Secukinumab for treating moderate to severe plaque psoriasis

The company did a systematic review of the literature to identify studies evaluating the clinical effectiveness and safety of secukinumab for treating people with moderate to severe chronic plaque psoriasis. It identified 5 relevant international, multicentre, phase 3, double‑blind, randomised, controlled trials: 3 superiority trials compared secukinumab with placebo (ERASURE; JUNCTURE; FEATURE) and 1 compared secukinumab with both placebo and etanercept (FIXTURE). Another trial (SCULPTURE) was a non‑inferiority trial comparing 2 different dosing regimens of secukinumab: a regular dose regimen of secukinumab compared with re‑treatment with secukinumab only at relapse.

The company did not find any other head‑to‑head studies, and therefore did a network meta‑analysis to compare secukinumab with all 5 comparators identified in the scope (best supportive care, etanercept, ustekinumab, adalimumab and infliximab). The company did not find any relevant non‑randomised controlled or observational studies.

The 4 placebo‑controlled trials (FIXTURE, ERASURE, JUNCTURE and FEATURE) had similar designs. Patients were stratified by either body weight alone (90 kg or more or less than 90 kg), or geographical location and body weight, and randomised to secukinumab 300 mg or 150 mg, or placebo. Secukinumab was given at weeks 0, 1, 2, 3, and 4, and then 4‑weekly. The FIXTURE trial also included an etanercept comparator arm (50 mg twice a week initially, then 50 mg once per week). At week 12, patients in the placebo arms who did not have a Psoriasis Area Severity Index (PASI) of 75 (that is, had not had a 75% reduction in the absolute PASI score from baseline; the primary outcome) were 're‑randomised' to either 150 mg or 300 mg secukinumab (at weeks 12, 13, 14 and 15 from baseline, followed by the same dose every month from week 16). All trials had a placebo‑controlled period lasting 12 weeks followed by a 40‑week maintenance period. However, patients completing JUNCTURE and FEATURE were followed for a further 156 weeks.

Patients were eligible for inclusion in the trials if they had moderate to severe chronic plaque psoriasis not adequately controlled by topical treatment, phototherapy or previous systemic therapy. Severity was defined as: the percentage body surface area affected by psoriasis, absolute PASI score, and the Investigator Global Assessment score (modified in 2011, hereafter referred to as IGA). IGA is a 5‑point scale that measures psoriasis severity ranging from 0 (clear) to 4 (severe disease). To be eligible, patients needed to have 10% or more of their body surface area affected by psoriasis, a PASI score of 12 or more, or an IGA score of 3 or more. Patient characteristics were generally similar across the trials, with some differences. For example, FIXTURE and ERASURE had a lower proportion of patients who had received prior biological treatments (range: 10.7% to 29.4% across trials and trial arms) than JUNCTURE and FEATURE (21.3% to 47.5% across trials and trial arms); and FEATURE and ERASURE had a higher proportion of patients for whom a prior biological had failed. Across trials, previous systemic therapies (including methotrexate) ranged from 33.9% (FEATURE, secukinumab 300 mg arm) to 62.6% (FIXTURE, etanercept arm).

The company presented clinical trial results for secukinumab both at 150 mg and 300 mg. The marketing authorisation, however, is only for 300 mg. Therefore, this final appraisal determination presents only the results for the licensed dose (300 mg; hereafter referred to as secukinumab).

The co‑primary outcome measures in all 4 placebo‑controlled trials were measured at week 12: PASI 75 (that is, a 75% reduction from baseline in PASI score), and an IGA score of 0 or 1 (indicating clear or almost clear of disease). The PASI response was used in the model and the network meta‑analysis. Therefore, this final appraisal determination presents results only for PASI outcomes. The company analysed the data using intention‑to‑treat methods. The company reported odds ratios for FIXTURE and ERASURE, and 'risk differences' (the difference in proportions of patients in whom the outcome was reached) for JUNCTURE and FEATURE. In all 4 placebo‑controlled trials, there were statistically significant improvements with secukinumab in the co‑primary outcomes compared with placebo. For example, across trials, at week 12, 75.9% to 86.7% of patients randomised to secukinumab had a PASI 75 response, compared with a 0% to 4.9% (p<0.0001 all trials) of patients randomised to placebo. There were also statistically significant improvements with secukinumab compared with etanercept. PASI 75 response was 77.1% with secukinumab compared with 44% with etanercept (p<0.0001). The company also noted that response to secukinumab for these outcomes continued to increase between week 12 and week 16 in the FIXTURE and ERASURE trials.

Secondary outcomes in the placebo‑controlled trials included assessing PASI 75 at different time points (weeks 16 and 52), different PASI responses, for example, at week 12, PASI 50/90/100 responses, and maintenance of PASI 75 and health‑related quality of life. The effectiveness of secukinumab for these secondary outcomes was consistent with the results for 12‑week PASI 75 in that there were improvements with secukinumab compared with placebo across the 4 placebo‑controlled trials (statistical significance was achieved for some outcomes, but the company did not perform or present statistical analyses for all outcomes). For example, week‑12 PASI 100 (that is, complete clearance of the disease) ranged from 24% to 43% for secukinumab (across all trials), was 4.3% for etanercept (FIXTURE trial) and ranged from 0% to 0.8% for placebo. In the FIXTURE trial at week 52, 36.2% of patients had a PASI 100, which was higher than with etanercept (9.9%). The company had not predefined this as an outcome, so did not do statistical analyses.

The company provided the results of analyses of primary and secondary trial outcomes for pre‑specified subgroups that included sex, age, weight, geographic location, age at diagnosis, disease duration, quality of life and previous experience of other treatments (biological and non‑biological). The company provided these as 'academic in confidence' but noted these were consistent with the main results.

The company presented results on the effect of secukinumab on the dermatology life quality index (DLQI) score in all 4 placebo‑controlled trials. In all the trials, secukinumab improved (that is, reduced) DLQI score at week 12 from baseline by between 10.4 to 11.6 points, which was higher than with placebo (1.1 to 1.9 points; p<0.001 for all trials other than FIXTURE, in which no p value was given). The company stated that these improvements were maintained at week 52 in the FIXTURE and ERASURE trials. The number of people with a week 12 DLQI response of 0 or 1 (that is, showing no impact on daily living) was statistically significantly higher for secukinumab in all trials than with placebo (p<0.001) and etanercept (p<0.001).

The company presented evidence on the absolute changes from baseline in quality of life with secukinumab compared with placebo based on a EQ‑5D visual analogue scale of 0 (worst possible health state) to 100 (best possible health state), which were statistically significantly higher with secukinumab than with comparators in all 4 placebo‑controlled trials.

The company compared secukinumab with the other biological comparators (adalimumab, etanercept, infliximab and ustekinumab) using a random effects network meta‑analyses. The network meta‑analysis presented for each treatment the effectiveness reflected by PASI response from baseline (less than 50%, 50% to 74%, 75% to 90%, 90% or more). The company identified 30 relevant trials for the network meta‑analyses from its systematic literature review, including the 5 relevant secukinumab trials (as described in sections 3.1 to 3.10) and 25 trials for the comparator treatments. However, the company did not include all trials in all its analyses. Doses included in the base‑care analysis are described in table 1. The company excluded some studies or arms that had tested irrelevant doses or comparators. However, it included a 150 mg dose arm for secukinumab and a 100 mg arm for etanercept (16‑week scenario only) to connect the network.

The base‑case network meta‑analyses presented the PASI response after 'induction' assessed at 10 weeks (infliximab), 12 weeks (secukinumab, etanercept and ustekinumab) or 16 weeks (adalimumab; see table 2). The company also presented 2 scenario analyses: a 12‑week analysis (assessing effectiveness at 12 weeks for each treatment) and a 16‑week analysis (repeating the base case, but using 16‑week assessment for secukinumab only). In the base case, the network meta‑analysis showed that secukinumab 300 mg was statistically significantly more effective than placebo, secukinumab 150 mg, etanercept and adalimumab in achieving a 50%, 75% and 90% reduction in PASI. There were no statistically significant differences when comparing secukinumab 300 mg with ustekinumab or infliximab. Results for the scenario analyses were consistent with the base case. The company stated that similar results were seen with sensitivity analyses including the following populations: DLQI of more than 10 (other than adalimumab, for which comparisons were not possible); duration of psoriasis; baseline PASI score; and prior exposure to biological drugs. The company also reported that it showed similar results with a meta‑regression adjusting for prior biological exposure.

All trials captured adverse effects of treatment. The most commonly reported adverse effects for secukinumab were nasopharyngitis, headache, diarrhoea, upper respiratory tract infection, itching and fever. In FIXTURE, ERASURE, FEATURE and JUNCTURE, the proportion of patients who developed any adverse event with 300 mg secukinumab (50.8% to 70.0%) was higher than with placebo (47.0% to 54.1%). The FIXTURE study showed similar rates of any adverse events at week 52 for secukinumab and etanercept (252.0 and 243.4 cases per 100 patient years respectively). The secukinumab 300 mg arm showed no more increase in safety‑related events than the 150 mg arm.

The ERG stated that the systematic review done by the company was of good quality, and appeared to be complete because it included the 4 main clinical trials (FIXTURE, ERASURE, JUNCTURE, and FEATURE). The ERG considered the clinical trials to be of good quality, and the review of clinical evidence was generally well‑conducted, with an appropriate assessment of bias. The ERG stated that patients in the clinical trials were appropriately randomised and allocated to treatment, and baseline demographics and disease characteristics were balanced across intervention groups. The ERG stated that there was strong evidence that secukinumab 300 mg is superior to placebo for PASI efficacy outcomes at week 12.

The ERG stated that the description of the network meta‑analysis, the method used to conduct the network meta‑analysis, and the method used to evaluate consistency between direct and indirect evidence, were all generally appropriate. However, the ERG noted several issues, including that the company had meta‑analysed 1 outcome only, and it would also have been possible to present meta‑analyses for quality of life.

The ERG reviewed the adverse event information presented in the company submission. The clinical adviser to the ERG had noted that, because biological treatments suppress the immune system, this may increase rates of cancer. The clinical adviser also stated that there is no evidence for an increase in lymphoma with biological treatment for people with psoriasis, that phototherapy can increase the risk of both melanoma and non‑melanoma skin cancer, and that prolonged ciclosporin treatment has been associated with lymphoma.

The company developed a new economic model that combined a decision tree with a Markov state‑transition model to compare secukinumab 300 mg with etanercept, ustekinumab (45 mg and 90 mg), adalimumab, infliximab and best supportive care. The model applied to adults with moderate to severe plaque psoriasis (defined as absolute values of PASI 10 or more and DLQI more than 10) whose disease had inadequately responded to at least 2 standard systemic therapies. The model had a 10‑year time horizon (1‑year decision tree followed by 9‑year Markov model) with 3 key time periods: an induction phase (10 to 16 weeks depending on the pharmaceutical agent); a post‑induction to 1‑year phase; and an annual phase (9 years). Within each phase, patients could be in 1 of 4 health states based on the response from baseline in PASI (PASI less than 50%, 51% to 74%, 75% to 90%, more than 90%), plus a state reflecting death. The company assigned resource use, costs and quality‑adjusted life years (QALYs) for each of these health states. The company conducted the analysis from the perspective of the NHS and Personal Social Services, and discounted costs and health effects at an annual rate of 3.5%.

Patients entered the decision tree at the beginning of the induction period, the duration being determined by which drug they were taking, (that is, 12 weeks for secukinumab, ustekinumab and etanercept, and 10 and 16 weeks respectively for infliximab and adalimumab). Patients remained on treatment for the whole induction period. If, after the induction period, a patient's disease had not responded (that is, their PASI had not improved by 75% or more) the biological treatment was stopped and the patient moved on to best supportive care and into the PASI less than 50 health state. In the best supportive care state, treatments included systemic therapies (90% received either methotrexate or ciclosporin) and phototherapy (16%), and all patients were assumed to receive day centre care (psoriasis day‑case admission), based on the costing template for NICE's guideline on psoriasis. Patients whose disease responded to treatment during the induction phase (that is, patients who had a PASI 75 or above) continued on treatment for 1 year. At 1 year, the company assumed a discontinuation rate based on the FIXTURE and ERASURE trials of 11.7% for patients who stopped treatments with biological drugs and moved to the PASI less than 50 state and who received treatment with best supportive care. All other patients progressed into the annual Markov model, with a 20% annual all‑cause discontinuation rate (based on expert opinion) applied beyond 1 year. The company assumed that, after the induction period, patients did not change PASI health states; patients either remained in their PASI health state and accrued the costs and benefits of that health state, or transitioned to best supportive care (PASI less than 50 health state) or death. The company noted that, although people with severe psoriasis have a reduced life expectancy because of cardiovascular disease, lymphoma and non‑melanoma skin cancer, it did not model disease‑specific mortality rates. Instead, the company used age‑specific all‑cause mortality rates from the general population.

The company's base‑case results for the cost effectiveness of secukinumab (based on a model updated with a correction to utility values, which had a minor impact on cost‑effectiveness results) were presented as an incremental analysis of secukinumab and other biological treatments. The incremental cost‑effectiveness ratio (ICER) for secukinumab compared with etanercept was £2,515 per QALY gained (incremental costs £573, incremental QALYs 0.22) in the incremental analysis and secukinumab dominated all other biological treatments (adalimumab, ustekinumab 45 mg and 90 mg, and infliximab). The ICER for secukinumab compared with best supportive care was £7,231 per QALY gained (incremental costs £2,752, incremental QALYs 0.38).

The ERG stated that the company appeared to have found the relevant evidence for economic evaluations and data on health‑related quality of life. The ERG's expert stated that, although people can take etanercept intermittently, in clinical practice, any effective biological is likely to be used continuously, as modelled by the company in its base case. The ERG calculated that intermittent dosing of etanercept would be given at a frequency of 1.33 doses per week, compared with 2 doses per week used assuming continuous dosing.

The ERG noted that the health economic model was similar to models used in previous NICE appraisals for psoriasis. Notably, previous models had assumed that people with severe psoriasis that fails to respond to a biological are then given best supportive care. However, the ERG's clinical adviser stated that this was not realistic because, in clinical practice, people would switch to another biological treatment, or add treatments (for example, methotrexate or phototherapy). The ERG stated that a more appropriate model would take into account treatment sequencing.

The ERG queried why the company had applied a discontinuation rate at the end of year 1 because the model already assumed that people whose disease had not responded to treatment stopped treatment. The ERG's clinical expert noted that a rate of 15% to 20% per year was a reasonable estimate for the proportion of patients who stop treatment annually beyond the first year.

The ERG noted that the company did not model an increased risk of mortality for people with psoriasis and did not account for deaths during the first year or after people had stopped treatment. The ERG considered that this may have biased the model against the more effective treatments.

The ERG noted the company's assumptions that people remain in a given health state from the end of the induction period for the duration of the model. However, the ERG also noted that data from the FIXTURE trial showed that over time most people with either a PASI 50 to 74 or PASI 75 to 90 at week 12 either improved or worsened over time. The ERG noted that most people with a PASI 90 response at week 12 maintained their response to 52 weeks.

The ERG considered the company's approach to modelling quality of life, noting that modelled utility values depend on a patient's health state. However, in reality, a treatment in itself may influence a patient's health‑related quality of life. The ERG considered that this would bias the cost‑effectiveness estimates against secukinumab. The ERG considered that the EQ‑5D was unlikely to capture the disutility associated with adverse events because the company had stratified the EQ‑5D data by PASI response, rather than by treatment arm. The ERG also noted that the company had calculated QALYs accrued in the first year assuming that the amount a patient's disease had responded by week 12 was maintained for up to 1 year. The ERG stated it would be more appropriate for people whose disease responds partially with a PASI 50 to 74 to accrue the quality of life for PASI 50 to 74 for the induction period (weeks 0 to 12), and then the quality of life associated with PASI less than 50 for the remaining period of the first year because people whose disease partially responds stop treatment from week 12.

The ERG noted a number of issues with the calculation of the costs of best supportive care. The ERG noted that the Fonia et al. (2010) study (see section 3.24) showed that biologicals had a lower impact on the use of healthcare resources (in particular, hospital admissions) than the estimates used by the company derived from the costing template for NICE's guideline on psoriasis.

The ERG considered it optimistic that all people receiving biological treatments could self‑administer subcutaneously after only 1 hour of training. It considered the estimate from NICE's technology appraisal guidance on etanercept and efalizumab for the treatment of adults with psoriasis, which assumed 3 × 1‑hourly sessions of training for self‑administration to be a more reasonable estimate, and used this in the ERG's base case.

The ERG conducted an exploratory analysis of the company's base case (based on the company's original base case, before the minor corrections for utility values, see section 3.28). The ERG:

• removed the costs of 5 intravenous infusions that the company had incorrectly attributed to secukinumab

• included the costs of serious adverse events for patients taking biologicals for the first year, which the company had omitted from the model

• updated the number of doses of secukinumab and ustekinumab because the ERG interpreted the licensing information for dosing differently to the company:

  • for secukinumab, the ERG interpreted the dosing as 4‑weekly, which would be 13 doses annually, and not 12 as modelled by the company

  • for ustekinumab, it stated that the post‑induction dose would be 3 doses, and not 4 as modelled by the company

• corrected mortality calculations within the cohort because the company had originally assumed that patients who stop treatment do not die

• revised the QALY calculations for people whose disease partially responds and therefore stop treatment, by applying the PASI less than 50 quality‑of‑life value for the post‑induction period

• removed the hospitalisation cost for people with a PASI 75 response remaining on drug therapy in the best supportive care arm

• removed the costs of hospitalisation in the first year among people with a PASI 50 to 74 response from week 0 to week 12 and instead calculated the costs from week 12 to week 52 (to remove hospitalisation costs included in the induction period)

• revised the utility values to reflect those supplied by the company to the ERG during the clarification phase of the appraisal

• revised the time a nurse needs to teach a patient to inject subcutaneous biologicals from 1 hour to 3 hours

• revised the mean patient weight to 83.3 kg, which the ERG took from the FIXTURE trial.

The ERG presented 2 different base cases reflecting 2 alternative sources of costs for the best supportive care. Both base cases incorporated the ERG's corrections to the model:

• Base case A: This scenario (the ERG's preferred scenario) used assumptions related to best supportive care based on Fonia et al. (2010; an average increase of 5 inpatient days, an average increase of 3 phototherapy sessions and no increase in the average number of day centre care attendances when compared with biological treatments).

• Base case B: This scenario used assumptions related to best supportive care based on the costing template for NICE's guideline on psoriasis and hospital episode statistics data (an average increase of 10.7 inpatient days, an average increase of 3.84 phototherapy sessions and an average increase of 5 day centre care attendances when compared with biological treatments).

The ERG's preferred exploratory base case generated an ICER of £52,760 per QALY gained (incremental costs £20,087, incremental QALYs 0.38) for secukinumab compared with best supportive care. Secukinumab extendedly dominated etanercept, adalimumab and ustekinumab, and secukinumab dominated infliximab (an option is 'extendedly dominated' when its ICER is higher than that of the next, more effective, option when compared with a common baseline). The ICERs for secukinumab, when compared directly with etanercept, adalimumab, ustekinumab 45 mg and ustekinumab 90 mg, were £42,367, £38,684, £26,321 and £17,717 per QALY gained respectively. The ERG's exploratory base case B used the company's preferred source of data for best supportive care and generated an ICER of £14,902 per QALY gained (incremental costs £5,673, incremental QALYs 0.38) for secukinumab compared with best supportive care. Secukinumab extendedly dominated etanercept and adalimumab, dominated ustekinumab and infliximab, and had an ICER compared with etanercept and adalimumab of £8,899 and £6,979 per QALY gained respectively.

The ERG presented sensitivity analyses to explore the impact of:

• intermittent dosing of etanercept (see section 3.32)

• alternative discontinuation rates

• increasing the mortality risk associated with psoriasis of 20%

• making the costs of best supportive care consistent throughout the model by reducing the costs of year 1 and 2

• alternative quality‑of‑life estimates from previous technology appraisals and EQ‑5D models submitted by the company.

The ERG presented results of its sensitivity analyses for base cases A and B. The ERG noted that changing the dosing of etanercept to intermittent dosing worsened the cost effectiveness of secukinumab compared with etanercept from £42,368 per QALY gained to £59,268 per QALY gained (base case A), and from £8,899 per QALY gained to £25,800 per QALY gained (base case B). Using utility values from NICE's technology appraisal on infliximab for the treatment of adults with psoriasis decreased the ICERs for secukinumab to less than £27,000 per QALY gained when using utility values (using base case A). The ERG noted that varying the price and effectiveness of the biologicals influenced cost effectiveness more than other variables.

Full details of all the evidence are in the committee papers.