Pirfenidone for treating idiopathic pulmonary fibrosis

Review objectives

4.1 The appraisal committee reviewed existing and new data on the clinical and cost effectiveness of pirfenidone, having considered evidence on the nature of idiopathic pulmonary fibrosis and the value placed on the benefits of pirfenidone by people with the condition, those who represent them, and clinical experts. It also took into account the effective use of NHS resources. The committee recognised that this appraisal reviewed NICE's previous technology appraisal guidance on pirfenidone to take into account new evidence relating to people with a forced vital capacity (FVC) above 80% predicted and to consider removing the stopping rule, that is, to no longer stop pirfenidone after an absolute decline of 10% or more in predicted FVC within any 12‑month period. The review considered the whole of pirfenidone's marketing authorisation.

Current practice

4.2 The committee discussed the treatments for idiopathic pulmonary fibrosis in current NHS practice. The committee heard the clinical experts explain that they offer treatment with pirfenidone or nintedanib to people with an FVC between 50% and 80% predicted; this reflects NICE's previous technology appraisal guidance on pirfenidone and nintedanib for treating idiopathic pulmonary fibrosis. The clinical experts noted that they would offer best supportive care to people with an FVC above 80% predicted because NICE does not recommend pirfenidone or nintedanib in this population. The committee noted comments received during consultation that clinicians would like to offer active treatments to people with an FVC above 80% predicted. The committee concluded that the current treatment options are nintedanib and pirfenidone for people with an FVC between 50% and 80% predicted, and best supportive care for those with an FVC above 80% predicted.

4.3 The committee discussed the stopping rule included in NICE's previous technology appraisal guidance on pirfenidone and nintedanib, and how this is implemented in clinical practice. It heard from clinical experts that they follow this stopping rule, that is, stopping treatment if there is an absolute decline of 10% or more in predicted FVC within any 12‑month period, and that they often confirm the drop in percent predicted FVC with repeat testing. The committee heard from the clinical experts during the committee meeting and from the comments received during consultation about the limitations of using percent predicted FVC to assess lung function in people with idiopathic pulmonary fibrosis. It heard that the disease course varies, both between patients and over time. It heard that there were periods of relative stability interspersed with acute exacerbations, and that acute exacerbations are associated with permanently reduced lung function and an increased risk of dying. The committee noted that the clinical experts could not suggest a better way of objectively defining treatment success or failure than using percent predicted FVC. It noted the limitations of FVC but understood that, in clinical practice, the wider patient characteristics would be taken into account in interpreting percent predicted FVC.

Comparators

4.4 The committee recalled that current NICE recommendations for treatment depend on percent predicted FVC, and discussed whether best supportive care or nintedanib were relevant comparators for pirfenidone. The committee was aware that nintedanib is recommended for a subgroup of the population, but recognised that the recommendation for nintedanib was based on the comparison of nintedanib with pirfenidone, rather than nintedanib with best supportive care. The committee therefore agreed it was more appropriate to compare pirfenidone with best supportive care and it did not consider the comparison with nintedanib further.

Clinical effectiveness

4.5 The committee considered the clinical evidence presented by the company. It noted that the evidence came from 4 randomised double-blind placebo-controlled phase III trials (CAPACITY 1, CAPACITY 2, ASCEND and SP3) and other observational data:

4.6 The committee noted that the inclusion criteria of the trials differed with respect to percent predicted FVC: the CAPACITY trials recruited patients with an FVC above 50% predicted and without an upper limit. ASCEND recruited patients with an FVC between 50% and 90% predicted, and the SP3 trial did not specify the range, but reported an average baseline FVC of 77% predicted. The committee understood that most of the data presented by the company (92% of patients in ASCEND and the CAPACITY trials) came from patients with an FVC up to 90% predicted. The committee concluded that the trial evidence was most generalisable to people with an FVC of up to 90% predicted.

4.7 The committee discussed whether the populations in the pirfenidone trials reflected people with idiopathic pulmonary fibrosis in current NHS practice. It understood that only 25% of patients across ASCEND and the CAPACITY trials had an FVC above 80% predicted, compared with 36% to 41% in UK practice (as a proportion of people with an FVC above 50% predicted; estimates are based on data from the British Thoracic Society idiopathic pulmonary fibrosis registry and comments from the company at the committee meeting). The committee concluded that people with an FVC above 80% predicted were under-represented in the clinical trials.

4.8 The committee discussed the clinical effectiveness of pirfenidone. It was aware that in NICE's previous technology appraisal for pirfenidone the committee concluded that 'pirfenidone seemed to have a modest but measurable effect on slowing the decline in lung function'. The committee acknowledged that the company had provided new long-term and mature data in this appraisal, mainly relating to mortality, for patients with an FVC up to 90% predicted. The committee considered the results of the ERG's meta-analyses. It agreed that pirfenidone reduced disease progression compared with placebo. It also agreed that there was evidence that it may reduce mortality. It concluded that it had not seen anything to contradict the conclusion in NICE's previous technology appraisal. The committee concluded that pirfenidone is effective in people with an FVC between 50% and 90% predicted.

4.9 The committee discussed whether the effect of pirfenidone on slowing the decline in lung function may differ according to a person's baseline percent predicted FVC. It considered the evidence for people with an FVC above or below 80% predicted, noting that:

Cost effectiveness

4.12 The committee considered the company's partitioned survival model, which had 3 mutually exclusive health states: progression-free, progressed and dead. The company's model assumed that the effectiveness of pirfenidone relative to best supportive care was the same regardless of a person's baseline percent predicted FVC. The committee considered this to be appropriate based on the data currently available (see section 4.9). The committee heard from the ERG that the company's model did not capture the progressive nature of idiopathic pulmonary fibrosis. The committee noted that the company, in choosing the model structure, made several clinically implausible assumptions. For example:

4.13 The committee discussed how the company estimated the potential long-term mortality benefit with pirfenidone over a patient's lifetime by extrapolating from relatively short trials. It noted that the choice of mortality parametric curve was a key driver of the cost-effectiveness results, and that the company estimated long-term mortality based on data from ASCEND, CAPACITY 1 CAPACITY 2 and RECAP. The committee agreed that this was appropriate. It noted that the company had modelled mortality using the Weibull distribution, and that the ERG had used the Gompertz distribution.

4.14 The committee discussed the company's modelling assumption that the mortality benefit of pirfenidone compared with best supportive care remains constant over a person's lifetime. It appreciated that the randomised trials were too short (that is, either 52 or 72 weeks) to provide evidence to support this assumption. The committee also noted that, although follow-up data for pirfenidone were collected for over 8 years in RECAP, there was no best supportive care group in the study and so no long-term relative effectiveness could be estimated from the study. The committee was also aware that the model was very sensitive to the assumptions around duration of treatment benefit. It heard from 1 clinical expert that the treatment benefit of pirfenidone was likely to be constant over a person's lifetime. The committee did not agree that this was plausible, based on advice from the ERG that the trials showed a reduction in treatment effect over time for mortality. In its response to the appraisal consultation document, the company used data from INOVA to support its argument that the treatment effect lasts for at least 8 years. The company provided a Kaplan–Meier plot and log-cumulative hazard plot for mortality based on data from the trials and INOVA. The committee noted that the log-cumulative hazard plots for pirfenidone and best supportive care were not parallel after 5 years. It determined that, although there was some maintained treatment effect, it was not constant. The committee concluded that the evidence did not justify assuming a constant mortality benefit for 8 years. It further concluded that it was reasonable to assume a constant benefit up to 5 years.

4.15 The committee discussed the population in the model. It recognised that the model was based on the pirfenidone clinical trials and therefore included fewer people with an FVC above 80% predicted than are seen in NHS clinical practice (see section 4.7). The committee considered whether the incremental cost-effectiveness ratios (ICERs) would be affected if the modelled population included a higher proportion of people with an FVC above 80% predicted, to better reflect NHS practice. It understood from the analyses presented that there was a consistent trend for higher ICERs in the population with an FVC above 80% predicted than in the population with an FVC between 50% and 80% predicted. For example, when assuming a 5‑year treatment effect and including the stopping rule, the ICER for pirfenidone compared with best supportive care was between:

4.16 The committee discussed whether the company's model appropriately incorporated the treatment stopping rule. The ERG explained that ICERs including the stopping rule for pirfenidone would likely be biased in favour of pirfenidone when compared with best supportive care. This was because, in the model, the stopping rule reduced pirfenidone costs without affecting treatment outcomes. The committee had concluded that analyses including a stopping rule for pirfenidone would underestimate the ICER because of the model structure. However, at its third meeting, the committee heard from the company that analyses including a stopping rule would overestimate the ICERs in scenarios where the mortality benefit of pirfenidone is not constant over a person's lifetime. The company had not submitted any evidence to support its statement and therefore the committee could not draw a firm conclusion about whether reducing the duration of treatment benefit under- or overestimated the ICERs with a stopping rule. The committee concluded that all ICERs including the stopping rule for pirfenidone were associated with uncertainty.

Cost-effectiveness results and conclusions

4.17 The committee considered the cost effectiveness of pirfenidone compared with best supportive care. The committee noted that, for the population specified in the marketing authorisation (people with 'mild to moderate' idiopathic pulmonary fibrosis, that is, people with an FVC above 50% predicted), the ICER (including the stopping rule) that most closely matched the committee's preferred assumptions was between £25,706 (Weibull) and £28,870 (Gompertz) per QALY gained when assuming a 5‑year treatment effect. The committee agreed that the estimate using the Gompertz survival curve was more plausible (see section 4.13). The committee then discussed the cost-effectiveness estimates for the group with an FVC between 50% and 90% predicted, because most of the data presented by the company was from this group (see section 4.6), noting that the ICERs were similar to those for the full population.

4.18 The committee recognised that, when above a most plausible ICER of £20,000 per QALY gained, judgements about the acceptability of a technology as an effective use of NHS resources must take account of the degree of certainty around the ICER, the innovative nature of the technology and whether its benefits have been adequately captured in the model. The committee recalled that all of the ICERs were uncertain, noting that:

4.19 The committee then discussed whether there was a subgroup for whom pirfenidone could be recommended. The committee was aware that pirfenidone was associated with a lower ICER in populations with a lower FVC (see section 4.15). It recalled that treatment decisions based on percent predicted FVC are currently implemented in clinical practice (see section 4.2). The committee was aware that, in NICE's previous technology appraisal guidance, pirfenidone was regarded as cost effective for people with an FVC between 50% and 80% predicted if the company provided pirfenidone with the discount agreed in the patient access scheme. It noted that the company had presented analyses for this group. The ICER (including the stopping rule) that most closely matched the committee's preferred assumptions for pirfenidone compared with best supportive care was between £24,933 (Weibull) and £27,780 (Gompertz) per QALY gained when assuming a 5-year treatment effect. The committee agreed that these ICERs were also subject to uncertainty relating to the stopping rule and duration of survival benefit, and recalled its conclusion that no sizeable health-related benefits had been excluded from the economic model (see section 4.18). However, the committee understood that pirfenidone was considered a reasonably innovative treatment at the time of the previous appraisal because it was the first drug to improve outcomes in idiopathic pulmonary fibrosis without the long-term side effects of immunosuppressants. It agreed that it had not seen any clinical evidence contradictory to that considered in NICE's previous technology appraisal on pirfenidone (see section 4.8), and therefore it was not minded to change this recommendation and withdraw an existing treatment option for people with moderate disease, despite the relatively high ICERs. The committee concluded that the recommendations in NICE's previous technology appraisal guidance on pirfenidone remained appropriate for people with idiopathic pulmonary fibrosis with an FVC between 50% and 80% predicted.

4.20 The committee considered the cost effectiveness of removing the requirement to stop treatment if a person's predicted FVC drops by an absolute value of 10% in the population with an FVC between 50% and 80% predicted. The committee was aware that removing the stopping rule increased the ICERs. Having concluded that the ICER for pirfenidone in this group was high, and associated with uncertainty, when a stopping rule is included (see section 4.19), the committee concluded that pirfenidone could not be considered cost effective without the stopping rule.

Potential equality issues

4.21 The committee noted the potential equality issue raised by consultees that restricting treatment based on percent predicted FVC could discriminate against:

The Pharmaceutical Price Regulation Scheme

4.22 The committee was aware of NICE's position statement on the Pharmaceutical Price Regulation Scheme (PPRS) 2014 and, in particular, the PPRS payment mechanism. It accepted the conclusion 'that the 2014 PPRS payment mechanism should not, as a matter of course, be regarded as a relevant consideration in its assessment of the cost effectiveness of branded medicines'. The committee heard nothing to suggest that there is any basis for taking a different view about the relevance of the PPRS to this appraisal. It therefore concluded that the PPRS payment mechanism was not relevant in considering the cost effectiveness of the technology in this appraisal.

Summary of appraisal committee's key conclusions