Abiraterone for castration-resistant metastatic prostate cancer previously treated with a docetaxel-containing regimen

Clinical effectiveness

4.4 The Committee considered the evidence submitted by the manufacturer on the clinical effectiveness of abiraterone. The Committee agreed with the manufacturer and the ERG that, although mitoxantrone was listed as a comparator in the scope, differences in the COU‑AA‑301 and the TROPIC trials made it difficult to indirectly compare mitoxantrone with abiraterone. The Committee heard from clinical specialists that participants in the COU‑AA‑301 trial were likely to be healthier than those who would receive abiraterone treatment in UK clinical practice. However, it acknowledged that, in response to the appraisal consultation document, a number of comments from clinical organisations suggested that participants in the COU‑AA‑301 trial would be similar to patients who would be considered for abiraterone treatment in UK clinical practice.

4.5 The Committee considered the different ways that progression‑free survival had been defined in the COU‑AA‑301 trial: radiographic evidence of progression‑free survival, 'modified' progression‑free survival and time to treatment discontinuation. The Committee noted that determining progression‑free survival from radiographic evidence was difficult because patients entered the study already with metastatic disease and could die of prostate cancer without any evidence of further radiological progression. The Committee heard from the clinical specialists that disease progression is not determined with a single measure and that they would also consider progression of a patient's symptoms. The Committee discussed whether discontinuing abiraterone treatment was an adequate proxy for progression of disease. The Committee heard from clinical specialists that in general patients stop treatment at approximately the same time as their disease progresses, but noted that some patients in the COU‑AA‑301 trial continued to take abiraterone after their disease had progressed. The Committee also noted that most patients in the COU‑AA‑301 trial discontinued abiraterone treatment for reasons other than disease progression. The Committee acknowledged uncertainty around this measure, but agreed that of the measures of disease progression in the COU‑AA‑301 trial, time to treatment discontinuation was the most reasonable indicator of disease progression.

4.6 The Committee discussed the results for the COU‑AA‑301 trial and noted that abiraterone was associated with a statistically significant improvement in median overall survival and median progression‑free survival (based on time to treatment discontinuation) compared with prednisolone for both the whole (intention‑to‑treat) population and the manufacturer's base‑case population (patients who had received only one prior docetaxel‑containing chemotherapy regimen), with an improvement in median overall survival of 4.6 and 5.3 months respectively, in the manufacturer's updated analysis. The Committee also noted that patients receiving abiraterone were more likely to experience an improvement in symptoms, including pain, functional status and fatigue. The Committee therefore concluded that the evidence demonstrated that abiraterone was an effective second‑line treatment for castration‑resistant metastatic prostate cancer.

4.7 The Committee considered the manufacturer's base‑case population of patients who had received one prior chemotherapy in the COU‑AA‑301 trial. The Committee noted that this subgroup did not match the population for which abiraterone is licensed (the therapeutic indication in the marketing authorisation does not stipulate only one prior chemotherapy) but probably reflected the population in England and Wales for whom abiraterone would be considered. The Committee noted that in the COU‑AA‑301 trial the relative median overall survival benefit for abiraterone was higher in the subgroup with one prior chemotherapy than in the subgroup with more than one prior chemotherapy, but tests for interaction showed that the difference between these two groups was not statistically significant. The Committee heard that the manufacturer considered the number of prior chemotherapies sufficiently important as a prognostic factor (in that more than one chemotherapy would imply a later stage of disease, more previous adverse reactions and more treatment‑resistant tumours) to include it as a stratification factor for randomisation. The Committee also heard from the manufacturer that the difference in relative median overall survival benefit for abiraterone between the one prior chemotherapy subgroup and the subgroup who received more than one prior chemotherapy was supported by results for progression‑free survival and other related outcomes from the COU‑AA‑301 trial. The difference was also supported by overall survival results from published studies of other second‑line treatments for castration‑resistant metastatic prostate cancer. The Committee accepted that this population was likely to reflect patients who would be treated with abiraterone in UK practice, and who would have better treatment outcomes because they have less advanced disease. Therefore, the Committee concluded that it was reasonable based on biological plausibility and the pre‑specification of this group in the COU‑AA‑301 trial (as a stratification factor) to accept this patient subgroup and its associated effectiveness data as the base‑case for the analysis.

4.8 The Committee considered the evidence on adverse reactions associated with abiraterone. The Committee was aware that abiraterone may cause hypertension, hypokalaemia and fluid retention as a consequence of an increased mineralocorticoid concentration. The Committee noted that adherence to abiraterone in the COU‑AA‑301 trial was generally high and reflective of abiraterone use in clinical practice, and that adverse reactions were generally manageable and reversible. The Committee also noted that abiraterone is not associated with the more severe adverse reactions that can occur with cytotoxic drugs such as mitoxantrone. The Committee heard from the clinical specialists that abiraterone is a well‑tolerated oral medication. The Committee concluded that abiraterone is generally safe and that any associated adverse reactions are tolerable.

Cost effectiveness

4.9 The Committee considered the manufacturer's economic model, the assumptions on which the parameters were based, and the critique and exploratory analyses conducted by the ERG. The Committee concluded that the model closely adhered to the NICE reference case for economic analysis.

4.10 The Committee considered the comparator treatments (placebo plus prednisolone and mitoxantrone with or without prednisolone) and the population included in the economic model. The Committee noted that in the base‑case analysis the manufacturer assumed that overall survival and progression‑free survival were the same for patients taking mitoxantrone and patients taking prednisolone. The Committee agreed that, in the absence of any direct or indirect comparative data and the absence of published evidence showing any survival benefit for mitoxantrone in a population without prior chemotherapy, this assumption was reasonable.

4.11 The Committee considered the overall survival curve used in the manufacturer's economic model. The Committee noted that estimates related to overall survival were taken from the 'updated' analyses, which the Committee preferred to the 'primary' analyses. The Committee also noted that the manufacturer had modelled overall survival using Kaplan–Meier survival analysis of patient‑level data from the COU‑AA‑301 trial up to a cut‑off point at which a small proportion (10%) of patients were still alive. Beyond this, the manufacturer had extrapolated overall survival using a constant hazard. The Committee considered that the Kaplan–Meier survival curves specifically reflected the COU‑AA‑301 trial population and whether a well‑fitting parametric distribution would be more applicable to all patients for whom abiraterone may be a potential therapy in clinical practice. The Committee noted that the 10% cut‑off chosen by the manufacturer for overall survival produced a relatively favourable ICER compared with other possible cut‑off points. The Committee heard from the manufacturer that fitting a specific parametric distribution to the overall survival curve was not necessary because survival data in the COU‑AA‑301 trial were almost complete and because additional analyses suggested that no single parametric function provides a better fit to the data than others. The Committee accepted that it may have been more appropriate to use a well‑fitting parametric curve to extrapolate overall survival, but was also sympathetic to the manufacturer's argument that it is appropriate to use the observed Kaplan–Meier data when trial data were almost complete.

4.12 The Committee considered the modelling of progression‑free survival in the manufacturer's base‑case analysis. The Committee noted that time to treatment discontinuation was used as a proxy for disease progression in the economic model. The Committee was aware that the manufacturer was unable to determine the proportion of patients among those who discontinued therapy who met criteria for disease progression. The Committee agreed with the ERG that using time to treatment discontinuation as a proxy for disease progression was acceptable because of concerns about using radiographic imaging to monitor disease progression in prostate cancer. However, the Committee considered that, although time to treatment discontinuation may provide a reliable estimate of treatment costs in the model, it was less clear whether it provided a reliable estimate of the QALY gains associated with being in the pre‑progression state. The Committee was also aware that disease is likely to progress before patients stop abiraterone. However, the Committee concluded that changing the duration of progression‑free survival did not significantly alter the ICERs associated with treatment with abiraterone relative to treatment with prednisolone.

4.13 The Committee noted the manufacturer's inconsistent approach to extrapolating progression‑free survival beyond the data observed in the trial, with extrapolation beyond the time at which 5% of the population remained for abiraterone, and no extrapolation for prednisolone. The Committee also considered the ERG's concerns that the shape of the Kaplan–Meier survival curve for treatment discontinuation in the prednisolone group was unusual, with a high proportion of patients discontinuing treatment over a narrow time period after approximately 60 days of treatment. The Committee agreed that this was unlikely to represent actual disease progression and that a well‑fitting parametric distribution should have been used. The Committee noted that when the manufacturer used a Weibull parametric distribution to extrapolate progression‑free survival beyond the 5% cut‑off for abiraterone, the base‑case ICER increased slightly from £46,800 to £47,200 per QALY gained. The Committee accepted that, although it would have been more appropriate to use a well‑fitting parametric curve to extrapolate progression‑free survival, data for time to treatment discontinuation in the COU‑AA‑301 trial were virtually complete and thus the impact on the ICER of using alternative assumptions was minimal.

4.14 The Committee considered the utility values used in the economic model for the pre‑progression and the post‑progression health states. The Committee was aware that the manufacturer had not provided EQ‑5D values for health states obtained directly from patients, which would have been in line with the preferred methods recommended by NICE, but had derived utility values for the pre‑progression state from an algorithm that mapped FACT‑P scores to EQ‑5D utility values from a separate cross‑sectional dataset of patients with castration‑resistant metastatic prostate cancer. The Committee also noted that this mapping algorithm produced utility values that differed according to treatment. The Committee was aware that patients contributing to the cross‑sectional dataset may have differed from the population in the COU‑AA‑301 trial and from patients who might receive abiraterone in the UK. The Committee also heard from the manufacturer that its mapping algorithm had not been externally validated. The Committee noted that the mapping algorithm resulted in pre‑progression utility values which were similar to or higher than utility values observed in the age‑matched general population. In the Committee's view this might not be reasonable because people with metastatic prostate cancer would be expected to have a poorer quality of life than people without prostate cancer. However, it heard from the manufacturer and consultees that, because patients in the COU‑AA‑301 trial had few comorbidities and had been fit enough to receive chemotherapy, it was not implausible that they would have a similar health‑related quality of life to people of the same age in the general UK population. The Committee also noted that the utility values for the pre‑progression state were slightly higher than those used in the appraisal of cabazitaxel for metastatic hormone‑refractory prostate cancer (TA255) (derived from interim analysis of a small study). The Committee acknowledged a sensitivity analysis from the manufacturer which showed that when a published utility value of 0.715 (from the UK subgroup of Sullivan et al. 2007) was assigned to the pre‑progression state, the ICER increased to £51,110 per QALY gained for abiraterone compared with prednisolone. However, the Committee was aware that the utility value taken from Sullivan et al. was based on a small patient subgroup and that this study may have included patients at different stages of prostate cancer. Additionally, the Committee also heard from one clinical specialist that the estimated utility gain for abiraterone compared with prednisolone may have been underestimated and, as a result, the ICER may have been overestimated. The Committee concluded that there was uncertainty about the validity of the utility values for the pre‑progression health state derived from the manufacturer's FACT‑P mapping algorithm, but that no other robust utility value for the pre‑progression health state was currently available.

4.15 The Committee noted that the COU‑AA‑301 trial did not collect FACT‑P data after patients stopped treatment and therefore the manufacturer identified a utility value for the post‑progression state from the literature (see section 3.18). This value was lower than that estimated by the ERG from the manufacturer's FACT‑P mapping algorithm for patients at the end of treatment. The Committee acknowledged that the utility value estimated from patients at the end of treatment in the COU‑AA‑301 trial was based on a small group of patients healthy enough to complete the questionnaire after treatment discontinuation. The Committee also heard from the manufacturer that this utility value was measured at the start of progression and would not reflect the mean utility experienced by a patient throughout progression up to the time of death. The Committee acknowledged that although patients were considered to be in the pre‑progression state for the purposes of the model, they already had metastatic disease and would be unlikely to have the decrease in utility modelled by the manufacturer when progressing further to the post‑progression state (defined only by stopping treatment). However, the Committee noted that a patient's health‑related quality of life could be very poor in the last months of life and that the post‑progression utility value should also reflect this. Additionally, the Committee heard from the manufacturer that the utility difference between the pre‑progression and post‑progression health states was within the range used in recent technology appraisals of treatments for metastatic and advanced solid tumours. The Committee noted that a smaller utility difference between the pre‑progression and post‑progression health states would increase the ICER. The Committee concluded that uncertainty remained about the true difference in utility values between the pre‑progression and post‑progression states in the economic model, but that no other robust utility values that correctly capture the changes in health‑related quality of life in progressed disease were currently available.

4.16 The Committee considered the assumptions related to resource use in the manufacturer's economic model. The Committee noted that the ERG was generally satisfied with these assumptions. The Committee also noted that, in response to the appraisal consultation document, the manufacturer had made corrections to resource use suggested by the ERG (see section 3.32). The Committee was informed by the ERG that the manufacturer had incorrectly applied a half‑cycle correction to the drug costs in the model, and that changing this increased the drug costs by approximately half the monthly cost. The Committee was satisfied that the change to the half‑cycle correction to drug costs would probably result in only small increases in the ICER.

4.17 The Committee considered the most plausible ICERs presented by the manufacturer and also by the ERG in its exploratory analyses. The Committee agreed that, for nearly all analyses presented, mitoxantrone was extendedly dominated by abiraterone. In the Committee's view, a reasonable starting point for its decision was the manufacturer's base‑case ICER for abiraterone plus prednisolone compared with prednisolone alone of £46,800 per QALY gained for the one prior chemotherapy subgroup (when the discount agreed in the patient access scheme was included). The Committee agreed that the ICER would increase by a small amount if the model correctly accounted for the half‑cycle correction to drug costs. The Committee noted that use of a lower utility value for the pre‑progression health state or the assumption of a smaller difference in utility between the pre‑progression and post‑progression health states would further increase the ICER. However, the Committee agreed that more reliable utility values for the pre‑progression and post‑progression health states were not available. The Committee also noted that the ICER would increase slightly if a parametric curve were used to model overall and progression‑free survival. However, the Committee agreed that it was acceptable to use the observed Kaplan–Meier data given the completeness of the survival data in the COU‑AA‑301 trial. The Committee therefore agreed that once these factors had been taken into account, the most plausible ICER was likely to be higher than the manufacturer's base‑case estimate for the one prior chemotherapy subgroup, but would be under £50,000 per QALY gained.

4.18 The Committee was mindful that the NICE 'Guide to the methods of technology appraisal' states that there should be a strong case for accepting an ICER above £30,000 per QALY gained and judgements should specifically take account of:

4.19 The Committee discussed whether the assessment of the change in health‑related quality of life had been inadequately captured in the economic analysis. The Committee considered that abiraterone may offer a step change in treatment because it is life‑extending rather than only palliative but that this element of innovation would already be accounted for when moving from an ICER of £20,000 to £30,000 per QALY gained. The Committee concluded that abiraterone offers a step change in treatment because it is an oral drug taken by patients at home, and is associated with few adverse reactions. The benefit related to being an oral drug was not captured in the analysis because the model applied the same utility benefit to abiraterone as to mitoxantrone. The Committee therefore acknowledged that abiraterone provides health‑related quality of life benefits other than those captured in the QALY calculation for patients currently receiving mitoxantrone, and that the ICER would decrease when these benefits were taken into consideration. The Committee heard that a portion of the profits from sales of abiraterone would contribute to publically funded medical research within the UK. However, the Committee concluded that although health benefits were likely to accrue from this research, these did not contribute to health‑related quality of life as defined within the 'Guide to the methods of technology appraisal'.

4.20 The Committee considered supplementary advice from NICE that should be taken into account when appraising treatments that may extend the life of patients with short life expectancy and that are licensed for indications that affect small numbers of people with incurable illnesses. For this advice to be applied, all the following criteria must be met:

4.21 The Committee discussed whether abiraterone fulfilled the criteria for consideration as a life‑extending, end‑of‑life treatment. For castration‑resistant metastatic prostate cancer that has progressed after first‑line chemotherapy, the Committee agreed that the first criterion related to life expectancy was fulfilled, because life expectancy from trials including patients randomised to best supportive care was less than 15 months. The Committee considered by how much abiraterone extended life. It noted that in the manufacturer's base‑case economic analysis the estimated mean overall survival gain for abiraterone was greater than 3 months (median overall survival gain 4.6 months; mean overall survival gain commercial in confidence). The Committee concluded that an improvement of more than 3 months in mean overall survival had been robustly demonstrated. The Committee was aware that the population included in the marketing authorisation was likely to be larger than that for cabazitaxel given that abiraterone can be administered at home, and that some patients who would be unlikely to tolerate chemotherapy might take abiraterone. The Committee understood from estimates obtained from the appraisal of cabazitaxel that approximately 3500 people with hormone‑refractory metastatic prostate cancer received docetaxel in England and Wales in 2011, and that, according to estimates provided by the manufacturer, approximately 70% of these (n=2,500) would be able to receive second‑line abiraterone treatment in line with the marketing authorisation. The Committee noted from the manufacturer's response to the appraisal consultation document that in the future abiraterone may be licensed for use earlier in the treatment of castration‑resistant prostate cancer but acknowledged that any future extensions to the marketing authorisation remain uncertain and should not be taken into consideration when estimating the number of patients currently eligible for treatment. In the Committee's view, abiraterone was licensed for a small population and therefore meets the criteria for an end‑of‑life treatment. The Committee concluded that the additional weight to be assigned to the original QALY benefits in this patient group fell within the range considered acceptable for an end‑of‑life treatment. The Committee therefore recommended abiraterone in combination with prednisolone or prednisone as an option for the treatment of castration‑resistant metastatic prostate cancer that has progressed on or after one docetaxel‑containing chemotherapy regimen.

4.22 The Committee discussed whether any equality issues required consideration in this appraisal. The Committee was aware that people who are undergoing or have completed male to female gender reassignment can develop prostate cancer. The Committee therefore concluded that this appraisal should refer to 'people' rather than to men. Furthermore, the Committee had been made aware that these patients may find it uncomfortable to attend male urology clinics; however, the Committee agreed that the treatment of prostate cancer would be likely to be provided in oncology clinics.

Summary of Appraisal Committee's key conclusions