by 6 weeks for all dosing schedules; however, the increasing dosing schedule showed a greater continued rate of improvement at 6 weeks than the decreasing or constant dosing schedules, which were increasing gradually [1 study; 58 participants; very low quality evidence]²⁶

Summary

The evidence suggests that acitretin may lead to remission or maximum response after approximately 20 weeks of treatment, and that an increasing dose may allow greater improvement than a deceasing or constant dosing schedule²⁶

10.14.4. Economic evidence

An economic evaluation should ideally compare all relevant alternatives. No applicable studies of good enough methodological quality were identified comparing all interventions of interest – acitretin, ciclosporin and methotrexate – in the treatment of patients with psoriasis.

Three studies²⁹⁴^,³⁷³^,⁴²⁷ were included that included the relevant comparison between ciclosporin and methotrexate and best supportive care. These are summarised in the economic evidence profiles below (Table 143, Table 144, Table 145 and Table 146). See also the full study evidence tables in Appendix I.

Table 143

Methotrexate versus ciclosporin versus best supportive care – economic study characteristics.

Table 144

Methotrexate versus best supportive care – economic summary of findings.

Table 145

Ciclosporin versus best supportive care – economic summary of findings.

Table 146

Ciclosporin versus methotrexate – economic summary of findings.

Five studies⁸⁷^,⁹⁵^,¹³⁶^,¹³⁸^,³⁰⁹ were selectively excluded due to their poor applicability and very serious methodological limitations. These are detailed in Appendix G.

No relevant economic evaluations comparing acitretin with either ciclosporin or methotrexate were identified.

Despite its limitations and partial applicability, the analysis by Opmeer has been included in this review because it is the only study to be based on prospectively collected resource use data associated with treatment with ciclosporin and methotrexate during both a trial period and follow-up. The analysis shows that the biggest difference in cost between the treatments is driven by the difference in drug cost during the first 16 weeks during which ciclosporin is more costly. During follow-up however, the difference between the two treatments becomes less significant due to the similar use of other therapies, such as UVB phototherapy, day care treatments and topicals after treatment with the systemic therapies has stopped. In clinical practice, it is unlikely that duration of treatment with these drugs will be identical. Ciclosporin is often given for a shorter duration than methotrexate due to the increased risk of nephrotoxicity with longer term use. Methotrexate is often given for a longer period as its maximum effectiveness may not even be observed by 16 weeks. Therefore, it is unlikely that the cost differences between ciclosporin and methotrexate would diminish as rapidly in clinical practice as the results of Opmeer and colleagues would suggest.

The studies by Sizto and Woolacott clearly show that treatment with methotrexate or ciclosporin to be cost saving compared to best supportive care or no treatment. They also demonstrate methotrexate to be cost saving compared to ciclosporin; that is, producing greater quality of life gains for less NHS resource. However, the limitations of these studies are potentially serious insofar as their conclusions about cost-effectiveness are based on a now incomplete evidence base and out-of-date unit costs. The Sizto analysis does not include all the relevant RCT data for ciclosporin (missing studies include by Van Joost and colleagues⁴¹³, Ellis and colleagues⁸⁵ and Guenther and colleagues¹³³) which is likely why it has performed more poorly compared to methotrexate than in the analysis by Woolacott and colleagues. The study by Woolacott includes clinical evidence published only up until April 2004, which means that it does not include the more recent RCTs by Ho and colleagues¹⁵⁰, Saurat and colleagues³⁵³ and Flytstrom and colleagues¹⁰⁴, the last in which ciclosporin is shown to be more effective than methotrexate. Additionally, the cost of ciclosporin has decreased by about one-third since these evaluations were undertaken.

10.14.4.1. New cost-effectiveness analysis

A full economic analysis was not prioritised for this question. Despite the existing economic evidence having some potentially serious limitations, the GDG believe that the conclusions of these analyses (i.e. that methotrexate is more cost-effective than ciclosporin) are still very likely to be true and that a new cost-effectiveness analysis is unlikely to inform recommendations further. On that basis, this question was not considered a high priority for de novo modelling and would only have been undertaken if other higher priority areas, such as topical therapies and second-line biological therapies, were deprioritised. Therefore, the GDG made their recommendations about which systemic treatments should be offered and when based on published clinical and cost-effectiveness evidence and a simple cost analysis presented briefly here.

Although the price of ciclosporin has fallen by one-third since Woolacott and colleagues undertook their economic analysis, the cost of methotrexate is still only a fraction of the cost of ciclosporin. Depending on the weight of the patient and dose of ciclosporin, methotrexate is around 95% to 98.5% less costly than ciclosporin.

For example, if a 75 kg patient is taking a dose of 2.5 mg/kg of ciclosporin, their weekly drug cost is approximately £23 which translates to an annual cost of £1,174. If this patient was taking methotrexate, his/her yearly cost would only be £36 based on a weekly dose of 15 mg. This means that every one week of treatment with ciclosporin costs the same as 32 weeks of MTX. At higher mg/kg doses or for heavier patients this ratio increases, with the one-week cost of a 75 kg patient on 4 mg/kg ciclosporin accruing the same drug costs as a patient receiving one year’s continuous treatment with MTX. Put another way, if a patient receives a 6 month course of ciclosporin (75 kg at 2.5 mg/kg) he/she would need to be in remission for more than 15 years to cost the same or less than continuous MTX.

10.14.4.2. Evidence statements

No cost-effectiveness analyses were identified comparing all three interventions of interest – acitretin, ciclosporin and methotrexate – in the treatment of patients with psoriasis.
Two cost-effectiveness analyses showed methotrexate and ciclosporin to be cost saving compared to best supportive care in the treatment of patients with moderate to severe plaque psoriasis. These studies are directly applicable and have potentially serious limitations.
Two cost-effectiveness analyses and one cost-minimisation analysis show methotrexate to be cost saving compared to ciclosporin in the treatment of patients with moderate to severe plaque psoriasis. Overall, the studies contributing to this evidence are partially or directly applicable and have potentially serious limitations.
No economic evidence is available to estimate the relative cost-effectiveness of acitretin.

10.15. Recommendations and link to evidence

Table

Responsibility for use of systemic therapy should be in specialist settings only. Certain aspects of supervision and monitoring may be delegated to other healthcare professionals and completed in non-specialist settings, in which case, such arrangements (more...)

Publication Details

Copyright

Apart from any fair dealing for the purposes of research or private study, criticism or review, as permitted under the Copyright, Designs and Patents Act, 1988, no part of this publication may be reproduced, stored or transmitted in any form or by any means, without the prior written permission of the publisher or, in the case of reprographic reproduction, in accordance with the terms of licences issued by the Copyright Licensing Agency in the UK. Enquiries concerning reproduction outside the terms stated here should be sent to the publisher at the UK address printed on this page.

The use of registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant laws and regulations and therefore for general use.

The rights of National Clinical Guideline Centre to be identified as Author of this work have been asserted by them in accordance with the Copyright, Designs and Patents Act, 1988.

Publisher

Royal College of Physicians (UK), London

NLM Citation

National Clinical Guideline Centre (UK). Psoriasis: Assessment and Management of Psoriasis. London: Royal College of Physicians (UK); 2012 Oct. (NICE Clinical Guidelines, No. 153.) 10, Systemic non-biological therapy.

Table 137Disease severity inclusion and dosing schedules of included studies

Reference ID	Disease severity	Comparison	Dose and schedule
Induction of remission
BERBIS 1989	Severe psoriasis (66.7% plaque, 9.1% guttate, 13.6% pustular, 3.0% erythrodermic, 4.5% palmoplantar pustulosis, 3.0% acrodermatitis continua)	Acitretin dosing	Acitretin: increasing (10 up to 50 mg/day) vs decreasing (50 down to 10 mg/day) or constant (30 mg/day) dose schedule Note: for this study the increasing dose arm was used as the control arm as this reflects current clinical practice in the UK
CHRISTOPHERS 1992	Severe generalised chronic plaque psoriasis PASI≥15	CSA dosing (induction)	CSA: 1.25 vs 2.5 mg/kg/day (initial doses but could be doubled if ineffective)
ELLIS 1986	Severe chronic large plaque-type psoriasis vulgaris >20% BSA involvement	CSA vs placebo (induction)	CSA: 14 mg/kg/day (plus open phase)
ELLIS 1991	Chronic plaque psoriasis affecting >25% BSA, or disabling psoriasis	CSA vs placebo (induction)	CSA: 3, 5 or 7.5 mg/kg/day (plus open dose adjustment phase)
ERKKO 1998	Clinically defined palmoplantar pustulosis of the palms and/or soles with at least 20 whitish-yellow pustules of diameter at least 1mm	CSA vs placebo for PPP	CSA: 1 mg/kg/day (1 month double blind); plus 11 month open phase (dose increased by 1 mg/kg/day if no response up to a maximum of 4 mg/kg/day)
FLYTSTROM 2008	Moderate-to-severe chronic plaque psoriasis: classified by physician and patient	MTX vs CSA	MTX: 7.5 mg/wk (3-divided dose) up to 15 mg/wk (plus folic acid) CSA: 3 mg/kg/d (divided into 2 doses) up to 5 mg/kg/d
GOLDFARB 1988	BSA>10% or disabling disease	Acitretin vs placebo	10, 25, 50 or 75 mg/day acitretin (plus open phase)
GUENTHER 1991	Large plaque psoriasis BSA ≥25% and PASI ≥12	CSA vs placebo (induction)	CSA: 2.5–5 mg/kg/day
GUMUSEL 2011	Moderate-to-severe psoriasis BSA >10% and PASI ≥10 and NAPSI >10	MTX vs CSA	MTX: 15 mg/wk (initial dose) reduced to 10 mg/wk after 3 months (plus folic acid) CSA: 5 mg/kg/d reduced to 2.5–3.5 mg/kg/d
HEULE 1988/VANJOOST 1988A	Chronic plaque psoriasis PASI≥20	CSA vs placebo (induction)	CSA: ~5–7 mg/kg/day (plus open phase)
HEYDENDAEL 2003	Moderate-to-severe chronic plaque psoriasis: PASI ≥8	MTX vs CSA	MTX: 15 mg/wk (3-divided dose) up to 22.5 mg/wk (folic acid use not specified) CSA: 3 mg/kg/d (divided into 2 doses) up to 5 mg/kg/d
HO 2010	BSA ≥20% Plaque psoriasis	MTX vs placebo	MTX: 2.5–5.0 mg/wk to assess safety then 10 mg/wk up to 30 mg/wk Folic acid supplement (MTX arm only)
KINGSTON 1987	BSA>20%	Acitretin vs placebo	10, 50 or 75 mg/day acitretin (plus open phase)
LASSUS 1987	Severe psoriasis (87.5% plaque, 5% pustular and 7.5% erythrodermic)	Acitretin vs placebo	10, 25 or 50 mg/day acitretin (plus open phase)
MEFFERT 1997	Psoriasis vulgaris PASI ≥8	CSA vs placebo (induction)	CSA: 1.25 or 2.5 mg/kg/day (plus open phase)
REITAMO 1993	Clinically defined palmoplantar pustulosis of the palms and/or soles with at least 20 whitish-yellow pustules of diameter at least 2mm	CSA vs placebo for PPP	CSA: 2.5 mg/kg/day (1 month double blind); plus 2 month open phase (dose increased by 1.25 mg/kg/day if no response)
SANDHU 2003	Severe psoriasis (73.3% plaque and 26.6% erythrodermic) BSA >40%	MTX vs CSA	MTX: 0.5 mg/kg/wk (folic acid use not specified) CSA: 3 mg/kg/d (divided into 2 doses) up to 4 mg/kg/d Doses tapered once PASI75 reached (maintenance)
SAURAT 2008	Moderate-to-severe plaque psoriasis: BSA ≥10% and PASI ≥10	MTX vs placebo	MTX: 7.5 mg increased to 25 mg/wk as needed and tolerated Folic acid supplement (both arms)
Maintenance of remission
CHAIDEMENOS 2007	Moderate-to-severe chronic plaque psoriasis PASI≥8	CSA regimens for maintenance	Intermittent CSA: abruptly stopped ciclosporin after induction, then received additional 12-week course on relapse Continuous CSA: tapered by 0.5mg/kg/day bi-monthly down to maintenance level (lowest marginally effective dose)
COLOMBO 2010	Chronic plaque psoriasis treated with continuous ciclosporin (severity not stated) Achieved remission (PASI75) during induction therapy	CSA vs placebo (maintenance)	CSA: 5 mg/kg/day at weekends only
ELLIS 1995	Chronic plaque psoriasis affecting >25% BSA, or disabling psoriasis	CSA vs placebo (maintenance)	CSA: 1.5 or 3 mg/kg/day (no dose adjustment)
HO 1999	Plaque psoriasis unresponsive to topical therapies (mean baseline PASI 24.5)	CSA regimens for maintenance	Intermittent CSA: abruptly stopped ciclosporin after induction, then received additional course on relapse Continuous CSA: tapered by 1 mg/kg daily each week until stopping within 4 weeks, then received additional course on relapse
HO 2001	Plaque psoriasis Requiring systemic therapy (mean BSA at baseline approximately 17%)	CSA regimens for maintenance	Intermittent CSA: abruptly stopped ciclosporin after induction, then received additional course on relapse Continuous CSA: tapered by 1 mg/kg daily each week until stopping within 4 weeks, then received additional course on relapse
LABURTE 1994	Severe chronic plaque psoriasis PASI ≥18	CSA dosing (induction and maintenance)	CSA: 2.5 vs 5.0 mg/kg/day (initial doses during phase 1); patients achieving remission entered a maintenance phase (2.5 vs 5.0 mg/kg/day: 5 mg tapered to 2.5 over 3 months and dose tapered in all from month 9–12)
OHTSUKI 2003	Severe psoriasis PASI>20	CSA regimens for maintenance	‘Continuous’ CSA: Following induction of remission with 3–5 mg/kg/day ciclosporin the dose was reduced by 0.5–1.0 mg/kg/day every week and maintained as the lowest effective dose (in the range 0.5–3 mg/kg/day) If relapse occurred, the dose was increased to 3–5 mg/kg/day until remission was achieved, and the same procedure was repeated. ‘Intermittent’ CSA: Following induction of remission with 3–5 mg/kg/day ciclosporin the dose was reduced by 0.5–1.0 mg/kg/day every other week followed by withdrawal. During withdrawal, topical steroids (10 g/day or less) of strong or medium potency were applied If relapse occurred, the dose was increased to 3–5 mg/kg/day until remission was achieved. Treatment was withdrawn on remission and topical steroids were again applied.
OZAWA 1999	Psoriasis vulgaris with PASI >20; psoriatic arthritis; generalised pustular psoriasis; erythrodermic psoriasis	CSA regimens for maintenance	‘Continuous’ CSA: Following induction of remission with 3–5 mg/kg/day ciclosporin the dose was reduced by 0.5–1.0 mg/kg/day every week and maintained as the lowest effective dose (in the range 0.5–3 mg/kg/day) If relapse occurred, the dose was increased to 3–5 mg/kg/day until remission was achieved, and the same procedure was repeated. ‘Intermittent’ CSA: Following induction of remission with 3–5 mg/kg/day ciclosporin the dose was reduced by 0.5–1.0 mg/kg/day every other week followed by withdrawal. During withdrawal, topical steroids (10 g/day or less) of strong or medium potency were applied If relapse occurred, the dose was increased to 3–5 mg/kg/day until remission was achieved. Treatment was withdrawn on remission and topical steroids were again applied.
SHUPACK 1997	BSA>12% or disabling psoriasis that impairs daily activities	CSA vs placebo (maintenance)	CSA: 3 mg/kg/day (with dose adjustment) Note: initial randomisation to 1.5 mg/kg arm stopped after 7 people were recruited owing to evidence suggesting lack of efficacy (so results reported for 3 mg/kg/day vs placebo only)
THACI 2002	Chronic plaque type psoriasis PASI ≥12.	CSA vs placebo (maintenance)	CSA: lowest effective dose

Table 138Evidence profile comparing methotrexate vs placebo for induction of remission

Quality assessment							Summary of findings
Quality assessment							No of patients		Effect		Quality
No of studies	Design	Limitations	Inconsistency	Indirectness	Imprecision	Other considerations	MTX	Placebo	Relative (95% CI)	Absolute
PASI90 – Incremental MTX dosing (7.5 up to 25 mg/wk) (follow-up 16 weeks)
1 Saurat 2008	randomised trials	no serious limitations	no serious inconsistency	serious^a	very serious^b	Folic acid also given	15/104 (14.4%)	6/52 (11.5%)	RR 1.25 (0.52 to 3.03)	29 more per 1000 (from 55 fewer to 234 more)	⊕○○○ VERY LOW
Clear/nearly clear on PGA – Incremental MTX dosing (7.5 up to 25 mg/wk) (follow-up 16 weeks)
1 Saurat 2008	randomised trials	no serious limitations	no serious inconsistency	serious^a	no serious imprecision	Folic acid also given	33/104 (31.7%)	6/52 (11.5%)	RR 2.75 (1.23 to 6.14)	202 more per 1000 (from 27 more to 593 more)	⊕⊕⊕○ MODERATE
PASI75 – Incremental MTX dosing (7.5 up to 25 mg/wk or 10 up to 30 mg/wk) (follow-up 4–6 months)
2 Ho 2010 Saurat 2008	randomised trials	no serious limitations^c	no serious inconsistency	serious^d	no serious imprecision	Folic acid also given	51/123 (41.5%)	13/69 (18.8%)	RR 2.26 (1.34 to 3.83)	237 more per 1000 (from 64 more to 533 more)	⊕⊕⊕○ MODERATE
PASI50 – Incremental MTX dosing (7.5 up to 25 mg/wk or 10 up to 30 mg/wk) (follow-up 4–6 months)
2 Ho 2010 Saurat 2008	randomised trials	no serious limitations^c	no serious inconsistency	serious^d	no serious imprecision	Folic acid also given	83/123 (67.5%)	20/69 (29%)	RR 2.33 (1.58 to 3.43)	386 more per 1000 (from 168 more to 704 more)	⊕⊕⊕○ MODERATE
PASI change/final score – Incremental MTX dosing (7.5 up to 25 mg/wk or 10 up to 30 mg/wk) (follow-up 4–6 months; better indicated by lower values)
2 Ho, 2010 Saurat, 2008	randomised trials	no serious limitations^e	no serious inconsistency	serious^d	no serious imprecision	Folic acid also given	123	69	-	MD 6.69 lower (9.48 to 3.90 lower)	⊕⊕⊕○ MODERATE
Severe adverse events – Incremental MTX dosing (7.5 up to 25 mg/wk) (follow-up 26 weeks)
1 Saurat, 2008	randomised trials	no serious limitations	no serious inconsistency	serious^a	very serious^b	Folic acid also given	1/110 (0.9%)	1/53 (1.9%)	RR 0.48 (0.03 to 7.55)	10 fewer per 1000 (from 18 fewer to 124 more)	⊕○○○ VERY LOW
Withdrawal due to toxicity – Incremental MTX dosing (7.5 up to 25 mg/wk) (follow-up 26 weeks)
1 Saurat, 2008	randomised trials	no serious limitations	no serious inconsistency	serious^a	very serious^b	Folic acid also given	6/110 (5.5%)	1/49 (2%)	RR 2.67 (0.33 to 21.61)	34 more per 1000 (from 14 fewer to 421 more)	⊕○○○ VERY LOW
Raised liver enzymes – Incremental MTX dosing (7.5 up to 25 mg/wk) (follow-up 26 weeks)
1 Saurat, 2008	randomised trials	no serious limitations	no serious inconsistency	serious^a	very serious^b	Folic acid also given	10/110 (9.1%)	4/53 (7.5%)	RR 1.2 (0.4 to 3.66)	15 more per 1000 (from 45 fewer to 201 more)	⊕○○○ VERY LOW

a: Data not given separately for the 2 placebo groups (subcutaneous and oral)

b: Confidence interval crosses the boundary for clinical significance in favour of both interventions, as well as line of no effect

c: Ho study (19.2% weighted) had unclear allocation concealment

d: Larger study (Saurat): data not given separately for the 2 placebo groups (subcutaneous and oral)

: Ho study (20.6% weighted) had unclear allocation concealment and a long follow-up (6 months)

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Ciclosporin	Methotrexate	Relative (95% CI)	Absolute	Quality
Clear/nearly clear (PASI90) - Incremental dose MTX (7.5 up to 15 mg/wk) (follow-up 12 weeks)
1 Flytstrom 2008	randomised trials	serious^a	no serious inconsistency	no serious indirectness	serious^b	Folic acid also given	9/31 (29%)	4/37 (10.8%)	RR 2.69 (0.91 to 7.88)	183 more per 1000 (from 10 fewer to 744 more)	⊕⊕○○ LOW
Clear/nearly clear (PASI90) - Incremental dose MTX (15 up to 22.5 mg/wk) (follow-up 16 weeks)
1 Heydendael 2003	randomised trials	serious^c	no serious inconsistency	no serious indirectness	very serious^d	none	14/42 (33.3%)	17/43 (39.5%)	RR 0.84 (0.48 to 1.48)	63 fewer per 1000 (from 206 fewer to 190 more)	⊕○○○ VERY LOW
Clearance - High dose MTX (0.5 mg/kg/wk) (follow-up 10 weeks)
1 Sandhu 2003	randomised trials	very serious^e	no serious inconsistency	serious^f	no serious imprecision	none	6/15 (40%)	13/15 (86.7%)	RR 0.46 (0.24 to 0.88)	468 fewer per 1000 (from 104 fewer to 659 fewer)	⊕○○○ VERY LOW
Time-to-remission - PASI75 - Incremental dose MTX (15 up to 22.5 mg/wk) (follow-up 16 weeks)
1 Heydendael 2003	randomised trials	serious^c	no serious inconsistency	no serious indirectness	serious^b	none	30/42 (71.4%)	26/43 (60.5%)	HR 1.63 (0.96 to 2.77)	175 more per 1000 (from 15 fewer to 319 more)	⊕⊕○○ LOW
Time-to-remission – PASI90 - Incremental dose MTX (15 up to 22.5 mg/wk) (follow-up 16 weeks)
1 Heydendael 2003	randomised trials	serious^c	no serious inconsistency	no serious indirectness	very serious^d	none	14/42 (33.3%)	17/43 (39.5%)	HR 0.87 (0.43 to 1.76)	41 fewer per 1000 (from 201 fewer to 192 more)	⊕○○○ VERY LOW
PASI75 - Incremental dose MTX (7.5 up to 15 mg/wk) (follow-up 12 weeks)
1 Flytstrom 2008	randomised trials	serious¹	no serious inconsistency	no serious indirectness	no serious imprecision	Folic acid also given	18/31 (58.1%)	9/37 (24.3%)	RR 2.39 (1.26 to 4.54)	338 more per 1000 (from 63 more to 861 more)	⊕⊕⊕○ MODERATE
PASI75 - Incremental dose MTX (15 up to 22.5 mg/wk) (follow-up 16 weeks)
1 Heydendael 2003	randomised trials	serious^g	no serious inconsistency	no serious indirectness	serious^b	none	30/42 (71.4%)	26/43 (60.5%)	RR 1.18 (0.87 to 1.61)	109 more per 1000 (from 79 fewer to 369 more)	⊕⊕○○ LOW
PASI50 - Incremental dose MTX (7.5 up to 15 mg/wk) (follow-up 12 weeks)
1 Flytstrom 2008	randomised trials	serious^a	no serious inconsistency	no serious indirectness	no serious imprecision	Folic acid also given	27/31 (87.1%)	24/37 (64.9%)	RR 1.34 (1.02 to 1.76)	221 more per 1000 (from 13 more to 493 more)	⊕⊕⊕○ MODERATE
Final PASI - High dose MTX (0.5 mg/kg/wk) (follow-up 12 weeks; better indicated by lower values)
1 Sandhu 2003	randomised trials	very serious^e	no serious inconsistency	serious^f^,^h	no serious imprecision	none	15	15	-	MD 3.9 higher (0.69 to 7.11 higher)	⊕○○○ VERY LOW
Final PASI - incremental dose MTX (within licensed range; maximum 22.5 mg/wk) (follow-up 12–16 weeks; better indicated by lower values)
2 Flytstrom 2008 Heydendael 2003	randomised trials	seriousⁱ	no serious inconsistency	serious^h	no serious imprecision	none	73	80	-	MD 1.62 lower (2.7 lower to 0.54 lower)	⊕⊕○○ LOW
Change in NAPSI – Decreasing MTX dose (15 mg/wk reduced to 10 mg/wk) (follow-up 6 months; better indicated by higher values)
1 Gumusel 2011	randomised trials	serious^j	no serious inconsistency	no serious indirectness	serious^b	Folic acid also given	19	18	-	MD 4.8 higher (3.73 lower to 13.33 higher)	⊕⊕○○ LOW
Elevated liver enzymes - MTX dose within licensed range (maximum 22.5 mg/wk) (follow-up 12–24 weeks)
3 Flytstrom 2008 Heydendael 2003 Gumusel 2011	randomised trials	seriousⁱ	no serious inconsistency	no serious indirectness^k	no serious imprecision	Folic acid also given in Flytstrom and Gumusel studies	0/92 (0%)	20/98 (20.4%)	RR 0.07 (0.01 to 0.38)	190 fewer per 1000 (from 127 fewer to 202 fewer)	⊕⊕⊕○ MODERATE
Elevated creatinine - Standard MTX dose range (maximum 15 mg/wk) (follow-up 12–24 weeks)
2 Flytstrom 2008 Gumusel 2011	randomised trials	serious^l	no serious inconsistency	no serious indirectness^m	no serious imprecision	Folic acid also given	8/50 (16%)	0/55 (0%)	RR 9.79 (1.32 to 72.65)	-	⊕⊕⊕○ MODERATE
Hypertension requiring treatment - Incremental dose MTX (15 up to 22.5 mg/wk) (follow-up 16 weeks)
1 Heydendael 2003	randomised trials	serious^c	no serious inconsistency	no serious indirectness	very serious^d	none	2/42 (4.8%)	0/43 (0%)	RR 5.12 (0.25 to 103.5)	-	⊕○○○ VERY LOW
Diastolic hypertension - High dose MTX (0.5 mg/kg/wk) (follow-up 12 weeks)
1 Sandhu 2003	randomised trials	very serious^e	no serious inconsistency	serious^f	very serious^d	none	4/15 (26.7%)	0/15 (0%)	RR 9 (0.53 to 153.79)	-	⊕○○○ VERY LOW
Withdrawal due to toxicity - Standard MTX dose range (maximum 15 mg/wk) (follow-up 12–16 weeks)
2 Flytstrom 2008 Gumusel 2011	randomised trials	seriousⁿ	no serious inconsistency	no serious indirectness	serious^b	Folic acid also given	6/50 (12%)	1/55 (1.8%)	RR 4.6 (0.84 to 25.16)	65 more per 1000 (from 3 fewer to 439 more)	⊕⊕○○ LOW
Withdrawal due to toxicity - Incremental dose MTX (15 up to 22.5 mg/wk) (follow-up 16 weeks)
1 Heydendael 2003	randomised trials	no serious risk of bias	no serious inconsistency	no serious indirectness	no serious imprecision	none	1/42 (2.4%)	12/43 (27.9%)	RR 0.09 (0.01 to 0.63)	254 fewer per 1000 (from 103 fewer to 276 fewer)	⊕⊕⊕⊕ HIGH
Remaining clear at 12 weeks (after tapering high dose MTX (0.5 mg/kg/wk)) (follow-up 12 weeks)
1 Sandhu 2003	randomised trials	very serious^e	no serious inconsistency	serious^f	serious^b	none	2/6 (33.3%)	13/13 (100%)	RR 0.37 (0.14 to 1.01)	630 fewer per 1000 (from 860 fewer to 10 more)	⊕○○○ VERY LOW
Mean change from baseline in DLQI - Incremental dose MTX (7.5 up to 15 mg/wk) (follow-up 8 weeks)
1 Flytstrom 2008	randomised trials	serious^a	no serious inconsistency	no serious indirectness	serious^o	Folic acid also given	31	37	MTX: 42% CSA: 71% p=0.0078	-	⊕⊕○○ LOW
Mean change from baseline in DLQI - Incremental dose MTX (7.5 up to 15 mg/wk) (follow-up 12 weeks)
1 Flytstrom 2008	randomised trials	serious^a	no serious inconsistency	serious^p	serious^o	Folic acid also given	31	37	NS difference		⊕○○○ VERY LOW
Median time to relapse - Incremental dose MTX (15 up to 22.5 mg/wk) (follow-up 8 weeks)
1 Heydendael 2003	randomised trials	serious^a	no serious inconsistency	no serious indirectness	serious^o	none	42	43	MTX: 4 weeks CSA: 4 weeks Note: NS difference in duration of PASI75 or PASI90 response (p = 0.43 and 0.34, respectively from log rank test)^p	-	⊕⊕○○ LOW

a: High differential drop out before treatment began (MTX = 9.8%; CSA = 27.9%) but baseline characteristics still matched; and differential drop out during treatment due to adverse events: MTX = 0; CSA = 12.9%

b: Confidence interval ranges from clinically important effect to no effect

c: Differential drop out rate: MTX = 27.9%; CSA = (2.4%) due in abnormal LFTs with high dose MTX

d: Confidence interval crosses the boundary for clinical significance in favour of both treatment, as well as line of no effect

e: Unclear allocation concealment, blinding and drop out rates

f: Methotrexate dosing not within current UK practice

g: Differential drop out rate in Heydendael study MTX = 27.9%; CSA = (2.4%) due in abnormal LFTs with high dose MTX

h: Surrogate outcome for change in PASI

i: Flytstrom: High differential drop out before treatment began (MTX = 9.8%; CSA = 27.9%) but baseline characteristics still matched; and differential drop out during treatment due to adverse events: MTX = 0; CSA = 12.9%. Differential drop out rate in Heydendael study MTX = 27.9%; CSA = (2.4%) due in abnormal LFTs with high dose MTX

j: Inadequate sequence generation and unclear blinding

k: Unclear definition of elevation of LFTs in Heydendael paper

l: 1/2 High differential drop out before treatment began (MTX = 9.8%; CSA = 27.9%) but baseline characteristics still matched; and differential drop out during treatment due to adverse events: MTX = 0; CSA = 12.9% 1/2 Inadequate sequence generation and unclear blinding

m: Unclear definition of elevation

n: 1/2 studies (69.2% weighted) inadequate sequence generation and unclear blinding

o: No range available

p: Only states non-significant - no data provided

q: Hazard ratio could not be calculated as numbers relapsing not reported

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Acitretin	Placebo	Relative (95% CI)	Absolute	Quality
PASI75 - 10 mg acitretin (follow-up 8 weeks)
2 Lassus 1988 Goldfarb 1988	randomised trials	very serious^a	no serious inconsistency	very serious^b	very serious^c	none	8/25 (32%)	6/32 (18.8%)	RR 1.46 (0.6 to 3.54)	86 more per 1000 (from 75 fewer to 476 more)	⊕○○○ VERY LOW
PASI75 - 25 mg acitretin (follow-up 8 weeks)
2 Lassus 1988 Goldfarb 1988	randomised trials	very serious^a	no serious inconsistency	very serious^b	serious^d	none	12/25 (48%)	6/32 (18.8%)	RR 2.13 (0.96 to 4.75)	212 more per 1000 (from 8 fewer to 703 more)	⊕○○○ VERY LOW
PASI75 - 50 mg acitretin (follow-up 8 weeks)
2 Lassus 1988 Goldfarb 1988	randomised trials	very serious^a	no serious inconsistency	very serious^b	no serious imprecision	none	16/31 (51.6%)	6/32 (18.8%)	RR 2.7 (1.26 to 5.81)	319 more per 1000 (from 49 more to 902 more)	⊕○○○ VERY LOW
PASI75 - 75 mg acitretin (follow-up 8 weeks)
1 Goldfarb 1988	randomised trials	very serious^e	no serious inconsistency	serious⁶	very serious^c	none	2/5 (40%)	1/12 (8.3%)	RR 4.8 (0.55 to 41.7)	317 more per 1000 (from 37 fewer to 1000 more)	⊕○○○ VERY LOW
Cheilitis - 10 mg acitretin (follow-up 8 weeks)
2 Lassus 1988 Goldfarb 1988	randomised trials	very serious^a	no serious inconsistency	very serious^g	no serious imprecision	none	17/23 (73.9%)	8/31 (25.8%)	RR 2.75 (1.39 to 5.44)	452 more per 1000 (from 101 more to 1000 more)	⊕○○○ VERY LOW
Cheilitis - 25 mg acitretin (follow-up 8 weeks)
2 Lassus 1988 Goldfarb 1988	randomised trials	very serious^a	no serious inconsistency	very serious^g	no serious imprecision	none	18/22 (81.8%)	8/31 (25.8%)	RR 3.06 (1.66 to 5.66)	532 more per 1000 (from 170 more to 1000 more)	⊕○○○ VERY LOW
Cheilitis - 50 mg acitretin (follow-up 8 weeks)
2 Lassus 1988 Goldfarb 1988	randomised trials	very serious^a	no serious inconsistency	very serious^g	no serious imprecision	none	27/29 (93.1%)	8/31 (25.8%)	RR 3.45 (1.92 to 6.2)	632 more per 1000 (from 237 more to 1000 more)	⊕○○○ VERY LOW
Cheilitis - 75 mg acitretin (follow-up 8 weeks)
1 Goldfarb 1988	randomised trials	serious^e	no serious inconsistency	serious^h	seriousⁱ	none	4/5 (80%)	3/12 (25%)	RR 3.2 (1.09 to 9.36)	550 more per 1000 (from 23 more to 1000 more)	⊕○○○ VERY LOW
Cheilitis - 10 mg acitretin (follow-up 6 months)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	no serious imprecision	none	16/20 (80%)	6/20 (30%)	RR 2.67 (1.32 to 5.39)	501 more per 1000 (from 96 more to 1000 more)	⊕○○○ VERY LOW
Cheilitis - 25 mg acitretin (follow-up 6 months)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	no serious imprecision	none	17/20 (85%)	6/20 (30%)	RR 2.83 (1.42 to 5.67)	549 more per 1000 (from 126 more to 1000 more)	⊕○○○ VERY LOW
Cheilitis - 50 mg acitretin (follow-up 6 months)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	no serious imprecision	none	19/20 (95%)	6/20 (30%)	RR 3.17 (1.61 to 6.23)	651 more per 1000 (from 183 more to 1000 more)	⊕○○○ VERY LOW
Hair loss - 10 mg acitretin (follow-up 8 weeks)
2 Lassus 1988 Goldfarb 1988	randomised trials	very serious^a	no serious inconsistency	very serious^g	very serious^c	none	0/23 (0%)	1/31 (3.2%)	RR 0.72 (0.03 to 15.26)	9 fewer per 1000 (from 31 fewer to 460 more)	⊕○○○ VERY LOW
Hair loss - 25 mg acitretin (follow-up 8 weeks)
2 Lassus 1988 Goldfarb 1988	randomised trials	very serious^a	no serious inconsistency	very serious^g	very serious^g	none	1/22 (4.5%)	1/31 (3.2%)	RR 2.4 (0.18 to 31.29)	45 more per 1000 (from 26 fewer to 977 more)	⊕○○○ VERY LOW
Hair loss - 50 mg acitretin (follow-up 8 weeks)
2 Lassus 1988 Goldfarb 1988	randomised trials	very serious^a	no serious inconsistency	very serious^g	no serious imprecision	none	8/29 (27.6%)	1/31 (3.2%)	RR 6.06 (1.13 to 32.6)	163 more per 1000 (from 4 more to 1000 more)	⊕○○○ VERY LOW
Hair loss - 75 mg acitretin (follow-up 8 weeks)
1 Goldfarb 1988	randomised trials	very serious^e	no serious inconsistency	serious^h	very serious^h	none	2/5 (40%)	1/12 (8.3%)	RR 4.8 (0.55 to 41.7)	317 more per 1000 (from 37 fewer to 1000 more)	⊕○○○ VERY LOW
Hair loss - 10 mg acitretin (follow-up 6 months)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	very serious^c	none	3/20 (15%)	2/20 (10%)	RR 1.5 (0.28 to 8.04)	50 more per 1000 (from 72 fewer to 704 more)	⊕○○○ VERY LOW
Hair loss - 25 mg acitretin (follow-up 6 months)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	very serious^c	none	3/20 (15%)	2/20 (10%)	RR 1.5 (0.28 to 8.04)	50 more per 1000 (from 72 fewer to 704 more)	⊕○○○ VERY LOW
Hair loss - 50 mg acitretin (follow-up 6 months)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	no serious imprecision	none	15/20 (75%)	2/20 (10%)	RR 7.5 (1.97 to 28.61)	650 more per 1000 (from 97 more to 1000 more)	⊕○○○ VERY LOW
Increased triglycerides - 10 mg acitretin (follow-up 8 weeks)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	very serious^c	none	2/18 (11.1%)	1/19 (5.3%)	RR 2.11 (0.21 to 21.32)	58 more per 1000 (from 42 fewer to 1000 more)	⊕○○○ VERY LOW
Increased triglycerides - 25 mg acitretin (follow-up 8 weeks)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	very serious^c	none	2/17 (11.8%)	1/19 (5.3%)	RR 2.24 (0.22 to 22.51)	65 more per 1000 (from 41 fewer to 1000 more)	⊕○○○ VERY LOW
Increased triglycerides - 50 mg acitretin (follow-up 8 weeks)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	very serious^c	none	2/18 (11.1%)	1/19 (5.3%)	RR 2.11 (0.21 to 21.32)	58 more per 1000 (from 42 fewer to 1000 more)	⊕○○○ VERY LOW
Increased triglycerides - 10 mg acitretin (follow-up 6 months)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	very serious^c	none	1/16 (6.3%)	1/19 (5.3%)	RR 1.19 (0.08 to 17.51)	10 more per 1000 (from 48 fewer to 869 more)	⊕○○○ VERY LOW
Increased triglycerides - 25 mg acitretin (follow-up 6 months)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	very serious^c	none	1/15 (6.7%)	1/19 (5.3%)	RR 1.27 (0.09 to 18.62)	14 more per 1000 (from 48 fewer to 927 more)	⊕○○○ VERY LOW
Increased triglycerides - 50 mg acitretin (follow-up 6 months)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	very serious^c	none	0/15 (0%)	1/19 (5.3%)	RR 0.42 (0.02 to 9.55)	31 fewer per 1000 (from 52 fewer to 450 more)	⊕○○○ VERY LOW
Increased liver enzymes - 10 mg acitretin (follow-up 8 weeks)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	very serious^c	none	2/18 (11.1%)	0/19 (0%)	RR 5.26 (0.27 to 102.66)	-	⊕○○○ VERY LOW
Increased liver enzymes - 25 mg acitretin (follow-up 8 weeks)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	no serious imprecision	none	0/17 (0%)	0/19 (0%)	not pooled	not pooled	⊕○○○ VERY LOW
Increased liver enzymes - 50 mg acitretin (follow-up 8 weeks)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	no serious imprecision	none	0/18 (0%)	0/19 (0%)	not pooled	not pooled	⊕○○○ VERY LOW
Increased liver enzymes - 10 mg acitretin (follow-up 6 months)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	very serious^c	none	1/16 (6.3%)	0/19 (0%)	RR 3.53 (0.15 to 81.11)	-	⊕○○○ VERY LOW
Increased liver enzymes - 25 mg acitretin (follow-up 6 months)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	very serious^c	none	3/15 (20%)	0/19 (0%)	RR 8.75 (0.49 to 157.34)	-	⊕○○○ VERY LOW
Increased liver enzymes - 50 mg acitretin (follow-up 6 months)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	very serious^c	none	2/15 (13.3%)	0/19 (0%)	RR 6.25 (0.32 to 121.14)	-	⊕○○○ VERY LOW
Increased cholesterol - 10 mg acitretin (follow-up 8 weeks)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	very serious^c	none	2/18 (11.1%)	3/19 (15.8%)	RR 0.7 (0.13 to 3.73)	47 fewer per 1000 (from 137 fewer to 431 more)	⊕○○○ VERY LOW
Increased cholesterol - 25 mg acitretin (follow-up 8 weeks)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	very serious^c	none	5/17 (29.4%)	3/19 (15.8%)	RR 1.86 (0.52 to 6.65)	136 more per 1000 (from 76 fewer to 892 more)	⊕○○○ VERY LOW
Increased cholesterol - 50 mg acitretin (follow-up 8 weeks)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	very serious^c	none	3/18 (16.7%)	3/19 (15.8%)	RR 1.06 (0.24 to 4.57)	9 more per 1000 (from 120 fewer to 564 more)	⊕○○○ VERY LOW
Increased cholesterol - 10 mg acitretin (follow-up 6 months)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	very serious^c	none	2/16 (12.5%)	1/19 (5.3%)	RR 2.38 (0.24 to 23.84)	73 more per 1000 (from 40 fewer to 1000 more)	⊕○○○ VERY LOW
Increased cholesterol - 25 mg acitretin (follow-up 6 months)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	very serious^c	none	0/15 (0%)	1/19 (5.3%)	RR 0.42 (0.02 to 9.55)	31 fewer per 1000 (from 52 fewer to 450 more)	⊕○○○ VERY LOW
Withdrawal due to toxicity (all doses) (follow-up 6 months)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	very serious^c	none	1/57 (1.8%)	0/19 (0%)	RR 1.03 (0.04 to 24.38)	-	⊕○○○ VERY LOW
Improvement in sign scores (follow-up 8 weeks; Better indicated by higher values)
1 Kingston 1987	randomised trials	very serious^k	no serious inconsistency	very serious^l	serious^m	none	10	11	50 or 75 mg/day showed significant improvement on every parameter (scaling, erythema, thickness and pustulation), whereas those receiving 0 or 10 mg/day did not Most patients needed daily doses ≥0.66 mg/kg to initiate remission		⊕○○○ VERY LOW
Final PASI (maintenance phase) (follow-up 6 months; Better indicated by lower values)
1 Lassus 1988	observational studiesⁿ	very serious^j	no serious inconsistency	no serious indirectness	serious^m	none	10, 25 or 50 mg 60	20	No significant difference in PASI score between the placebo, 10, 25 and 50 mg groups		⊕○○○ VERY LOW
Change in PASI (follow-up 8 weeks; Better indicated by higher values)
1 Lassus 1988	randomised trials	very serious^e	no serious inconsistency	very serious^j	serious^o	none	25 or 50 mg 40	40	Significantly greater reduction in PASI on 25 and 50 mg/day compared with placebo (p<0.05) No significant difference between 25 and 50 mg The mean percentage decrease in PASI score in the 10 mg group was greater than in the placebo group, but did not differ significantly from any other group		⊕○○○ VERY LOW
Adverse events (follow-up 6 months; Better indicated by lower values)
1 Kingston 1987	observational studiesⁿ	very serious^k	no serious inconsistency	serious^p	serious^o	dose response gradient^q	21		More side effects at higher doses %of those receiving ≥0.66 mg/kg with: Cheilitis & mucosal dryness: 89 % Palmoplantar peeling: 86% Alopecia : 58%		⊕○○○ VERY LOW

a: 2/2 unclear allocation concealment and blinding not explained fully

b: Unclear reporting of baseline characteristics and in Lassus trial steroids administered on request (numbers using differed between the groups); Goldfarb data is surrogate outcome measure of >75% global improvement

c: Confidence interval crosses the boundary for clinical significance in favour of both interventions, as well as line of no effect

d: Confidence interval ranges from clinically important effect to no effect

e: Unclear allocation concealment and blinding not explained fully

f: Unclear reporting of baseline characteristics and data are surrogate outcome measure of >75% global improvement

g: Unclear reporting of baseline characteristics and in Lassus trial steroids administered on request (numbers using differed between the groups)

h: Unclear reporting of baseline characteristics

i: Serious imprecision according to GDG discussion (confidence interval ranges from clinically important harm to no clinically important harm)

j: Disease severity at baseline not reported and steroids administered on request (numbers using differed between the groups)

k: Unclear baseline characteristics; high drop-out rate (38.1%) and numbers in each arm not given

l: Surrogate outcome for change in PASI and placebo and 10 mg group combined

m: No numerical data

n: Open extension phase of RCT with dose adjustment

o: Insufficient information to analyse precision

p: Surrogate outcome measure for serious adverse events

q: There were more side effects at higher doses

Quality assessment							Summary of findings
Quality assessment							No of patients		Effect		Quality
No of studies	Design	Limitations	Inconsistency	Indirectness	Imprecision	Other considerations	Decreasing acitretin dosing schedule	Increasing acitretin dosing schedule	Relative (95% CI)	Absolute	Quality
% change in PASI (follow-up 6 weeks; better indicated by higher values)
1 Berbis, 1989	randomised trials	serious^a	no serious inconsistency	serious^b	serious^c	none	19	21	-	MD 6.8 higher (Decreasing: 67.1% Increasing: 62.7%)	⊕○○○ VERY LOW
Cheilitis (follow-up 6 weeks)
1 Berbis, 1989	randomised trials	serious^a	no serious inconsistency	serious^b	no serious imprecision	none	21/21 (100%)	21/21 (100%)	RR 1 (0.91 to 1.09)	0 fewer per 1000 (from 90 fewer to 90 more)	⊕⊕○○ LOW
Hair loss (follow-up 6 weeks)
1 Berbis, 1989	randomised trials	serious^a	no serious inconsistency	serious^b	serious^d	none	6/21 (28.6%)	1/21 (4.8%)	RR 6 (0.79 to 45.63)	238 more per 1000 (from 10 fewer to 2125 more)	⊕○○○ VERY LOW
Withdrawal due to toxicity (follow-up 6 weeks)
1 Berbis, 1989	randomised trials	serious^a	no serious inconsistency	serious^b	very serious^e	none	2/21 (9.5%)	0/20 (0%)	RR 4.77 (0.24 to 93.67)	0 more per 1000 (from 0 fewer to 0 more)	⊕○○○ VERY LOW
Serious adverse events (follow-up 6 weeks; better indicated by lower values)
1 Berbis, 1989	randomised trials	serious^a	no serious inconsistency	serious^f	serious^g	none	20	19	-	See Table 139	⊕○○○ VERY LOW

a: Unclear allocation concealment

b: Higher proportion of men in group 1 and more with pustular and guttate psoriasis in group 3

c: No SD provided

d: Confidence interval ranges from clinically important effect to no effect

e: Confidence interval crosses the boundary for clinical significance in favour of both interventions, as well as line of no effect

f: Analysing different doses within each randomised group (not the randomised comparison)

g: Insufficient data to analyse precision

Table 139Summary of non-analysed data for increasing vs decreasing acitretin dosing

Study	Total N	Follow-up	Result					Treatment favoured
Severe clinical adverse reactions

Berbis	42	6 weeks	Treatment period	Increasing dose		Decreasing dose		Low dose
			Treatment period	Dose (mg/d)	N′/n	Dose (mg/d)	N′/n
			Week 0–2^*	10	0/21	50	9/21
			Week 3–4	30	3/20	30	5/20
			Week 5–6^**	50	8/20	10	2/19

*: Increasing vs decreasing: p<0.01

**: Increasing vs decreasing: p =0.06

Quality assessment							Summary of findings
Quality assessment							No of patients		Effect		Quality
No of studies	Design	Limitations	Inconsistency	Indirectness	Imprecision	Other considerations	Constant acitretin dosing schedule	Increasing acitretin dosing schedule	Relative (95% CI)	Absolute	Quality
% change in PASI (6 weeks) (follow-up 6 weeks; better indicated by higher values)
1 Berbis, 1989	randomised trials	serious^a	no serious inconsistency	serious^b	serious^c	none	19	21	-	MD 6.8 lower (Constant 55.9% Increasing: 62.7%)	⊕○○○ VERY LOW
Cheilitis (follow-up 6 weeks)
1 Berbis, 1989	randomised trials	serious^a	no serious inconsistency	serious^b	no serious imprecision	none	23/23 (100%)	21/21 (100%)	RR 1 (0.92 to 1.09)	0 fewer per 1000 (from 80 fewer to 90 more)	⊕⊕○○ LOW
Hair loss (follow-up 6 weeks)
1 Berbis, 1989	randomised trials	serious^a	no serious inconsistency	serious^b	very serious^d	none	2/23 (8.7%)	1/21 (4.8%)	RR 1.83 (0.18 to 18.7)	40 more per 1000 (from 39 fewer to 843 more)	⊕○○○ VERY LOW
Withdrawal due to toxicity (follow-up 6 weeks)
1 Berbis, 1989	randomised trials	serious^a	no serious inconsistency	serious^b	very serious^d	none	3/22 (13.6%)	0/20 (0%)	RR 6.39 (0.35 to 116.57)	0 more per 1000 (from 0 fewer to 0 more)	⊕○○○ VERY LOW

a: Unclear allocation concealment

b: Higher proportion of men in group 1 and more with pustular and guttate psoriasis in group 3

c: No SD provided

d: Confidence interval crosses the boundary for clinical significance in favour of both interventions, as well as line of no effect

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Ciclosporin	Placebo	Relative (95% CI)	Absolute	Quality
Clear/nearly clear on PGA - CSA 3 mg/kg/day (follow-up 8 weeks)
1 Ellis 1991	randomised trials	serious^a	no serious inconsistency	no serious indirectness	no serious imprecision	none	9/25 (36%)	0/25 (0%)	RR 19.00 (1.17 to 309.77)	-	⊕⊕⊕○ MODERATE
Clear/nearly clear on PGA - CSA 5 mg/kg/day (follow-up 8 weeks)
1 Ellis 1991	randomised trials	serious^a	no serious inconsistency	no serious indirectness	no serious imprecision	none	13/20 (65%)	0/25 (0%)	RR 33.43 (2.11 to 530)	-	⊕⊕⊕○ MODERATE
Clear/nearly clear on PGA - 7.5 mg/kg/day (follow-up 8 weeks)
1 Ellis 1991	randomised trials	serious^a	no serious inconsistency	no serious indirectness	no serious imprecision	none	12/15 (80%)	0/25 (0%)	RR 40.63 (2.58 to 640.1)	-	⊕⊕⊕○ MODERATE
Clearance - CSA 14 mg/kg/day (follow-up 4 weeks)
1 Ellis 1986	randomised trials	serious^a	no serious inconsistency	no serious indirectness	very serious^b	none	2/11 (18.2%)	0/10 (0%)	RR 4.58 (0.25 to 85.33)	-	⊕○○○ VERY LOW
PASI 75 - CSA 1.25 mg/kg/day (follow-up 10 weeks)
1 Meffert 1997	randomised trials	serious^c	no serious inconsistency	no serious indirectness^d	very serious^b	none	4/41 (9.8%)	2/43 (4.7%)	RR 2.1 (0.41 to 10.84)	51 more per 1000 (from 27 fewer to 458 more)	⊕○○○ VERY LOW
PASI 75 - CSA 2.5–3.0 mg/kg/day (follow-up 8–10 weeks)
2 Meffert 1997 Ellis 1991	randomised trials	serious^e	no serious inconsistency	no serious indirectness^d	no serious imprecision	none	16/69 (23.2%)	3/68 (4.4%)	RR 6.24 (1.94 to 20.11)	231 more per 1000 (from 41 more to 843 more)	⊕⊕⊕○ MODERATE
PASI 75 - CSA 5 mg/kg/day (follow-up 8 weeks)
1 Ellis 1991	randomised trials	serious^a	no serious inconsistency	no serious indirectness^d	no serious imprecision	none	12/20 (60%)	1/25 (4%)	RR 15.00 (2.13 to 105.79)	560 more per 1000 (from 45 more to 1000 more)	⊕⊕⊕○ MODERATE
PASI 50 CSA 2.5–7 mg/kg/day (follow-up 4–10 weeks)
2 Guenther 1991 van Joost 1988	randomised trials	very serious^f	no serious inconsistency	no serious indirectness	no serious imprecision	none	20/22 (90.9%)	1/21 (4.8%)	RR 12.97 (2.77 to 60.81)	570 more per 1000 (from 84 more to 1000 more)	⊕⊕○○ LOW
Mean % change in PASI - CSA 2.5 mg/kg/day (follow-up 10 weeks; Better indicated by higher values)
1 Meffert 1997	randomised trials	serious^c	no serious inconsistency	no serious indirectness	no serious imprecision	none	41	39	-	MD 45.1 higher (30.34 to 59.86 higher)	⊕⊕⊕○ MODERATE
Mean % change in PASI - CSA 1.25 mg/kg/day (follow-up 10 weeks; Better indicated by higher values)
1 Meffert 1997	randomised trials	serious^c	no serious inconsistency	no serious indirectness	no serious imprecision	none	40	39	-	MD 21.3 higher (5.7 to 36.9 higher)	⊕⊕⊕○ MODERATE
Hypertension CSA 2.5–14 mg/kg/day (follow-up 8–10 weeks)
2 Guenther 1991 Ellis 1986	randomised trials	serious^g	no serious inconsistency	no serious indirectness	very serious^b	none	9/23 (39.1%)	7/21 (33.3%)	RR 1.15 (0.61 to 2.17)	50 more per 1000 (from 130 fewer to 390 more)	⊕○○○ VERY LOW
Decreased GFR - CSA 3 mg/kg (follow-up 8 weeks)
1 Ellis 1991	randomised trials	serious^a	no serious inconsistency	no serious indirectness	very serious^b	none	4/12 (33.3%)	0/9 (0%)	RR 6.92 (0.42 to 114.19)	-	⊕○○○ VERY LOW
Decreased GFR - CSA 5 mg/kg (follow-up 8 weeks)
1 Ellis 1991	randomised trials	serious^a	no serious inconsistency	no serious indirectness	very serious^b	none	5/10 (50%)	0/9 (0%)	RR 10 (0.63 to 158.87)	-	⊕○○○ VERY LOW
Decreased GFR - CSA 7.5 mg/kg (follow-up 8 weeks)
1 Ellis 1991	randomised trials	serious^a	no serious inconsistency	no serious indirectness	serious^h	none	9/12 (75%)	0/9 (0%)	RR 14.62 (0.96 to 222.24)	-	⊕⊕○○ LOW
Withdrawal due to toxicity CSA 5–14 mg/kg/day (follow-up 4 weeks)
2 Ellis 1986 van Joost 1988	randomised trials	seriousⁱ	no serious inconsistency	no serious indirectness	no serious imprecision	none	0/21 (0%)	0/20 (0%)	-	-	⊕⊕⊕○ MODERATE
Change in PASI CSA 3.0–7.5 mg/kg/day (follow-up 8 weeks; Better indicated by lower values)
1 Ellis 1991	randomised trials	serious^a	no serious inconsistency	no serious indirectness	serious^j	none	60	25	-	PASI improved significantly in all groups receiving CSA compared to placebo (P<0.001 for each),	⊕⊕○○ LOW
Change in PASI CSA 3.0–7.5 mg/kg/day (follow-up 8 weeks; Better indicated by lower values)
1 Ellis 1991	randomised trials	serious^a	no serious inconsistency	no serious indirectness	serious^j	none	60	25	-	NS difference in PASI score between 5 and 7 mg/kg (P>0.4), but each better than the response in the group receiving the lowest dose (P<0.01 for each comparison).	⊕⊕○○ LOW

a: Unclear allocation concealment

b: Confidence interval crosses the boundary for clinical significance in favour of both interventions, as well as line of no effect

c: Unclear method of randomisation and allocation concealment

d: These data were derived from a published review

e: 2/2 unclear allocation concealment; 1/2 unclear method of randomisation

f: Unclear allocation concealment in 2/2 studies; 1/2 high differential dropout in placebo group (8/11 withdrawn due to treatment failure by week 6)

g: Unclear method of randomisation and allocation concealment in 2/2 studies

h: Confidence interval ranges from clinically important effect to no effect

i: 2/2 unclear allocation concealment

j: Insufficient data to analyse precision

Quality assessment							Summary of findings
Quality assessment							No of patients		Effect		Quality
No of studies	Design	Limitations	Inconsistency	Indirectness	Imprecision	Other considerations	Ciclosporin low dose	Ciclosporin high dose	Relative (95% CI)	Absolute	Quality
PASI 75 – initial CSA dose 1.25 vs 2.5 mg/kg (follow-up 12–36 weeks)
1 Christophers, 1992	randomised trials	very serious^a	no serious inconsistency	very serious^b	serious^c	none	68/109 (62.4%)	78/108 (72.2%)	RR 0.86 (0.72 to 1.04)	101 fewer per 1000 (from 202 fewer to 22 more)	⊕○○○ VERY LOW
PASI 75 - CSA 2.5 vs 5.0 mg/kg (follow-up 12 weeks)
1 Laburte, 1994	randomised trials	very serious^a	no serious inconsistency	serious^b	no serious imprecision	none	57/119 (47.9%)	117/132 (88.6%)	RR 0.54 (0.44 to 0.66)	408 fewer per 1000 (from 301 fewer to 496 fewer)	⊕○○○ VERY LOW
Elevated creatinine - CSA 1.25 mg/kg vs CSA 2.5 mg/kg (follow-up 12–36 weeks)
1 Christophers, 1992	observational studies^f	serious^d	no serious inconsistency	no serious indirectness	very serious5	none	1/109 (0.9%)	9/183 (4.9%)	RR 0.19 (0.02 to 1.45)	40 fewer per 1000 (from 48 fewer to 22 more)	⊕○○○ VERY LOW
Elevated creatinine - CSA 2.5 mg/kg vs CSA 5 mg/kg (follow-up 12–36 weeks)
1 Christophers, 1992	observational studies	serious^d	no serious inconsistency	no serious indirectness	serious ^f	none	9/183 (4.9%)	8/60 (13.3%)	RR 0.37 (0.15 to 0.91)	84 fewer per 1000 (from 12 fewer to 113 fewer)	⊕○○○ VERY LOW
Hypertension - CSA 1.25 mg/kg vs CSA 2.5 mg/kg (follow-up 12–36 weeks)
1 Christophers, 1992	observational studies^f	serious4	no serious inconsistency	no serious indirectness	serious ^f	none	12/109 (11%)	38/183 (20.8%)	RR 0.53 (0.29 to 0.97)	98 fewer per 1000 (from 6 fewer to 147 fewer)	⊕○○○ VERY LOW
Hypertension - CSA 2.5 mg/kg vs CSA 5 mg/kg (follow-up 12–36 weeks)
1 Christophers, 1992	observational studies^f	serious^d	no serious inconsistency	no serious indirectness	very serious^e	none	38/183 (20.8%)	16/60 (26.7%)	RR 0.78 (0.47 to 1.29)	59 fewer per 1000 (from 141 fewer to 77 more)	⊕○○○ VERY LOW
Elevated uric acid (>400 micromol/L) - CSA 1.25 mg/kg vs CSA 2.5 mg/kg (follow-up 12–36 weeks)
1 Christophers, 1992	observational studies^f	serious^d	no serious inconsistency	no serious indirectness	serious^c	none	21/109 (19.3%)	51/183 (27.9%)	RR 0.69 (0.44 to 1.08)	86 fewer per 1000 (from 156 fewer to 22 more)	⊕○○○ VERY LOW
Elevated uric acid (>400 micromol/L) - CSA 2.5 mg/kg vs CSA 5 mg/kg (follow-up 12–36 weeks)
1 Christophers, 1992	observational studies^f	serious^d	no serious inconsistency	no serious indirectness	serious ^f	none	51/183 (27.9%)	26/60 (43.3%)	RR 0.64 (0.44 to 0.93)	156 fewer per 1000 (from 30 fewer to 243 fewer)	⊕○○○ VERY LOW
PASI75 (dose increases) (follow-up 12–36 weeks)
1 Christophers, 1992	observational studies^f	serious^d	no serious inconsistency	no serious indirectness	serious^h	dose response gradientⁱ	109		See Table 140	-	⊕○○○ VERY LOW

a: Unclear allocation concealment, unblinded and unclear dropout rate

b: Patients did not receive the randomised dose for the full induction period

c: Confidence interval ranges from clinically important effect to no effect

d: Unclear drop-out rates and outcomes reported as percentages but the denominators were sometimes unclear due to patients moving between dosage groups

e: Confidence interval crosses the boundary for clinical significance in favour of both treatment, as well as line of no effect

f: Serious imprecision according to GDG discussion (confidence interval ranges from clinically important harm to no clinically important harm)

g: Non-randomised comparison within RCT

h: Not analysed in MA because non-randomised comparison

i: Increasing dose increased the chance of PASI75

Table 140Summary of non-analysed data for ciclosporin dosing increments for induction

Study	Total N	Follow-up	Result
PASI75

Christophers 1992	109	12–36 weeks	Initial dose 1.25mg/kg/day
			Remission on to 1.25mg/kg/day	Remission after increased 2.5mg/kg/day	Remission after increased again to 5mg/kg/day
			19/109 (17.4%)	27/90 (30.0%)	22/63 (34.9%)

	108	12–36 weeks	Initial dose 2.5mg/kg/day
			Remission on 2.5mg/kg/day	Remission after increased to 5mg/kg/day
			60/108 (55.6%)	18/48 (37.5%)

Quality assessment							No of patients		Effect		Quality
No of studies	Design	Risk of bias	Inconsistency	Indirectness	Imprecision	Other considerations	Ciclosporin	Placebo	Relative (95% CI)	Absolute	Quality
PASI 75 – CSA 5 mg/kg/day at weekends only (follow-up 24 weeks)
1 Colombo 2010	randomised trials	very serious^a	no serious inconsistency	no serious indirectness	serious^b	none	85/127 (66.9%)	33/62 (53.2%)	RR 1.26 (0.97 to 1.64)	138 more per 1000 (from 16 fewer to 341 more)	⊕○○○ VERY LOW
Mean final PASI – CSA 5 mg/kg/day at weekends only (follow-up 24 weeks; better indicated by lower values)
1 Colombo 2010	randomised trials	very serious^a	no serious inconsistency	serious^c	no serious imprecision	none	127	62	-	MD 1.5 lower (4.14 lower to 1.14 higher)	⊕○○○ VERY LOW
Maintaining at least mild psoriasis following PASI75 – CSA three-times weekly (follow-up 12 weeks)
1 Thaci 2002	randomised trials	serious^d	no serious inconsistency	no serious indirectness	serious^b	none	14/31 (45.2%)	5/22 (22.7%)	RR 1.99 (0.84 to 4.71)	225 more per 1000 (from 36 fewer to 843 more)	⊕⊕○○ LOW
Time-to-relapse – CSA three-times weekly (follow-up 12 weeks)
1 Thaci 2002	randomised trials	serious^d	no serious inconsistency	no serious indirectness	no serious imprecision	none	17/42 (40.5%)	29/51 (56.9%)	HR 0.45 (0.24 to 0.82)	254 fewer per 1000 (from 70 fewer to 386 fewer)	⊕⊕⊕○ MODERATE
Time-to-relapse – CSA 3 mg/kg/day (follow-up 24 weeks)
1 Shupack 1997	randomised trials	serious^d	no serious inconsistency	no serious indirectness	no serious imprecision	Median time CSA 3mg/kg/day: >24 weeks Placebo or CSA 1.5mg/kg/day: 6 weeks	35/83 (42.2%)	40/48 (83.3%)	HR 0.30 (0.19 to 0.49)	418 fewer per 1000 (from 249 fewer to 545 fewer)	⊕⊕⊕○ MODERATE
Mean time to relapse (weeks) - CSA 1.5 mg/kg/day (follow-up up to 4 months; better indicated by higher values)
1 Ellis 1995	randomised trials	serious^a	no serious inconsistency	no serious indirectness	serious^b	none	20	20	-	MD 2 higher (0.77 lower to 4.77 higher)	⊕⊕○○ LOW
Relapse rate - CSA 1.5 mg/kg/day (follow-up up to 4 months)
1 Ellis 1995	randomised trials	serious^a	no serious inconsistency	serious^e	serious^b	none	14/20 (70%)	18/20 (90%)	RR 0.78 (0.56 to 1.07)	198 fewer per 1000 (from 396 fewer to 63 more)	⊕○○○ VERY LOW
Mean time to relapse (weeks) - CSA 3 mg/kg/day (follow-up up to 4 months; better indicated by higher values)
1 Ellis 1995	randomised trials	serious^a	no serious inconsistency	no serious indirectness	no serious imprecision	none	21	20	-	MD 5 higher (2.23 to 7.77 higher)	⊕⊕⊕○ MODERATE
Relapse rate - CSA 3 mg/kg/day (follow-up up to 4 months)
1 Ellis 1995	randomised trials	serious^a	no serious inconsistency	serious^e	no serious imprecision	none	8/21 (38.1%)	18/20 (90%)	RR 0.42 (0.24 to 0.74)	522 fewer per 1000 (from 234 fewer to 684 fewer)	⊕⊕○○ LOW
Relapse rate – CSA 5 mg/kg/day at weekends only (follow-up up to 24 weeks)
1 Colombo 2010	randomised trials	very serious^a	no serious inconsistency	no serious indirectness	serious^b	none	42/127 (33.1%)	29/62 (46.8%)	RR 0.71 (0.49 to 1.02)	136 fewer per 1000 (from 239 fewer to 9 more)	⊕○○○ VERY LOW
Withdrawal due to toxicity – CSA 5 mg/kg/day at weekends only (follow-up 24 weeks)
1 Colombo 2010	randomised trials	very serious^f	no serious inconsistency	no serious indirectness	very serious^g	none	8/160 (5%)	2/79 (2.5%)	RR 1.98 (0.43 to 9.08)	25 more per 1000 (from 14 fewer to 205 more)	⊕○○○ VERY LOW
Severe adverse events – CSA 5 mg/kg/day at weekends only (follow-up 24 weeks)
1 Colombo 2010	randomised trials	very serious^a	no serious inconsistency	no serious indirectness	very serious^g	none	1/160 (0.6%)	0/79 (0%)	RR 1.49 (0.06 to 36.18)	-	⊕○○○ VERY LOW
Elevated serum creatinine – CSA 5 mg/kg/day at weekends only (follow-up 24 weeks)
1 Colombo 2010	randomised trials	very serious^a	no serious inconsistency	no serious indirectness	very serious^g	none	8/160 (5%)	3/79 (3.8%)	RR 1.32 (0.36 to 4.83)	12 more per 1000 (from 24 fewer to 145 more)	⊕○○○ VERY LOW
Elevated serum creatinine – CSA three-times weekly (at 2 consecutive visits) (follow-up 12 weeks)
1 Thaci 2002	randomised trials	serious^d	no serious inconsistency	no serious indirectness	no serious imprecision	none	0/42 (0%)	0/51 (0%)	not pooled	not pooled	⊕⊕⊕○ MODERATE
Change in PASI – CSA three-times weekly (follow-up 12 weeks; better indicated by lower values)
1 Thaci 2002	randomised trials	serious^d	no serious inconsistency	no serious indirectness	serious^h	none	42	51	-	Mean PASI increase CSA: 2.7 to 9.9 Placebo: 3.0 to 11.9	⊕⊕○○ LOW
Median time to relapse – CSA three-times weekly (follow-up 12 weeks; better indicated by higher values)
1 Thaci 2002	randomised trials	serious^d	no serious inconsistency	no serious indirectness	seriousⁱ	none	42	51	-	CSA: 98 days Placebo: 69 days	⊕⊕○○ LOW
Time to relapse – CSA 5 mg/kg/day at weekends only (follow-up 24 weeks; better indicated by higher values)
1 Colombo 2010	randomised trials	very serious^a	no serious inconsistency	serious^j	serious	none	160	79	p = 0.0233 (favours CSA)	-	⊕○○○ VERY LOW

a: Unclear method of randomisation and unclear allocation concealment and high dropout rate (30% - figures reported were per protocol)

b: Confidence interval ranges from clinically important effect to no effect

c: Surrogate measure for change in PASI

d: Unclear allocation concealment

e: Surrogate for time to relapse

f: Unclear method of randomisation and unclear allocation concealment and high dropout rate (30%)

g: Confidence interval crosses the boundary for clinical significance in favour of both treatment, as well as line of no effect

h: No range or SD around change scores

i: No range stated

j: Only p-value provided

Quality assessment							Summary of findings
Quality assessment							No of patients		Effect		Quality
No of studies	Design	Limitations	Inconsistency	Indirectness	Imprecision	Other considerations	Continuous CSA	Intermittent (abrupt stop) CSA	Relative (95% CI)	Absolute	Quality
Clear/nearly clear (PASI90) (follow-up 9 months)
1 Chaidemenos, 2007	randomised trials	very serious^a	no serious inconsistency	no serious indirectness	no serious imprecision	none	14/24 (58.3%)	4/21 (19%)	RR 3.06 (1.19 to 7.87)	392 more per 1000 (from 36 more to 1309 more)	⊕⊕○○ LOW
PASI75 (follow-up 9 months)
1 Chaidemenos, 2007	randomised trials	very serious^a	no serious inconsistency	no serious indirectness	serious^b	none	22/24 (91.7%)	13/21 (61.9%)	RR 1.48 (1.04 to 2.12)	297 more per 1000 (from 25 more to 693 more)	⊕○○○ VERY LOW
PASI50 (follow-up 9 months)
1 Chaidemenos, 2007	randomised trials	very serious^a	no serious inconsistency	no serious indirectness	no serious imprecision	none	23/24 (95.8%)	20/21 (95.2%)	RR 1.01 (0.89 to 1.14)	10 more per 1000 (from 105 fewer to 133 more)	⊕⊕○○ LOW
Increased serum creatinine (follow-up 9 months)
1 Chaidemenos, 2007	randomised trials	very serious^a	no serious inconsistency	no serious indirectness	very serious^c	none	2/24 (8.3%)	2/21 (9.5%)	RR 0.88 (0.13 to 5.68)	11 fewer per 1000 (from 83 fewer to 446 more)	⊕○○○ VERY LOW
Hypertension (follow-up 9 months)
1 Chaidemenos, 2007	randomised trials	very serious^a	no serious inconsistency	no serious indirectness	very serious^c	none	1/24 (4.2%)	0/21 (0%)	RR 2.64 (0.11 to 61.54)	0 more per 1000 (from 0 fewer to 0 more)	⊕○○○ VERY LOW
Time-to-relapse (follow-up 1 year)
1 Ho, 1999	randomised trials	very serious^d	no serious inconsistency	no serious indirectness	serious ^b	Median time-torelapse Continuous: 113 days Intermittent: 109 days	173	192	HR 0.77 (0.61–0.98)	0 fewer per 1000 (from 0 fewer to 0 fewer)	⊕○○○ VERY LOW
Median time to relapse (follow-up 2 years; Better indicated by higher values)
1 Ho, 2001	randomised trials	very serious^d	no serious inconsistency	no serious indirectness	serious^e	none	30	46	-	Continuous : 119.5 days Intermitten t: 115 days	⊕○○○ VERY LOW

a: Quasi-randomised and inadequate allocation concealment

b: Serious imprecision according to GDG discussion (confidence interval ranges from clinically important benefit/harm to no clinically important benefit/harm

c: Confidence interval crosses the boundary for clinical significance in favour of both interventions, as well as line of no effect

d: Unclear allocation concealment and unblended

e: No range stated

Quality assessment							Summary of findings
Quality assessment							No of patients		Effect		Quality
No of studies	Design	Limitations	Inconsistency	Indirectness	Imprecision	Other considerations	Intermittent (taper to cessation) CSA	Continuous CSA	Relative (95% CI)	Absolute	Quality
Percentage change in PASI (follow-up 48 months; better indicated by higher values)
1 Ozawa, 1999	randomised trials	very serious^a	no serious inconsistency	serious^b	no serious imprecision	none	20	17	-	MD 9.3 higher (6.05 to 12.55 higher)	⊕○○○ VERY LOW
Final PASI (follow-up >48 months; better indicated by lower values)
1 Ohtsuki, 2003	randomised trials	very serious^d	no serious inconsistency	serious^e	no serious imprecision	none	16	15	-	MD 3.56 higher (2.37 to 4.75 higher)	⊕○○○ VERY LOW
Withdrawal due to toxicity (follow-up 48 months)
1 Ozawa, 1999	randomised trials	very serious^a	no serious inconsistency	serious^c	very serious	none	2/33 (6.1%)	1/35 (2.9%)	RR 2.12 (0.20 to 22.31)	32 more per 1000 (from 23 fewer to 609 more)	⊕○○○ VERY LOW
Hypertension (follow-up 1 year)
1 Ohtsuki, 2003	randomised trials	serious^d	no serious inconsistency	serious^e	very serious^b	none	10/61 (16.4%)	6/61 (9.8%)	RR 1.67 (0.65 to 4.3)	66 more per 1000 (from 34 fewer to 325 more)	⊕○○○ VERY LOW
Increased creatinine (follow-up 1 year)
1 Ohtsuki, 2003	randomised trials	serious^d	no serious inconsistency	serious^e	very serious^b	none	3/61 (4.9%)	2/61 (3.3%)	RR 1.5 (0.26 to 8.66)	16 more per 1000 (from 24 fewer to 251 more)	⊕○○○ VERY LOW
Hyperuricaemia (follow-up 1 year)
1 Ohtsuki, 2003	randomised trials	serious^d	no serious inconsistency	serious^e	very serious5	none	6/61 (9.8%)	3/61 (4.9%)	RR 2 (0.52 to 7.64)	49 more per 1000 (from 24 fewer to 327 more)	⊕○○○ VERY LOW
Increased liver enzymes (follow-up 1 year)
1 Ohtsuki, 2003	randomised trials	serious^d	no serious inconsistency	serious^e	very serious^b	none	3/61 (4.9%)	0/61 (0%)	RR 7 (0.37 to 132.7)	0 more per 1000 (from 0 fewer to 0 more)	⊕○○○ VERY LOW

a: High dropout rate (continuous: 32%; intermittent: 29.5%) and patients lost due to relapse or remission not counted in analysis; unclear allocation concealment and blinding

b: Confidence interval crosses the boundary for clinical significance in favour of both interventions, as well as line of no effect

c: No baseline data except PASI score so unclear if groups are balanced

d: High dropout in both groups: 45/61 in intermittent group and 46/61 in continuous group (reasons in each group unclear); but data available for all for adverse event outcomes; unblended

e: Many patients in intermittent group restarted ciclosporin earlier than in the protocol (so regimen more like the continuous treatment than planned).

Quality assessment							Summary of findings
Quality assessment							No of patients		Effect		Quality
No of studies	Design	Limitations	Inconsistency	Indirectness	Imprecision	Other considerations	Ciclosporin low dose	Ciclosporin high dose	Relative (95% CI)	Absolute	Quality
Severe adverse events (follow-up 18 months)
1 Laburte, 1994	randomised trials	very serious^a	no serious inconsistency	no serious indirectness	no serious imprecision	none	2/119 (1.7%)	17/132 (12.9%)	RR 0.13 (0.03 to 0.55)	112 fewer per 1000 (from 58 fewer to 125 fewer)	⊕⊕○○ LOW
Hypertension (follow-up 18 months)
1 Laburte, 1994	randomised trials	very serious^a	no serious inconsistency	no serious indirectness	very serious^b	none	17/119 (14.3%)	20/132 (15.2%)	RR 0.94 (0.52 to 1.71)	9 fewer per 1000 (from 73 fewer to 108 more)	⊕○○○ VERY LOW
Elevated creatinine (follow-up 18 months)
1 Laburte, 1994	randomised trials	very serious^a	no serious inconsistency	no serious indirectness	no serious imprecision	none	26/119 (21.8%)	73/132 (55.3%)	RR 0.4 (0.27 to 0.57)	332 fewer per 1000 (from 238 fewer to 404 fewer)	⊕⊕○○ LOW
Elevated uric acid (follow-up 18 months)
1 Laburte, 1994	randomised trials	very serious^a	no serious inconsistency	no serious indirectness	very serious^b	none	5/119 (4.2%)	8/132 (6.1%)	RR 0.69 (0.23 to 2.06)	19 fewer per 1000 (from 47 fewer to 64 more)	⊕○○○ VERY LOW
Change in PASI (follow-up 18 months; Better indicated by lower values)
1 Laburte, 1994	randomised trials	very serious^a	no serious inconsistency	no serious indirectness	serious^c	none	119	132	-	2.5 mg: +1.7 5.0 mg: +2.7 See Table 141	⊕○○○ VERY LOW

a: Unclear method of randomisation, unclear allocation concealment, unblinded study

b: Confidence interval crosses the boundary for clinical significance in favour of both interventions, as well as line of no effect

c: No SD provided

Table 141Summary of non-analysed data for ciclosporin in the maintenance of remission

Study	Total N	Followup	Result				Treatment favoured
Change in PASI (during maintenance phase)

Laburte 1994	251	18 months		2.5 mg group	5 mg group	2.5 mg non-responders	No clear difference (1 PASI point)
			Beginning of maintenance	4.2 (n=52)	3.6 (n=116)	3.9 (n=41)
			End of maintenance	5.9 (n=40)	6.3 (n=79)	8.3 (n=25)
			Change	+1.7	+2.7	+4.4

Quality assessment							Summary of findings
Quality assessment							No of patients		Effect		Quality
No of studies	Design	Limitations	Inconsistency	Indirectness	Imprecision	Other considerations	CSA	Placebo	Relative (95% CI)	Absolute	Quality
Improvement (follow-up 4 weeks)
2 Erkko, 1998 Reitamo, 1993	randomised trials	serious^a	no serious inconsistency	no serious indirectness	no serious imprecision	none	30/46 (65.2%)	10/50 (20%)	RR 3.22 (1.78 to 5.85)	444 more per 1000 (from 156 more to 970 more) NNT = 2	⊕⊕⊕○ MODERATE
Hypertension (follow-up 4 weeks)
1 Erkko, 1998	randomised trials	serious^a	no serious inconsistency	no serious indirectness	very serious^b	none	1/27 (3.7%)	0/31 (0%)	RR 3.43 (0.15 to 80.83)	0 more per 1000 (from 0 fewer to 0 more)	⊕○○○ VERY LOW
Serum creatinine increased (follow-up 4 weeks)
1 Reitamo, 1993	randomised trials	serious^a	no serious inconsistency	no serious indirectness	no serious imprecision	none	0/19 (0%)	0/19 (0%)	-	-	⊕⊕⊕○ MODERATE
Hypertension (follow-up 2–12 months)
1 Erkko, 1998	observational studies3	serious^d	no serious inconsistency	serious^e	serious^f	none	7/27 (25.9%)	0/31 (0%)	RR 17.14 (1.02 to 286.86)	0 more per 1000 (from 0 more to 0 more)	⊕○○○ VERY LOW
Serum creatinine increased (follow-up 2–12 months)
1 Erkko, 1998	observational studies^c	serious^d	no serious inconsistency	serious^e	very serious^b	none	2/27 (7.4%)	0/31 (0%)	RR 5.71 (0.29 to 114.05)	0 more per 1000 (from 0 fewer to 0 more)	⊕○○○ VERY LOW
Improvement (open phase) (follow-up 4 months)
1 Reitamo, 1993	observational studies^c	serious^d	no serious inconsistency	serious^e	serious^b	none	10/14 (71.4%)	10/14 (71.4%)	RR 1 (0.63 to 1.6)	0 fewer per 1000 (from 264 fewer to 429 more)	⊕○○○ VERY LOW
Relapse rate (open phase) (follow-up 4 months)
1 Reitamo, 1993	observational studies^c	serious^d	no serious inconsistency	serious^g	very serious^b	none	0/19 (0%)	2/13 (15.4%)	RR 0.14 (0.01 to 2.7)	132 fewer per 1000 (from 152 fewer to 262 more)	⊕○○○ VERY LOW
Relapse rate (withdrawal phase) (follow-up 6 months)
1 Reitamo, 1993	observational studies^c	serious^d	no serious inconsistency	serious^g	very serious^b	none	6/10 (60%)	8/12 (66.7%)	RR 0.9 (0.47 to 1.72)	67 fewer per 1000 (from 353 fewer to 480 more)	⊕○○○ VERY LOW

a: Unclear allocation concealment and blinding

b: Confidence interval crosses the boundary for clinical significance in favour of both interventions, as well as line of no effect

c: Open phase of RCT

d: Unclear if still matched for demographic characteristics

e: Open phase of trial (patients originally randomised to placebo received ciclosporin if no response)

f: Serious imprecision according to GDG discussion (confidence interval ranges from clinically important harm to no clinically important harm)

g: Surrogate outcome for time-to-relapse and follow-up after open phase of trial (patients originally randomised to placebo received ciclosporin if no response)

Quality assessment							Summary of findings
Quality assessment							No of patients	Effect	Quality
No of studies	Design	Limitations	Inconsistency	Indirectness	Imprecision	Other considerations	Ciclosporin	Effect	Quality
Median time to 70% or 90% reduction in BSA (follow-up 16 weeks; better indicated by lower values)
1 Shupack 1997	observational studies^a	no serious limitations^b	no serious inconsistency	no serious indirectness	serious^c	none	181	Median time to 70% reduction in BSA: 8 weeks Median time to 90% reduction in BSA: 12 weeks	⊕○○○ VERY LOW
Median time to 75% reduction in BSA (follow-up 12 weeks; better indicated by lower values)
1 Ho 1999	observational studies^a	no serious limitations^b	no serious inconsistency	no serious indirectness	serious^c	none	365	Median time to satisfactory clinical response (≥75% reduction in BSA): 9.7 weeks	⊕○○○ VERY LOW
Mean time to PASI80 (follow-up to remission; better indicated by lower values)
1 Ozawa 1999	observational studies^a	no serious limitations ^d	no serious inconsistency	serious^e	serious^f	none	37	Mean time to remission (decrease in PASI of 80%): 15.4 weeks (4.7 months in continuous group and 3.0 months in intermittent group – but both received the same dose schedule during the induction period)	⊕○○○ VERY LOW
Mean time to maximum response (mean PASI); (follow-up 12 weeks; better indicated by lower values)
1 Flytstrom 2008	observational studies^a	no serious limitations^g	no serious inconsistency	no serious indirectness	very serious^h	none	31	Mean PASI score still decreasing at 12 weeks CSA response greatest over the first 4 weeks By 12 weeks the mean % improvement in PASI was 72%	⊕○○○ VERY LOW
Mean time to maximum response (mean PASI); (follow-up 24 weeks; better indicated by lower values)
1 Gumusel 2011	observational studies^a	no serious limitations^g	no serious inconsistency	no serious indirectness	very serious^h	none	17	Maximal response based on PASI score appeared to be at 16 weeks	⊕○○○ VERY LOW
Mean time to maximum response (mean PASI); (follow-up 16 weeks; better indicated by lower values)
1 Heydendael 2003	observational studies^a	no serious limitations^g	no serious inconsistency	no serious indirectness	very serious^h	none	42	Maximal response based on PASI score appeared to be at 12 weeks By 16 weeks the mean % improvement in PASI was 72%	⊕○○○ VERY LOW
Mean time to maximum response (mean % improvement in PASI); CSA (follow-up 24 weeks; better indicated by lower values)
1 Christophers 1992	randomised trials	very seriousⁱ	no serious inconsistency	serious^j	very serious^h	none	Remaining on 1.25 mg/kg/d: 26 Remaining on 2.5 mg/kg/d: 68	Mean % change in PASI beginning to plateau at 8–12 weeks in both dose groups (approaching PASI75 at higher dose by this time point)	⊕○○○ VERY LOW

a: Although the data are taken from randomised trials the benefit of control data is not being utilised as considerations are being made based on single interventions without reference to the comparator arm

b: Non-randomised, non-comparative induction period of maintenance trial

c: No range given for median time

d: Non-comparative induction period of maintenance trial

e: Mean is inappropriate for time-to-event data

f: No SD given for mean

g: Non-comparative data from RCT

h: Results interpreted from graphical representation of data

i: Unclear allocation concealment, unblinded and unclear dropout rate

j: Data based only on those who did not require dose escalation (24% of 1.25 mg/kg group and 62% of 2.5 mg/kg group)

Quality assessment							Summary of findings
Quality assessment							No of patients	Effect	Quality
No of studies	Design	Limitations	Inconsistency	Indirectness	Imprecision	Other considerations	Methotrexate	Effect	Quality
Mean time to maximum response (% change in PASI); MTX (follow-up 16 weeks; better indicated by lower values)
1 Saurat 2008	observational studies^a	no serious limitations^b	no serious inconsistency	no serious indirectness	very serious^c	none	110	Maximal response not achieved during 16 week trial	⊕○○○ VERY LOW
Mean time to maximum response (mean PASI); MTX (follow-up 24 weeks; better indicated by lower values)
1 Ho 2010	observational studies^a	no serious limitations^b	no serious inconsistency	no serious indirectness	very serious^c	none	20	Response beginning to plateau at 4–6 months based on mean PASI score over time,, but there is still a very gradual continued improvement over this period	⊕○○○ VERY LOW
Mean time to maximum response (mean PASI); MTX (follow-up 12 weeks; better indicated by lower values)
1 Flytstrom 2008	observational studies^a	no serious limitations^b	no serious inconsistency	no serious indirectness	very serious^c	none	37	Mean PASI score still decreasing at 12 weeks By 12 weeks the mean % improvement in PASI was 58%	⊕○○○ VERY LOW
Mean time to maximum response (mean PASI); MTX (follow-up 16 weeks; better indicated by lower values)
1 Heydendael 2003	observational studies^a	no serious limitations^b	no serious inconsistency	no serious indirectness	very serious^c	none	43	Maximal response based on PASI score appeared to be at 12 weeks By 16 weeks the mean % improvement in PASI was 64% Note: HR for time-to PASI75 0.61 (0.36 to 1.04) in favour of CSA; HR for time-to PASI90 = 1.15 in favour of MTX	⊕○○○ VERY LOW
Mean time to maximum response (mean PASI); MTX (follow-up 24 weeks; better indicated by lower values)
1 Gumusel 2011	observational studies^a	no serious limitations^b	no serious inconsistency	no serious indirectness	very serious^c	none	17	Maximal response based on PASI score appeared to be at 8 weeks	⊕○○○ VERY LOW

a: Although the data are taken from randomised trials the benefit of control data is not being utilised as considerations are being made based on single interventions without reference

b: Non-comparative data from RCT

c: Results interpreted from graphical representation of data

Quality assessment							Summary of findings
Quality assessment							No of patients	Effect	Quality
No of studies	Design	Limitations	Inconsistency	Indirectness	Imprecision	Other considerations	Acitretin	Effect	Quality
Mean time to maximum response (% improvement in BSA); acitretin (follow-up 24 weeks; better indicated by lower values)
1 Goldfarb 1988	observational studies^a	no serious limitations^b	no serious inconsistency	no serious indirectness	very serious^c	none	17	Improvement in global score and %BSA based on pooled data for all doses of acitretin was maximal at 20 weeks based on graphical presentation of change over time (0–24 weeks)	⊕○○○ VERY LOW
Mean time to maximum response (mean % improvement in PASI); acitretin (follow-up 24 weeks; better indicated by lower values)
1 Lassus 1987	randomised trials	serious^d	no serious inconsistency	serious^e	very serious^c	none	60 (20 in each group)	Mean % improvement in PASI score was still increasing at 2 months on 10, 25 and 50 mg/day acitretin	⊕○○○ VERY LOW
Mean time to maximum response (mean % improvement in PASI); acitretin (follow-up 24 weeks; better indicated by lower values)
1 Berbis 1989	randomised trials	serious^f	no serious inconsistency	serious^g	very serious^c	none	Increasing dose: 21 Constant dose: 19 Decreasing dose: 19	All dosing schedules: mean % change in PASI still increasing at 6 weeks Increasing dosing schedule: greater rate of % improvement in PASI still apparent at 6 weeks than the decreasing or constant dosing schedules (which were increasing more gradually)	⊕○○○ VERY LOW

a: Although the data are taken from randomised trials the benefit of control data is not being utilised as considerations are being made based on single interventions without reference

b: Non-comparative data from RCT

c: Results interpreted from graphical representation of data

d: Unclear allocation concealment and blinding not explained fully

e: Disease severity at baseline not reported and steroids administered on request (numbers using differed between the groups)

f: Unclear allocation concealment

g: Higher proportion of men in increasing dose group and more with pustular and guttate psoriasis in decreasing dose group

Table 142Absolute data on time to maximum effect or time to remission

Study	Total N	Follow-up	Intervention	Result	Notes
Time to remission
Shupack	181	16 weeks	CSA	Median time to 70% reduction in BSA: 8 weeks Median time to 90% reduction in BSA: 12 weeks	Non-randomised induction period of maintenance trial (CSA: 5 mg/kg)
Ho 1999	365	12 weeks	CSA	Median time to satisfactory clinical response (≥75% reduction in BSA): 9.7 weeks	Non-randomised induction period of maintenance trial (CSA: 2.5–5 mg/kg)
Ozawa 1999	37	To remission (maximum not stated)	CSA	Mean time to remission among responders (decrease in PASI of 80%): 15.4 weeks (4.7 months in continuous group and 3.0 months in intermittent group – but both received the same dose schedule during the induction period)	Induction period of maintenance trial (CSA: 3–5 mg/kg)
Time to maximum response (based on graphical representation)
Saurat 2008	158	16 weeks	MTX vs placebo	MTX maximal response not achieved during 16 week trial (curve for mean % improvement in PASI had reached 54.3% but still increasing gradually)	MTX: 7.5 mg increased to a maximum of 25 mg/wk as needed and tolerated Folic acid supplement
Ho 2010	36	24 weeks	MTX vs placebo	MTX response beginning to plateau at 4–6 months based on mean PASI score over time, but there is still a very gradual continued improvement over this period The mean % improvement in PASI had reached 73.9% by 6 months	MTX: 2.5–5.0 mg/wk to assess safety then 10 mg/wk up to 30 mg/wk Folic acid supplement
Flytstrom 2008	68	12 weeks	MTX vs CSA	Mean PASI scores for both MTX and CSA still decreasing gradually at 12 weeks CSA response appears to be more rapid, with greater improvement over the first 4 weeks By 12 weeks the mean % improvement in PASI was 58% in MTX group and 72% in CSA group	MTX: 7.5 mg/wk (3-divided dose) up to 15 mg/wk (plus folic acid) CSA: 3 mg/kg/d (divided into 2 doses) up to 5 mg/kg/d
Heydendael 2003	62	16 weeks	MTX vs CSA	Maximal response based on PASI score appeared to be at 12 weeks for both MTX and CSA, with the PASI score increasing slightly between 12 and 16 weeks By 16 weeks the mean % improvement in PASI was 64% in MTX group and 72% in CSA group	MTX: 15 mg/wk (3-divide dose) up to 22.5 mg/wk CSA: 3 mg/kg/d (divided into 2 doses) up to 5 mg/kg/d
Goldfarb 1988	37	24 weeks	Acitretin dosing	Improvement in global score and % BSA based on pooled data for all doses of acitretin were maximal at 20 weeks based on graphical presentation of change over time (0–24 weeks) The % BSA decreased from 35% to 13% by 24 weeks	10, 25, 50 or 75 mg/day acitretin (plus open phase)
Lassus 1987	80	8 weeks	Acitretin dosing	Mean % change in PASI score was still increasing at 2 months (based on graphical representation of % change in PASI) on 10, 25 and 50 mg/day acitretin	10, 25 or 50 mg/day acitretin (plus open phase) Patients using potent steroid concomitantly
Berbis 1989	58	6 weeks	Acitretin dosing schedule	The increasing dosing schedule of acitretin appeared to still be effecting a greater rate of % improvement in PASI at 6 weeks than the decreasing or constant dosing schedules, which were also still improving, although more gradually By 6 weeks the mean % improvement in PASI was approximately 55–65% across the three regimens	Acitretin: 10 up to 50 mg/day vs 50 down to 10 mg/day vs 30 mg/day
Christophers 1992	217	12 weeks	CSA	Mean % change in PASI was beginning to plateau at 8–12 weeks (and the response was approaching PASI75 at higher dose by this time point)	Doses initially 1.25 or 2.5 mg/kg (data based on those who did not require dose escalation: 24% of 1.25 mg/kg group and 62% of 2.5 mg/kg group)
Gumusel 2011	34	24 weeks	MTX vs CSA	Mean PASI scores reached maximum response for MTX at 8 weeks and CSA at 16 weeks	MTX: 15 mg/wk (single dose) for first 3 months then 10 mg/wk (single dose) for second 3 months (plus folic acid) CSA: 5 mg/kg/d (divided into 2 doses) for first 3 months then 2.5–3.2 mg/kg/d for second 3 months

Table 143Methotrexate versus ciclosporin versus best supportive care – economic study characteristics

Study	Limitations	Applicability	Other comments
Opmeer 2004²⁹⁴	Potentially serious limitations^(a)	Partially applicable^(b)	Cost-minimisation analysis of an RCT (Heydendael 2003¹⁴⁷) Patients with moderate to severe psoriasis Time horizon: 12 wks treatment; 36 wks follow-up Comparators: methotrexate and ciclosporin Costs: Direct medical costs (medication, diagnostic procedures, laboratory tests, visits to healthcare providers, therapies used during follow-up)
Sizto 2009³⁷³	Potentially serious limitations^(c)	Directly applicable^(d)	Decision analytic model Patients with moderate to severe psoriasis Treatment effects: probabilities of PASI 50, 75 and 90 estimated through systematic review and network meta-analysis of RCTs²³ Time horizon: not stated Comparators: methotrexate and ciclosporin and best supportive care^(e) Costs: Drugs and monitoring (excludes cost of dermatology and GP visits)
Woolacott 2006⁴²⁷	Potentially serious limitations^(f)	Directly applicable^(g)	Decision analytic model Patients with moderate to severe psoriasis Treatment effects: probabilities of PASI 50, 75 and 90 estimated through systematic review and network meta-analysis of RCTs (by the same authors) Time horizon: up to 10 years Comparators: methotrexate and ciclosporin and best supportive care^(h) Costs: Drugs, monitoring, outpatient visits, inpatient visits

a: Short time horizon (1 year); assumption informing treatment effects based on single RCT, not entire evidence base; relatively old cost estimates (1999/2000); no sensitivity analysis reported

b: Costing perspective is Dutch society: some uncertainty about applicability of Dutch estimates of resource use and unit costs; cost-minimisation method

c: Time horizon not stated; systematic review and network meta-analysis does not include all recent and relevant studies of ciclosporin and methotrexate; estimates of long-term effectiveness/withdrawal of treatments not stated; excludes important costs of outpatient dermatology and GP visits; funded by Abbott laboratories (makers of Adalimumab – biological therapy included in the analysis)

d: No discounting rates reported for costs or effects

e: Best supportive care not defined explicitly, but cost £117 per year.

f: Analysis was mainly focused on evaluation of etanercept and efalizumab – ciclosporin and methotrexate were evaluated as part of one probabilistic scenario analysis; systematic review and network meta-analysis does not include all recent and relevant studies of ciclosporin and methotrexate; cost of ciclosporin has decreased by one-third since analysis was undertaken

g: Discounting rates were 6% for costs and 1.5% for benefits instead of 3.5% for both

h: Best supportive care defined as two outpatient visits per year, an annual cost of £113.

Table 144Methotrexate versus best supportive care – economic summary of findings

Study	Incremental cost	Incremental effects	ICER	Uncertainty
Sizto 2009	£3,844 ^(a)	−129 QALYs ^(b)	Dominates	Costs 95% CI: −5049 to −2722 QALYs 95% CI: 0.078 to 0.185 Visual inspection of 95% confidence interval ellipses indicates that methotrexate dominates best supportive care in 100% of simulations
Woolacott 2006	− £4,223^(c)	0.126 QALYs	Dominates	Cost 95% CI: −4604 to −3224 QALYs 95% CI: 0.072 to 0.182 At thresholds of £20K and £30K per QALY, methotrexate has a 100% probability of being more cost-effective than best supportive care.

a: 2005/06 UK Pounds; does not include costs of outpatient of GP visits

b: Time horizon not reported

c: 2004/05 UK Pounds

Table 145Ciclosporin versus best supportive care – economic summary of findings

Study	Incremental cost	Incremental effects	ICER	Uncertainty
Sizto 2009	− £1987 ^(a)	0.079 QALYs ^(b)	Dominates	Costs 95% CI: −3313 to −597 QALYs 95% CI: 0.044 to 0.116 Visual inspection of 95% confidence interval ellipses indicates that ciclosporin dominates best supportive care in 100% of simulations
Woolacott 2006	− £452 ^(c)	0.122 QALYs	Dominates	Cost 95% CI: −795 to 41 QALYs 95% CI: 0.072 to 0.175 Probability of being more cost-effective than best supportive care could not be determined from the study report.

a: 2005/06 UK Pounds; does not include costs of outpatient of GP visits

b: Time horizon not reported

c: 2004/05 UK Pounds

Table 146Ciclosporin versus methotrexate – economic summary of findings

Study	Incremental cost	Incremental effects	ICER	Uncertainty
Opmeer 2004	£1,013 ^(d)	Assumed same ^(e)		Visual inspection of box and whisker plot indicate that costs accrued during treatment were significantly different between strategies, but this did not hold during 36 weeks follow-up
Sizto 2009	£1,857 ^(f)	−0.05 QALYs ^(g)	Dominated	Costs 95% CI: 1736 to 2125 QALYs 95% CI: −0.034 to −0.069 95% CI ellipses overlap, but visual inspection indicates that methotrexate dominates ciclosporin in approximately 80% of simulations
Woolacott 2006	£3,771 ^(h)	−0.004 QALYs	Dominated	Cost 95% CI: 3265 to 3809 QALYs 95% CI: 0 to −0.007 At a threshold of £20K per QALY, methotrexate has a 100% probability of being more cost-effective than ciclosporin; at £30K per QALY, this probability is 99%

d: Converted from 1999 Dutch Euros

e: Cost minimisation approach assumes the clinical outcomes are the same for both strategies

f: 2005/06 UK Pounds; does not include costs of outpatient of GP visits

g: Time horizon not reported

h: 2004/05 UK Pounds

Recommendations on systemic therapy	Systemic therapy General recommendations 74. Responsibility for use of systemic therapy should be in specialist settings only. Certain aspects of supervision and monitoring may be delegated to other healthcare professionals and completed in non-specialist settings, in which case, such arrangements should be formalised. 75. When offering systemic therapy, tailor the choice of agent and dosing schedule to the needs of the individual and include consideration of: the person’s age disease phenotype, pattern of activity and previous treatment history disease severity and impact the presence of psoriatic arthritis (in consultation with a rheumatologist) conception plans comorbidities the person’s views. 76. Be aware of the benefits of, contraindications to and adverse effects associated with systemic treatments. Explain the risks and benefits to people undergoing this treatment (and their families or carers where appropriate), using absolute risks and natural frequencies when possible^rrr. Support and advice should be provided by healthcare professionals who are trained and competent in the use of systemic therapies. 77. When reviewing response to systemic therapy, take into account: disease severity compared with baseline (for example, PASI baseline to endpoint score) control of psoriatic arthritis disease activity (in consultation with a rheumatologist if necessary) the impact of the disease on the person’s physical, psychological and social wellbeing the benefits versus the risks of continued treatment the views of the person undergoing treatment (and their family or carers where appropriate). 78. Monitor people using systemic treatment for all types of psoriasis in accordance with national and local drug guidelines and policy. Take appropriate action in the event of laboratory abnormalities or adverse events. 79. Offer adjunctive topical therapy to people with psoriasis using systemic therapy to optimise treatment outcomes. 80. Offer people with psoriasis who are starting treatment with a systemic non-biological or biological drug the opportunity to participate in long-term safety registries (for example the British Association of Dermatologists Biologic Interventions Register). Systemic non-biological therapy 81. Offer systemic non-biological therapy to people with any type of psoriasis if: it cannot be controlled with topical therapy and it has a significant impact on physical, psychological or social wellbeing and one or more of the following apply: - psoriasis is extensive (for example, more than 10% of body surface area affected or a Psoriasis Area and Severity Index (PASI)^sss score of more than 10) or - psoriasis is localised and associated with significant functional impairment and/or high levels of distress (for example severe nail disease or involvement at high-impact sites) or - phototherapy has been ineffective, cannot be used or has resulted in rapid relapse (rapid relapse is defined as greater than 50% of baseline disease severity within 3 months). Choice of drugs 82. Offer methotrexate^ttt as the first choice of systemic agent for people with psoriasis who fulfil the criteria for systemic therapy (see recommendation 81) except in the circumstances described in recommendations 84 and 92. 83. In people with both active psoriatic arthritis and any type of psoriasis that fulfils the criteria for systemic therapy (see recommendation 81) consider the choice of systemic agent in consultation with a rheumatologist. 84. Offer ciclosporin^uuu as the first choice of systemic agent for people who fulfil the criteria for systemic therapy (see recommendation 81) and who: need rapid or short-term disease control (for example a psoriasis flare) or have palmoplantar pustulosis or are considering conception (both men and women) and systemic therapy cannot be avoided. 85. Consider changing from methotrexate to ciclosporin (or vice-versa) when response to the first-choice systemic treatment is inadequate. 86. Consider acitretin for adults, and in exceptional cases only for children and young people, in the following circumstances: if methotrexate and ciclosporin are not appropriate or have failed or for people with pustular forms of psoriasis.
	Drug regimens 87. Use incremental dosing of methotrexate (for example, starting with an initial dose of 5–10 mg once a week) and gradually increase up to an effective dose and a maximum of 25 mg a week. Assess the treatment response after 3 months at the target dose of methotrexate and stop treatment if the response is inadequate (for example, a decrease of less than 75% in PASI score or a decrease of less than 50% in PASI score and 5 points in DLQI score). 88. Use the lowest possible therapeutic dose of methotrexate to maintain remission. 89. Use 2.5–3 mg/kg a day of ciclosporin^uuu. Escalate to 5 mg/kg a day after 4 weeks only when there is no response to the lower dose or when rapid disease control is necessary (for example in severe unstable disease). Assess the treatment response after 3 months at the optimum dose of ciclosporin and stop treatment if the response is inadequate (for example, less than a 75% decrease in PASI score or less than a 50% decrease in PASI score and less than 5 points in DLQI score). 90. Use the lowest possible therapeutic dose of ciclosporin to maintain remission for up to 1 year. Consider other treatment options when disease relapses rapidly on stopping ciclosporin therapy (rapid relapse is defined as greater than 50% of baseline disease severity within 3 months of stopping treatment). Do not use ciclosporin continuously for more than 1 year unless disease is severe or unstable and other treatment options, including systemic biological therapy, cannot be used. 91. Use incremental dosing of acitretin to minimise mucocutaneous side effects and achieve a target dose of 25 mg daily in adults. Consider dose escalation to a maximum of 50 mg daily when no other treatment options are available. Assess the treatment response after 4 months at the optimum dose of acitretin and stop treatment if the response is inadequate, for example: in plaque-type psoriasis, less than a 75% decrease in PASI score or less than a 50% decrease in PASI score and less than 5 points in DLQI score in pustular forms of psoriasis, not achieving clear or nearly clear on the static Physician’s Global Assessment.
Recommendations on methotrexate and risk/monitoring for hepatotoxicity	See sections 11.4 and 12.4.
Future research recommendations	18. In people with psoriasis, are there any clinical (for example, demographic or phenotypic) or laboratory (for example genetic or immune markers) that identify people who will respond to treatment with, or who will remain in remission following, treatment with methotrexate or ciclosporin? 19. In people with psoriasis, including pustular forms, what is the efficacy, optimal dosing, safety and cost-effectiveness of systemic non-biological agents for maintenance therapy (moderate to long-term outcomes are important)? 20. What is the most effective, safe and cost effective methotrexate dosing regimen to treat psoriasis and what is the role of folic acid in reducing efficacy or improving safety of methotrexate? 21. In children with psoriasis, what are the clinical effectiveness, safety, tolerability and cost effectiveness of methotrexate, ciclosporin and acitretin? 22. In people with palmoplantar pustulosis, what are the clinical effectiveness, safety, tolerability and cost effectiveness of acitretin and methotrexate? 23. In people with psoriasis, does early intervention with systemic treatments improve the long-term prognosis of psoriasis severity, comorbidities (including psoriatic arthritis), or treatment-related adverse effects, and are there any clinical (for example demographic or phenotypic) or laboratory (for example genetic or immune) biomarkers that can be used to identify those most likely to benefit from this treatment approach?
Relative values of different outcomes	The GDG agreed to prioritise the following outcomes when considering the evidence: PASI75 (or clear/nearly clear) Time to relapse Time to remission Serious adverse events Withdrawal due to toxicity Of the outcomes listed as priorities, the GDG were particularly interested in data from long-term studies. When considering the evidence, the GDG chose outcome measures that reflect impact on quality of life (as indicated by a change in the Dermatology Life Quality Index (DLQI), and objective assessments of skin involvement, namely the ‘physician’s global evaluation’ of clear/nearly clear, and various measures derived from the Psoriasis Area and Severity Index (PASI) including final PASI and improvement in the PASI as reported by PASI 75 and PASI 50 (i.e. 75% and 50% improvement from baseline respectively). Achievement of a PASI 75 and clear/nearly clear is an accepted ‘gold standard’ indicator of clinical effectiveness and tends to be reported in trials. PASI 50 is related to PASI 75, but has been specifically included as an indication of the minimum level of efficacy required to continue with therapy. These efficacy outcomes are also consistent with the NICE defined treatment response criteria for biological therapy where therapy can be continued only in those who achieve either a PASI 75, or PASI 50 and a fall in the DLQI of 5 points. The GDG looked for evidence of efficacy in both the short term (12–16 weeks, induction of remission) as well as in the longer term, and relapse rates following cessation of treatment. Clearly the toxicity and tolerability of systemic treatments are major considerations in relation to drug choice, and the adverse effects of each of the interventions are detailed in the relevant drug -specific Summary of Product Characteristics (SPC). However, the comparative toxicities of the different drugs are important given there may need to be a trade off between effectiveness and side effects. The GDG therefore looked for evidence of general drug toxicity (drug withdrawal and development of severe adverse effects) and the drug-specific side effects related to each of the interventions reviewed.
Trade off between clinical benefits and harms	In relation to trade offs between benefits and harms the GDG considered both stable and unstable disease, induction of remission and maintenance of remission together with efficacy differences between drugs, long term maintenance compared to intermittent dosing, concomitant drug use, side effects, adverse events and dosing. Induction of remission is clearly important, but for most patients with stable chronic plaque psoriasis, given the long term nature of the condition and the negative impact on wellbeing, maintenance of remission is of greater importance. Long term safety is also, for the same reason, very important as systemic therapies are likely to be required over many years since none of the interventions to date have been shown to be disease modifying. The GDG agreed that the expected benefits and risks of therapy should be clearly communicated to patients and monitoring arrangements are imperative to achieve optimal outcomes and minimise risk to patients. The GDG noted that national policy documents are available or in development from the British Association of Dermatologists for acitretin and methotrexate. The GDG were aware that there is variation in practice around dosing of systemic therapies, and in some instances, under-dosing may contribute to poor outcomes. Also, benefit and harm tend to be dose related and so the GDG agreed specification of dose would be helpful and reviewed the evidence with this in mind. When considering induction of remission, the ciclosporin 5mg/kg/day dose is more effective than the 2.5–3mg/kg/day dose, but is associated with greater clinically significant toxicity and drug withdrawal. The GDG agreed that dose escalation should be recommended only when a lower dose had failed or when rapid achievement of disease control necessary (such as severe/unstable disease). Studies on longer term ‘maintenance’ regimens were only available for ciclosporin. Low dose (1.5mg/kg/day) or intermittent (twice or three times weekly dosing) showed no clinically relevant benefit in terms of disease control or toxicity, compared to placebo. Disease control was better using continuous therapy compared to intermittent ‘courses’ of ciclosporin for up to a year; there was no difference in toxicity although clinically relevant nephrotoxicity (i.e. >30% rise in creatinine from baseline) and new onset hypertension occurred in over 27% and 12% of all patients treated, respectively by 1 year. Two studies addressed intermittent (taper to withdrawal) versus continuous long term use of ciclosporin for up to 4 years but around a third of participants in each arm dropped out, so the evidence is limited by high risk of attrition bias. The GDG noted from their clinical experience that abrupt cessation of ciclosporin can cause rebound flare that may be worse than baseline disease severity although no evidence for this was found in the studies. The GDG agreed that use of continuous ciclosporin is clinically appropriate based on good efficacy, and limited toxicity, up to one year. By 18 months of continuous therapy, unacceptably high rates of nephrotoxicity occur. The GDG agreed therefore that continuous treatment for longer than one year could not be routinely recommended except for patients who cannot use any other treatment option and have severe or unstable disease. For most patients treatment with ciclosporin should be discontinued at or around one year, with repeat courses possible in the event of relapse. It was also noted that in patients who relapse rapidly, alternative treatment options should be considered given the chronicity of psoriasis and the evidence that showed that with repeated courses of ciclosporin, time to develop clinically relevant elevations of creatinine became shorter with each course. For methotrexate, efficacy outcomes across the different studies were variable; pooled analysis indicates that methotrexate is as effective as ciclosporin by 12–16 weeks (PASI 75) although in two studies [Flytstrom, Ho] where a low initiating dose (2.5mg–7.5mg) and folic acid were used, methotrexate appeared to be less effective than ciclosporin. Risk of abnormal liver function tests and discontinuation of therapy were highest in studies when the starting dose of 15mg or greater per week (without folic acid) compared to incremental dosing from lower doses (2.5 to 10mg depending on the study). The side effect profile of methotrexate and ciclosporin differed, but there was no clinically significant difference between the two interventions with respect to overall drug withdrawal rates, serious adverse effects or relapse rates. The GDG noted from one study¹⁰⁴ that the improvement in DLQI was more rapid with ciclosporin than methotrexate, but by the end of the 12-week trial the DLQI scores were similar in both groups. The GDG considered that for patients with stable disease requiring long term disease management and/or where there was associated psoriatic arthritis, methotrexate should be used first line based on its efficacy and safety in the short term, low cost, and the known toxicity profile of ciclosporin in the longer term. The GDG considered the evidence around dosing regimens for methotrexate insufficient to make any changes to current practice (incremental dosing, concomitant folic acid), and agreed that any benefit to starting at a therapeutic dose of methotrexate in terms of reduced time to treatment effect was outweighed by the possibility of increased risk of liver dysfunction even though this trend may have been confounded by lack of folic acid co-therapy. The GDG noted that there was variation in practice in relation to the dose and frequency of folic acid supplementation but that this was beyond the scope of the guideline and should be used in accordance with guidance in the BNF. Data on acitretin indicated dose-related mucocutaneous toxicity occurring in the majority of people treated; efficacy appeared to be similar across all doses (25mg, 50mg, 75mg). In addition, it was noted that acitretin is teratogenic, and needs to be discontinued for 3 years before conception in women. The GDG agreed that the clinical utility of acitretin was limited due to the uncertainty about clinical efficacy, poor tolerability and, in view of data on elevated risk of cardiovascular disease in psoriasis, associated hyperlipidaemia. In the absence of evidence, the clinical experience of the GDG noted that it may be helpful in a subset of patients, particularly hyperkeratotic forms of localised hand and foot psoriasis and pustular forms of psoriasis. The GDG agreed therefore that acitretin should be retained as a treatment option given the paucity of treatments available. Data on the time to maximum effect for ciclosporin indicated that across different dosages no/little further response was seen after 12 weeks, therefore, the GDG decided to use this as the time to assess response and stop treatment if the treatment is not effective. Additionally, the graphical data demonstrated that the average reduction in PASI was 50% by 4 weeks, therefore the GDG recommended that this time point should be used to assess initial response to determine whether dose escalation is required. For methotrexate, the data suggested that maximum effect was seen after 16 to 24 weeks of treatment, although the higher initial dose of 15 mg/kg in one study (Heydendael) appeared to achieve maximal response after 12 weeks of treatment. The GDG reviewed the graphical data and discussed that the time to maximum response depends on the dosing schedule and is most dependent upon the duration of treatment at the target dose when incremental dosing is used. Therefore, the GDG agreed that review of response to determine whether methotrexate should be discontinued should be performed following 3 months treatment at the target dose. The dosing regimen for methotrexate varied across the studies reviewed; it was noted that studies with a lower starting dose (e.g. 5mg, 7.5mg, 10mg) had a lower incidence of abnormal liver function tests than did those studies where target dose was used (e.g. 15mg). However, it was also noted that use of folic acid may confound these data (i.e. studies employing the incremental dosing regimen also used co therapy with folic acid which may ameliorate liver toxicity). The GDG were also aware of other RCT data (using biological therapy as the comparator) using incremental dosing where the incidence of liver function tests and other abnormalities was comparable in the two arms of the study. Bone marrow suppression is dose related. Most of the studies reviewed used a target dose of between 15 and 20mg. The GDG therefore felt that given incremental dosing only delayed reaching the target dose by 2–3 weeks, that methotrexate is usually initiated with a view to long term disease control, and that it may be associated with improved safety profile, it was justifiable to recommend use of incremental dosing, with the recommended target dose (25 mg weekly) based on the RCT efficacy data reviewed. The GDG noted that the target dose in the studies reviewed ranged between 15 and 30mg weekly. The GDG agreed to recommend 25mg weekly as the target dose, given this was in line with expert opinion and the BNF. For acitretin the time-to-effect data suggested that maximum response rates were achieved ay 4 months. Although the data were limited this was in line with the clinical experience of relevant members of the GDG and so the GDG decided to use this as the time to assess response and stop treatment if the treatment is not effective. As acitretin may be useful for pustular forms of psoriasis it was felt to be important to include a relevant assessment tool for this population in the recommendation.
Economic considerations	No economic evidence was available to compare the cost-effectiveness of all systemic non-biological therapies – acitretin, ciclosporin and methotrexate. Two cost-utility analyses suggest that in a population with moderate to severe psoriasis, both methotrexate and ciclosporin are cost saving compared to best supportive care or no treatment. These two analyses plus a cost-minimisation analysis also indicate that methotrexate is cost saving compared to ciclosporin. Although each analysis had potentially serious limitations, largely due to a broadening evidence base since they were originally undertaken, the GDG believed that the conclusions arising from these analyses were still likely to be true. A simple cost analysis considering only drug acquisition costs was performed and it showed that methotrexate is around 95% to 98.5% less costly than ciclosporin. It also showed that depending on dose per kilogram and patient weight, one week of treatment with ciclosporin could cost as much as 6 months’ to a year’s worth of methotrexate. Given this substantial cost difference, the uncertainty as to which is more effective (conflicting evidence) and the fact that ciclosporin has potentially serious complications if used in the long term, the GDG considered methotrexate to represent the best value for NHS resource in the population of patients for whom it is a reasonable treatment option (i.e. patients potentially requiring long term treatment and without contraindications to methotrexate). They also considered methotrexate likely to be the optimal systemic non-biological therapy in the treatment of psoriasis patients with concomitant psoriatic arthritis. For patients who cannot take methotrexate or for whom rapid control of psoriasis is the primary goal, the GDG considered short term treatment with ciclosporin to represent an efficient use of NHS resources. There was no economic evidence for the use of systemic non-biological therapies in the treatment of palmoplantar pustulosis, but the clinical evidence suggest that it is more effective than placebo/no treatment, although it carries an increased risk of hypertension. Given the ciclosporin was found to be cost saving compared to best supportive care in moderate to severe plaque psoriasis, the GDG considered it likely to be cost-effective in the treatment of palmoplantar pustulosis. That is, they believed that any additional costs for ciclosporin treatment in this group are likely to be justified by its additional benefits compared to no treatment. There was no economic evidence to inform the GDG of the cost-effectiveness of acitretin. Based on the clinical evidence and their clinical experience, they judged acitretin unlikely to be more cost-effective than either methotrexate or ciclosporin. Therefore, they decided it should be reserved only for patients for whom neither of these other systemic non-biological agents were suitable.
Quality of evidence	The GDG noted that most of the studies addressed treatment of plaque psoriasis, but the baseline disease severity in these studies represented the population likely to be offered treatment in the UK. . Only one study addressed treatment of palmoplantar pustulosis (using ciclosporin) and one nail psoriasis (Gumusel) and this was small and of low quality owing to an inadequate method of randomisation. There were no studies in children. Research recommendations were formulated for these important groups. In the absence of paediatric evidence and given the importance of adequate treatment for this high need group, the GDG agreed that the recommendations on use of systemic non-biological drugs could be extrapolated to children. The GDG were aware that drug metabolism may be different in children but that paediatric dermatologists would take this into account when prescribing. The GDG had no reason to believe that there would be any biological implausibility for differences in drug metabolism or that the harm profile was likely to be different in the paediatric population. In relation to offering paediatric systemic therapy the GDG noted that non-biologic systemic therapies are currently not licensed for the treatment of psoriasis in children of less than 16 years. Ultimately the prescriber must take responsibility for using drugs outside of their licensed indications. When managing psoriasis in children and young people, treatment choice should be carefully considered to avoid or minimise long-term sequelae. The available data were mostly short term (up to 16 weeks), and related to induction of remission. Trials on ‘maintenance’ regimens were limited to ciclosporin with no data on methotrexate or acitretin. This lack of data constitutes a major gap in evidence given that psoriasis is a chronic disease. The GDG were aware of long term registries that aim to address this shortfall in data and agreed that clinicians should talk to patients about contributing data to these registries and encourage participation whenever feasible. Further research is warranted to evaluate efficacy, optimised dosing and safety of systemic non-biological agents in psoriasis including pustular forms for both induction and maintenance. The following points were noted in relation to the studies assessing methotrexate: There was marked heterogeneity in findings for all methotrexate-related efficacy outcome measures between the studies, mostly accounted for by differences in dosing regimens. Additional confounders include variation in duration of treatment and concomitant use of folic acid. The Sandhu study (low quality) demonstrated the highest efficacy for methotrexate across all the studies and used an initial dose of 0.5mg/kg. The Heydendael study analysed the final PASI score by ANCOVA to take account of baseline differences. The quality of evidence for ciclosporin vs. placebo for maintenance was lowered by the following factors: Variation between the studies in dosing schedules and definitions of relapse The Ozawa and Ohtsuki studies, while having a long (i.e.48 months) follow-up period for maintenance regimens on ciclosporin, also had very high drop-out rates so are likely to represent an underestimate of toxicity rates. The Shupack and Ellis studies used a ciclosporin dose of 3mg/kg for maintenance. This dose is an induction dose. The mean time to relapse reported in the Ellis 1995 study for the 3 mg/kg/day ciclosporin group was likely to be an underestimation because follow-up was restricted to a maximum of 4 months by the protocol but most had not relapsed at this time point. Therefore, the true mean time to relapse is likely to be longer. There was a high (30%) drop-out rate in the Colombo study and a per protocol analysis was reported for efficacy and relapse outcomes owing to the high drop-out rate (largely due to sun exposure and unwillingness to continue when improvement in the psoriasis was seen). The RCT investigating initial dosing of ciclosporin (1.25mg or 2.5mg) [Christophers et al] allowed dose escalation in non responders and did not maintain randomisation. The studies for acitretin vs. placebo were small and of low quality, and included a mixture of psoriasis phenotypes including chronic plaque, guttate and pustular forms. The Lassus study, which compared different dosing regimens of acitretin, allowed concomitant use of potent steroid in all 3 trial arms so the efficacy of acitretin alone is unclear.
Other considerations	The GDG agreed that the initiation and monitoring of systemic therapy required specialist supervision to ensure optimal outcomes, and that it was therefore important to define which groups of people with psoriasis should be offered this treatment to ensure rapid, and appropriate referral. The GDG agreed that systemic therapy should be reserved for people where topical therapy cannot be used (for example, the area affected is extensive), or where it is likely to be ineffective (for example nail disease which may lead to functional impairment) or where control cannot be achieved (for example rapid relapse or failure to clear with potent corticosteroids), or for forms of psoriasis that, although limited in extent, have a significant adverse impact on quality of life/wellbeing. It was noted also that systemic therapy can be associated with potentially very serious adverse effects in the short term whilst the longer term adverse effect profile is largely unknown, that there is a significant burden on both the person affected in terms of requirement to take regular systemic therapy and have appropriate monitoring, and also health service providers and that to date there is no evidence that treating psoriasis per se has any impact on overall disease prognosis or on associated morbidities. The GDG therefore agreed that use of systemic therapy could not be justified unless the psoriasis was having an important impact on the individual’s quality of life. In addition the patient members of the GDG agreed that escalation of therapy from topical to systemic non-biological was common if the psoriasis was have a significant impact on their physical, psychological or social wellbeing. The GDG considered the clinical as well as statistical significance of the findings. The GDG agreed that: Older people are more likely to develop nephrotoxicity with ciclosporin. Conception plans should be taken into account when choosing which systemic non-biological therapy to use; for example ciclosporin may be relatively favoured over methotrexate in men or women of childbearing potential. Presence of psoriatic arthritis should be considered when treating psoriasis with systemic therapy. No evidence was available on systemic therapy in children; clinical expertise within the GDG noted that a higher dose of 5mg/kg/day of ciclosporin is needed to be effective in children. All the systemic non-biological interventions are of variable efficacy and may lead to clinically significant toxicity including rarely, life threatening events. The GDG noted that in some instances of poor response to oral methotrexate, a switch to subcutaneous administration may improve responses either due to improved adherence or bioavailability. However the GDG could not make a specific national recommendation in the absence of high quality evidence in psoriasis. Supplementary topical therapy is commonly required to achieve optimal control of psoriasis with systemic non-biological therapy. This clinical opinion is supported by the evidence, as most of the studies allowed at least emollients and mild or moderate potency steroids to be used, with potent steroids allowable in some. A recommendation to encourage use of concomitant therapy was felt to be important in order to optimise outcomes.

rrr: See Appendix S for details of the risk-benefit profiles of interventions recommended in this guideline.

sss: The PASI is also available from the British Association of Dermatologists website.

ttt: At the time of publication (October 2012), methotrexate did not have UK marketing authorisation for this indication in children and young people. The prescriber should follow relevant professional guidance, taking full responsibility for the decision. The patient (or their parent or carer) should provide informed consent, which should be documented. See the General Medical Council’s Good practice in prescribing medicines – guidance for doctors for further information.

uuu: At the time of publication (October 2012), ciclosporin did not have UK marketing authorisation for this indication in children and young people under 16 years of age. The prescriber should follow relevant professional guidance, taking full responsibility for the decision. The patient (or their parent or carer) should provide informed consent, which should be documented. See the General Medical Council’s Good practice in prescribing medicines – guidance for doctors for further information.