4 Evidence submissions

The evaluation committee (section 7) considered evidence submitted by the manufacturer of ataluren, a review of this submission by the evidence review group (ERG) and evidence submitted by clinical experts, patient experts and NHS England. It also considered a proposed managed access agreement that was developed by several stakeholders.

Nature of the condition

4.1 Evidence from patient experts and patient groups highlighted the substantial impact of Duchenne muscular dystrophy (DMD) on the quality of life of people with the condition and their families:

  • People with DMD have a loss of motor function until eventually they become wheelchair dependent, making it difficult to participate in normal activities at home or at school with siblings, family and friends. Parents and carers describe the frustrations experienced by their child when they cannot take part in games with their peers. Often, younger children do not understand the implications of the disease and why it makes them different.

  • As the disease progresses, people with DMD lose the ability to breathe unaided and need assisted ventilation. Scoliosis develops as the back muscles weaken, for which surgery is needed. Parents and carers of people with DMD describe the importance of maintaining their child's ability to walk for as long as possible because loss of walking is an indication of disease progression.

  • Parents and carers of people with DMD describe the emotional impact of the short life expectancy of people with DMD. They describe the sadness, anxiety and depression of knowing their child will probably die at a young age. The devastating impact of the disease and its prognosis often leads to isolation from friends and family members.

  • Parents and carers described the financial impact of looking after a person with DMD. They described giving up work to look after their child full time. In addition, out-of-pocket expenses can be very high (for example, moving house to ensure the home is wheelchair accessible).

Clinical evidence

Study 007

4.2 The clinical evidence in the company's original submission focused on the safety and efficacy of ataluren, which was investigated in a phase IIb double-blind randomised placebo-controlled trial (Study 007). Study 007 included 174 male patients with nonsense mutation DMD aged 5 years and older. Patients were recruited from 37 study sites in 11 countries and included 7 patients from the UK. They were randomised to ataluren at a total daily dosage of 40 mg/kg (n=57) or 80 mg/kg (n=60), or to placebo (n=57), all for 48 weeks. The primary outcome was change in the patient's ability to walk on a hard, flat surface measured using the 6‑minute walk distance (6MWD). The study compared the mean change in 6MWD from baseline to week 48 measured in the placebo group with that in the ataluren group. The secondary outcomes included change in proximal muscle function measured by timed function tests, and change in force exerted during knee flexion and extension. Quality of life was assessed using the Pediatric Quality of Life (PedsQL) inventory, which contains 4 scales: physical, emotional, social and school functioning.

4.3 The pre-specified subgroups in Study 007 were: age (less than 9 years and 9 years and older), corticosteroid use (yes or no) and baseline 6MWD (350 m or less and greater than 350 m). The company conducted a post-hoc subgroup analysis in patients who were classified as being in the decline phase (n=31 in the placebo group and n=32 in the ataluren group). The decline phase was defined as patients aged 7–16 years with a baseline 6MWD test of 80% or less of that predicted and, to minimise heterogeneity, a baseline 6MWD of 150 m or more on a stable dose of corticosteroids. The decline phase was considered clinically important because patients younger than 7 years tend to increase their 6MWD over 48 weeks because of normal developmental improvements in walking.

4.4 The intention-to-treat analysis showed no statistically significant difference between ataluren and placebo in the change in 6MWD from baseline to 48 weeks. In the corrected intention-to-treat analysis, baseline values for 2 patients (1 taking placebo and 1 taking ataluren 80 mg/kg) were replaced by their values at screening because the patients had lower-limb injuries before the baseline test. In this analysis, there was a mean observed difference at 48 weeks of 31.3 m between ataluren 40 mg/kg and placebo (−12.9 m and −44.1 m respectively). In a mixed model for repeated measures analysis, the estimated mean difference between ataluren 40 mg/kg and placebo was 31.7 m (95% confidence interval [CI] 5.1 to 58.3, p=0.0197). No effect was seen in the ataluren 80 mg group.

4.5 In the post-hoc subgroup analysis for patients in the decline phase, patients having ataluren experienced a statistically significantly smaller reduction in 6MWD compared with patients having placebo (difference in mean change in 6MWD of 45.6 m, p=0.0096). In the pre-specified group of patients with a baseline 6MWD of less than 350 m, there was a statistically significantly smaller reduction in 6MWD in the ataluren group compared with the placebo group (difference in mean change in 6MWD of 59.8 m, p=0.0053).

4.6 There were no statistically significant differences in quality of life between the ataluren and placebo groups. The company stated there was a positive trend towards improved quality of life with ataluren 40 mg/kg daily in the physical functioning subscale. The company submission also described a positive effect on school functioning and a negative trend in emotional and social subscales.

4.7 The company reported that the number of adverse events was similar in the ataluren and placebo treatment groups in Study 007. None of the patients stopped treatment with ataluren or withdrew from the study because of a treatment-related adverse event, and there were no deaths reported. The most common treatment emergent adverse events reported were: gastrointestinal disorders (73.7% of patients in the ataluren 40 mg/kg group and 37% in the placebo group), vomiting (56.1% in the ataluren 40 mg/kg group and 45% in the placebo group) and diarrhoea (19.3% in the ataluren 40 mg/kg group and 28.3% in the placebo group).

Study 020

4.8 In its response to consultation, the company responded to all the committee's requests described in the evaluation consultation document, including the results of the multicentre randomised double-blind placebo-controlled confirmatory study (PTC124‑GD‑020‑DMD; Study 020).

4.9 Study 020's intention-to-treat population included 228 male patients with nonsense mutation DMD aged between 7 years and 16 years. At baseline, patients had to have a 6MWD of more than 150 m but 80% or less of that predicted for weight and height. Patients were randomised to ataluren or to placebo for 48 weeks. After this, patients eligible for treatment could have ataluren through an open-label extension study. Patients were stratified based on age, duration of corticosteroid use and baseline 6MWD. The primary outcome was mean change in 6MWD from baseline to week 48 in the placebo group compared with the ataluren group. The secondary outcomes included timed function tests and the North Star Ambulatory Assessment, a clinician-reported outcome instrument consisting of 17 items designed to measure ambulatory function in people with DMD. Quality of life was assessed using the Paediatric Outcomes Data Collection Instrument (PODCI) and Activities of Daily Living Questionnaire. The company stated that pre-specified analyses included a meta-analysis of Study 020 (n=228) and the ambulatory decline phase subgroup of the corrected intention-to-treat population from Study 007 (n=63), and a subgroup analysis of patients from Study 020 with baseline 6MWD of 300–400 m (n=99). The company also presented a meta-analysis of results from a subgroup of patients with a baseline 6MWD of 300–400 m in Study 007 (n=44) and Study 020 (n=99).

4.10 The primary outcome results from Study 020 showed a15 m difference in change in 6MWD at 48 weeks favouring ataluren over placebo in the intention-to-treat population, which was not statistically significant (p=0.213). In the subgroup of patients with a baseline 6MWD of 300–400 m, there was a statistically significant benefit of 47 m for ataluren compared with placebo (p=0.007). The company noted that Study 020 has contributed to improving the clinical knowledge of DMD and clinical trial design for treatments for this condition. Based on this, it considered that, for a 48‑week trial such as Study 020 using 6MWD as a primary outcome, the optimum range in the 6MWD at baseline to detect a significant difference is 300–400 m. The company added that patients with a baseline 6MWD of more than 400 m appear to be too stable and those with a baseline 6MWD of less than 300 m appear to have such severe muscle loss it is not possible to detect a statistically significant treatment effect. Results for the 6MWD from the meta-analysis of the intention-to-treat population from Study 020 and the ambulatory decline phase subgroup of the corrected intention-to-treat population from Study 007 showed a benefit of ataluren compared with placebo (difference in 6MWD of 22 m, p=0.015). The meta-analysis for the subgroup of people with a baseline 6MWD of 300–400 m from Studies 007 and 020 also showed a benefit of ataluren compared with placebo, with a difference of 45 m in the 6MWD (p<0.001). The company reported that, in the subgroup analysis of patients with a baseline 6MWD of 300–400 m, none of the 47 patients (0%) lost their ability to walk in the ataluren group compared with 4 out of 52 patients (8%) in the placebo group at 48 weeks. The number of people who lost their ability to walk in the intention-to-treat population was provided by the company as commercial in confidence and cannot be reported here.

4.11 Results for timed function tests in the intention-to-treat population of Study 020 were provided by the company as commercial in confidence and cannot be presented here. Results for the North Star Ambulatory Assessment in the intention-to-treat population did not show a statistically significant difference for ataluren compared with placebo (p=0.27). In the subgroup of patients with a baseline 6MWD of 300–400 m, ataluren showed benefits compared with placebo in the timed function tests (10 m run/walk, −2.1 seconds, p=0.066; 4 stair climb, −3.6 seconds, p=0.003; and 4 stair descend, −4.3 seconds, p<0.001) and the North Star Ambulatory Assessment (p=0.04).

4.12 The company provided graphical results from Study 020 suggesting improvements in quality of life based on changes in 2 dimensions of the PODCI in patients with a baseline 6MWD 300–400 m having ataluren compared with placebo. It also noted that results with the Activities of Daily Living Questionnaire showed that people having ataluren had greater improvement or less deterioration in walking, stair-climbing and upper extremity activities of self-care than people having placebo (results were presented as commercial in confidence and cannot be reported here). The company also provided the results of a survey on the quality of life of caregivers (n=6). The results showed that DMD had a serious impact on multiple aspects of their life including emotional wellbeing and mental health, personal care and the ability to maintain relationships. The results showed that caregivers felt tired, depressed and anxious and that, in many cases, at least another family member in addition to both parents were involved in giving care (for example, siblings and grandparents).

Economic evidence

Overview of company's cost–consequence models

4.13 Over the course of the evaluation, the company presented several cost–consequence analyses comparing the licensed dose of ataluren (40 mg/kg daily) with best supportive care in people aged 5 years or older who could walk:

  • The first model, which used data from Study 007, was part of the original submission (see sections 4.15–4.19).

  • The second model was submitted in response to consultation and used data from Study 020 as well as Study 007 (see sections 4.20 and 4.21).

  • A third model was later submitted to show how the financial components of the proposed managed access agreement would affect the value of ataluren (see sections 4.35 and 4.36).

4.14 Each of the company's Markov models had 6 states, representing the progression of DMD from the ambulatory phase to the non-ambulatory phases and death. The cycle length was 3 months. In the first model, the time horizon of the model was limited to the last point when 1 or more patients were in the ambulatory state (because only patients who could walk had treatment). In the second and third models, a lifetime time horizon was adopted. The analysis was carried out from the perspective of the NHS and personal social services, and costs and benefits were discounted at a rate of 3.5% per year (except in the second model, which used a rate of 1.5%).

Company's first cost–consequence model

4.15 To inform the best supportive care transition probabilities for loss of walking, the company used Kaplan–Meier estimates from the literature to derive time-dependent transition probabilities based on patient age. Ricotti et al. (2013) reported long-term outcomes of boys with DMD in the UK, comparing daily and intermittent use of corticosteroids. In this study, loss of walking with daily corticosteroid use occurred at a median age of 14 years. The company considered it reasonable to assume that these data were representative of the placebo arm in Study 007. In its original base case, the company used a Weibull function to fit the data.

4.16 To inform the transition probabilities for ataluren compared with placebo in its first model, a linear regression of the values of 6MWD from week 24 to week 48 of Study 007 against time was done. The regression analysis was performed on the data from week 24 to week 48 because the company deemed it to be more representative of the long-term treatment effect of ataluren. The company suggested that this was a conservative assumption because ataluren had a greater benefit compared with best supportive care in improving 6MWD in the first 24 weeks of the study. The linear extrapolation suggested that loss of walking would occur in the best supportive care group at week 313 (6.0 years) and at week 733 (14.1 years) in the ataluren group, which equated to a difference of 420 weeks (8.1 years). The company fitted a Weibull curve and shifted the best supportive care curve to the right so that the difference in median time to loss of walking between ataluren and best supportive care was 8.1 years (that is, the same as that predicted by linearly extrapolating Study 007 data). In its response to clarification, the company explored fitting alternative parametric models.

4.17 The company model included health-related quality-of-life data from the literature to inform the utility values in the cost–consequence analysis (Landfeldt et al. 2014) for patients and carers. It explained that it did not use the PedsQL inventory data from Study 007 because the algorithm used to map the data to EQ‑5D was adapted from a study by Khan et al. (2014), which was conducted in a healthy population. The company said that no loss of utility for adverse events had been included in the company model because there were no significant differences in the incidence of adverse events between the ataluren and placebo arms in Study 007.

4.18 The company estimated that the total cost per year of treatment with ataluren (list price) for an average 8‑year‑old child weighing 26 kg is £246,448. To calculate the cost per patient in the cost–consequence analysis, an age–weight curve from the Royal College of Paediatrics and Child Health was used to estimate the annual increase in weight for the cohort, with a starting age of 8.5 years. The company assumed no additional costs for monitoring. Health-state costs were taken from a published study (Landfeldt et al. 2014) and were converted using the UK 2012 purchasing power parity (OECD, 2015) and then inflated to 2014 costs using the consumer price index for health (ONS, 2015). For patients in the ambulatory health state, the total costs were £9,605. For patients in a non-ambulatory health state, the total costs were £23,600. In the non-ambulatory and ventilation-assisted health state, the total costs were also £23,600. In the non-ambulatory with scoliosis (with or without assisted ventilation) health states, the total costs ranged from £25,058 to £46,043.

4.19 In the company's original base case, best supportive care was associated with £235,207 in costs and 2.39 quality-adjusted life years (QALYs) over the lifetime of the model. Ataluren, at list price, was associated with £5,092,540 in costs and 6.15 QALYs, amounting to an incremental cost of £4,857,333 and an additional 3.77 QALYs compared with best supportive care. The incremental costs when applying the initial patient access scheme price for ataluren are confidential and no longer valid because the company subsequently increased the discount and included further financial components in the proposed managed access agreement.

Company's second cost–consequence model

4.20 After consultation the company provided a response to all the committee's requests described in the evaluation consultation document. The company's second cost–consequence model included the following changes compared with its first model:

  • updated parametric curves used to extrapolate time to loss of walking, time to scoliosis, time to ventilation assistance and time to death

  • an assumption that patients do not develop scoliosis after puberty

  • continued treatment costs with ataluren for 6 months following loss of walking

  • a lifetime time horizon

  • a change in the discount rate for costs and outcomes to 1.5%, which the company considered appropriate because it believed that ataluren significantly delays loss of walking by 7–12 years up to age 30 years

  • an increased disutility associated with scoliosis from 0.1 to 0.3

  • full caregiver disutilities ascribed to 3 caregivers rather than 1

  • different utility values for people in the ataluren and best supportive care groups after loss of walking (the company assumed that, because ataluren prolongs the time that patients are able to walk, loss of walking would occur after puberty at a stage of greater physical development in the ataluren group leading to a better quality of life even after walking is lost compared with patients who stop being able to walk at a younger age)

  • costs for ventilation assistance

  • an assumption that people who are able to walk cannot die from DMD-related causes

  • data from Study 020 and 2 different extrapolation methods:

    • linear extrapolation of 6MWD

    • stepped decline in 6MWD.

      The linear extrapolation method used the 6MWD results for the ataluren and placebo groups from a meta-analysis that included patients from Study 007 and Study 020. The company noted that, based on this approach, there was a difference of 12.2 years in time to loss of walking for ataluren compared with placebo. The stepped decline extrapolation method used a weighted average of the 6MWD results in Study 007 and Study 020 for the subgroups with a baseline 6MWD of 400 m or more, 300–400 m, and 300 m or less. However, the company suggested that, because the 6MWD was not sensitive enough to show differences in the treatment effect of ataluren in the subgroups of patients with a baseline 6MWD 300 m or less and of 400 m or more, it was more appropriate to use the following assumptions for its extrapolation rather than data from the trials:

  • In the subgroup of patients with a baseline 6MWD of 300 m or less the company assumed that the treatment effect with ataluren would be 20% to account for the improved function with ataluren compared with placebo seen in the timed function tests and the North Star Ambulatory Assessment.

  • In the subgroup of patients with a baseline 6MWD of 400 m or more, the company assumed a rate of decline in the best supportive care group taken from the average of the 48-week decline in the 300–400 m and 400 m or more subgroups after 2 years. This was because it considered the results seen in the trials do not appropriately represent the natural history of the condition. Based on this approach, there was a difference of 7.1 years in time to loss of walking for ataluren compared with placebo.

    The company argued that all these amendments were based on clinical data from Study 007 and Study 020, the committee's considerations about the original model and additional amendments based on further understanding of the natural history of the condition and the clinical effectiveness of ataluren.

4.21 The base-case results from the company's updated model suggested that ataluren provided 8.19 and 11.75 additional QALYs compared with best supportive care (using the stepped decline and linear extrapolation methods respectively), with associated incremental costs using the list price of £5,532,819 and £8,400,164 (using the stepped decline and linear extrapolation methods respectively). The incremental costs when applying the patient access scheme for ataluren are confidential and are no longer valid because the company subsequently increased the discount and included further financial components in the proposed managed access agreement.

Company's budget impact model

4.22 The company presented budget impact analyses to predict the cost of ataluren to the NHS and personal social services. In its original submission, the company estimated that, in year 1, a total of 35 people would have treatment, rising to 65 in year 5. The budget impact in year 1 (using ataluren's list price) was estimated to be about £8,625,680, rising to £16,019,120 in year 5. The results of the original budget impact analysis that incorporated the patient access scheme are confidential and are no longer valid because the company subsequently increased the discount and included further financial components in the proposed managed access agreement. The company submitted a further budget impact model in tandem with the proposed managed access agreement (see section 4.37).

Evidence review group review

Clinical effectiveness

4.23 The ERG noted that the evidence in the company's original submission reflected the decision problem and considered most of the analyses to be appropriate. The ERG noted several limitations in the clinical-effectiveness evidence presented by the company, including:

  • The follow-up time in Study 007 (48 weeks) was potentially too short to measure important outcomes (for example, mortality).

  • A summary of serious adverse events from 4 ongoing and 5 completed company-sponsored clinical trials suggested that several of these, including femur fractures, were more common with ataluren than with placebo. However, the ERG was unclear if this was because of longer exposure in the ataluren group.

4.24 The ERG reviewed the evidence from Study 020 submitted by the company in response to the committee's consultation request. It considered that there seemed to be selective reporting bias. It also noted that the company had not provided information on the minimal clinically important differences for the North Star Ambulatory Assessment, the PODCI and the Activities of Daily Living Questionnaire and that, in general, there was a lack of statistical test reporting in the company's analyses.

Cost effectiveness

4.25 The ERG noted the lack of evidence available on the long-term follow-up of people with DMD and considered that the company's use of external studies to inform model transition probabilities was valid. However, the ERG considered that there were issues with the methods used to extrapolate the data for the models, and investigated these in its exploratory analyses. In addition, the ERG noted that it may not be clinically plausible to assume that the treatment benefit of ataluren over best supportive care remained the same over time.

ERG exploratory analyses using the company's first model

4.26 The ERG's preferred scenario used a lifetime horizon and included the costs for continuing treatment with ataluren 6 months after loss of walking. It did so because the additional treatment costs had not been included in the company's base-case analysis, even though the company submission said that people would continue to have treatment for up to 6 months following loss of walking. The ERG's preferred scenario also included the best-fitting parametric curves to inform the clinical parameter transition probabilities. Flexible parametric models were selected for all transitions other than for the ambulatory to non-ambulatory state. For transitions to the non-ambulatory state, a log-normal model was used: although a flexible parametric model gave the best statistical fit, the ERG stated that its predictions may not be clinically plausible.

4.27 In the ERG's preferred scenario analysis, best supportive care was associated with £199,194 in costs and 3.80 QALYs over the lifetime of the model. Ataluren, at list price, was associated with £5,744,175 in costs and 6.86 QALYs, amounting to an incremental cost of £5,544,981 and an additional 3.05 QALYs compared with best supportive care.

ERG exploratory analyses using the company's second model

4.28 The ERG reviewed the company's cost–consequence analyses using the second model and expressed concerns with some of the company's assumptions. The ERG considered that it was appropriate to explore the following assumptions and amendments:

  • alternative parametric curves to extrapolate time to scoliosis and time to ventilation assistance

  • that patients could develop scoliosis after puberty (because, although the rate of scoliosis would be expected to be lower after puberty, the ERG was uncertain whether the risk would be completely eliminated)

  • a discount rate for costs and outcomes of 3.5% (because the ERG was unclear whether ataluren provided benefits that would restore the patient to full health and whether these benefits are likely to be maintained in the long term)

  • disutility because of scoliosis of 0.1 as per the company's original model (because there was no evidence on which to base applying the specific value assumed by the company, which was based on expert opinion)

  • caregiver disutilities applied for 2 caregivers rather than 3 (because the ERG was unclear why the same disutility for 1 primary caregiver should be applied to all caregivers)

  • the same utility values for people in the ataluren and best supportive care groups after loss of walking (because the ERG was unclear from the source the company had based this assumption on whether quality-of-life data from people who have just lost their ability to walk can be assumed to apply for their whole life time and to be different for patients having ataluren and best supportive care).

4.29 The ERG also expressed concerns about the 2 methods used by the company to extrapolate 6MWD. It considered that the linear extrapolation method was not appropriate because it used rates of decline in 6MWD from the subgroup of people with a baseline 6MWD of 300–400 m and applied them to the whole population. The ERG preferred the stepped decline approach but had concerns that the company's method of replacing some 6MWD trial data with assumed values favoured ataluren. To correct for this, the ERG presented 2 alternative approaches:

  • stepped decline approach 1:

    • using trial data in the subgroup of patients with a baseline 6MWD of 300 m or less

    • taking the average of the 6MWD data from the subgroup of patients with baseline 6MWD of 400 m or more and 300–400 m and applying it to the subgroup of patients with baseline 6MWD of 400 m or more after 2 years in both treatment groups

  • stepped decline approach 2:

    • using trial data in the subgroup of patients with a baseline 6MWD of 300 m or less

    • applying the numerical increase in decline rate after 2 years seen in the placebo group to the ataluren group for the subgroup of patients with a baseline 6MWD of 400 m or more after 2 years.

4.30 The results from the ERG's exploratory analyses, when applying its preferred assumptions (see sections 4.28 and 4.29), showed that ataluren provided between 2.03 and 6.41 additional QALYs compared with best supportive care (using the stepped decline approach 2 and the linear extrapolation methods respectively). The incremental costs for ataluren compared with best supportive care that incorporated the patient access scheme for ataluren are confidential and are no longer valid because the company subsequently increased the discount and included further financial components in the proposed managed access agreement.

ERG's exploratory analyses using the company budget impact model

4.31 Using the company's original budget impact model, the ERG explored changing the average weight of people having treatment to the average weight of people occupying the ambulatory health state in the cost–consequence model (39 kg in the best supportive care group and 53 kg in the ataluren group). Using the list price and an average weight of 39 kg, the budget impact in year 1 was estimated to be about £13,456,065, rising to £24,989,835 in year 5. The corresponding results using an average weight of 53 kg were £18,286,450 and £33,960,550 respectively. The ERG's exploratory results incorporating the patient access scheme are confidential and are no longer valid because the company subsequently improved the discount and included further financial components in the proposed managed access agreement.

Proposed managed access agreement

4.32 The company asked for and received permission from NICE to submit an improved patient access scheme and to begin discussions about a managed access agreement. Facilitated by NICE, a proposed managed access agreement was developed and submitted by a group of stakeholders comprising the company, NHS England, patient community experts and clinical experts. The proposed managed access agreement takes effect once stakeholders have confirmed they are in agreement with the proposed method of supporting delivery of this guidance. If agreed, the proposed managed access agreement would remain in force for up to 5 years after publication of this guidance. The proposed managed access agreement states that, if NICE does not recommend ataluren for NHS funding when the guidance is reviewed, NHS England funding for ataluren will no longer be available for any patient and treatment will stop. Those involved in the proposed managed access agreement will ensure that any patient receiving ataluren whose treatment is funded by NHS England under this proposed managed access agreement is made aware of these funding limitations and, on signing the patient agreement, accepts them on the agreed terms.

4.33 The proposed managed access agreement states that ataluren will be considered as a treatment option for all patients aged 5 years and older with DMD resulting from a nonsense mutation and who are able to walk 10 steps unaided:

  • Patients should only start treatment once a full set of standard baseline criteria has been obtained and they have signed the proposed managed access patient agreement.

  • Patients should stop treatment no later than 6 months after becoming fully non-ambulant (defined as no longer able to stand even with support and entirely dependent on wheelchair use for all indoor and outdoor mobility, unless this is because of an accident or an intercurrent illness).

4.34 Data will be collected from all patients when starting ataluren treatment and at all subsequent clinic visits and will be entered into the NorthStar database. Patients will be monitored according to the standard NorthStar criteria and Child Health Utility 9D quality-of-life data will be collected annually in those aged 7 years and older. Patients receiving ataluren (n=50) will be compared with an historical natural history population and a matched control group of the same age and North Star Ambulatory Assessment (n=100) to assess response to treatment. The historical data and matched control group will be identified from patients included in the NorthStar registry. Over the first 2 years of the proposed managed access agreement, the cohort of patients receiving ataluren is expected to have a decline on the North Star Ambulatory Assessment scale of around 4 points less than the matched control cohort (the anticipated values for each cohort are academic in confidence and cannot be presented here). The company said it calculated the expected decline using similar extrapolations to those seen in Study 020, as well as natural history data (Ricotti et al, 2015). At this time, the ataluren and matched control group declines will be evaluated and the extrapolation over the full 5‑year period will be confirmed or recalibrated.

Company's third cost–consequence model

4.35 In tandem with the proposed managed access agreement, the company submitted a third cost–consequence model. The third model was broadly similar to the second but included:

  • an improved confidential patient access scheme

  • other confidential financial components to enhance the value proposition to the NHS

  • some alternative modelling assumptions:

    • the same parametric survival models for scoliosis and ventilation assistance as in the ERG's exploratory analyses using the company's second model

    • discount rates of 3.5%

    • a decline after 2 years in the ataluren group with a baseline 6MWD of more than 400 m that was assumed to be the midpoint of the declines in the groups with a baseline 6MWD of more than 400 m and 300–400 m

    • disutility for scoliosis of 0.2.

4.36 In its base case, the company assumed that patients would receive ataluren or best supportive care for 5 years (the projected duration of the proposed managed access agreement) then all patients would receive best supportive care. The company also did scenario analyses in which patients receiving ataluren continued treatment until loss of walking and others that adopted the ERG's preferred assumptions (these were subsequently ratified by the ERG). All results were provided using linear extrapolation and stepped-decline extrapolation methods (see table 1). Incremental QALYs for ataluren compared with best supportive care ranged from 1.913 to 8.562. Total and incremental costs incorporating the patient access scheme and other financial components in the proposed managed access agreement are confidential and cannot be presented here.

Table 1 Results using the company's third cost–consequence model

Linear

Stepped

Ataluren

BSC

Incremental

Ataluren

BSC

Incremental

Base case: Company assumptions; 5‑year ataluren treatment; PMAA treatment costs

QALYs

4.087

−0.914

5.001

2.977

−0.914

3.891

Scenario: Company assumptions; ataluren until loss of walking; PMAA treatment costs

QALYs

7.647

−0.914

8.562

3.647

−0.914

4.561

Scenario: ERG assumptions; 5-year ataluren treatment; PMAA treatment costs

QALYs

3.694

0.803

2.890

2.716

0.803

1.913

Scenario: ERG assumptions; ataluren until loss of walking; PMAA treatment costs

QALYs

7.212

0.803

6.409

3.192

0.803

2.389

Abbreviations: BSC, best supportive care; ERG, evidence review group; PMAA, proposed managed access agreement; QALY, quality-adjusted life year.

4.37 The company also submitted a budget impact model that included the patient access scheme and other confidential financial components in the proposed managed access agreement. The results are confidential and cannot be presented here.

4.38 Full details of all the evidence are in the submissions received for this evaluation, and in the ERG report, which are all available in the committee papers.