Enzalutamide for treating metastatic hormone-relapsed prostate cancer before chemotherapy is indicated

Clinical effectiveness

4.6 The Committee discussed the estimates for overall survival (OS) for enzalutamide compared with placebo from the PREVAIL trial. The Committee noted that the trial had been unblinded early for benefit, and that the company had presented data from the interim analysis of OS (on which the decision to stop the trial was made), and from what would have been the final analysis (according to the study protocol) after the study had been unblinded for 6 months. The Committee noted that, at both time points, OS was longer with enzalutamide than with placebo and that the differences between enzalutamide and placebo were statistically significant (p<0.05). The Committee was aware that, once the disease progressed, people on the study drug in PREVAIL could move on to subsequent treatments, and that the company considered that some of these treatments (such as cabazitaxel and sipuleucel‑T, cytotoxic chemotherapy other than docetaxel, and investigational treatments) prolonged life but were unlikely to be used in England at this position in the treatment pathway. The company also noted that in clinical practice people would not have enzalutamide or abiraterone if they already had taken enzalutamide. The Committee noted that most people in PREVAIL went on to have docetaxel after disease progression, which reflects the treatment pathway in England. However, it agreed that, although currently offered via the Cancer Drugs Fund, cabazitaxel is not recommended for prostate cancer in NICE's technology appraisal on cabazitaxel for hormone-refractory metastatic prostate cancer previously treated with a docetaxel-containing regimen and that the marketing authorisation for sipuleucel‑T has been withdrawn. The Committee also agreed that cabazitaxel and sipuleucel‑T prolong life and, if these were disproportionately taken by patients in the placebo group after progression or unblinding in PREVAIL, the survival estimates for enzalutamide compared with placebo would be biased against enzalutamide. The Committee accepted that people whose cancer progresses on enzalutamide would not receive subsequent abiraterone or enzalutamide and having both enzalutamide and abiraterone in the trial did not reflect clinical practice in England. The Committee concluded that enzalutamide improves OS compared with placebo and that it is appropriate to adjust the results for subsequent life-extending treatments not available in the NHS that people received in PREVAIL.

4.7 The Committee discussed the methods the company used to adjust its OS estimates, noting that the company had presented results using the inverse probability of censoring weights (IPCW) method and the 'two‑stage method'. The Committee was aware of other possible complex methods for adjustment, including marginal structural models and rank‑preserving structural nested failure time models. It heard that the company had pre‑specified adjusting for treatment switching using the IPCW and two‑stage methods in its statistical analysis plan; the Committee considered it good practice to pre‑specify the methods of adjustment. It further noted the inherent assumption in both methods was that there were no unmeasured confounders affecting the association between moving on to another treatment and dying. The Committee appreciated that the company provided the list of covariates identified as potential confounders in response to the appraisal consultation document, and considered the list to be generally appropriate. The Committee was aware that both methods improved the effectiveness of enzalutamide compared with placebo, but that the IPCW method improved it considerably more. The Committee heard the company's rationale for preferring the IPCW over the two‑stage method. One reason given was that it needed fewer assumptions. The Committee heard from the company that the two‑stage method involves using a 'second baseline' that, in this case, assesses patients' characteristics when their disease progressed and they switched to their second treatment. However, the company stated that there was a period of about 6 weeks between a patient's disease progressing and them starting a new treatment, and during this time their characteristics may have changed. The Committee identified potential issues with adjusting OS using either the two‑stage or IPCW method. First, these methods need proportional hazards between the treatment arms, but the PREVAIL data did not appear to meet this criterion. Second, the company had adjusted only for non‑NHS, life-extending treatments that were taken second‑line in PREVAIL, and not for treatments taken third‑line, because of insufficient data. The Committee noted that, although the issue of non‑NHS, life-extending third‑line treatment would apply to any method of adjustment, it meant that people who received active, non‑NHS, third‑line treatments in PREVAIL may have survived longer than would be expected in clinical practice in England. The Committee concluded that it was unclear which method of adjustment provided estimates that represented the true difference in survival between enzalutamide and placebo, but the true value was likely to lie nearer to the estimates from the IPCW method than the two‑stage method and unadjusted estimates.

4.8 The Committee discussed the estimates for progression-free survival (PFS) for enzalutamide from PREVAIL. It noted that the company had used a radiographic measure of progression as its primary outcome, but the company considered that time to treatment discontinuation (TTD) was the most appropriate endpoint to reflect PFS in clinical practice. The Committee heard from the clinical experts that the measures of progression used in clinical practice include Response Evaluation Criteria in Solid Tumors radiographic criteria and measuring prostate specific antigen (PSA) levels. The Committee noted that TTD had been used as a proxy for PFS in other appraisals of hormone‑relapsed prostate cancer and may reflect staying on treatment until confirmed progression. However, it also noted that people may stop treatment before disease progression if they have severe side effects. The clinical experts noted that, because enzalutamide is relatively well‑tolerated, few people stop taking it because of side effects. The Committee recognised that TTD better captured the costs of treatment than radiographic disease progression. The Committee concluded that enzalutamide had been shown to delay disease progression using either measure. It considered that, although a TTD estimate includes people who stop treatment before disease progression, because enzalutamide is well‑tolerated, the number of people stopping before progression would be low. Overall, the Committee concluded that TTD was a relevant proxy to estimate disease progression and provided the advantage of better capturing costs.

Cost effectiveness

4.9 The Committee considered the structure of the company's economic model. It agreed that people taking enzalutamide before chemotherapy would not get abiraterone after chemotherapy. It agreed with the company that patients on best supportive care before chemotherapy would have an active treatment (such as enzalutamide or abiraterone) after chemotherapy. It noted that the company had applied the survival estimates from PREVAIL to the whole model, meaning that the duration of treatment with docetaxel or third‑line active treatment (for people who first had best supportive care before docetaxel) did not affect how long patients were modelled to live. The Committee could not judge whether the modelled TTD with docetaxel or enzalutamide (when taken after docetaxel) reflected that seen in clinical practice because the company had not presented clinical data to show that its modelled estimates were plausible. The Committee noted that the company's survival modelling approach meant that the risk of death was related to time rather than health state. So, at any point in time the risk of death was the same for people whose disease had not progressed on multiple treatments, as those who had progressed, which the Committee and clinical experts considered implausible. The Committee concluded that the model structure was appropriate in terms of the sequence of treatments people would have in clinical practice in England. However, there was still uncertainty about whether the time spent on treatments after enzalutamide in the model reflected clinical practice.

4.10 The Committee noted that the company assumed in the model that more than 80% of people in both treatment arms would go on to have docetaxel, but heard from the clinical experts that in clinical practice in England this figure would be around 40%. The Committee understood that that both the company and Evidence Review Group (ERG) had done sensitivity analyses around the proportion of people taking docetaxel. However, the Committee noted that because of the way OS was applied in the model (that is, modelled OS was independent of modelled duration of treatment) these sensitivity analyses only captured the costs of varying the proportion of people who had docetaxel and not the impact of OS. The Committee concluded that the model overestimated the proportions of people who would go on to have docetaxel compared with clinical practice, but how this impacted the modelled survival and cost effectiveness estimates was unclear.

4.11 The Committee noted that, to estimate the mean life extension associated with enzalutamide, the company needed to extrapolate OS from the trial data in its model. The Committee noted that, at both the September 2013 and June 2014 data cut-offs, most of the trial population were still alive. The Committee agreed that it was better to extrapolate OS from the June 2014 cut‑off because the trial patients were followed up for longer and also because this reflected the planned final analysis for OS. However, the Committee acknowledged that there was still uncertainty because less than half of the trial population had died at this cut‑off. The Committee was aware that the company had selected the parametric curve for extrapolating OS by testing the fit of various parametric curves to the trial data, both statistically and by using predicted 5- and 10‑year survival rates as a measure of face validity. It was concerned that the company had not further checked the validity of the extrapolated data. The Committee noted that this was particularly important because of the immaturity of the trial data and because of the small population at risk at the end of the trial follow‑up (those who had not died or had been otherwise censored). This meant that a large proportion of the estimated survival benefit was based on the extrapolated period rather than the trial data. The Committee noted that the Weibull distribution, used in the company's base case, gave a more conservative estimate than the log‑logistic and log‑normal curves the company had dismissed because they gave implausible 5- and 10‑year survival estimates (section 3.17). The Committee noted that, in response to the appraisal consultation document, the company confirmed that it had used the AFFIRM study to model transition probabilities in the model, but had stated that it could not use data from AFFIRM to validate the modelled post‑docetaxel survival estimates for enzalutamide because the follow‑up period in AFFIRM was not long enough. The Committee concluded that enzalutamide increased survival compared with best supportive care, and that the company had chosen a conservative model, but the Committee was uncertain about the extent of the survival benefit with enzalutamide over the period after trial follow‑up had ended.

4.12 The Committee noted that, in addition to extrapolating OS, the company had extrapolated TTD from PREVAIL. The Committee agreed that it was more appropriate to extrapolate the data on TTD from the September 2013 cut‑off data from PREVAIL, rather than the June 2014 cut‑off, because the June 2014 estimates may be biased (favouring enzalutamide) because of unblinding. The Committee noted the company's comment that bias because of unblinding would be minimal because only 7.2% of people were still on placebo after September 2013. The Committee preferred using the September 2013 cut‑off data for TTD because the reduced potential for bias outweighed the benefit of the additional data provided by the June 2014 cut‑off data. The Committee concluded that the most appropriate data cut‑offs from PREVAIL to model were June 2014 for OS and September 2013 for TTD.

4.13 The Committee discussed the utility values that had been calculated from EQ‑5D data collected in PREVAIL for best supportive care and enzalutamide. It noted that, in PREVAIL, quality of life had decreased over time while people had best supportive care or enzalutamide, but it did so to a lesser extent with enzalutamide. Because quality of life as measured by EQ‑5D and Functional Assessment of Cancer Therapy – Prostate (FACT‑P) decreased over time, the Committee did not consider the company's approach of adding a utility increment for enzalutamide to the estimated utility value before treatment had started to be appropriate. The Committee preferred the approach suggested by the ERG, in which the utility decrement over time seen with best supportive care and enzalutamide was subtracted from the starting utility value. The Committee noted that the company used an estimate from the literature for the utility experienced when taking enzalutamide after docetaxel, rather than using its own estimate reflecting data from AFFIRM that it had presented in NICE's technology appraisal on enzalutamide for metastatic hormone-relapsed prostate cancer previously treated with a docetaxel-containing regimen. The Committee accepted the company's revision to its base case, which used a utility value from AFFIRM that was adjusted for the drop in quality of life with best supportive care seen in the trial. The Committee also accepted the company's approach of using a utility gain for enzalutamide from a source other than AFFIRM, because this was accepted as a plausible utility uplift with enzalutamide in a NICE's technology appraisal on enzalutamide for metastatic hormone-relapsed prostate cancer previously treated with a docetaxel-containing regimen. The Committee also noted that the utility assumed by the company for people having palliative care (0.500) did not match the value reported in the reference (Sandblom et al. 2004) cited by the company. The company responded to the appraisal consultation document that in its model, it had used a weighted average of utility values from Sandblom et al. to estimate utility values for people with a life expectancy similar to people modelled to be having palliative care. The Committee noted that the company did not give the formula it used to get the weighted value. The Committee concluded that its preferred utility values were those proposed by the ERG for the stable disease health state, and those based on AFFIRM presented by the company for people having enzalutamide after docetaxel.

4.14 The Committee considered how the company had applied the enzalutamide patient access scheme (PAS) in the model. It was aware that there is an existing simple discount PAS for enzalutamide, which was agreed as part of the appraisal of NICE's technology appraisal guidance on enzalutamide after docetaxel. It noted that if the current appraisal recommended enzalutamide, then a new simple PAS with an increased discount would apply to enzalutamide used either before or after docetaxel. The Committee noted that the company used the new increased PAS to model the costs of enzalutamide before docetaxel, and the existing PAS to model the costs of enzalutamide after docetaxel (for people who have best supportive care before docetaxel). The Committee also noted that the ERG agreed with the company's approach, but additionally provided a scenario in which the new increased PAS was applied to all costs of enzalutamide in the model to show what the costs for each treatment option would be if enzalutamide were recommended. The Committee recognised that, if the current appraisal did not recommend enzalutamide before docetaxel, then the existing PAS would apply for patients having enzalutamide after docetaxel. Therefore, the Committee concluded that the company's approach to applying the PAS in the model was appropriate.

4.15 The Committee considered that the following modelling assumptions were the most plausible:

4.16 The Appraisal Committee considered whether it should take into account the Pharmaceutical Pricing Regulation Scheme (PPRS) 2014, and in particular the PPRS Payment Mechanism, when appraising enzalutamide. The Committee noted NICE's position statement in this regard, and accepted '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 queried why the company, in its original submission (before the revised PAS submission) for the appraisal, had presented a scenario analysis that included a 10.36% price rebate to reflect the PPRS. The company clarified that this percentage had been calculated by the Department of Health and applied to all Astellas' products. The company stated in its response to the appraisal consultation document that it repays the PPRS into a 'Health General Cash' account. It acknowledged that the rebate would be unlikely to be returned to commissioning groups relative to the amount of enzalutamide prescribed in the population covered by its marketing authorisation. The Committee agreed that there was no detailed and transparent justification of how the PPRS would directly affect the acquisition cost of enzalutamide to the NHS (in a way that would represent a nationally available price reduction). It also heard from the company that the 10.36% rebate level was 'likely to remain for 3 years'. The Committee did not accept that this could function as the guarantee needed for this to be acceptable as a 'nationally available price reduction', as envisaged in the guide to the methods of technology appraisal 2013. In summary, the Committee heard nothing to suggest that there is any basis for taking a different view on the relevance of the PPRS to this appraisal of enzalutamide. It therefore concluded that the PPRS payment mechanism was irrelevant for considering the cost effectiveness of enzalutamide.

4.17 The Committee noted that the company did not propose that enzalutamide taken before docetaxel meets the end‑of‑life criteria. The Committee nevertheless considered whether enzalutamide met these criteria. It noted that, in both the placebo and enzalutamide arms of PREVAIL the median OS was more than 30 months and as such, the mean life expectancy at this point in the treatment pathway was more than 24 months. The first end‑of‑life criterion is that the treatment is indicated for patients with a short life expectancy, normally less than 24 months. Because enzalutamide did not meet this criterion, the Committee did not consider the other criteria. It concluded that enzalutamide did not meet end‑of‑life criteria for treating metastatic hormone‑relapsed prostate cancer in people for whom chemotherapy is not yet indicated.

4.18 The Committee discussed whether enzalutamide was innovative and whether it had substantial, demonstrable and distinctive benefits adequately captured in the modelling of the QALYs. The Committee noted that enzalutamide offers people with hormone‑relapsed disease a first‑line active treatment before cytotoxic chemotherapy. The Committee noted that enzalutamide was the only treatment option for people with visceral disease, for whom abiraterone is not available through the Cancer Drugs Fund, or with severe liver dysfunction for whom abiraterone is contraindicated, or who cannot take corticosteroids. It further noted the comments, received in response to the appraisal consultation document, that enzalutamide is an important treatment option for people who have tried abiraterone but have stopped taking it because of severe side effects. It considered that, although enzalutamide is not a new treatment, it is the only treatment that can give these benefits at this position in the treatment pathway, and so is innovative. The Committee noted that the patient experts stated that delaying chemotherapy was of great importance to patients. The Committee was aware that delaying chemotherapy may mean that some people would no longer be eligible for chemotherapy. However, it noted that, despite this possibility, people wanted pre‑chemotherapy treatments to be available to them. The Committee considered whether the model captured the benefits of delaying chemotherapy. The Committee agreed that the model predicted that people having enzalutamide had more time with better utility than people on best supportive care. However, the Committee agreed that the benefit of delaying chemotherapy may not have been fully captured by the utility values included in the modelling, and that accounting for this would have reduced the ICER. The Committee concluded that enzalutamide was innovative and this should be considered in its decision‑making.

4.19 The Committee noted that the NICE guide to the methods of technology appraisal states that, if a technology has a most plausible ICER above £30,000 per QALY gained, the Committee will need to identify an increasingly stronger case for supporting the technology as an effective use of NHS resources. The Committee noted that:

Summary of Appraisal Committee's key conclusions