3 The manufacturer's submission
3.1 The manufacturer described nine trials that provided evidence on the clinical effectiveness of prucalopride in people with chronic constipation. There were three pivotal phase III randomised, double-blind, placebo-controlled trials in adults (aged 18–65 years) with chronic constipation (PRU-INT-6, PRU-USA-11 and PRU-USA-13), one phase III, randomised, double-blind, placebo-controlled trial in older people (65 years or older, PRU-INT-12), one trial in adults (18 years or older) with opioid-induced constipation (PRU-INT-8), one retreatment study (PRU-USA-28) and three extended, open-label, single-arm, observational studies (PRU-INT-10, PRU-USA-22 and PRU-INT-17). The key clinical evidence presented by the manufacturer was derived from the three pivotal trials, which reported the efficacy of prucalopride compared with placebo in adults, and PRU-INT-12, which reported the efficacy of prucalopride compared with placebo in older people. The number of people randomised to PRU-INT-6, PRU-USA-11, PRU-USA-13 and PRU-INT-12 was 720, 628, 651 and 305 respectively. Approximately 90% of people in the pivotal trials were women. The manufacturer also presented other trials, which reported additional safety considerations and response rates (see section 3.8). The manufacturer's submission stated that people were enrolled in the pivotal trials and PRU-INT-12 if they had a history of chronic constipation (defined as no more than two spontaneous complete bowel movements per week) and one or more of the following for at least 6 months before the screening visit:
straining during at least 25% of bowel movements
very hard or hard stools in at least 25% of bowel movements
sensation of incomplete evacuation for at least 25% of bowel movements.
3.2 There was a 2-week run-in period for each pivotal trial and for PRU-INT-12, in which no laxative medication (except for rescue medication) was allowed. People in the pivotal trials were then randomised 1:1:1 to prucalopride 2 mg, prucalopride 4 mg or placebo. People in PRU-INT-12 were also randomised to prucalopride 1 mg. If people had not had a bowel movement for 3 days or more, they could receive a single dose of 15 mg bisacodyl as rescue medication (medications used for quick relief of symptoms). If a bowel movement did not occur, the dose of bisacodyl could be increased; if there was still no bowel movement after this, an enema could be administered. In the pivotal trials people were treated for 12 weeks and in PRU-INT-12 people were treated for 4 weeks. Data were collected at 4- and 12-week time points in the pivotal trials and at 4 weeks in PRU-INT-12.
3.3 The primary outcome measure in the pivotal trials was three or more spontaneous complete bowel movements per week which was evaluated over the first 4 weeks of treatment and averaged over the full 12 weeks of the trial. The proportion of people with an average increase of one or more spontaneous complete bowel movements per week compared with baseline was measured as a secondary outcome in the trials. The proportion of people treated with prucalopride 2 mg in the pivotal trials who had three or more spontaneous complete bowel movements per week during weeks 1–4 ranged from 23.7% to 32.1%, compared with 9.8% to 11.5% for placebo (all p ≤ 0.001). During weeks 1–12, the proportion of people treated with prucalopride 2 mg who had three or more spontaneous complete bowel movements per week ranged from 19.5% to 28.9% compared with 9.6% to 13.0% for placebo (all p ≤ 0.01).
3.4 The proportion of people treated with prucalopride 2 mg in the pivotal trials who had an average increase of one or more spontaneous complete bowel movements per week (the secondary outcome measure) during weeks 1–4 ranged from 41.0% to 56.5% compared with 20.9% to 25.5% for placebo (all p ≤ 0.001). During weeks 1–12 of treatment, the proportion of people who had an average increase of one or more spontaneous complete bowel movements per week ranged from 38.1% to 50.3% for prucalopride 2 mg compared with 20.9% to 27.5% for placebo (all p ≤ 0.001).
3.5 In PRU-INT-12 the proportion of people treated with prucalopride who had an average of three or more spontaneous complete bowel movements per week during weeks 1–4 was 39.5% for prucalopride 1 mg and 32.0% for prucalopride 2 mg, compared with 20.0% for placebo (p ≤ 0.05). In addition, the proportion of people treated with prucalopride who had an average increase of one or more spontaneous complete bowel movements per week during weeks 1–4 was 61.1% for prucalopride 1 mg and 56.9% for prucalopride 2 mg compared with 33.8% for placebo (p ≤ 0.05).
3.6 The manufacturer's submission reported quality-of-life data from the pivotal trials, which were derived from Patient Assessment of Constipation – Symptoms (PAC-SYM) and Patient Assessment of Constipation – Quality of Life (PAC-QOL) scores. All pivotal trials showed a significantly greater improvement in PAC-QOL scores for people treated with prucalopride compared with placebo at weeks 1–4 and weeks 1–12 (both p < 0.001 compared with placebo). Statistically significant improvements in PAC-SYM scores were also seen in all three trials at weeks 1–4 (p ≤ 0.001 compared with placebo) and in all trials except PRU-INT-6 at weeks 1–12 (p ≤ 0.05). PRU-INT-12 also reported quality-of-life data for older women derived from PAC-SYM and PAC-QOL scores. Statistically significant improvements in PAC-SYM and PAC-QOL scores were shown for prucalopride 1 mg compared with placebo at week 4 (both p ≤ 0.05). Improvements in PAC-SYM and PAC-QOL scores for prucalopride 2 mg compared with placebo were seen at week 4 but they did not reach statistical significance.
3.7 Surveys of the SF-36 mental component summary and the SF-36 physical component summary were taken during the run-in period and at weeks 4 and 12 of the pivotal trials. No trials showed statistically significantly greater improvements in SF-36 scores for prucalopride 2 mg compared with placebo at week 12. A statistically significant improvement in the SF-36 physical component summary at week 4 was only seen in the PRU-INT-6 study for prucalopride 2 mg compared with placebo (p ≤ 0.05). Additional evidence provided by the manufacturer in response to the appraisal consultation document suggested, however, that when only the cohort of patients who responded to treatment was compared with placebo, a statistically significant difference between the effect of prucalopride and placebo was seen. The SF-36 data were not used in further sections of the manufacturer's submission.
3.8 The following three single-arm extension studies were designed to assess the long-term tolerability and safety of prucalopride:
PRU-INT-10: included people from PRU-INT-6 (pivotal trial) and PRU-INT-12 (trial in older people).
PRU-USA-22: included people from PRU-USA-3 (phase II dose–response trial), PRU-USA-11 and PRU-USA-13 (pivotal trials), PRU-USA-21 (phase II dose–response trial), PRU-USA-25 (phase III dose–titration trial), PRU-USA-27 (opioid-induced chronic constipation trial) and PRU-USA-28 (phase III retreatment trial).
PRU-INT-17: included people from PRU-INT-8 and PRU-INT-14 (both opioid-induced chronic constipation trials).
Studies PRU-INT-10, PRU-USA-22 and PRU-INT-17 had durations of 24, 36 and 12 months respectively. People received prucalopride doses ranging from 0 to 4 mg. Results from these studies reported that prucalopride treatment was associated with an improvement in constipation from baseline at all time points (this was statistically significant in PRU-INT-10 and PRU-USA-22) and a decrease in the use of laxatives. At 12 months, on average, less than 50% of people remained in these trials. The reasons for stopping treatment included insufficient treatment response (18%), withdrawal of consent (15%) and adverse events (9%). However for the three trials, most people (approximately 45%) discontinued treatment because the previous trial sponsor decided to stop the prucalopride development programme worldwide.
3.9 The manufacturer reported that prucalopride was generally well tolerated and that the majority of adverse events in the clinical trials were mild or moderate. In PRU-INT-6, 80.8% of people in the prucalopride 2 mg arm reported at least one adverse event, compared with 66.0% in the placebo arm. The incidence of serious adverse events was 2.1% in both the prucalopride and placebo arms. The most frequently reported adverse events included headache, nausea and abdominal pain. The incidence of diarrhoea in the prucalopride 2 mg arm (13.0%) was more than twice that of the placebo arm (5.4%). The adverse event profiles in the PRU-USA-11 and PRU-USA-13 trials were similar to those in the PRU-INT-6 trial. The onset of these adverse events was most frequently reported on the day after the start of treatment ('day one') and the duration was short. The manufacturer reported that when day one was excluded from the analysis, the incidence of adverse events was comparable among the treatment groups.
3.10 The manufacturer developed a decision analytic model based on patient-level data from the clinical trials. All data from the included trials, for men and women, were used in the model, however all analyses presented by the manufacturer were derived using data from women only. The model compared prucalopride 1 mg daily (for older women) and prucalopride 2 mg daily (for adult women) with placebo for up to 52 weeks. In both arms, bisacodyl as rescue medication was allowed, and if it was used, any bowel movements in the following 48 hours were not included in the analysis. In the base case, results were presented for all women (that is, adult women and older women). Treatment duration was 4 weeks, after which women could only continue treatment if they had three or more spontaneous complete bowel movements per week.
3.11 Two additional analyses were presented. One incorporated data for adult women only and one incorporated data for older women only. For the first 12 weeks, the model for adult women included randomised controlled trial data for all women treated with prucalopride 2 mg. Additional observational trial data were incorporated up to a further 40 weeks after the initial trial period. The model in older women incorporated randomised controlled trial data for women treated with prucalopride 1 mg in the first 4 weeks followed by observational data for up to 1 year.
3.12 No discounting was applied in the model because both costs and utility values were modelled for 52 weeks. The only costs incorporated in the economic model were the list prices of prucalopride 2 mg (£2.13 per tablet) and prucalopride 1 mg (£1.38 per tablet). Costs and utility values for placebo plus rescue therapy were not included in the model. The manufacturer assumed that women would take their treatment for only part of the year (220 days). Adverse events and their associated costs were not included in the model. The manufacturer acknowledged that the rates of adverse events were comparable between prucalopride and placebo and therefore they considered that including these events would not affect the outcome of the analysis.
3.13 Clinical data incorporated in the model were derived from the three pivotal trials, two trials in older people (PRU-INT-12 and PRU-USA-26) and the extension studies. Patient characteristics from these studies were used to inform the disease states in the model. Further patient characteristics were obtained from other trials not fully described in the manufacturer's submission, including three additional dose–response trials (PRU-INT-1, PRU-INT-2 and PRU-USA-3) and two phase II trials (PRU-FRA-1 and PRU-GBR-4). No methods or results for these trials were included in the submission. PAC-SYM and PAC-QOL data from the clinical trials were mapped to EQ-5D through SF-36 scores using the generalised least squares regression method. People who had chronic constipation who did not respond to prucalopride were assumed to have no quality-adjusted life year (QALY) gain.
3.14 The manufacturer's base case presented an average cost-effectiveness ratio because no cost for the comparator was included in the model. The average cost of prucalopride for all women was £498 with an average QALY gain of 0.0316, resulting in an average incremental cost-effectiveness ratio (ICER) of £15,700 per QALY gained. The average cost of prucalopride for adult women (18–64 years) was £622 with an average QALY gain of 0.0369, resulting in an ICER of £16,800 per QALY gained. The average cost of prucalopride for older women (65 years or older) was £403 with an average QALY gain of 0.0342, resulting in an ICER of £11,700 per QALY gained.
3.15 The manufacturer also presented an analysis that included all women who had an additional bowel movement per week (the secondary outcome measure in the pivotal trials). The manufacturer estimated that, for all women, the annual cost per person to reach this secondary outcome would be £498 with an average QALY gain of 0.0277, resulting in an ICER of £18,000 per QALY gained. For adult women, the cost would be £622 with an average QALY gain of 0.0342, resulting in an ICER of £18,000 per QALY gained. The cost for older women was £403 with a QALY gain of 0.0255, resulting in an ICER of £15,800 per QALY gained.
3.16 The manufacturer presented probabilistic sensitivity analyses for all women, adult women and older women, with and without an adjustment for baseline severity of constipation. The probabilistic sensitivity analysis results showed that the probabilities of the ICERs for prucalopride exceeding £20,000 per QALY gained were approximately 45%, 44% and 47% for all women, adult women and older women respectively. The probabilities of the ICERs for prucalopride exceeding £30,000 per QALY gained were approximately 40%, 36% and 45% for all women, adult women and older women respectively. The manufacturer reported that the main factors affecting cost effectiveness were:
the effect of constipation severity at baseline on treatment effectiveness (that is, if the treatment effect is assumed to be the same regardless of baseline severity, the probability of prucalopride being cost effective at £20,000 per QALY gained is increased)
the ability to identify women whose constipation did not respond to prucalopride at a very early stage of treatment
the acquisition cost of prucalopride
the utility values derived from mapping PAC outcome measures (PAC-SYM and PAC-QOL) to EQ-5D scores.
3.17 The ERG reviewed the evidence submitted by the manufacturer on the clinical and cost effectiveness of prucalopride. It noted that the three pivotal trials (PRU-INT-6, PRU-USA-11 and PRU-USA-13) formed the basis of the manufacturer's assessment of clinical effectiveness. The ERG was unclear how people from the original trials were selected for the extension studies because no baseline data were provided in the manufacturer's submission. The ERG considered it possible that the people in the extension studies had constipation that was not necessarily refractory to laxative treatment. The ERG further noted that the extension studies included both older people and people with opioid-induced chronic constipation and that the results were not separated. The ERG was also concerned that the high rate of withdrawal from the extension studies (more than 50% of people at 12 months) was likely to have resulted in people who were relatively more satisfied with their treatment continuing with treatment compared with those dropping out.
3.18 Overall, the ERG noted that there was a considerable quantity of clinical-effectiveness evidence in adults that suggested an improvement in chronic constipation for people treated with prucalopride compared with placebo. The ERG calculated the weighted average of the effect of prucalopride across the pivotal trials and estimated that 28% of people reached the primary outcome of three or more spontaneous complete bowel movements per week after treatment with prucalopride 2 mg compared with 10.6% of people treated with placebo after 14 weeks. After 112 weeks, 23.8% of people treated with prucalopride 2 mg reached the primary outcome compared with 11.4% of people treated with placebo.
3.19 The ERG was uncertain whether the population in the trials reflected the population covered by the marketing authorisation or decision problem for prucalopride. It noted that in the three pivotal trials, 17% of people at baseline answered that they had found their previous laxative treatment adequate and may not have been eligible for the trials (that is, not laxative refractory). The ERG further considered that people who have one or two bowel movements per week while on laxative treatment were likely to be having beneficial effects from laxatives and therefore their constipation may not have been refractory to laxatives. It also considered that any two of the criteria used alone by the manufacturer to describe chronic constipation (section 3.1 above) would be unlikely to be sufficient evidence of treatment failure with laxatives.
3.20 The ERG considered that the comparator used in the pivotal trials (placebo plus rescue medication with bisacodyl) did not represent standard clinical practice for chronic constipation. It suggested that a more appropriate comparator would have been a variety of oral laxative treatments, at the discretion of the treating clinician. It further commented that the manufacturer's submission did not consider some of the comparators outlined in the decision problem, including invasive procedures (such as rectal interventions) and bowel surgery.
3.21 The ERG assessed the manufacturer's cost-effectiveness analysis and considered its methodological approach acceptable. It noted that the manufacturer's decision to exclude the cost of the comparator from the analysis was conservative. However, the ERG was concerned that precise details of the trials used to inform the inputs in the economic model were not given or did not fully correspond with those described in the manufacturer's submission. It noted that five trials used for the economic model (PRU-INT-1, PRU-INT-2, PRU-USA-3, PRU-FRA-1 and PRU-GBR-4) were not fully described in the submission.
3.22 The ERG noted that quality of life data from PAC-QOL and PAC-SYM scores were mapped to the EQ-5D using SF-36 scores obtained from the trials. The ERG was concerned that the SF-36 data did not directly contribute to EQ-5D scores, even though these results were available from the trials, and no sensitivity analysis was undertaken by the manufacturer to test the impact of using SF-36 results.
3.23 The ERG noted that the manufacturer's model only allowed for variation in the response rate and mean treatment rates to be analysed. It also noted that no explicit allowance was made for withdrawal from treatment at any time after 4 weeks and that the assumption that the last measured QALY gain was sustained for the rest of the year was not tested in the model.
3.24 The ERG noted there were more adverse events in the prucalopride arms than in the placebo arms of the trials. It was concerned that adverse events, including rare events, and their associated costs were not included in the model.
3.25 The ERG ran the manufacturer's model using alternative scenarios and assumptions including the following:
Assuming that people who responded to treatment with prucalopride would receive treatment for a mean of 220 days or 365 days.
Using response rates taken from pooled trial estimates at week 4 calculated in the effectiveness review.
Allowing for the possibility that adverse events may be higher in the prucalopride arm than the placebo arm by increasing costs by 5% and reducing QALY gain by 5% in the prucalopride arm.
Reducing the effectiveness (QALY) of prucalopride and placebo uniformly by 25%, 50% and 75% to allow for possible variation in the regression method used to calculate the QALYs.
The ERG concluded that the results from its sensitivity analysis were not significantly different from those provided by the manufacturer.
3.26 Full details of all the evidence are in the manufacturer's submission and the ERG report.