4 Evidence and interpretation
The Appraisal Committee (appendix A) considered evidence from a number of sources (see appendix B).
4.1 Clinical effectiveness
4.1.1 Etanercept
4.1.1.1 The Assessment Report included three double-blind RCTs that evaluated etanercept at a dose of 25 mg or 50 mg twice a week. In total, 1347 patients were enrolled in the RCTs. The Assessment Group deemed all three of the RCTs to be of 'good' quality. At the end of each trial there was an open-label and/or non-randomised extension period.
4.1.1.2 All three trials reported statistically significant differences in favour of etanercept 25 mg over placebo as measured by PASI 50, PASI 75 and PASI 90 at 12 weeks. There was no statistically significant heterogeneity in these results and therefore the results of the trials were pooled. The pooled relative risks were PASI 50 RR 5.41 (95% CI 4.10 to 7.14), PASI 75 RR 10.69 (95% CI 6.15 to 18.57) and PASI 90 RR 18.55 (95% CI 5.24 to 65.69). The trial that continued to 24 weeks also showed a statistically significant difference in favour of etanercept for PASI outcomes. Etanercept 50 mg twice a week also showed statistically significant benefits over placebo in both of the RCTs. The pooled relative risks were PASI 50 RR 6.48 (95% CI 4.84 to 8.66), PASI 75 RR 14.80 (95% CI 8.40 to 26.06) and PASI 90 RR 38.12 (95% CI 9.43 to 154.06). The number of participants who were rated by the physician as being 'clear or almost clear' of psoriasis showed a statistically significant superiority of etanercept at both doses and time-points. The percentage change in PASI and DLQI echoed this, but data were highly skewed and therefore not pooled. At the end of treatment in all three trials there were statistically significant differences in favour of etanercept in the patient measures of disease status.
4.1.1.3 In one of the RCTs, patients on etanercept continued taking the drug after the 12-week placebo-controlled period for a further 24 weeks (total treatment period of 36 weeks). The Assessment Group concluded that the results at 36 weeks did not indicate any lessening of response over time.
4.1.1.4 The manufacturer also provided the results of the follow-up after discontinuation for two of the RCTs. Although insufficient data were presented for the smaller trial (only 18% of patients were followed), 3-month post-treatment follow-up data were presented for 409 patients who had responded to etanercept treatment in the larger study. Only one of these 409 participants experienced a rebound of their psoriasis to more than 125% of their baseline PASI score. The larger RCT also provided data on patients who were re-treated. Across all doses, the mean difference in PASI score after 12 weeks re-treatment compared with the initial 12-week treatment was –0.5 (95% CI –1.09 to 0.09), indicating that there was no statistically significant difference.
4.1.1.5 Post hoc analyses of one of the RCTs examined the efficacy of etanercept in patients who had received prior systemic therapy or phototherapy for psoriasis (89% of patients). Of the 521 patients who had received prior systemic therapy or phototherapy, 337 (65%) had failed to respond to at least one of these treatments. The proportion of patients achieving PASI 75 at week 12 in the etanercept 25 mg twice weekly group was 28% (18/64) in patients whose psoriasis had not failed to respond to prior therapy and 37% (41/112) in patients whose psoriasis had failed to respond to prior therapy. These data have only been published in abstract form, although for licensing purposes, the manufacturer supplied subgroup analyses by prior systemic therapy for all three studies. The EMEA scientific discussion indicated that there was no significant difference in PASI 75 response between patients who had received previous systemic therapy and those who had not.
4.1.1.6 The Assessment Group concluded that, in general, etanercept appeared to be well tolerated during short-term and longer-term use. Injection site reaction was the most common adverse event. Across the three trials, a total of 232/415 (56%) people on etanercept 25 mg experienced one or more adverse events compared with 211/414 (51%) people taking placebo (RR 1.10, 95% CI 0.97 to 1.25). The numbers of participants classed as having a serious adverse event were 6/415 (1%) and 4/414 (1%) for the etanercept and placebo groups, respectively (RR 1.50, 95%CI, 0.43 to 5.26). In the two groups, a total of 9/415 (2%) and 10/414 (2%) withdrew because of adverse events (RR 0.90, 95% CI 0.97 to 2.19). There were no statistically significant differences in adverse event rates between the patients taking etanercept 50 mg and those taking placebo.
4.1.2 Efalizumab
4.1.2.1 A total of five randomised controlled trials (RCTs) that studied efalizumab at a dose of 1 mg/kg once a week were included in the Assessment Report. Across all doses, a total of 3130 participants were randomised. Each trial was placebo-controlled and of 12 weeks duration. Inadequacies in the reporting of the trials meant that the quality of four of the trials could not be properly assessed by the Assessment Group. Outcomes collected included the proportion of patients achieving PASI 50, PASI 75 and PASI 90; PASI score; DLQI score; physician global assessments; itching score; and PsA frequency and severity.
4.1.2.2 Not all of the data for each outcome were available. All five trials reported a statistically significant benefit of efalizumab over placebo in the numbers of participants achieving a PASI 50 (pooled relative risk [RR] 3.9, 95% confidence interval [CI] 3.3 to 4.6). Four of the trials reported a PASI 75, and these results were also statistically significant in favour of efalizumab for each trial (pooled RR 7.4, 95% CI 5.2 to 10.7). The one trial reporting a PASI 90 found no statistically significant differences between efalizumab and placebo (RR 5.3, 95% CI 0.9 to 31.7). Only one trial reported mean change from baseline in PASI score; it was 52% for efalizumab-treated patients compared with 19% for placebo-treated patients (p < 0.001). DLQI outcomes were reported in four trials, all of which reported a statistically significant greater reduction in patients taking efalizumab; the data could not be pooled due to lack of measures of variance.
4.1.2.3 The submission from the manufacturer presented a subgroup analysis of trial IMP24011 (n = 793), which included 526 participants who were deemed 'high-need' because they were resistant to, intolerant of or had contraindications to, at least two currently available systemic agents. The numbers achieving a PASI 75 in the whole population were 166/529 (31%) receiving efalizumab and 11/264 (4%) receiving placebo (RR 7.5, 95% CI 4.2 to 13.6). In the 'high-need' subgroup of individuals, 103/342 (30%) of participants who received efalizumab achieved a PASI 75 compared with 6/184 (3%) who received placebo (RR 10.9, 95% CI 4.6 to 26.2). The manufacturer concluded that the efficacy of efalizumab in the high-need group was similar to its efficacy across all treated patients.
4.1.2.4 There were no RCTs that had a duration of longer than 12 weeks, although two of the RCTs had 24 week response data from extension periods. Fifteen-month response data were available from an ongoing uncontrolled study, which indicated that response was maintained in patients who continued treatment.
4.1.2.5 Pooled data from the five RCTs indicated that there were no statistically significant differences between the efalizumab 1 mg/kg and placebo groups in the number of patients experiencing one or more adverse events, those that had a serious event and those that withdrew due to adverse events. Efalizumab was only licensed in late 2004 and therefore limited data were available relating to long-term adverse events. The SPC states that the safety of efalizumab has not been tested beyond 12 weeks, but the Assessment Group identified data from two 24-week periods, and in one study 339 participants were entered into a 3-year open-label follow-up.
4.1.2.6 The Assessment Group undertook a formal evidence synthesis with the aim of bringing together the clinical evidence regarding the efficacy of etanercept, efalizumab and other treatments. This evidence synthesis generated parameter estimates (response rates) used in the economic modelling. The evidence synthesis found that efalizumab was less effective than etanercept 25 mg, and both were less effective than infliximab, methotrexate and ciclosporin. The 50 mg dose of etanercept was found to be more effective than the 25 mg dose.
4.2 Cost effectiveness
4.2.1 Published economic evaluations
4.2.1.1 The Assessment Group did not identify any published economic evaluations that considered efalizumab. The Assessment Group identified only one published economic evaluation of etanercept that met its inclusion criteria. The base-case analysis found UVB phototherapy to be the most cost-effective option, followed by methotrexate. Of the three biological therapies examined (infliximab, etanercept and alefacept), infliximab was found to be the most cost effective, although it was still less cost effective than non-biological treatments. The analysis, however, had limited usefulness for decision making primarily because it was US-based and the results were not expressed as incremental costs per quality-adjusted life year (QALY).
4.2.2 Etanercept – manufacturer's model
4.2.2.1 The manufacturer of etanercept developed a Markov model using pooled data from the three registration RCTs. It aimed to assess the cost effectiveness of etanercept 25 mg (twice weekly), etanercept 50 mg (twice weekly) and an option of topical therapy only. A mapping exercise was used to estimate the relationship between DLQI (measured in the trials) and utility. Average improvement in utility was multiplied by the time between visits to estimate QALY gain between the visits. The manufacturer modelled a 12-week time horizon and longer-term outcomes (comprising eight 12-week treatment periods) by means of extrapolation. No discounting on costs and effects appears to have been applied and adverse events were not explicitly included.
4.2.2.2 The short-term (12-week) analysis estimated that the ICER for etanercept 25 mg over no systemic therapy was almost £125,000; the ICER for etanercept 50 mg was substantially higher. In the longer-term (96-week) analysis, the ICER for intermittent 25 mg etanercept treatment versus placebo was estimated to be £37,200. When people with relatively severe psoriasis (PASI greater than 10) and a poor quality of life (DLQI greater than 15) at baseline were considered in the 96-week analysis, the ICERs for etanercept versus systemic therapy declined markedly. For intermittent 25 mg etanercept therapy, the ICER in this subgroup was found to be around £24,000 per QALY, respectively.
4.2.3 Efalizumab – manufacturer's model
4.2.3.1 The manufacturer developed a decision-tree-based model, which compared the cost effectiveness of efalizumab treatment with that of topical therapies (based on calcipotriol and betamethasone) over a 10-year timeframe. Utilities (based on a time trade-off approach) were obtained from a study of 87 people with psoriasis. In order to estimate QALYs, the number of successfully treated years was multiplied by the difference in utility between a PASI 50 responder and a non-responder. There were no published PASI 50 response rates for topical therapies in people with moderate to severe psoriasis; therefore the rates associated with the placebo arms of the efalizumab RCTs (which permitted concomitant topical therapy use) were used. With the exception of a small proportion of patients who discontinue therapy for reasons unrelated to efficacy or adverse events, it was assumed that patients responding at 12 weeks (in terms of PASI 50) would continue to respond for a further 10 years. Resource use due to adverse events was included in the model.
4.2.3.2 In the base-case scenario, costs were discounted by 6% and outcomes were discounted by 1.5%. Over the 10-year time horizon, expected 'quality-adjusted response years' with efalizumab were estimated to be 1.39 versus 0.36 for a treatment strategy beginning with topical therapy. The incremental cost/QALY results from the deterministic and probabilistic analyses were around £25,600 and £25,800, respectively. The data were subject to a number of deterministic sensitivity analyses; the incremental cost-effectiveness ratio (ICER) remained below £30,000 under most scenarios.
4.2.4 The Assessment Group model
4.2.4.1 The Assessment Group developed its own model for assessing the cost effectiveness of etanercept and efalizumab. The main analysis compared etanercept (intermittent 25 mg and 50 mg, and continuous 25 mg), efalizumab (continuous) and supportive care without DMARDs or biological therapies. Utilities were estimated by mapping the mean change in DLQI score (conditional on PASI response) to changes in EQ-5D (a non-disease specific instrument for describing and valuing HRQoL). When modelling intermittent etanercept treatment, it was assumed that the time between 12-week treatment cycles would be 29 days, resulting in 3.2 treatment cycles per year. This was based on the median duration of PASI 75 response as reported in an unpublished etanercept re-treatment study. Annual discount rates of 6% on costs and 1.5% on outcomes were applied in the analyses. Adverse events were not directly modelled. Decision uncertainty was examined using probabilistic sensitivity analysis.
4.2.4.2 The base-case analysis showed that supportive care is the only cost-effective strategy until the threshold reaches £70,000 per QALY. The ICER for intermittent low-dose (25 mg) etanercept was found to be £65,320 per QALY gained. The ICER for intermittent high-dose (50 mg) etanercept treatment was substantially higher. Efalizumab was dominated in the analysis by intermittent etanercept 25 mg. The results of several alternative scenarios presented indicated that the cost effectiveness of efalizumab and etanercept varied considerably according to baseline DLQI and whether it was assumed that all non-responders were hospitalised for 21 days annually. In all cases, the ICERs of the biological agents were found to be lower than in the base-case; but efalizumab was less cost-effective than intermittent etanercept 25 mg. In the scenario that considered both poor baseline quality of life and hospitalisation for non-responders, the ICER for intermittent etanercept 25 mg was £14,460 per QALY gained.
4.3 Consideration of the evidence
4.3.1 The Committee reviewed the data available on the clinical and cost effectiveness of etanercept and efalizumab, having considered evidence on the nature of the condition and the value placed on the benefits of etanercept and efalizumab by people with psoriasis, those who represent them, and clinical experts. It was also mindful of the need take account of the effective use of NHS resources.
4.3.2 The Committee accepted that the RCTs demonstrated the clinical effectiveness of etanercept and efalizumab in people with moderate to severe plaque psoriasis. The Committee noted that the inclusion criteria for the RCTs did not wholly reflect the population for which these technologies are currently licensed because their psoriasis had not necessarily failed to respond to other treatment options. The Committee noted, however, that one of the efalizumab trials did include a subgroup of patients whose psoriasis had failed to respond to previous therapy and that post hoc analyses of one of the etanercept trials examined the effectiveness of this technology in people whose condition had failed to respond to at least one prior systemic therapy. The Committee also heard from the clinical experts that, in clinical practice, these drugs were used according to the licensed indications and were as effective for people who had not responded to other available treatments as for those who were treatment naïve. The Committee considered, therefore, that the overall baseline characteristics of the patients included in the trials indicated that they were a population with relatively severe psoriasis.
4.3.3 The Committee considered carefully the clinical effectiveness of etanercept and efalizumab on the basis of dose administered and duration of therapy within their respective licensed indications. It noted that, according to the SPCs, the maximum duration of etanercept therapy within any one cycle of treatment is 24 weeks and that treatment should be stopped in non-responders at 12 weeks. However, no stipulations are made in the SPC as to the period needed to be allowed between successive treatment cycles. The experts advised the Committee that people who had responded to etanercept sometimes relapsed within weeks of stopping therapy and therefore in some people the interval between successive cycles could be short. The experts also indicated that in their clinical experience 12 weeks was a sufficient period of time in which to determine whether a patient was likely to respond to treatment with either etanercept or efalizumab. This experience was in accordance with the recent British Association of Dermatologists guidelines. The experts also advised that people who responded at this stage would be suitable for either further cycles of etanercept or continuous therapy with efalizumab.
4.3.4 In considering the economic modelling the Committee recognised that there was considerable uncertainty in the estimates of cost effectiveness that had been produced. This uncertainty related principally to estimates of the efficacy of the alternative interventions and treatment regimens and the evidence on long-term outcomes. In addition, the Committee noted the differences in modelling approaches adopted by the manufacturers and the Assessment Group, and their respective weaknesses and strengths.
4.3.5 Noting this uncertainty in the economic modelling, the Committee concluded it was unlikely that these interventions would be cost effective except in people who had very poor quality of life and who would be likely to require hospital admission for treatment. Testimony from the clinical experts and consultees suggested that these people would be those with severe disease as defined by a PASI of 10 or more and DLQI of more than 10, who had not responded to standard systemic therapies. The Committee noted that the SPCs for etanercept and efalizumab state that the interventions are indicated in adults who have failed to respond to, who have a contraindication to, or are intolerant to other systemic therapy including ciclosporin, methotrexate and/or PUVA.
4.3.6 The Committee noted that in all cost-effectiveness scenarios of the Assessment Group model efalizumab was dominated by intermittent 12-weekly cycles of etanercept 25 mg (that is, efalizumab treatment was more costly and slightly less effective). The Committee noted that there were data supporting the continued efficacy of etanercept when administered on an intermittent basis according to its licensed indications. In particular it noted that in the re-treatment study discussed in 4.1.2.4 there was no statistically significant difference in mean absolute PASI score after 12 weeks of re-treatment compared with the initial 12 weeks of treatment. In addition, the Committee was also aware that the Assessment Group model took into account data from the manufacturers relating to a median duration of PASI 75 response of 29 days as reported in an unpublished etanercept re-treatment study. On this basis the Committee was persuaded that intermittent treatment with etanercept would be clinically effective in a population of adults with psoriasis, and also more cost effective than treatment with continuous efalizumab, principally because of the treatment-free periods that characterise intermittent therapy.
4.3.7 The Committee did, however, consider that because the two drugs had different mechanisms of action, it was possible that people whose psoriasis had not responded to etanercept might well have a response to efalizumab. Given that other treatment options available for these people were limited, the Committee concluded that efalizumab should be recommended for use in the group of people whose psoriasis had failed to respond to etanercept (using the response criteria in Section 1.2) or who were either intolerant of or had definite contraindications to etanercept. TheCommittee was also persuaded that patients who have been shown to experience severe relapse early after discontinuation of etanercept therapy might not be suitable for intermittent treatment. Under these circumstances the choice between rapid reinstitution of intermittent etanercept and changing to continuous efalizumab should be made on clinical grounds as either approach was likely to be equally cost effective.
4.3.8 The Committee additionally considered the cost effectiveness of the use of the 50 mg twice-weekly dose of etanercept in the subgroup of people who had very poor quality of life. It was aware that within the SPC, the 50 mg dose is allowable only for a period of 12 weeks for etanercept-naïve patients and should not be used for repeat cycles of therapy in patients who respond initially and then subsequently relapse. The experts commented that the 50 mg dose may be of use in patients who had particularly severe psoriasis in order to achieve more rapid control of their symptoms. The Committee was not persuaded that this very severe subgroup had been sufficiently defined (in terms of PASI score or DLQI at baseline) or that their potential for an enhanced response was supported by trial data. There was also no evidence that an initial 12-week period of 50 mg achieved better longer-term outcomes than an initial 25 mg dose. The Committee noted that the Assessment Group's economic model had shown that the (unlicensed) use of the intermittent 50 mg twice-weekly regimen of etanercept was not cost effective. The Committee was aware that the licensed scenario of an initial 12 weeks treatment with 50 mg etanercept twice-weekly in etanercept naïve patients had not been directly modelled. However, the Committee was persuaded that the economic modelling had shown that the use of the higher dose of etanercept for 12 weeks was not a cost-effective option compared with the intermittent use of 25 mg twice weekly: the increased drug acquisition cost would substantially outweigh the benefits obtained from an initial 12 weeks at a higher dose. The Committee therefore concluded that etanercept should not be recommended at doses exceeding 25 mg twice weekly.
4.3.9 The Committee considered the most appropriate criteria for assessment of response to treatment. It was aware that the EMEA recognises that reduction of baseline PASI score by 75% (PASI 75) is an indicator that severe psoriasis has responded to treatment. Additionally the British Society for Rheumatology guidelines on the use of these agents in PsA also recommends collecting PASI data at baseline and using the PASI 75 as an indicator of response. Therefore, the Committee considered that a criterion for adequacy of response for both drugs assessed at 12 weeks should be the PASI 75. It was however persuaded that the evidence and the expert testimony indicated that there were a number of individuals who, on the basis of assessment of improvement in quality of life, would derive significant benefit from etanercept or efalizumab, but might have failed to achieve a PASI 75 after 12 weeks of treatment. The Committee therefore considered that the assessment of response should also include a measure of quality of life improvement as defined by a change from the baseline DLQI at initiation of treatment. It concluded that it would also be appropriate for individuals to continue on treatment if they had achieved a PASI 50 response providing they had also achieved a five-point reduction in their DLQI score from when treatment was initiated. This view was supported by recent guidelines from the British Association of Dermatologists on the use of biological interventions in psoriasis. The Committee also concluded that if people had failed to reach these response criteria at 12 weeks then treatment should be discontinued. It further concluded that re-treatment in these patients would not be appropriate as there was no evidence to suggest that the response would differ in subsequent cycles.
4.3.10 The Committee considered the possibility of adverse events with etanercept and efalizumab both in the short and longer term; it noted that there was currently little information on the use of these drugs in people with psoriasis beyond the duration of the RCTs. It was also aware that these drugs could increase risk of malignancy and that people with psoriasis were potentially already at an increased risk of skin cancer as a result of therapeutic exposure to ultraviolet radiation. Additionally the possibility of recrudescence of latent tuberculosis has been associated with the use of etanercept. Consequently the Committee agreed with the experts' advice that a register should be established in order to collect information on long-term outcomes (including adverse events) in patients with psoriasis treated with cytokine inhibitors. The Committee therefore strongly supported the rapid establishment of the proposed British Association of Dermatologists Biological Interventions Register (BADBIR). It also considered that, in order to ensure that all relevant data were captured, procedures should be implemented to allow cross-referencing of BADBIR with information from people with PsA enrolled in the British Society for Rheumatology biologicals register.
4.3.11 The Committee also discussed who should initiate treatment and monitor the use of etanercept and efalizumab in patients with psoriasis. The Committee took note of the content of the SPC for each drug. In addition, it heard evidence from experts who strongly argued that treatment strategies should take into account the fact that certain people will have concomitant PsA. The Committee therefore agreed that the use of etanercept and efalizumab should be initiated and supervised only by specialist physicians experienced in the diagnosis and treatment of psoriasis. Furthermore, if a person has concomitant psoriasis and PsA their treatment should be managed by collaboration between a rheumatologist and a dermatologist.