The Appraisal Committee considered evidence submitted by the manufacturer of infliximab and a review of this submission by the Evidence Review Group (ERG).
The manufacturer developed a systematic review of the literature on the use of infliximab and comparator drugs in the target population. Two randomised controlled trials (RCTs) comparing infliximab with placebo were included in the review. For the comparator drug, ciclosporin, two RCTs were included. Searches carried out by the ERG identified no additional RCTs for either infliximab or ciclosporin. The ERG noted that the total number of participants in the studies was small.
The primary outcome in both the review of clinical effectiveness and the economic evaluation was the avoidance of colectomy. In the larger of the two infliximab studies (n=45) 67% and 29% of patients treated with placebo and infliximab, respectively, had a colectomy within the first 3 months. Within 12 months, 71% and 42% of patients treated with placebo and infliximab, respectively, underwent colectomy. In the smaller infliximab study (n=11), all three participants treated with placebo underwent colectomy whereas none of the three patients treated with infliximab at the licensed dose did so within the first 3 months. Both studies included people with severe acute ulcerative colitis that had not responded to intravenous corticosteroids. The larger study also included people with moderately severe ulcerative colitis. Infliximab was given as a single infusion at a dose of approximately 5 mg per kg in the larger study. In the smaller study, patients were randomly assigned to single doses of 5 mg per kg, 10 mg per kg or 20 mg per kg. Neither study used a multiple-dose regimen as specified in the SPC for infliximab.
The two studies investigating ciclosporin differed from each other in the populations included and in the comparator used. In one study, ciclosporin was compared with placebo in people with acute severe ulcerative colitis that had not responded to corticosteroids. In this respect it was similar to the infliximab studies. In this study (n=20) 44% of people treated with placebo and 27% of those treated with ciclosporin underwent colectomy within the first 3 months. The second study, by D'Haens and coworkers, was different in that it compared ciclosporin with intravenous corticosteroid treatment in people who had not already received treatment with intravenous corticosteroids (n=30). In this study, 20% of people treated with corticosteroids and 21% of those treated with ciclosporin underwent colectomy within the first 3 months, and 40% of the corticosteroid-treated group and 36% of the ciclosporin-treated group underwent colectomy within 12 months.
To compare the effectiveness of infliximab with ciclosporin in the absence of a study comparing them directly, a mixed-treatment comparison (MTC) model was used to synthesise the relative treatment effects in respect of colectomy outcomes observed in the trials. The objective was to develop probabilities of colectomy that could be used in an economic evaluation comparing infliximab with ciclosporin. The probabilities of a patient undergoing colectomy were estimated to be 0.67, 0.23 and 0.58 for placebo, infliximab and ciclosporin, respectively, for the first 3 months. The respective probabilities during months 4 to 12 were 0.14, 0.27 and 0.18 for placebo, infliximab and ciclosporin. The ERG pointed out that it is likely that the model does not appropriately estimate the true effects of the different treatment options, particularly with respect to the effectiveness of ciclosporin. The estimate of colectomy rate in the MTC was nearly twice that actually observed in the RCTs of ciclosporin. In the ERG's view it was not appropriate to include the study by D'Haens and coworkers in this analysis because neither the population nor the comparator treatment was in line with the other infliximab and ciclosporin RCTs. In a further analysis undertaken by the ERG the exclusion of this study from the analysis of colectomy rate for months 0 to 3 reduced the estimated rate for ciclosporin from 0.58 to 0.48. According to the ERG, however, these probabilities were still much higher than would be expected in practice.
Neither of the infliximab RCTs reported any deaths and the frequency of adverse events appeared to be comparable between the infliximab and placebo groups. In these studies, two patients treated with infliximab had serious adverse events that required prolonged hospitalisation, and one had long-lasting bleeding. In the ciclosporin studies, no deaths were reported. Adverse events reported in the ciclosporin groups included paresthesias, grand mal seizure and headaches.
A decision analytical model was used to simulate the progression of hypothetical cohorts of patients. The structure of this model was informed by the infliximab (first infusion was 5 mg per kg including concomitant intravenous corticosteroids) and ciclosporin (4 mg per kg daily, intravenously) RCTs, information on current UK clinical practice and expert opinion. People with severe active ulcerative colitis hospitalised for an acute exacerbation of the disease were considered in the economic evaluation. It was assumed that these patients had already received treatment with corticosteroids for 72 hours, and that this had not improved their condition adequately. They were tracked as they received one of four treatment strategies:
Infliximab: the first infusion was 5 mg per kg on day 4, including concomitant intravenous corticosteroid treatment for an additional 7 days during the hospital stay. In addition, responders also received two 5-mg per kg doses of infliximab at weeks 2 and 6 following the first infusion.
Ciclosporin: patients were given a 4-mg per kg daily dose of intravenous ciclosporin starting on day 4 for 7 days. Following discharge from hospital, ciclosporin responders were switched to oral ciclosporin (2 mg per kg per day) for 3 months.
Standard care: patients continued treatment with intravenous corticosteroids for an additional 7 inpatient days.
Surgical intervention.
The time horizon used in the base case was 1 year. The course of the disease was represented by post-hospitalisation outcomes including medical remission, surgical remission and surgical complications. Treatment outcomes were characterised in the model as short term (0 to 3 months) and medium term (4 to 12 months). In the first 3 months, treatment with infliximab, ciclosporin or standard care either caused the ulcerative colitis to respond to treatment and go into remission, or the treatment failed and patients underwent colectomy. For the rest of the base-case analysis (4 to 12 months), patients whose disease went into initial remission either stayed in remission or the response was lost and they underwent surgery. An analysis extrapolated up to 10 years was also conducted to address the long-term treatment effect. Overall, the ERG considered that the structure of the manufacturer's economic model appropriately addressed the acute phase of the disease. However, the model did not take into account the costs and disutilities associated with adverse events. The ERG noted this was especially important because trials of infliximab in patients with ulcerative colitis described 'serious adverse events'. Mortality was also not included in the model.
The parameter estimates used in the model were obtained from clinical studies of infliximab and ciclosporin, and other sources. Although the baseline risk of disease progression was estimated using the placebo arm from the larger infliximab study, this was inconsistently stated in the manufacturer's submission. The primary source for the base-case utilities was an unpublished study based on the Health Outcomes Data Repository (HODaR), which was conducted using the EQ-5D instrument in patients with ulcerative colitis in south Wales. These were supplemented with utilities from another study in which utilities were estimated using the time trade-off method. Estimates of healthcare resource use and concomitant medication use were based on the opinions of a panel of UK gastroenterologists. The costs of all drugs were calculated based on the average doses used in the clinical trials and on pack sizes listed in the BNF. Drug administration costs were obtained from the NHS reference costs.
Univariate sensitivity analyses were performed considering variations in treatment effect, patient weight, utility estimates, infliximab administration cost, hospitalisation period and infliximab infusion doses. Probabilistic sensitivity analyses were conducted to assess the uncertainty of the cost-effectiveness estimates by assigning distributions around the primary outcome (colectomy), secondary outcome (post-surgery complications), utility estimates and unit costs. The ERG considered that the sensitivity analyses were appropriate, although they noted that the colectomy rates associated with the alternative treatment arms were not varied as part of the univariate analyses. The incremental cost-effectiveness ratios (ICERs) were shown by the ERG to be most sensitive to these colectomy rates. Also, the probabilistic sensitivity analysis placed distributions around selected parameters only.
The base-case cost-effectiveness estimates presented in the manufacturer's submission were £11,589 per additional quality-adjusted life year (QALY) gained for infliximab compared with standard care; £18,425 per additional QALY gained for infliximab compared with ciclosporin; and £13,407 per additional QALY gained for infliximab compared with surgery. These ICERs after minor corrections by the ERG rose to £12,307, £19,922 and £14,427, respectively. When the ERG excluded the study by D'Haens and coworkers (which the ERG considered was included in the analysis inappropriately; see 3.3 and 3.4), the ICER for infliximab versus ciclosporin increased considerably from £19,922 to £48,367 per QALY gained. Following consultation, further analyses were presented by the manufacturer to reflect the uncertainty around the estimate of the colectomy rate during months 4 to 12 following treatment with ciclosporin. In the original analysis, the colectomy rate for this period was based on the study by D'Haens and coworkers. If this study was excluded, then there was no other estimate available for this parameter. The manufacturer presented analyses assuming a high value (0.48, the same as the estimate for months 0 to 3 following ciclosporin treatment) and a low value (0.143, based on the estimate for months 4 to 12 following standard care). When the high value was assumed, the ICER for infliximab versus ciclosporin was reduced to £9,300 per QALY. Conversely when the low value was assumed, the ICER increased to £52,000 per QALY.
Full details of all the evidence are in the manufacturer's submission and the ERG report.