Enzalutamide for treating hormone-sensitive metastatic prostate cancer

3.1 The clinical and patient experts noted that people with hormone-sensitive metastatic prostate cancer have limited treatment options. NICE's guideline on prostate cancer recommends androgen deprivation therapy (ADT) alone, and docetaxel with prednisolone or prednisone (from now, docetaxel) plus ADT. The patient experts explained that, when people are first diagnosed with metastatic prostate cancer, they may have no or few symptoms. They also explained that some people perceive that treatment with docetaxel worsens quality of life and choose to have ADT alone, even though the long-term outcomes may be worse than with docetaxel plus ADT. So, because enzalutamide plus ADT is generally better tolerated than docetaxel plus ADT, and is more effective than ADT alone, people would welcome it as an option at this point in the treatment pathway. The committee concluded that some people with hormone-sensitive metastatic prostate cancer would welcome the option of treatment with enzalutamide plus ADT.

3.2 NICE's rapid guideline on the delivery of systemic anticancer treatments during the COVID-19 pandemic aims to:

3.3 The committee was aware that the NICE scope included as comparators, ADT alone and docetaxel plus ADT. The clinical experts explained that both are offered in the NHS to people with hormone-sensitive metastatic prostate cancer. The committee was also aware of the ongoing NICE technology appraisals on abiraterone plus ADT for newly diagnosed high-risk hormone-sensitive metastatic prostate cancer and apalutamide for treating prostate cancer. The committee concluded that neither abiraterone plus prednisone (from now, abiraterone) and ADT nor apalutamide plus ADT were comparators because they are not routinely commissioned at this point in the treatment pathway. The Cancer Drugs Fund's clinical lead noted that around two-thirds of people presenting with hormone-sensitive metastatic prostate cancer in England have ADT alone. Of these people, some are not fit enough for docetaxel, and some choose not to have it because of its adverse events (see section 3.1). NHS England's docetaxel commissioning policy states that someone may not be fit enough for docetaxel if they have:

3.4 Under NHS policy, people who have docetaxel plus ADT for hormone-sensitive prostate cancer for up to 6 cycles can have docetaxel again (for up to 10 cycles) for hormone-relapsed prostate cancer. This is because the benefit of docetaxel is not exhausted. Other treatment options for hormone-relapsed metastatic prostate cancer include both enzalutamide and abiraterone when chemotherapy is not yet clinically indicated, or after a docetaxel-based regimen. In summary, enzalutamide has a licence for 4 positions in the treatment pathway (including 1 for non-metastatic prostate cancer). However, the Cancer Drugs Fund clinical lead explained that NHS England commissions each of enzalutamide and abiraterone only once in the treatment pathway. This is because there is no evidence of clinical benefit for using one after the other. It means that people who have enzalutamide plus ADT for hormone-sensitive metastatic prostate cancer cannot have enzalutamide or abiraterone later in the treatment pathway. It also means that people who have docetaxel plus ADT first for hormone-sensitive metastatic prostate cancer have more treatment options than people who have enzalutamide plus ADT first. This is because they can have either enzalutamide or abiraterone, and can also have docetaxel again. The sequence of follow-on treatments when the cancer is hormone-relapsed may vary from person to person. Possible treatments include:

3.5 Two randomised controlled trials, ARCHES and ENZAMET, have investigated the clinical effectiveness of enzalutamide plus ADT for treating hormone-sensitive metastatic prostate cancer:

3.6 The baseline characteristics of the people in ARCHES and ENZAMET were similar. However, more people in ARCHES had Gleason scores equal to or greater than 8, or high-volume disease. The proportion of people with high-volume disease in ARCHES was similar to that in STAMPEDE, an entirely UK-based trial assessing the best way to treat newly diagnosed advanced prostate cancer. The clinical experts agreed that people with high-volume disease have poorer prognoses than people with low-volume disease. However, they disagreed on whether disease volume modifies the relative effectiveness of treatment. One clinical expert noted that, in STAMPEDE, volume of disease did not alter treatment effectiveness. The committee noted that the evidence for enzalutamide plus ADT was based on a relatively fit population. It specifically excluded people with an Eastern Cooperative Oncology Group performance status of 2 or above, significant cardiovascular or renal disease, and other conditions. So, it may not be representative of some of the people who cannot have docetaxel. The committee appreciated that these issues could have added uncertainty to the results of the economic modelling. It concluded that the trials were broadly generalisable to NHS practice.

3.7 ARCHES compared enzalutamide plus ADT with ADT alone while ENZAMET compared it with conventional NSAAs plus ADT. The committee highlighted that using NSAAs does not reflect UK clinical practice and the company acknowledged this. However, the company did not think that it would affect the generalisability of the results to NHS clinical practice. The clinical experts confirmed that conventional NSAAs are not used in the NHS in this setting. They explained that evidence suggested the combination may be more effective than ADT alone, but that adverse events are increased when adding NSAAs to ADT. The committee concluded that the results of both trials were appropriate for decision making.

3.8 The 2 trials measured progression-free survival differently. In ENZAMET, it was defined based on clinical progression by radiographic imaging, symptoms attributable to cancer progression or starting another treatment for prostate cancer. This was broader than in ARCHES, in which progression-free survival was defined based on radiographic disease progression by an independent blinded and central review. The company chose only to model progression-free survival from ARCHES (see section 3.13). The clinical experts explained that there is more than a single way in clinical practice to assess progression-free survival, and that different centres might use different definitions. They confirmed that other measures might include serum prostate specific antigen. The committee concluded that it was appropriate to consider progression-free survival from both trials, and that the definition from ENZAMET better reflected NHS practice.

3.9 The company presented data from planned final analyses for progression-free survival from ARCHES and ENZAMET. However, the trials are ongoing for the endpoint of overall survival. In ARCHES, enzalutamide plus ADT improved progression-free survival compared with ADT alone. The time to median progression-free survival was not reached for enzalutamide plus ADT and, for ADT alone, was 19 months. The hazard ratio was 0.39 (95% confidence interval [CI] 0.30 to 0.50). In ARCHES, cancer progressed in 16% of people in the treatment group and in 35% in the control group. In ENZAMET, enzalutamide plus ADT improved progression-free survival compared with conventional NSAAs plus ADT (hazard ratio [HR] 0.34, 95% CI 0.26 to 0.44). The number of events was not available for ENZAMET. The median follow up was 14.4 months in ARCHES and 37.0 months in ENZAMET. At the same time as doing final analyses for progression-free survival, the investigators did interim data analyses for overall survival. In ARCHES, 84 deaths had occurred at the time of the interim analysis out of the 342 deaths specified in the statistical analysis plan for the final analysis. In ENZAMET, 245 deaths had occurred at the time of the interim analysis out of the specified 470 deaths. At the time of the interim analysis, most people were still alive in both trials, and median overall survival could not be estimated in any treatment arm. Interim analyses from both trials suggested that enzalutamide plus ADT improved overall survival (ARCHES: HR 0.81, 95% CI 0.53 to 1.25; ENZAMET: HR 0.53, 95% CI 0.37 to 0.74). The confidence interval for the estimate from ENZAMET included the possibility of no effect. The company presented an analysis for overall survival based on unadjusted pooling of patient-level data from both trials. The committee did not support unadjusted pooling of results (see section 3.11). The ERG did not use the unadjusted pooled data to estimate overall survival. Instead, it modelled overall survival using hazard ratios from the company's network meta-analysis applied to the ADT-alone overall-survival curve. The committee concluded that enzalutamide plus ADT delayed time to progression compared with ADT alone, but that the data for overall survival were immature. This meant that the size of the overall benefit of enzalutamide plus ADT compared with ADT alone was uncertain.

3.10 There are no trials that directly compare enzalutamide plus ADT with docetaxel plus ADT. The company did a network meta-analysis that included ARCHES and ENZAMET, 3 trials of docetaxel plus ADT compared with ADT alone (STAMPEDE1, CHAARTED and GETUG) and 6 trials of conventional NSAAs plus ADT compared with ADT alone (DAPROC, EORTC 30853, INTERGROUP STUDY 0036, SWOG‑8894, Janknegt 1993 and Zalcberg 1996). The results suggested longer progression-free survival for enzalutamide plus ADT compared with docetaxel plus ADT. For overall survival, the point estimate suggested a benefit for enzalutamide plus ADT but included the possibility of no effect. None of the results can be reported here because the company considers them confidential. The committee concluded that enzalutamide plus ADT extended progression-free survival when compared with docetaxel plus ADT, but evidence on overall survival was uncertain.

3.11 The committee discussed the implications of pooling data from the 2 trials, ARCHES and ENZAMET, with different treatments for the control arm. It was aware that the company's network meta-analysis (see section 3.10) included 6 trials of conventional NSAAs plus ADT compared with ADT alone. The results of the network meta-analysis also showed a benefit of NSAAs plus ADT compared with ADT alone, which included the possibility of no effect. To avoid using the pooled overall-survival data, the ERG estimated overall survival using hazard ratios from the company's network meta-analysis applied to the ADT-alone overall-survival curve. The committee concluded that, when comparing enzalutamide plus ADT to ADT alone, the network meta-analysis should inform the treatment effect for overall survival. It was aware that the data from ARCHES for ADT alone and ENZAMET for conventional NSAAs plus ADT would still need to be pooled to provide a reference curve for applying treatment effect hazard ratios.

3.12 The company presented a partitioned survival model that included 3 main health states: hormone-sensitive disease, hormone-relapsed disease and death. The hormone-sensitive health state included on- and off-treatment sub-states. The hormone-relapsed health state included 3 sub‑states for follow-on treatments. The committee concluded that the model structure was appropriate for decision making.

3.13 Both ARCHES and ENZAMET provided data on progression-free survival, but ENZAMET had a longer duration. According to the company, progression-free survival in ARCHES closely resembled that of ENZAMET. Therefore, it used patient-level data from ARCHES to model progression-free survival for enzalutamide plus ADT and ADT alone. It considered that it could not combine data from ENZAMET and ARCHES for progression-free survival because of the different ways in which this outcome was measured. The committee recalled that the definition of progression-free survival in ENZAMET more closely reflected that used in NHS clinical practice (see section 3.8). However, the hazard ratios from both trials were similar (ARCHES: HR 0.39 and ENZAMET: HR 0.34; see section 3.9). As such, the committee considered that using 1 trial instead of another was unlikely to have had a large effect on the cost-effectiveness results. It concluded that it was reasonable to use data from ARCHES if using data from only a single trial to model progression-free survival.

3.14 To extrapolate progression-free survival beyond the trial duration and over the lifetime horizon defined in the model, the company used data only from ARCHES (see section 3.13). It fitted a log-normal distribution to both arms of the trial. The committee noted that median follow up in ARCHES was only 14.4 months, and that cancer had not progressed in most people at the final analysis for progression-free survival. This increased the uncertainties associated with estimating average progression-free survival by treatment arm. The immaturity of the data also meant that most distributions fitted well to the observed trial data. The company based its choice of distribution, the log-normal distribution, on input from clinical experts. It externally validated its choice using data from long-term survival on ADT from STAMPEDE and GETUG. The ERG commented that, when extrapolating progression-free survival using the log-normal distribution, the 5- and 10‑year estimates for progression-free survival for ADT seemed implausibly low. The clinical experts considered that around 20% of people who take enzalutamide plus ADT remain progression free at 5 years, which drops to 10% at 10 years. Both values are lower than those suggested in the company's model. The ERG suggested that:

3.15 To model overall survival for ADT alone and enzalutamide plus ADT, the company pooled data from ARCHES and ENZAMET (see section 3.9). It extrapolated beyond the trial duration and over the lifetime horizon defined in the model by fitting a Weibull distribution to both arms. The company based its choice of a Weibull distribution on input from clinical experts, and by externally validating its choice using data from STAMPEDE, CHAARTED and GETUG. The ERG considered that the predictions for how long people survive who take ADT alone were reasonably consistent with long-term data from STAMPEDE. However, it was concerned that 10- and 20‑year figures reflecting the proportion of people still alive after taking enzalutamide plus ADT were implausibly high with the Weibull distribution. The clinical experts estimated that overall survival with enzalutamide plus ADT would be around 10% to 20% at 10 years, and 0% to 5% at 20 years. The company's modelling suggested that a greater proportion of people would be alive at 20 years than estimated by the clinical experts. The committee noted that, because of the immaturity of the data, most distributions provided similar predictions. Only the Gompertz distribution predicted lower survival for enzalutamide plus ADT than the company's choice. The ERG explained that the Gompertz distribution may have underpredicted survival on enzalutamide plus ADT at later years. The ERG preferred using the hazard ratio from the network meta-analysis applied to the ADT alone curve because it gave better predictions than other curves for enzalutamide plus ADT survival after around 10 years. The committee agreed with the ERG. It concluded that using hazard ratios for enzalutamide plus ADT compared with ADT alone from the network meta-analysis (see section 3.11) estimated the relative treatment effect for enzalutamide plus ADT compared with ADT alone better than the company's approach. This was because it accounted for the different comparators in ARCHES and ENZAMET better than the company's approach of unadjusted pooling. However, the committee was again concerned that this approach implied that the treatment effect would be expected to continue indefinitely.

3.16 To model overall survival with docetaxel plus ADT, the company applied hazard ratios from the network meta-analysis for docetaxel plus ADT compared with ADT alone to the ADT curve. This predicted a survival benefit for docetaxel plus ADT compared with ADT alone, reflecting trial evidence. The company's model also predicted a survival benefit with enzalutamide plus ADT compared with docetaxel plus ADT. The point estimate from the network meta-analysis favoured enzalutamide plus ADT, but the credible interval included 1, the possibility of no effect. The committee noted that people who take enzalutamide plus ADT have fewer life-extending treatment options later (see section 3.4). Therefore, it considered that there might be no survival benefit with enzalutamide plus ADT compared with docetaxel plus ADT. To explore this uncertainty, the ERG provided scenario analyses modelling no survival benefit for enzalutamide plus ADT compared with docetaxel plus ADT. The committee concluded that, given the uncertainty around the overall-survival estimate, it was appropriate to consider these analyses.

3.17 The committee acknowledged that people with hormone-sensitive metastatic prostate cancer:

3.18 The company's model predicted that the benefit for overall survival with enzalutamide plus ADT compared with ADT alone or docetaxel plus ADT lasted for the model's 30‑year time horizon. This was the case whether extrapolated survival curves were used to estimate enzalutamide's treatment effect compared with ADT or the hazard ratio from the network meta-analysis. Data from STAMPEDE showed that there was an initial survival benefit at 5 years with docetaxel plus ADT compared with ADT alone (49% compared with 37%). However, the ERG highlighted that there was no difference in actual overall survival after 8.5 years (23% compared with 22%). This may have been because people on ADT alone go on to have other life-extending treatments and so 'catch-up' (see section 3.4 and section 3.17). One clinical expert explained that the effect of early systemic treatment lasts for a long time, so catching up might be unlikely. He noted that there were longer follow-up data for abiraterone plus ADT than for enzalutamide plus ADT, and that he thought that both have a similar mechanism of action. These data for abiraterone showed that more people remained alive on abiraterone plus ADT than on docetaxel plus ADT or on ADT alone beyond 5 years. The ERG presented scenario analyses in which the hazards of survival for enzalutamide plus ADT and the comparators were the same after 8.5 years. The committee concluded that, in the absence of long-term data for enzalutamide plus ADT, the ERG's scenarios in which the hazard ratios were equalised between treatment options after 8.5 years were useful for assessing the uncertainty.

3.19 In the company's model, people could be on or off treatment before disease progression. The ERG was concerned that people who were off treatment before disease progression would maintain the same quality of life as people on treatment, but at no additional cost. In ARCHES, around 12% of people stopped treatment before disease progression. According to the company, only about half of them stopped because of adverse events, while others withdrew consent. The clinical experts explained that, in clinical practice, few people would stop having enzalutamide plus ADT before disease progression because it is generally well tolerated. The committee considered that withdrawing consent is specific to trials and would not be reflected in clinical practice. The ERG noted that, in the company's model, there was a substantial gap between the curves for progression-free survival and time to stopping treatment. The ERG proposed extrapolating the time to stopping treatment using the log-logistic rather than exponential distribution. It considered that this aligned more closely with the progression-free survival curve. The ERG further noted that, if enzalutamide plus ADT progression-free survival was estimated by applying hazard ratios from the network meta-analysis, the time to stopping treatment could not be modelled separately from progression-free survival. The committee concluded that the time to stopping treatment should have closely resembled progression-free survival. The committee agreed with the ERG using hazard ratios to estimate progression-free survival. So, it also concluded that, in this case, progression-free survival should be used to model treatment discontinuation.

3.20 The committee's first meeting occurred before the European Medicines Agency (EMA) granted a marketing authorisation. At this point, the committee agreed that its preferred approach to modelling would:

3.21 After the committee's first meeting, the appraisal was paused. After the EMA granted a positive opinion, the company updated its commercial offer and acknowledged the committee's preferred assumptions. The committee, in a second closed meeting, considered the ERG's base case, which reflected its preferred assumptions, plus a scenario in which the hazards of survival were the same at 8.5 years for all comparators. The committee was aware that, because of the model the company chose, it adjusted only for costs, and not for the effects of subsequent treatments for hormone-relapsed metastatic prostate cancer to match NHS practice, (see section 3.18).

3.22 Because of confidential discounts for therapies taken during the hormone-relapsed metastatic stage, none of the cost-effectiveness results can be reported here. The ERG presented analyses reflecting the committee's preferred assumptions and scenarios (see section 3.20). In these analyses, the incremental cost-effectiveness ratios were within the range that NICE usually considers an acceptable use of NHS resources (£20,000 to £30,000 per quality-adjusted life year gained). The committee was aware that it had not seen data for people who could take enzalutamide plus ADT, but who could not have docetaxel plus ADT (see section 3.6) and for whom ADT alone is the only NHS treatment option. However, the committee took into account the uncertainties in these people around the relative effectiveness, baseline risk of dying and health-related quality of life. It then concluded that estimates of cost effectiveness were sufficiently low to account for this uncertainty for people who could not have docetaxel plus ADT. The committee therefore concluded that it could recommend enzalutamide plus ADT for routine commissioning.

3.23 The committee noted that, as in previous appraisals for technologies for treating prostate cancer, its recommendations should apply to trans women as well as to men. No other equality issues were raised during the scoping process or in the submissions for this appraisal.

3.24 The company considered enzalutamide to be innovative because it is an oral treatment and needs less monitoring than docetaxel plus ADT. It discussed whether enzalutamide reflects a 'step change' in treatment and whether the model captured the benefits of treatment. The committee recognised that many individuals who have not had enzalutamide plus ADT for hormone-sensitive metastatic prostate cancer have the option of getting it at 2 different points later in the treatment pathway for hormone-relapsed metastatic prostate cancer. It concluded that enzalutamide plus ADT, despite its associated advantages, is not a step change in the treatment of hormone-sensitive metastatic prostate cancer, but that the model captured the relevant benefits.

3.25 The company did not make a case for enzalutamide plus ADT meeting NICE's end of life criteria. NICE's advice about life-extending treatments for people with a short life expectancy did not apply.

3.26 Early trial results suggested that enzalutamide plus ADT increases progression-free and overall survival compared with ADT alone. Also, the results of an indirect comparison suggested that, compared with docetaxel plus ADT, enzalutamide plus ADT increases progression-free survival but its comparative effect on overall survival is unclear. The cost-effectiveness estimates are below what NICE considers an acceptable use of NHS resources. Therefore, the committee concluded that enzalutamide plus ADT is recommended for hormone-sensitive metastatic prostate cancer.