Prasugrel with percutaneous coronary intervention for treating acute coronary syndromes

The manufacturer and the Assessment Group both identified 1 randomised controlled trial (TRITON-TIMI 38) and 1 publication related to the core clinical cohort from TRITON-TIMI 38 (Wiviott et al. 2011) from a systematic search of the literature. Both the randomised controlled trial and the publication were also considered for the original appraisal of prasugrel (NICE's technology appraisal guidance 182 on prasugrel for the treatment of acute coronary syndromes with percutaneous coronary intervention). The Assessment Group commented that no new evidence had been identified since the original appraisal was published.

TRITON-TIMI 38 was a randomised double-blind trial that compared prasugrel with clopidogrel in 13,608 patients with moderate- to high-risk acute coronary syndromes (unstable angina, non-ST-segment-elevation myocardial infarction [NSTEMI] or ST-segment-elevation myocardial infarction [STEMI]) who were scheduled to have percutaneous coronary intervention. Patients were given aspirin (at a recommended daily dose of between 75 mg and 162 mg) in combination with the drugs studied. Patients were randomised to receive a loading dose of 60 mg prasugrel followed by 10 mg prasugrel daily or a loading dose of 300 mg clopidogrel followed by 75 mg clopidogrel daily for up to 15 months (the median treatment period was 14.5 months).

The primary efficacy end point for TRITON-TIMI 38 was a composite of the rate of non-fatal myocardial infarction, non-fatal stroke, and death from cardiovascular causes during the entire follow-up period. A range of secondary composite end points were also included. Major safety end points included thrombolysis in myocardial infarction (TIMI) major bleeding not related to coronary artery bypass graft (CABG), non-CABG-related TIMI life-threatening bleeding, and TIMI major bleeding (a fall in haemoglobin of 5 g/100 ml or more) or minor bleeding (a fall in haemoglobin of 3 g to less than 5 g/100 ml).

Wiviott et al. (2011) reported the results for a core clinical cohort of the TRITON‑TIMI 38 population (n=10,084). This cohort consisted of patients younger than 75 years, weighing 60 kg or more, and with no history of stroke or transient ischaemic attack. This subpopulation was described in NICE's technology appraisal guidance 182 on prasugrel for the treatment of acute coronary syndromes with percutaneous coronary intervention as the 'target population' or 'licensed population' for treatment with prasugrel.

The Assessment Group noted that there were no trials directly comparing prasugrel with ticagrelor. It considered performing an indirect comparison between prasugrel and ticagrelor using data from TRITON-TIMI 38 and the PLATO trial, with clopidogrel as the common comparator.

NICE's technology appraisal guidance on ticagrelor for the treatment of acute coronary syndromes recommends ticagrelor as a treatment option (in combination with low-dose aspirin) for up to 12 months in adults with acute coronary syndromes, including people with STEMI who are to be treated with percutaneous coronary intervention, people with NSTEMI or people with unstable angina. These recommendations were based on a single randomised controlled trial (PLATO). The PLATO trial was an international, multicentre, double-blind, double-dummy phase 3 trial comparing ticagrelor plus aspirin with clopidogrel plus aspirin in 18,624 patients admitted to hospital with acute coronary syndromes with or without STEMI. Patients were randomised to the trial irrespective of planned intervention and therefore the patient population included patients with acute coronary syndromes who were to be medically managed as well as those who were to have percutaneous coronary intervention. The trial follow up was for 12 months, but the trial protocol stipulated that once the requisite number of events (1780) had accrued, patients had to leave the trial after their 6‑month or 9‑month visit.

In the overall trial population, a statistically significant benefit of ticagrelor was found for the primary composite end point (ticagrelor 9.8% and clopidogrel 11.67%, hazard ratio 0.84 [95% CI 0.77 to 0.92], p<0.001). When the individual components of the composite end point were disaggregated, the reduction in the primary end point was driven by statistically significant reductions in death from vascular causes (hazard ratio 0.79 [95% CI 0.69 to 0.91], p=0.001) and myocardial infarction (hazard ratio 0.84 [95% CI 0.75 to 0.95], p=0.005). There were no statistically significant differences between the 2 arms of the trial for the end points of major bleed (primary safety end point) and major fatal or life‑threatening bleed. However, statistically significant differences in favour of clopidogrel were evident for the end points of total major and minor bleed (hazard ratio 1.11 [95% CI 1.03 to 1.20], p=0.008) and non-CABG-related major bleed (hazard ratio 1.19 [95% CI 1.02 to 1.38], p=0.03).

The Assessment Group stated that the TITRON-TIMI 38 and PLATO trials were not comparable, and so a comparison between prasugrel and ticagrelor based on these trials was inappropriate. It noted that TRITON-TIMI 38 and PLATO were similar in many ways: both trials were conducted in a population with acute coronary syndromes, used clopidogrel as a comparator and reported the same primary composite efficacy end point (death from cardiovascular causes, non‑fatal myocardial infarctions, or non‑fatal stroke during the follow‑up period). However, there were substantial differences in several characteristics of the studies. TRITON-TIMI 38 included patients with acute coronary syndromes who were early invasively managed and scheduled for percutaneous coronary intervention within 72 hours of symptom onset, whereas PLATO included a broad acute coronary syndromes population with symptom onset within 24 hours. TRITON-TIMI 38 only allowed a 300 mg loading dose of clopidogrel whereas PLATO allowed a 300 mg or 600 mg loading dose. TRITON-TIMI 38 had a 15‑month follow‑up period, whereas PLATO had a 12‑month follow‑up period.

The Assessment Group highlighted that the manufacturer of prasugrel had not provided an indirect comparison of prasugrel with ticagrelor in its submission for this appraisal. Although the Assessment Group considered an indirect comparison to be inappropriate, it identified 4 publications reporting an indirect comparison of prasugrel with ticagrelor based on TRITON-TIMI 38 and the PLATO trials. These publications reported that there were no statistically significant differences in overall death, non-fatal myocardial infarction, non-fatal stroke or their composite. Prasugrel was associated with a statistically significantly lower risk of stent thrombosis, and ticagrelor was associated with a statistically significantly lower risk of any major bleeding and major bleeding associated with cardiac surgery.

The Assessment Group identified an ongoing trial designed to assess whether ticagrelor is superior to prasugrel in patients with acute coronary syndromes and planned invasive strategy (ISAR-REACT 5). This study is due to complete in October 2018 (the final data collection date for primary outcome measure is October 2016). The results of this study should allow a formal comparison of the efficacy of prasugrel compared with ticagrelor.

A professional group commented that although evidence supporting the use of prasugrel in STEMI treated by primary percutaneous coronary intervention is unclear, the data supporting the use of prasugrel for STEMI treated by a variety of means (as still occurs in the UK) remain strong. As such, there is no robust new evidence to challenge the original guidance defining the patient population for whom prasugrel should be a treatment option. The group stated that the benefit of prasugrel therapy is limited to patients younger than 75 years, weighing more than 60 kg and without a history of transient ischaemic attack or stroke. It commented that data produced after NICE's technology appraisal guidance 182 on prasugrel for the treatment of acute coronary syndromes with percutaneous coronary intervention do not support extending the use of prasugrel to all patients with non-ST-elevation acute coronary syndromes, whether treated medically or by urgent revascularisation. It noted evidence from subgroup analyses for efficacy of prasugrel over clopidogrel in the reduction of stent thrombosis in patients receiving an intracoronary stent for treatment of acute coronary syndromes.

The professional group commented that since the publication of NICE's technology appraisal guidance 182 on prasugrel for the treatment of acute coronary syndromes with percutaneous coronary intervention, few further data have become available about prasugrel in the context of non-ST-elevation acute coronary syndromes and that no new data are available for STEMI. It stated that although prasugrel reduces major adverse cardiac events and stroke in patients with acute coronary syndromes at high risk of receiving percutaneous coronary intervention with stents, there is no evidence that prasugrel reduces mortality. The professional group commented that prasugrel should not be offered to people aged 75 years or older, weighing less than 60 kg or with a past history of transient ischaemic attack or stroke. However, it stated that prasugrel should remain a treatment option in patients with STEMI treated by primary percutaneous coronary intervention or by other means and in patients with diabetes and acute coronary syndromes of any variety (STEMI, NSTEMI or unstable angina).

The manufacturer submitted an economic model similar to the model described in NICE's technology appraisal guidance 182 on prasugrel for the treatment of acute coronary syndromes with percutaneous coronary intervention. The economic model had a Markov model structure with 2 phases. The first phase spanned the duration of the TRITON-TIMI 38 trial and the second phase extrapolated outcomes and costs beyond the trial, up until death or a time horizon of 40 years. Rather than using data from the trial directly in the model, separate risk equations for primary end point events were obtained from individual patient data from the TRITON-TIMI 38 trial. These risk equations were then used to model events and hospitalisation. Mortality was modelled based on adjustment of population life tables, to reflect the impact on longer-term mortality of the events modelled over the short term. Patients entered the model at the point of experiencing an acute coronary syndromes event, immediately before having percutaneous coronary intervention. Some aspects of the submitted economic model had been updated based on feedback during the original appraisal of prasugrel. These revisions included:

• the sensitivity analysis encompassing the entire population as opposed to a 'typical' patient profile

• scenario analysis using the Evidence Review Group's suggestions for utility values, amended long-term relative risk of mortality and reduced incidence of non-fatal myocardial infarction

• the (reduced) price of generic clopidogrel

• updated costs.

Although the TRITON-TIMI 38 trial included a health‑related quality of life sub‑study, the manufacturer stated that it was not possible to provide robust health‑related quality of life estimates because of the very small numbers of patients with events included in the analysis. Therefore, the manufacturer conducted a systematic review of the literature to identify studies relevant to the modelled trial population. Mean utility decrements for acute coronary syndromes (0.041) and stroke or myocardial infarction (0.052) were taken directly from a US study (Sullivan et al., 2006) that was designed to produce a specific list of preference weights for use in economic evaluations; the study used the US version of the EQ‑5D. To calculate utility weights for use in the economic evaluation, background UK population norms (free of disease) varied by age and sex, as described by Kind et al. (1999), were applied to all patients in the study. The utility decrements for acute coronary syndromes and stroke or myocardial infarction were then used alongside these background utility estimates. Finally, the manufacturer assumed that for a major bleed a decrement of 25% of the population (utility) norm was applicable for a 14‑day period.

The key categories of cost estimates in the manufacturer's submission were related to hospitalisations and drug costs. Only hospitalisations related to end points or to serious adverse events needing re-hospitalisation and potentially related to the acute coronary syndrome condition or the percutaneous coronary intervention were included in the manufacturer's cost analysis. Re-hospitalisations were valued at a weighted average unit cost per hospitalisation (using NHS reference costs) and differences in hospitalisation rates were applied by geographic location. The weighted average unit cost per hospitalisation was £3,070 for clopidogrel and £3,081 for prasugrel. Patients were assumed to be treated with either aspirin and clopidogrel or aspirin and prasugrel for 12 months. The cost for prasugrel was calculated at £10.20 per day for the loading dose and £1.70 per day for the maintenance dose (MIMS August 13, based on £47.56 per pack of 28 tablets). The cost of clopidogrel was calculated at £0.24 per day for the loading dose and £0.07 per day for the maintenance dose (NHS Drug Tariff, based on £1.83 per pack of 28 75 mg tablets). The cost of aspirin was calculated as £0.01 per day for the maintenance dose.

The manufacturer considered 5 subgroups in its cost-effectiveness analysis, resulting in the following incremental cost-effectiveness ratios (ICERs):

• For the all acute coronary syndromes licensed population (excluding prior stroke or transient ischaemic attack and including patients who are now recommended to be treated with a 5 mg maintenance dose), the ICER was £11,660 per quality‑adjusted life year (QALY) gained.

• For the core clinical cohort (excluding prior stroke or transient ischaemic attack and those weighing less than 60 kg, or aged 75 years or older), the ICER was £11,796 per QALY gained.

• For the unstable angina or NSTEMI licensed population (excluding patients with prior stroke or transient ischaemic attack and including patients who are now recommended to be treated with a 5 mg dose), the ICER was £15,452 per QALY gained.

• For the STEMI licensed population (excluding patients with prior stroke or transient ischaemic attack and including patients who were recommended to be treated with a 5 mg dose), the ICER was £6,987 per QALY gained.

• For the acute coronary syndromes licensed population with diabetes (excluding patients with prior stroke or transient ischaemic attack and including patients who were recommended to be treated with a 5 mg dose) the ICER was £4,675 per QALY gained.

The manufacturer carried out a series of one-way deterministic sensitivity analyses on the core clinical cohort population (excluding prior stroke or transient ischaemic attack and those 75 years or older or weighing less than 60 kg). The following changes to the model resulted in ICERs of more than £13,000 per QALY gained (an increase of more than £1,000 from the base-case ICER):

• discounting at 6% for both costs and effects (ICER £16,475 per QALY gained)

• relative risks for all-cause mortality associated with clinical events reduced by 50% (ICER £20,619 per QALY gained)

• clopidogrel pre-loading adjustment set at 70% (ICER £13,959 per QALY gained).

The manufacturer did not carry out probabilistic sensitivity analysis.

The manufacturer also carried out a series of scenario analyses for each of the following:

• using alternative values obtained from the Health Outcomes Data Repository (HODaR) database

• amending the long-term relative risks of mortality (by ignoring the impact of acute coronary syndromes before events that occurred in the TRITON-TIMI 38 trial)

• reducing the incidence of non-fatal myocardial infarctions such that the underlying rate was 50% of that of the trial.
  
  The results of the scenario analyses showed that when using alternative utility values, relative risks for mortality and myocardial infarction, the ICER for prasugrel compared with clopidogrel remained below £20,000 per QALY gained.

The Assessment Group provided a critique of the manufacturer's economic model. It stated that in the long-term component of the model, there was an assumption that differences established between the prasugrel and clopidogrel treatment arms of the TRITON-TIMI 38 trial will be preserved indefinitely at the level observed at the end of the trial. However, the Assessment Group considered that there is no reason to believe that further serious non-fatal events will not continue to occur to patients in both cohorts. It stated that if events during the trial are presumed to influence later survival, then it is also likely that any such events in subsequent periods will also have important effects. Because active treatment with clopidogrel or prasugrel will have stopped, it can be expected that event rates will be similar in both treatment arms. The Assessment Group commented that as a result of this process, it is likely that over time the disease histories of patients will converge, and so any initial advantage for either treatment will progressively decrease.

The Assessment Group carried out a systematic review to identify existing economic evaluations of prasugrel. Of the 15 potentially eligible references identified, none of the papers met the full inclusion criteria that had been set. The review identified the 3 studies included in the manufacturer's review of cost‑effectiveness evidence – two from a US perspective and one which used the economic model originally submitted during NICE's technology appraisal guidance 182 on prasugrel for the treatment of acute coronary syndromes with percutaneous coronary intervention but these were excluded by the Assessment Group.

The Assessment Group developed a 2-phase economic model: a short-term statistical model of the data from the TRITON-TIMI 38 trial and a long-term model projecting outcomes and costs at the end of the first phase up to a maximum of 40 years. The model compared dual antiplatelet therapy for 12 months from time of percutaneous coronary intervention with either clopidogrel in combination with low-dose aspirin or prasugrel in combination with low-dose aspirin. As a result of the lack of clinical evidence available, the Assessment Group did not compare prasugrel with ticagrelor.

The Assessment Group accepted the manufacturer's statistical model for the initial phase (up to 12 months) but replaced the long-term projection with a more detailed representation of subsequent cardiovascular events, accumulating patient histories, alteration in health states and associated care costs, as well as health-related quality of life.

The Assessment Group's decision model assessed 4 mutually exclusive subgroups of the core clinical cohort (that is, all patients with acute coronary syndromes, treated with percutaneous coronary intervention, excluding those with a history of stroke or transient ischaemic attack, those aged 75 years or older or weighing less than 60 kg). The 4 subgroups were:

• treated for STEMI and with diagnosed diabetes

• treated for STEMI and without diagnosed diabetes

• treated for unstable angina or NSTEMI and with diagnosed diabetes

• treated for unstable angina or NSTEMI and without diagnosed diabetes.
  
  Specific clinical data relating to patients with STEMI, unstable angina or NSTEMI or diabetes in the core clinical cohort were not available from the manufacturer's submission or the most recent publication. In place of these data, the Assessment Group extracted the outcomes from the manufacturer's short-term model for the 4 mutually exclusive subgroups of the core clinical cohort and used these as the initial conditions for surviving patients entering the Assessment Group's long-term state-transition model.

For the long-term projection, the Assessment Group used a modified version of the economic model described in NICE's technology appraisal guidance on clopidogrel and modified-release dipyridamole for the prevention of occlusive vascular events. This model used data provided by the manufacturer of clopidogrel from the CAPRIE clinical trial, supplemented by data provided by the manufacturer of dipyridamole from the PROFESS clinical trial. The Assessment Group stated that the myocardial infarction subpopulation analysis submitted for NICE's technology appraisal guidance on clopidogrel and modified-release dipyridamole for the prevention of occlusive vascular events addresses issues similar to those in this appraisal and was based largely on data from the myocardial infarction subpopulation in the CAPRIE trial. In order to reduce the time in generating model results from an individual patient simulation approach, the Assessment Group restructured the economic model submitted for NICE's technology appraisal guidance on clopidogrel and modified-release dipyridamole for the prevention of occlusive vascular events into a long-term Markov chain for the purpose of this appraisal.

In the Assessment Group's long‑term model the initial health state was determined by the worst previous event (none, myocardial infarction or stroke), the number of prior events (none, 1, 2, or 3 or more) and whether the event was disabling or not. Moving into another health state was determined by whether the patient experienced non-fatal myocardial infarction, non‑fatal haemorrhagic stroke (disabling or non‑disabling), or ischaemic stroke or transient ischaemic attack (disabling or non‑disabling). The patient may also have experienced no event during the year, and therefore stay in the same health state, or may have died if they experienced fatal myocardial infarction, fatal haemorrhagic stroke, fatal ischaemic stroke or transient ischaemic attack, other vascular death, or non‑vascular death.

The main source of data used to populate the Assessment Group's long-term model was the CAPRIE clinical trial. The Assessment Group stated that its clinical adviser had confirmed that CAPRIE data was an appropriate trial source for extrapolating long-term vascular events and that no better source had become available since 2010. The CAPRIE trial was a double-blind placebo comparison of clopidogrel with aspirin involving 19,185 patients with atherosclerotic vascular diseases manifested as either ischaemic stroke, myocardial infarction or symptomatic peripheral arterial disease. Only CAPRIE data from 5,741 patients with myocardial infarctions and without a history of other vascular events were used to populate the Assessment Group's long-term model. The manufacturer of clopidogrel carried out extensive re-analyses of the CAPRIE trial data as requested by the Assessment Group in order to estimate independent event hazards adjusted to age, sex and event history.

The Assessment Group's economic model included costs calculated from the pack price for 28 tablets of clopidogrel, prasugrel and low-dose aspirin as stated in the NHS Drug Tariff (November 2013; clopidogrel and low-dose aspirin) and the BNF (October 2013, edition 66; prasugrel). Clopidogrel has a pack price of £1.71, giving a cost of £0.24 per loading dose, £18.43 for a 12‑month supply (adjusted for treatment duration), and £29.37 for the total dual antiplatelet therapy cost in year 1. Prasugrel has a pack price of £47.56, giving a cost of £10.19 per loading dose, £511.67 for a 12‑month supply (adjusted for treatment duration) and £532.56 for the total dual antiplatelet therapy cost in year 1. Low-dose aspirin has a pack price of £0.82, giving a cost of £10.70 for a 12‑month supply and annual maintenance cost.

The Assessment Group used the same unit costs as used in NICE's technology appraisal guidance on clopidogrel and modified-release dipyridamole for the prevention of occlusive vascular events, updated to 2012 prices using the Hospital and Community Health Services inflation index. The unit cost of a fatal myocardial infarction is £2,373.68 (standard error £121.11, 95% CI £2,136.31 to £2,611.05), a non-fatal myocardial infarction is £6,165.21 (standard error £314.55, 95% CI £5,548.69 to £6,781.73), a fatal stroke is £9381.43 (standard error £478.64, 95% CI £8,443.29 to £10,319.57), a non‑fatal non‑disabling stroke is £6,858.64 (standard error £349.93, 95% CI £6,172.77 to £7,544.50), a non-fatal disabling stroke is £14,602.70 (standard error £754.04, 95% CI £13,142.43 to £16,062.97), and a non‑vascular or other vascular death is £2,407.50 (standard error £122.83, 95% CI £2,166.75 to £2,648.25). The annual cost in the event-free or myocardial infarction‑only health state is £618.03 (standard error £31.53, 95% CI £556.23 to £679.84). In the non-disabling stroke health state it is £1,804.06 (standard error £92.04, 95% CI £1,623.66 to £1,984.47) and in the disabling stroke health state it is £5,537.72 (standard error £282.54, 95% CI £4,983.95 to £6,091.50).

The Assessment Group obtained the continuing health-state EQ-5D utility value for patients who were event-free or suffered a non-fatal myocardial infarction (but no strokes) and who were alive 12 months after percutaneous coronary intervention from the economic sub-study of the PLATO clinical trial. The values were based on the weighted average of patients meeting those criteria. Utility parameters reflecting sex differences and mild versus severe strokes for patients suffering at least 1 stroke or transient ischaemic attack were obtained from a study of EQ-5D observations as part of the Oxford Vascular Study (OXVASC). The Assessment Group used an annual loss of utility estimated from the UK population EQ-5D norms, which was calculated by fitting a linear regression trend line to all people aged 35 years or older. The results were used to adjust the initial health state utilities of each subgroup for the differences in mean age between the TRITON-TIMI 38 cohort and OXVASC data. This decrement was also applied annually to the results of the model to reflect the average decline of utility score with increasing age.

The Assessment Group commented that 7 events in its model would be expected to result in additional utility decrement in the first year of follow up during early recovery. The Assessment Group identified a specific value for non-fatal myocardial infarction using an analysis of UK Prospective Diabetes Study trial results which compared utility values for events occurring within 12 months against those occurring earlier. Sources for non-fatal stroke parameters gave contradictory figures, suggesting that there is no clear additional early disutility effect beyond the long-term continuing effect of a stroke. As a result, the Assessment Group set these parameters to zero and conducted one-way sensitivity analyses on these parameters instead. No sources were identified for utility values for fatal myocardial infarction, fatal stroke, other vascular death or non-vascular death. The Assessment Group assigned these parameters a notional value of −0.1 and conducted sensitivity analyses.

The Assessment Group discounted costs and outcomes annually at 3.5% and carried out one-way sensitivity analyses using 0% and 6% discount rates for both costs and outcomes. The Assessment Group's model generated results at the end of every year from trial randomisation. However, deterministic results were reported at 1, 5, 10, 20 and 40 years, and probabilistic results at 5 and 40 years.

The Assessment Group assumed that follow-up secondary prophylaxis was limited to low-dose aspirin. The Assessment Group stated that it made this assumption for convenience and to avoid the possibility of obscuring the primary comparison between prasugrel and clopidogrel use. For the same reason, the Assessment Group did not incorporate other post-stroke and post-myocardial infarction care including surgery and other medication options. The Assessment Group did not incorporate the adverse effects of aspirin therapy, the possibility that people stop aspirin treatment or the risk of bleeding events associated with long-term prophylaxis, because all these issues would affect patients in both treatment arms and the incremental difference would be expected to be minimal.

The Assessment Group considered that it was inappropriate to calculate an ICER for the overall core clinical cohort. It reported separate results from the model for the 4 patient subgroups: STEMI with diabetes, STEMI without diabetes, unstable angina or NSTEMI with diabetes and unstable angina or NSTEMI without diabetes. For each subgroup, deterministic cost-effectiveness results were presented at 1, 15, 10, 20 and 40 years after the initial percutaneous coronary intervention. Probabilistic cost-effectiveness results were presented at 5‑year and 40‑year follow up.

For the STEMI with diabetes subgroup, the ICER was £31,915 per QALY gained at year 1 of follow up (incremental cost £230, incremental QALYs 0.007). The ICER decreased to £4,603 per QALY gained at year 5 (incremental cost £269, incremental QALYs 0.059), £2,139 per QALY gained at year 10 (incremental cost £275, incremental QALYs 0.129), and at year 40 it was £1,640 per QALY gained (incremental cost £447, incremental QALYs 0.272). The Assessment Group conducted one-way sensitivity analyses, the results of which indicated that uncertainty from individual model parameters had a minor effect on the ICER in this subgroup. Varying discount rates for costs and outcomes had the largest effect on the ICER, but it remained within the range of £1,000 to £2,500 per QALY gained. Probabilistic analysis at 40 years of follow up for this subgroup resulted in a higher estimated ICER of £1,732 per QALY gained (incremental cost £515 and incremental QALYs 0.297).

For the STEMI without diabetes subgroup, the ICER at year 1 of follow up was £224,302 per QALY gained (incremental cost £422, incremental QALYs 0.002). The ICER decreased to £29,607 per QALY gained at year 5 of follow up (incremental cost £465, incremental QALYs 0.016), £13,370 per QALY gained at year 10 (incremental cost £482, incremental QALYs 0.036), and at year 40 the ICER had decreased to £6,626 per QALY gained (incremental cost £555, incremental QALYs 0.084). The Assessment Group conducted one-way sensitivity analyses, the results of which indicated that uncertainty from individual model parameters had a minor effect on the ICER in this subgroup. Varying discount rates for costs and outcomes had the largest effect on the ICER, but it remained within the range of £4,000 to £9,000 per QALY gained. Probabilistic analysis at 40 years of follow up resulted in a higher estimated ICER of £7,073 per QALY gained obtained from small incremental cost and QALY estimates (incremental cost £609 and incremental QALYs 0.086).

For the unstable angina or NSTEMI with diabetes subgroup, the ICER at year 1 of follow up was £76,856 per QALY gained (incremental cost £259, incremental QALYs 0.003). At year 5 of follow up, the ICER for prasugrel decreased to £2,846 per QALY gained (incremental cost £96, incremental QALYs 0.034). From year 10 and beyond prasugrel dominated (that is, was less costly and more effective than) clopidogrel (at year 40 of follow up: incremental cost −£77, incremental QALYs 0.176). The Assessment Group undertook one-way sensitivity analyses, which indicated that uncertainty from event incidence and fatality rates had the largest effect on the estimated ICER (ranging between −£1,000 and £400 per QALY gained). Probabilistic analysis at 40 years of follow up confirmed that prasugrel dominated clopidogrel with a small net cost saving and positive incremental benefit (incremental cost −£120 and incremental QALYs 0.191).

For the unstable angina or NSTEMI without diabetes subgroup the ICER for prasugrel compared with clopidogrel was £1,101,662 per QALY gained at the end of the first year, as a result of the inclusion of the full additional cost of treatment (incremental cost £413, incremental QALYs 0.00037). The ICER for prasugrel decreased to £52,288 per QALY gained at year 5 of follow up (incremental cost £346, incremental QALYs 0.007), £14,276 per QALY gained at year 10 (incremental cost £280, incremental QALYs 0.020) and at year 40 the ICER decreased to £4,667 per QALY gained (incremental cost £248, incremental QALYs 0.053). The Assessment Group undertook one-way sensitivity analyses, which indicated that uncertainty from event incidence and fatality rates had the largest effect on the estimated ICER (ranging between £2,500 and £6,500 per QALY gained). Probabilistic analysis at 40 years of follow up resulted in a lower estimated ICER of £4,154 per QALY gained (incremental cost £212 and incremental QALYs 0.051).

Table 1 illustrates the differences in the cost-effectiveness estimates for the 3 different models: the Evidence Review Group's exploratory analyses from NICE's technology appraisal guidance 182 on prasugrel for the treatment of acute coronary syndromes with percutaneous coronary intervention, the manufacturer's model, and the Assessment Group's model for the current appraisal.

The Committee discussed the clinical management of acute coronary syndromes. The Committee heard from the clinical specialists that current management is in line with NICE's technology appraisal guidance 182 on prasugrel for the treatment of acute coronary syndromes with percutaneous coronary intervention as well as NICE's technology appraisal guidance on ticagrelor for the treatment of acute coronary syndromes, Myocardial infarction with ST-segment-elevation: The acute management of myocardial infarction with ST-segment-elevation (now replaced by NICE's guideline on acute coronary syndromes) and Unstable angina and NSTEMI: early management (now replaced by NICE's guideline on acute coronary syndromes). The Committee understood that, in England, treatment options for people with STEMI are prasugrel in combination with aspirin, ticagrelor in combination with low-dose aspirin or clopidogrel in combination with low-dose aspirin, along with percutaneous coronary intervention followed by dual antiplatelet treatment. It also understood that people in England with NSTEMI are offered treatments depending on their Global Registry of Acute Coronary Events (GRACE) or TIMI score. Medical management using aspirin is an option for people at the lowest risk of future adverse cardiovascular events, whereas people at higher risk are offered percutaneous coronary intervention along with either ticagrelor, or clopidogrel and subsequent dual antiplatelet therapy with clopidogrel and aspirin. In patients with NSTEMI and diabetes, prasugrel is an alternative to clopidogrel or ticagrelor. The Committee heard from patient experts and clinical specialists that, overall, prasugrel is a useful addition to the treatment options available. It has a potentially key advantage over clopidogrel because of its faster antiplatelet action. However, the Committee also noted that prasugrel increased the chance of (potentially fatal) bleeding compared with clopidogrel. The Committee also heard from the clinical specialists that there was variation in opinion among clinicians in England as to which of prasugrel, ticagrelor or clopidogrel should be considered the standard treatment for all patients with STEMI who have a percutaneous coronary intervention because of the chance of increased bleeding with these treatments. The Committee recognised that antiplatelet therapy such as prasugrel was a valued treatment option.

The Committee considered the economic models submitted by the manufacturer and the Assessment Group. The Committee considered the cost‑effectiveness results for the comparison of prasugrel and clopidogrel. The Committee noted that the manufacturer and the Assessment Group had presented cost‑effectiveness results (for the 10 mg dose of prasugrel only) for different populations and so the results could not be directly compared, although all of the results for prasugrel 10 mg compared with clopidogrel were below £20,000 per QALY gained at a time horizon of 40 years. The Committee then discussed which of the cost-effectiveness data it should consider in its decision making. Bearing in mind its considerations on clinical effectiveness (see sections 4.3.7, 4.3.8, 4.3.9, and 4.3.10), the Committee agreed that an advantage of prasugrel over clopidogrel was plausible in patients without an increased risk of bleeding (under the age of 75 and weighing more than 60 kg) if they had ST‑segment‑elevation or had diabetes (with STEMI or NSTEMI). However, in people with NSTEMI without diabetes, it was less certain whether there was an advantage of prasugrel over clopidogrel. The Committee noted that the Assessment Group had presented cost‑effectiveness results for these subgroups and concluded that these results were the most appropriate for the cost-effectiveness decision‑making.

The Committee considered the cost-effectiveness results from the Assessment Group's economic model. It noted that the Assessment Group had provided cost‑effectiveness results for each of the subgroups at time horizons of 1, 5, 10 and 40 years. It noted that all of the ICERs at 10 or 40 years were below £20,000 per QALY gained (ranging from prasugrel dominating [that is, it was less costly and more effective than clopidogrel] to £14,276 per QALY gained at 10 years, and from prasugrel dominating to £6,626 per QALY gained at 40 years), however, at 5 years the ICERs for people with STEMI without diabetes and NSTEMI without diabetes were over £20,000 per QALY gained (£29,607 and £52,288 per QALY gained respectively). The Committee noted that the QALY gains for prasugrel over the 40‑year time horizon for the STEMI without diabetes and the unstable angina or NSTEMI without diabetes subgroups were small (0.084 and 0.053 respectively), as was the difference in costs between prasugrel and clopidogrel treatment (STEMI without diabetes: clopidogrel total cost £21,167, prasugrel total cost £21,722, incremental cost £555; UA/NSTEMI without diabetes: clopidogrel total cost £20,328, prasugrel total cost £20,576, incremental cost £248). It accepted that, as a result, the cost effectiveness of prasugrel was highly sensitive to changes in key model assumptions. The Committee therefore considered which time horizon was the most appropriate for its decision-making. The Committee noted that the clinical data used in the second phase of the Assessment Group's economic data were obtained from the CAPRIE study, which had a maximum follow up of 3 years, and that these data were used to extrapolate results of the health economic analysis up to 40 years. The Committee agreed that, although the extrapolation of short-term clinical data over longer time horizons could only increase overall uncertainty, it is necessary in economic modelling and longer time horizons are generally preferable. The Committee noted that the time horizon within which the results would fall to within the range usually considered to be cost effective by NICE (£20,000 to £30,000 per QALY gained) is likely to be much less than 40 years. The Committee considered that despite uncertainty in the ICERs arising from this 40 year extrapolation, the results were sufficiently robust to permit their use, and concluded that the 40 year time horizon was the most appropriate for decision making.

The Committee considered the ICERs for a 10 mg dose of prasugrel compared with clopidogrel in patients with STEMI with and without diabetes. The Committee noted that, for patients with STEMI and diabetes, the 40‑year ICER from the Assessment Group's model was £1,600 per QALY gained. For patients with STEMI and without diabetes, the 40‑year ICER from the Assessment Group's model was £6,600 per QALY gained. The Committee concluded that prasugrel was a cost-effective use of NHS resources compared with clopidogrel for treating people with STEMI with and without diabetes.

The Committee considered the ICER for a 10 mg dose of prasugrel compared with clopidogrel in patients with unstable angina or NSTEMI with and without diabetes. The Committee noted that for patients with unstable angina or NSTEMI and diabetes, prasugrel dominated clopidogrel at the 40‑year time horizon. For patients with unstable angina or NSTEMI without diabetes, the 40‑year ICER from the Assessment Group's model was £4,700 per QALY gained. It accepted that the difference between the TRITON-TIMI 38 trial and clinical practice in the time delay between diagnosis and treatment meant that there is uncertainty about whether the results of the TRITON-TIMI 38 trial can be generalised to patients with unstable angina or NSTEMI in England. However, taking into account the slight advantage of prasugrel over clopidogrel beyond the period when preloading is relevant, and that the ICERs are well within the range usually considered to be cost effective by NICE (£20,000 to £30,000 per QALY gained), the Committee concluded that prasugrel can be considered a cost-effective use of NHS resources compared with clopidogrel for treating people with unstable angina or NSTEMI with and without diabetes.

Regarding the cost‑effectiveness of prasugrel compared with ticagrelor, the Committee noted that neither the Assessment Group nor the manufacturer had included ticagrelor in their economic modelling. Given the lack of a cost‑effectiveness analysis, the Committee concluded that it was unable to calculate a precise ICER for this comparison. It agreed, however, that given the similar cost‑effectiveness of prasugrel compared with clopidogrel and that of ticagrelor compared with clopidogrel, and their similar treatment costs, it was reasonable to accept that prasugrel is similarly cost‑effective to ticagrelor.