Abiraterone for treating newly diagnosed high-risk hormone-sensitive metastatic prostate cancer

3.1 The clinical experts explained in the first committee meeting in May 2018 that, in clinical practice, people with newly diagnosed hormone-sensitive metastatic prostate cancer have androgen deprivation therapy (ADT) alone or docetaxel plus ADT plus the oral corticosteroid prednisolone (from now on, 'docetaxel in combination'). NICE's guideline on prostate cancer recommends ADT in the form of continuous luteinising hormone-releasing hormone agonists, bilateral orchidectomy (removal of the testicles) or bicalutamide with ADT. The guideline also recommends docetaxel. Docetaxel is now licensed for hormone-sensitive metastatic prostate cancer, and NHS England commissions it for up to 6 cycles at this point in the treatment pathway. Docetaxel is administered intravenously with oral prednisolone 5 mg twice daily for 3 weeks. The clinical experts explained that orchidectomy and bicalutamide are rarely used in the NHS. The committee was aware that NICE's technology appraisal guidance on enzalutamide plus ADT for treating hormone-sensitive metastatic prostate cancer and apalutamide plus ADT for hormone-sensitive metastatic prostate cancer started during this appraisal, and enzalutamide plus ADT was recommended. Because the final guidance for enzalutamide plus ADT was not published by the time of the fifth committee meeting, it did not become a comparator in this appraisal. The committee agreed that ADT includes luteinising hormone-releasing hormone agonists. It concluded that ADT alone and docetaxel in combination were appropriate comparators to abiraterone plus ADT plus 5 mg of the oral corticosteroid prednisone (from now on, 'abiraterone in combination').

3.2 The committee recognised its obligation to appraise technologies across their marketing authorisations. This meant that, if abiraterone in combination was not cost effective across its marketing authorisation, the committee could consider a narrower population if there was a case for this. Although the company's original submission was for the full population covered in the marketing authorisation, it noted at the time that there was a group of people for whom docetaxel was clinically unsuitable or who chose not to have docetaxel. The company proposed abiraterone in combination as an alternative treatment to ADT alone for this group, which it termed 'chemotherapy ineligible' in its submission for the third committee meeting. The Cancer Drugs Fund's clinical lead noted in the third meeting that, at that time (January 2020), up to two-thirds of people presenting with hormone-sensitive metastatic prostate cancer in England had ADT alone. The committee noted that this included both people for whom docetaxel is contraindicated or unsuitable (the terminology agreed by the company, experts and committee in the fourth meeting) and people who choose not to have docetaxel. A patient expert explained that there is an unmet need for an alternative treatment option to ADT alone for this group. The committee recognised that most people who choose to have ADT alone rather than docetaxel in combination may wish to avoid the adverse events associated with docetaxel. The committee considered that this unmet need was in part addressed by NICE recommending enzalutamide plus ADT at this point in the pathway, which occurred during the course of this appraisal. The committee agreed that patient choice was important but that, for people who could have docetaxel, the comparators should be docetaxel in combination and ADT alone. If clinically and cost effective, abiraterone would be recommended as an option for the populations who can and cannot have docetaxel (from now, referred to as the 'whole population'). The committee agreed that it would be appropriate to define a group of people for whom docetaxel is unsuitable. It also agreed that the clinical and cost effectiveness of abiraterone in combination in this group should be considered if it was not cost effective for the whole population.

3.3 In its fourth meeting, the committee set out to understand more clearly how experts define people for whom docetaxel is contraindicated or unsuitable. The clinical and patient experts had previously explained that, although there are contraindications for docetaxel, defining the group for whom it is unsuitable is complicated. The committee had been aware that NHS England's clinical commissioning policy statement for docetaxel in combination with ADT indicates that someone may not be fit enough for docetaxel if they have a poor overall performance status (World Health Organization [WHO] performance 3 to 4), pre-existing peripheral neuropathy, poor bone marrow function or a life-limiting illness. The policy also states that docetaxel should be used with caution in people with a WHO performance status of 2 and that there are few absolute contraindications for docetaxel therapy. The Cancer Drugs Fund's clinical lead explained that many factors besides a person's performance status may affect whether they could have docetaxel. One of these is patient choice after hearing the risks and benefits of each available treatment. In the fourth meeting, the clinical experts explained that, while creating an exhaustive list of criteria for this group is unfeasible, developing a framework would be possible. The clinical lead for the Cancer Drugs Fund explained that a clinician assesses whether docetaxel is suitable for someone based on contraindications, fitness, comorbidities and preference. An oncologist would identify and discuss with a patient the individual risks and benefits associated with any treatment option before starting treatment. People for whom docetaxel is contraindicated or unsuitable would include:

3.4 The clinical experts explained that people who have docetaxel as first-line treatment in the hormone-sensitive setting can have docetaxel again for up to an additional 10 cycles in the hormone-relapsed setting. This is because the benefit of docetaxel is not exhausted when used for only 6 cycles. The Cancer Drugs Fund's clinical lead explained that abiraterone and enzalutamide are commissioned by NHS England only once in the treatment pathway. This is because there is no evidence of substantial clinical benefit for enzalutamide after abiraterone or for abiraterone after enzalutamide. The committee understood that people who have abiraterone in combination for hormone-sensitive prostate cancer have fewer options for active follow-on treatments (when the disease becomes hormone refractory) than people who start with ADT alone, or docetaxel in combination. This is because people who started with abiraterone cannot have abiraterone or enzalutamide later in the treatment pathway. The committee noted that the sequence of follow-on treatments when disease becomes hormone refractory may vary from person to person, but that ADT would be continued alongside subsequent therapies. It considered that possible follow-on treatments include:

3.5 Two randomised controlled trials have investigated the clinical effectiveness of abiraterone in combination in hormone-sensitive metastatic disease:

3.6 To compare abiraterone in combination with docetaxel in combination, the company was concerned that results from the subgroup of people with metastatic disease in STAMPEDE did not reflect the population included in the licence for abiraterone for the proposed indication (see section 3.5). The company further stated that STAMPEDE was not statistically powered to detect a difference in survival in this post-hoc analysis. The company instead developed a network meta-analysis which, as well as including the direct data from STAMPEDE, included several other trials. The company argued that additional trials contributed information to the estimated treatment effect of abiraterone compared with docetaxel. The trials included in the network were:

3.7 Abiraterone in combination improved both progression-free and overall survival compared with ADT alone in LATITUDE and in people with high-risk metastatic disease in STAMPEDE. The size of the relative improvement for abiraterone plus ADT compared with ADT alone was similar in the 2 trials. In LATITUDE, median progression-free survival was 14.8 months with ADT alone and 33.0 months with abiraterone in combination (hazard ratio [HR] 0.47, 95% confidence interval [CI] 0.39 to 0.55). Based on the planned final analysis of overall survival, the median overall survival with ADT alone was 36.5 months compared with 53.3 months with abiraterone in combination (HR 0.66, 95% CI 0.56 to 0.78). In STAMPEDE, at a median follow up of 3.3 years, the HR for progression-free survival in the high-risk metastatic subgroup was 0.46 (95% CI 0.36 to 0.59), and for overall survival was 0.54 (95% CI 0.41 to 0.70). Data from STAMPEDE shared after the appeal showed that the HR for overall survival was maintained after 8 years of follow up (HR 0.54, 95% CI 0.43 to 0.69). The committee noted that the survival benefit reported in LATITUDE was larger than that seen in clinical trials using abiraterone in hormone-relapsed disease, in which median overall survival was extended by 4.4 months compared with ADT alone. However, it acknowledged that its remit was to appraise abiraterone in the proposed indication and not to identify the best order and sequence of treatments. The committee concluded that abiraterone in combination improved both progression-free and overall survival compared with ADT alone. However, it noted that there was uncertainty about the magnitude of the long-term survival gain with abiraterone in combination. This was because of potential differences in the proportion of people who had life-extending treatments after disease progression on ADT in LATITUDE and STAMPEDE compared with clinical practice (see section 3.9).

3.8 In people with metastatic disease in STAMPEDE, abiraterone in combination improved progression-free survival compared with docetaxel in combination (HR 0.69, 95% CI 0.50 to 0.95). However, for overall survival, the HR favoured docetaxel (HR 1.13, 95% CI 0.77 to 1.66). In the company's updated base case, rather than use the results reflecting a direct comparison from STAMPEDE, it used the results of the indirect network meta-analysis that included data from LATITUDE, CHAARTED, GETUG‑AFU 15 and STAMPEDE. This showed similar results to the direct comparison for progression-free survival. However, the point estimate for overall survival favoured abiraterone, but the credible interval included 1, that is, the possibility of no difference in benefit of 1 treatment over the other. The company considered these values to be academic in confidence so they cannot be reported here. Considering the direct and indirect comparisons, the committee concluded that abiraterone in combination improves progression-free survival, but not overall survival compared with docetaxel in combination.

3.9 The committee recognised that life-extending treatments offered when the disease is no longer hormone sensitive (that is, is hormone relapsed) affects life-expectancy. Follow-on treatments in the unblinded UK STAMPEDE trial were expected to reflect what people would have in NHS clinical practice, for example, not getting abiraterone twice. This was because the choice of next treatment depended on knowing the first treatment. In STAMPEDE, people were aware of their treatment but, in the blinded LATITUDE trial, people were not. The committee noted that the trials differed from UK clinical practice in 2 ways:

3.10 No evidence was presented for abiraterone's relative effectiveness compared with ADT alone, specifically for the group for whom docetaxel is contraindicated or unsuitable. LATITUDE and STAMPEDE, the key clinical trials of abiraterone in this indication (see section 3.5), only included people with adequate haematological function, and an ECOG or WHO performance status of 0, 1 or 2 (meaning they were reasonably fit). Also, they could not have any condition that would interfere with them taking part in the trial. As part of the initial protocol, all people recruited to STAMPEDE had to be able to have docetaxel because it was 1 of the arms in the trial. A clinical expert explained that, in 2013, that arm closed but the trial continued to recruit to the ADT alone and abiraterone-in-combination arms (among others). This meant people recruited to the trial from this point could have included people for whom docetaxel was contraindicated or unsuitable. They noted that James et al. (2017) presented an analysis showing that the HR for overall survival for abiraterone in combination compared with ADT was 0.69 (95% CI 0.53 to 0.90) between November 2011 and January 2013. This was when the docetaxel arm of the trial was open. After the docetaxel arm closed, the HR for data collected between April 2013 and January 2014 was 0.59 (95% CI 0.44 to 0.78). The ERG highlighted, and the committee agreed, that this did not provide the evidence specifically for the group for whom docetaxel was contraindicated or unsuitable. This was because the 2013 to 2014 data would also have included people who could have docetaxel. The committee heard during its fourth meeting that STAMPEDE enrolled people who could take abiraterone but could not have docetaxel. However, in response to the consultation after the committee's fourth meeting, the company and consultees commented that data specific to this subgroup could not be separated from the metastatic, high-risk cohort in STAMPEDE. This was because the trial did not collect data on all the baseline characteristics specified in the suggested clinical framework describing people who cannot or should not have docetaxel (see section 3.3). The committee was aware that many of these characteristics were similar to the STAMPEDE exclusion criteria. It noted that stakeholders had explored several further data sources for people who are 'chemotherapy ineligible' but identified none that would resolve the uncertainty about clinical and cost effectiveness in this population. The company agreed that it was reasonable for the committee to request data specific to people who cannot or should not have docetaxel. However, the company did not consider it ethical to conduct a new clinical trial specifically in this group because trials have already shown that abiraterone is superior to ADT alone in the whole population. Acknowledging the uncertainties in the existing data, the company preferred to use the treatment effect from the whole population to represent the group for whom docetaxel is contraindicated or unsuitable. The committee concluded that assessing data specific to the relevant population was preferred, but acknowledged that this evidence was not available. It therefore considered the treatment effect from the LATITUDE whole population for people who cannot or should not have docetaxel in its decision making, while acknowledging this uncertainty.

3.11 No data were presented specifically for the group of people who cannot have docetaxel (see section 3.10). So, the committee discussed the uncertainties associated with applying the treatment effect of abiraterone in combination for the whole trial population to people who cannot or should not have docetaxel. It did this by looking at the treatment effect in people whose baseline characteristics meant they were unlikely to be able to have docetaxel. It was aware that older people (see section 3.23) and people with poorer levels of performance status would be less likely to have docetaxel. It noted the available evidence from LATITUDE and STAMPEDE on subgroups based on age and performance status. For the comparison of abiraterone in combination with ADT alone from STAMPEDE, the committee noted effect modification by age. For overall survival, abiraterone was not as effective in people 70 years and over (HR 0.94, 95% CI 0.69 to 1.29) compared with people under 70 years (HR 0.51, 95% CI 0.40 to 0.65; test for interaction p value 0.003) (James et al. 2017). The company argued that these results were not representative of the population covered by the licence for the proposed indication. This was because they included people with non-metastatic prostate cancer, who are generally younger and have earlier stage disease. The committee noted a similar pattern in LATITUDE, in which the HR for overall survival for abiraterone in combination compared with ADT alone was: 0.65 (95% CI 0.50 to 0.84) for people under 65 years; 0.68 (95% CI 0.55 to 0.83) for people 65 years and over and 0.86 (95% CI 0·62 to 1·21) for people 75 years and over (Fizazi et al. 2019). In response to the second consultation, the company stated that the test for interaction based on age was non-significant, suggesting that it was possible that age had no effect on treatment response. However, it also acknowledged in the fifth committee meeting that the test may be underpowered to detect a difference. The committee also noted a HR of 0.64 (95% CI 0.50 to 0.75) in people with an ECOG status of 0 or 1, and a HR of 1.42 (95% CI 0.65 to 3.08) for those with an ECOG status of 2 in LATITUDE (Fizazi et al. 2019). It recognised that age alone would not determine whether a person could have docetaxel, but that age was associated with decreased docetaxel use based on data provided by BUG. It also heard that people with a poor performance status may not get abiraterone. The committee agreed that abiraterone appears less effective among people with characteristics shared by people who cannot or should not have docetaxel based on subgroup data. However, it noted that subgroup data should be interpreted with caution when based on data from a small group of people (40 out of 1,159 people in LATITUDE had a performance status of 2). The committee recalled that the subgroups did not specifically reflect the population for whom docetaxel is contraindicated or unsuitable. Overall, the committee concluded it was not possible to determine the size of the effectiveness for people for whom docetaxel is contraindicated or unsuitable. It therefore chose to look at its preferred effect measures from existing data (the treatment effect seen in the whole LATITUDE population; see section 3.10), acknowledging the uncertainty.

3.12 Both the relative benefits of treatment with abiraterone compared with ADT alone, and the baseline risks of progression and dying, may differ between populations who can and cannot have docetaxel. The committee recognised that many of the risk factors for not having docetaxel (for example, age and poor performance status) are also risk factors for dying. It concluded that any modelling should take this into account (see section 3.15).

3.13 The company considered that the results of LATITUDE could be generalised to people who cannot or should not have docetaxel in its modelling (see section 3.15). The clinical experts explained that, with some exceptions, people with hormone-sensitive disease for whom docetaxel is contraindicated or unsuitable would not have docetaxel after their cancer progressed to being hormone relapsed. People in LATITUDE and STAMPEDE could go on to have docetaxel after abiraterone in combination or ADT alone. This did not reflect the treatment pathway for people who cannot have docetaxel, and yet was reflected in the trial results. The evidence of clinical effectiveness from LATITUDE does not reflect the treatment pathway for people for whom docetaxel is contraindicated or unsuitable. The committee therefore again agreed that it had not been presented with data on the effectiveness of abiraterone in combination compared with ADT in people for whom docetaxel is contraindicated or unsuitable. It concluded that estimates of overall survival from LATITUDE included the effect of taking docetaxel when indicated for hormone-relapsed metastatic disease, which would not apply to people who cannot or should not have docetaxel.

3.14 The company provided 2 models. In its original submission, it provided a multistate Markov model. The committee deemed that this did not provide plausible estimates of post-progression or overall survival and did not generate valid estimates of cost effectiveness. In its submission for the third committee meeting, the company provided a partitioned survival model. Both models were split into 2 phases:

3.15 For the third committee discussion, the company provided estimates of the cost effectiveness of abiraterone in combination compared with ADT alone for a population it referred to as 'chemotherapy ineligible' (see section 3.2). The committee noted that the company based these estimates on data for the whole population. The committee recognised that the clinical data used in the model may not be generalisable to the 'chemotherapy ineligible' group. It also recognised that the modelled treatment pathway did not reflect the treatments people would have for whom docetaxel is contraindicated or unsuitable. Specifically:

3.16 The committee agreed with the company that the hazards of progression and death for abiraterone in combination compared with ADT alone from LATITUDE were not proportional. It concluded that it was appropriate to fit curves to each arm separately. During the committee's third meeting, the company presented results for both progression-free and overall survival. It used the log-logistic distribution for each modelled treatment arm (which the company considered plausible but optimistic) and the Weibull distribution (which it considered plausible but pessimistic). The committee considered that the Weibull curves were plausible for progression-free survival. It noted that results from STAMPEDE for abiraterone in combination compared with ADT (submitted by BUG after the appeal for the fourth meeting) represented an 8‑year follow up. These suggested an ongoing benefit with abiraterone in combination compared with ADT alone (HR 0.54, 95% CI 0.43 to 0.69). The results supported using the log-logistic distribution to extrapolate overall survival from LATITUDE. At the third meeting, the ERG highlighted that a consequence of the model was that the company assumed the treatment effect to be maintained over the long term. However, in clinical practice, it may wane and the ERG provided scenarios to adjust for this at that time. The committee noted that the longer-term data from STAMPEDE could be used to determine the validity of these adjustments. It concluded that the progression-free survival extrapolation using the Weibull distribution was broadly appropriate for the overall population. However, it did not see evidence of extrapolating outcomes in people for whom docetaxel is contraindicated or unsuitable. The committee recognised that these people, being on average older, may have a different pattern of mortality. It further concluded that the predicted overall survival based on extrapolations using the log-logistic distribution was plausible for the overall trial population. The committee would have preferred to see further extrapolations exploring alternative time points for equalising the hazard using 2 plausible curves: 1 for the whole population, and another for people for whom docetaxel is contraindicated or unsuitable.

3.17 The company considered separately the effects on quality of life of adverse effects and of being on treatment. The sources of these data are in table 1.

The company used different approaches to estimate the effect on quality of life of having abiraterone in combination or ADT alone than it did to estimate the effect with docetaxel in combination. It sourced utility values for being on abiraterone in combination from EQ‑5D results from LATITUDE, and for being on docetaxel in combination from a separate preference study of the general public that it had carried out. The NICE methods guide states that EQ‑5D is the preferred measure of health-related quality of life. The committee noted that STAMPEDE collected EQ‑5D data for a UK population randomised to abiraterone in combination, to docetaxel in combination and to ADT alone. In response to the first consultation and in the third committee meeting, the company confirmed that it did not request or have access to these data. The ERG carried out a scenario using the disutility estimate for docetaxel from the economic evaluation of docetaxel in combination in NICE's guideline for prostate cancer. The ERG derived the disutility value from EQ‑5D data collected in STAMPEDE (whole population and metastatic subgroup). The company stated that the ERG's and company's scenarios were consistent, irrespective of the disutility source used. The committee considered that the effectiveness data from the metastatic subgroup from STAMPEDE were generalisable to the higher-risk population under appraisal (see section 3.5). However, it considered that it was plausible that the level of risk affects quality of life. It concluded that it was preferable to use EQ‑5D data from the subgroup of people from STAMPEDE with high-risk hormone-sensitive metastatic prostate cancer to assess quality of life. It further noted that comparable data were available for abiraterone in combination, docetaxel in combination and ADT alone. The committee then went on to consider evidence submitted by BUG after the appeal. This showed the results of a quality-of-life study from STAMPEDE using data relevant to people with metastatic disease but not limited to high-risk disease. The data used a different measure of quality of life to EQ-5D. Results supported an improved quality of life for abiraterone in combination compared with docetaxel in combination. However, the difference was clinically meaningful according to predefined criteria only for the first 6 months after starting treatment. The committee considered that the findings supported the company's modelling that showed a worse quality of life with docetaxel in combination compared with abiraterone in combination. It also agreed that the ERG's estimate was likely to be broadly appropriate. The committee concluded that, in the absence of STAMPEDE trial data from the company, the company's and ERG's approaches were broadly consistent with each other and acceptable for decision making.

3.18 In response to the committee's second meeting, the company revised the treatment pathways in the hormone-relapsed state to reflect NHS market shares of treatments for hormone-relapsed disease. It based its estimates of market shares on the opinion of 4 clinicians, which the committee concluded may not reflect the actual market shares in UK clinical practice. The company assumed that:

3.19 The committee agreed that its preferred approach to modelling would reflect the company's base case with the following assumptions:

3.20 The prices of abiraterone made by Janssen and technologies made by company's other than Janssen are confidential. This meant that estimates of the ICERs could not be presented in the final appraisal determination that followed the committee's third meeting. In this meeting, NICE asked the committee to consider abiraterone at list price for this indication (and abiraterone at a discount for its other indications). This was because the company and NHS had not agreed a discounted price. The final appraisal determination stated that the cost-effectiveness estimates without a commercial arrangement were considerably higher than the range normally considered a cost-effective use of NHS resources. The company appealed, and the appeal panel concluded that this was not transparent. At the time of the fourth committee meeting, the company agreed a confidential discounted price for abiraterone but noted it did not wish NICE to publish a narrow range of ICERs. This was because the cost-effectiveness estimates using list prices had been published previously and the company stated that this would allow back calculation of its discount for abiraterone. The committee addressed this appeal point (see section 3.21) by stating a figure above which the ICER lies. It supported transparency in reporting cost effectiveness, and agreed that this did not fully address the appeal panel's suggestion. However, the company acknowledged in the fifth meeting that it had not agreed to a range of cost-effectiveness estimates being available for publication for the reasons stated above. So, the committee could not provide a more accurate representation of the ICER. The committee concluded that it strongly supported being more transparent in reporting estimates of cost effectiveness when possible without compromising confidential commercial arrangements.

3.21 The committee noted its duty to appraise technologies across their marketing authorisation. At the confidential discount chosen by the company, the company's base case with the appraisal committee's preferred modelling assumptions was:

3.22 The committee recognised that there was an unmet need for another treatment option to ADT in people for whom docetaxel is contraindicated or unsuitable. However, it agreed that this need was lessened by enzalutamide plus ADT having been recommended for this indication. It noted that the benefits of abiraterone being an oral treatment that could be taken at home were not captured in the model. It also acknowledged the increasing need for oral treatments during the COVID‑19 pandemic. The committee concluded that abiraterone may be associated with benefits unaccounted for in the modelling.

3.23 The committee noted that, as in previous NICE technology appraisals of technologies for treating prostate cancer, its recommendations should apply to anyone with a prostate. It also noted that, in clinical practice, older people are less likely to have docetaxel than younger people based on NHS data from appellants presented at the fourth committee meeting. The clinical experts explained that, although docetaxel is more likely to be contraindicated or unsuitable for older people, age itself will not determine whether a person could or should have docetaxel in clinical practice. The committee was aware that making recommendations by age to reflect people who cannot or should not have docetaxel could discriminate against younger people for whom docetaxel is contraindicated or unsuitable. It noted the appeal panel's conclusions that: 'the current reasoning around the failure to define this subgroup does not address the fact that the subgroup will tend to comprise older men'. It also noted that the appeal panel: 'wishes to be clear that although equality legislation requires this subgroup to be more fully considered it does not necessarily follow that in this case, after appropriate consideration, special provision will need to be made for them'. The committee concluded that its recommendations should apply to all people with prostate cancer and not discriminate on the basis of age.

3.24 The committee noted that 22 points were discussed in the appeal, 16 of which were dismissed by the appeal panel and 6 of which were upheld. The upheld points related to transparency (see section 3.20), equality issues (see section 3.23) and defining a population for whom docetaxel is contraindicated or unsuitable (see section 3.3). The appeal panel stated that the committee should: