Oral azacitidine for maintenance treatment of acute myeloid leukaemia after induction therapy

3.1 Acute myeloid leukaemia (AML) is a rapidly progressing cancer of the blood and bone marrow that is usually diagnosed in older people. Treatment options for AML include chemotherapy and a stem cell transplant (see section 3.2). The committee understood the substantial psychological, social and physical impact of living with AML for people with the condition and their carers and families. The patient expert described how being diagnosed with AML and the potential prospect of living for only a few months, had a significant emotional impact on them and their family. The clinical experts explained how intensive chemotherapy is associated with morbidity and that each treatment cycle often requires a prolonged hospital stay over several weeks. They explained that because a person's immune system is likely to weaken with chemotherapy, this increases their likelihood of contracting a life-threatening infection, which is an additional worry for people having treatment. The clinical experts explained that for people who are well enough to have intensive chemotherapy, a stem cell transplant remains the most effective treatment option. The committee understood that the decision to have a transplant depends on a person's fitness, choice, response to chemotherapy, and donor availability. The clinical experts highlighted that most people with AML are aged over 60 and are often unable to have a transplant because of co-morbidities and frailty. They also explained that there is a lack of donor availability for people from a minority ethnic family background. Therefore, some people cannot have, or do not want to have a stem cell transplant. The clinical experts explained that the risk of relapse in people who do not have a transplant is around 70% to 80%. They added that this would most likely occur within the first year after reaching complete remission. They highlighted that there are no effective treatment options after relapse in this population and that their prognosis and quality of life is poor. The committee noted that while stem cell transplants have the potential to be curative, they are associated with significant morbidity and mortality. The patient expert described how their treatment in preparation for a stem cell transplant was painful and invasive and made them feel too unwell to carry out their regular activities. The committee heard how the financial implications of having a transplant, such as regular travel to hospital and having to take time off work, can have a significant impact on quality of life for people with AML, their carers and families. It understood that oral azacitidine would benefit people who cannot have, or do not want to have a stem cell transplant because it can be taken at home. The committee concluded that people with AML would welcome a new treatment option that would improve their life expectancy and quality of life.

3.2 Treatment for AML depends on whether a person is able to have intensive chemotherapy. If intensive chemotherapy is unsuitable, low dose chemotherapy may be given. If intensive chemotherapy is suitable, induction chemotherapy is initially given to achieve complete remission followed by consolidation chemotherapy. After induction or consolidation chemotherapy, some people may be able to have a stem cell transplant. People who have FLT3-mutation-positive AML may also have concomitant treatment with midostaurin during induction and consolidation chemotherapy and continue with midostaurin alone as maintenance treatment to prolong their remission (see NICE's technology appraisal guidance on midostaurin for untreated acute myeloid leukaemia). The committee discussed the company's positioning of oral azacitidine as a maintenance treatment for people who are in complete remission after induction chemotherapy, with or without consolidation chemotherapy, and who cannot have or do not want a stem cell transplant. The final scope for this appraisal included established clinical management without oral azacitidine (which may include a watch and wait strategy with best supportive care, low dose cytarabine or subcutaneous azacitidine) as a comparator. The committee noted that the company had not included low dose cytarabine and subcutaneous azacitidine as part of established clinical management. The company's clinical experts considered that these treatments are not used as maintenance treatment in the population who would be considered for treatment with oral azacitidine. The committee discussed the stakeholder comments that low dose cytarabine and subcutaneous azacitidine are not used routinely after induction and consolidation chemotherapy but are used when intensive chemotherapy is unsuitable. It noted the company's response to technical engagement which referenced data from the Haematological Malignancy Research Network (HMRN), which is an ongoing UK population-based cohort. The company presented subgroup data from the HMRN which was aligned to the eligibility criteria of the key clinical trial of oral azacitidine (see section 3.4). This indicated that very few people had maintenance treatment with low dose cytarabine and subcutaneous azacitidine (actual figures are confidential and cannot be reported here). Therefore, the committee concluded that these treatments would not likely be used routinely as maintenance treatment in people who are in complete remission.

3.3 The committee noted that the company's clinical experts considered that around 25% of people with AML have FLT3-mutation-positive AML. Most of these people will have a stem cell transplant after reaching remission, leaving around 10% who would likely have maintenance therapy with midostaurin. The clinical experts explained that most people with FLT3-mutation positive AML would have targeted treatment with midostaurin during induction, consolidation and maintenance treatment and would be unlikely to switch to oral azacitidine. The committee heard how midostaurin is often not well tolerated, so having an alternative treatment option such as oral azacitidine would be important for this population. It recognised that the proportion of people with FLT3-mutation-positive AML who would have oral azacitidine in clinical practice would likely be small, but that it was still a relevant population. Therefore, the committee concluded that midostaurin was a relevant comparator for people with FLT3-mutation-positive AML.

3.4 The clinical evidence came from QUAZAR AML‑001 (from now, referred to as QUAZAR), a phase 3, double-blind, randomised controlled trial that compared oral azacitidine plus best supportive care (from now, referred to as oral azacitidine) with placebo plus best supportive care (from now, referred to as placebo). The company considered that the placebo arm represented the watch and wait comparator. The population included adults with AML in complete remission after intensive induction chemotherapy with or without consolidation chemotherapy, who were not able to have a stem cell transplant. The company reported data from the trial's first data cut (July 2019, median follow up 41.2 months) for all outcomes. It also reported data from a second data cut (September 2020, median follow up 51.7 months) for the primary outcome of overall survival. In the intention-to-treat (ITT) population, oral azacitidine increased median overall survival compared with placebo from 14.8 months to 24.7 months (hazard ratio 0.69; 95% confidence interval 0.56 to 0.86, p value 0.0008). In the ITT population, oral azacitidine increased median relapse-free survival compared with placebo from 4.8 months to 10.2 months (hazard ratio 0.65; 95% confidence interval 0.52 to 0.81, p value 0.0001). The committee welcomed the mature trial data from QUAZAR and concluded that oral azacitidine improves overall survival and relapse-free survival compared with placebo.

3.5 The QUAZAR trial included 148 sites and included a large number of people from Europe (known as the EU-subgroup) and a small number of people from the UK (the actual numbers are considered confidential by the company and cannot be reported here). The EU-subgroup was part of the company's pre-specified subgroup analyses. The company considered that the baseline characteristics of people in QUAZAR aligned to the AML population in the UK who cannot have a transplant, with some exceptions:

3.6 The ERG noted that in QUAZAR, most people had 1 or no cycles of pre-trial consolidation therapy. The committee understood that 20% of people in the trial did not have any pre-trial consolidation therapy (based on the ITT population). The ERG considered that previous NICE technology appraisals for untreated AML imply that consolidation therapy is standard practice, but the number of cycles of consolidation is unclear. In response to technical engagement, the company's clinical experts suggested that there is variability in the number of consolidation cycles and that up to 60% of people in the UK are likely to have only 1 or no cycles in routine practice. The company presented data from the HMRN report which highlighted that in the NHS, a proportion of people whose cancer responded to intensive induction therapy did not have any cycles of consolidation chemotherapy (actual numbers are confidential and cannot be reported here). The ERG considered that there is some disparity between the HMRN report findings and the NHS website which suggests that all people with AML have consolidation therapy. It noted that the European Society for Medical Oncology's (ESMO) clinical practice guideline on acute myeloid leukaemia in adult patients (2020) recommends that people should have consolidation therapy after reaching complete remission after induction treatment. The ERG considered that consolidation therapy is expected so it initially preferred to use a subpopulation of the EU-subgroup who had had at least 1 cycle of consolidation therapy for its base case. This subpopulation was known as the EU-consolidation subgroup. The committee noted that the EU-consolidation subgroup slightly increased the ICER. The clinical experts explained that in clinical practice, many people can only have a single course of consolidation chemotherapy, because of delayed blood count recovery, toxicity or patient choice. They described how optimum best practice includes 3 to 4 cycles of consolidation chemotherapy; however, this is difficult to achieve particularly in an older population. The clinical expert described that around 20% of people will have the optimum number of consolidation cycles. They explained that the QUAZAR trial reflects the use of consolidation chemotherapy in clinical practice. The committee noted that data from the trial suggested that overall survival was improved irrespective of whether a person had consolidation treatment. The committee concluded that although most people would likely have at least 1 cycle of consolidation therapy, there may be people who would not have any consolidation treatment after induction chemotherapy. It concluded that the number of cycles of pre-trial consolidation therapy in QUAZAR likely reflects NHS clinical practice.

3.7 The company did not identify any direct evidence comparing the efficacy of oral azacitidine to midostaurin as maintenance treatment in people with FLT3-mutation-positive AML. Therefore, it did an anchored indirect treatment comparison using data from QUAZAR and a phase 3, randomised, placebo-controlled study of midostaurin plus standard chemotherapy in adults with newly diagnosed FLT3-mutation-positive AML (known as RATIFY). Because individual patient data was available from QUAZAR, the company matched the population in QUAZAR to the eligibility criteria in RATIFY, so that only people with FLT3-mutation-positive AML in complete remission were included in the analysis. The company considers the results of the indirect treatment comparison to be confidential and so they cannot be reported here. However, the company noted significant differences between QUAZAR and RATIFY including differences in:

3.8 The company presented a partitioned survival model with 3 health states: relapse-free survival, relapse and death. In the relapse-free health state, people could either be on or off treatment with oral azacitidine (the same utility value was applied). All people in the watch and wait plus best supportive care arm (from now, referred to as best supportive care), were assumed to be off treatment. The model included a cycle length of 28 days with a half-cycle correction over a lifetime time horizon. In the model, stem cell transplant was not included as a separate health state but was implicitly included in the modelling through the survival analysis of the QUAZAR ITT population. The company considered that because oral azacitidine has a marketing authorisation for people who cannot have a transplant, it would be unlikely in clinical practice that people will go on to have a transplant after oral azacitidine unless they had relapsed. In the trial, 6.3% of people in the oral azacitidine arm and 13.7% in the placebo arm had a transplant after stopping treatment. Costs and a temporary disutility associated with stem cell transplant were included in the model. The committee noted that the ERG's preference was to include stem cell transplant as a health state in the model. In response to technical engagement, the company stated that the QUAZAR trial did not collect sufficient data to allow modelling of stem cell transplant as a separate health state and this data is not available in the literature. Instead, it provided a scenario analysis which calculated a weighted average utility value for each treatment arm in the relapse health state. This was to account for the potential long-term health-related quality-of-life benefits of a transplant. The committee noted that the company's scenario analysis suggested only a small impact on the ICER. The ERG considered that it was unclear whether the model captured the long-term benefits of a stem cell transplant on a person's health-related quality of life (including after the company's scenario analysis). The committee agreed with the ERG's preference to remove the temporary disutility associated with a transplant, given that no benefit in health-related quality of life after having a transplant had been included in the model. It noted that the removal of this disutility had a small impact on the ICER. The committee considered the ERG's scenario analysis which included a utility increment for people who went on to have a stem cell transplant and noted that this slightly increased the ICER. The committee considered that it would have preferred the company to have included stem cell transplant as a health state in the model in line with NICE's technology appraisal guidance on midostaurin for untreated acute myeloid leukaemia. It noted that the company did not provide an updated model or any new evidence to support its approach to modelling a stem cell transplant in response to consultation. Therefore, the committee concluded that there is still uncertainty about whether the company's model captures the long-term benefits after a stem cell transplant.

3.9 The company considered that the QUAZAR trial did not capture health-related quality of life after relapse, except in some people who may have had an extended dose of oral azacitidine. This included people with a bone marrow blast count of 6% to 15% but not those with advanced disease (blast count greater than 15%). The company considered that using a utility value derived from this small cohort of people would be inappropriate and may overestimate the quality of life for people who relapse. Therefore, the company calculated the relapse utility based on a study by Joshi (2019) which used the composite time trade-off method to elicit utility values for AML from the UK general population. In the model, relapse utility was calculated as the difference between the relapse-free survival and relapse utilities in Joshi (2019) which was then applied to the relapse-free utility from QUAZAR. The ERG noted that the sample size in Joshi (2019) was small which resulted in a large standard error. The committee understood that the company considered alternative sources for relapse utilities including studies by Stein (2019) and Tremblay (2018). The ERG considered that both sources were not ideal because utility values were derived from US populations. In response to technical engagement, the company explained why the study by Joshi (2019) had been selected. It considered the utility elicitation methodology to be preferred by NICE, utility values were from individuals in the UK, and the utility value was clinically plausible. The ERG was unclear why the company preferred the composite time trade-off methodology, given that it was not part of the reference case in NICE's guide to the methods of technology appraisal 2013. In line with NICE's technology appraisal guidance on midostaurin for untreated acute myeloid leukaemia, the ERG used Tremblay (2018) to calculate health-related quality of life after relapse in its base case. This was because it considered that the method used to elicit utility values was more appropriate than Joshi (2019). The committee noted the company's scenario analyses using relapse utility values from Tremblay (2018) and Stein (2019) had a small impact on the ICER in the EU-subgroup. The company did not provide any new evidence to support its preferred approach to calculating utility after relapse in response to the appraisal consultation document. The committee agreed with the ERG's approach and concluded that health-related quality of life after relapse should be calculated using data from Tremblay (2018).

3.10 The committee understood that the company had adjusted health state utility values for age in the model. It discussed the ERG's critique that the relapse-free survival utility was higher than the age-adjusted population norm in the UK. The committee considered that this did not make sense because it would mean that people with AML had a better quality of life than people without the disease. It noted that the ERG had asked the company to provide a scenario analysis capping the utility at general population levels which slightly increased the ICER (based on the ITT population). The company did not provide any new evidence on utility for the relapse-free survival health state in response to consultation. The committee noted the ERG's analysis which capped the utility value for the relapse-free survival health state at general population levels in the EU-subgroup. The results of the analysis suggested only a small impact on the ICER. The committee concluded that the utility value for the relapse-free survival health state in the model should be capped at age- and sex-matched general population levels.

3.11 The company used joint models (joint models apply a single distribution to both treatment arms) to estimate overall survival and relapse-free survival based on QUAZAR trial data. It used the generalised gamma accelerated failure time model for overall survival and the log-logistic accelerated failure time model for relapse-free survival in the ITT population. The company considered that the survival analyses in the EU-subgroup were aligned with the assessment for the ITT population. The committee considered that the joint modelled curves showed that the survival function was being underestimated in the model for the comparator arm (best supportive care) when compared with the Kaplan–Meier curves from the trial (based on the ITT population). It considered that this may be because accelerated failure time models assume that the treatment effect is constant over the entire lifetime of the model. In response to consultation, the company stated that the joint models selected in its base case did not overestimate the expected treatment benefit with oral azacitidine for the EU-subgroup. It presented a scenario analysis exploring the impact of using the best-fitting individual models for the EU-subgroup (generalised gamma models for overall survival and log-logistic models for relapse-free survival). The committee noted that the company's scenario analysis suggested only a small impact on the ICER. It discussed the ERG's scenario analysis which explored the impact of using the same individual parametric models and the committee's preferred assumptions from the first meeting, which also had a small impact on the ICER. The committee noted the ERG's critique that the company's joint and individual modelling results were comparable and that the impact of choosing between these approaches was likely minor. It was reassured that both approaches reflected the trial data and resulted in similar extrapolations. Therefore, the committee concluded that the company's joint modelling approach was appropriate for estimating overall survival and relapse-free survival in the EU-subgroup.

3.12 The company's base-case analysis assumed no treatment effect waning for oral azacitidine. The committee initially considered that it was highly optimistic to assume a constant treatment benefit with oral azacitidine based on the observed trial data. The committee understood that the company had presented a scenario analysis which tried to explore the impact of treatment effect waning. It did this using individual log-normal models for overall survival and log-logistic models for relapse-free survival (based on the ITT population). The results of the scenario analysis increased the ICER by 11%. The committee considered that this was not explored fully and it would be preferable to include an assumption that the relative treatment effectiveness would decline over time. Therefore, the committee considered that it would be helpful to see a range of scenarios for both the overall population (based on the EU-subgroup) and for people with FLT3-mutation-positive AML including:

3.13 The committee considered the advice about life-extending treatments for people with a short life expectancy in NICE's guide to the methods of technology appraisal 2013. The committee agreed that based on the trial evidence and the views of clinical experts, the overall survival gain with oral azacitidine would likely be more than 3 months. The company's model suggested that there was an increase in mean overall survival of 12.8 months (median 9.9 months in QUAZAR) for the ITT population. For the EU-subgroup, the company's model suggested that there was an increase in mean overall survival of 16.2 months (median overall survival gain was greater than 3 months in QUAZAR but the exact figure is confidential so cannot be reported here). The committee agreed that oral azacitidine meets the criterion for a life-extending treatment because it increases overall survival by more than 3 months.

3.14 The company confirmed that mean estimates were not available from QUAZAR, but the median overall survival for people who had placebo was 14.8 months (ITT population). The company's original base case for the ITT population predicted a mean overall survival of 33.6 months for people having best supportive care. The company's revised base case using the EU-subgroup predicted a mean overall survival of 31.5 months for people having best supportive care (median overall survival from QUAZAR for the EU-subgroup was also less than 24 months but the exact figure is confidential and so cannot be reported here). The committee acknowledged that using extrapolation models to estimate mean values involves assumptions and uncertainty. However, the committee noted that the mean estimates were higher than 24 months and that the cost-effectiveness analyses are based on mean survival estimates. The clinical experts explained that a significant number of people with AML are unable to have a stem cell transplant and that they have a high risk of relapse in the first year. After relapse, they would have palliative treatment within 12 months and a life expectancy of around 3 months at this point. The clinical experts explained that around 20% of people may be cured after intensive chemotherapy, and reach long-term survival. The ERG considered that because some people will live for much longer, this will skew the survival distribution where the mean is often higher than the median. The committee recalled that while the short-term prognosis for most people would be poor, a proportion of people (around 20%) would be cured and would therefore be expected to live beyond 2 years. It initially considered that the short life expectancy criterion had not been met. The committee considered comments received in response to consultation from the company, a clinical expert and a patient group which strongly reiterated that life expectancy for most people in this population is less than 24 months. The company highlighted the appeal decision in NICE's technology appraisal guidance on avelumab for maintenance treatment of locally advanced or metastatic urothelial cancer after platinum-based chemotherapy (TA788). The appeal panel for TA788 concluded that it would be unreasonable to state that life expectancy was not normally less than 24 months given that the modelled mean life expectancy indicated that most people (65%) did not survive after 24 months. In this appraisal, the company highlighted that a similar proportion in the EU-subgroup did not survive beyond 24 months (the exact figure is considered confidential by the company and so cannot be reported here). The committee believed that the best estimate of life expectancy came from the mean survival for the relevant patient population, based on the decision model submitted by the company. However, it accepted a consultation comment that the NICE methods guide does not specifically state how this criterion should be assessed. It noted the appeal panel's conclusion for TA788 that the totality of evidence should be considered when assessing whether a technology meets the short life expectancy criterion. The committee took into consideration the mean and median survival estimates, clinical opinion from the first committee meeting and consultation comments from all stakeholders. It also recalled that the data from the QUAZAR trial was mature and this reduced the uncertainty in the results. The committee discussed that although there are different ways to estimate life expectancy, it is likely that the population who would be considered for treatment with oral azacitidine would live on average less than 24 months. Therefore, it accepted that the short life expectancy criterion was met and concluded that oral azacitidine meets the criteria to be considered a life-extending treatment at the end of life.

3.15 NICE's guide to the methods of technology appraisal 2013 notes that above a most plausible ICER of £20,000 per quality-adjusted life year (QALY) gained, judgements about the acceptability of a technology as an effective use of NHS resources will take into account the degree of certainty around the ICER. The committee will be more cautious about recommending a technology if it is less certain about the ICERs presented. At the first meeting, the committee considered that there was uncertainty around some of the assumptions used in the model which made the cost-effectiveness results uncertain. It considered that analyses which explored the use of individual models for extrapolating overall survival and treatment effect waning would reduce the uncertainty that the expected treatment benefit with oral azacitidine had been overestimated in the model (see section 3.11 and section 3.12). At the second committee meeting, the company and ERG presented these analyses which suggested that the cost-effectiveness results are robust to changes in survival modelling assumptions (see section 3.11 and section 3.12). The committee was therefore reassured that this uncertainty had now been adequately addressed. The ERG updated its base-case assumptions to align with the committee's preferences which remained unchanged from the first committee meeting. This included the following assumptions:

3.16 The committee discussed the potential equality issues raised during the appraisal. It noted a stakeholder comment that some people may struggle financially to have current treatment (such as a transplant) because of the cost of regular travel to hospital and reduced income from having to take time off work. Having a transplant may be especially difficult for people with caring responsibilities because of the significant time commitment needed. The stakeholder considered that these people should not be denied treatment and oral azacitidine would be a viable alternative to a transplant. The committee recognised that because oral azacitidine can be taken at home it may be more convenient and reduce the amount of time spent in hospital compared with having a transplant. It noted comments from stakeholders that many people with AML who are in complete remission are unable to have a transplant because of a lack of donor availability. Therefore, this results in an inequity of access to curative treatment and disproportionately affects people with a minority ethnic family background. The clinical experts explained that after a transplant there is a 70% reduction in the risk of relapse. They explained that in people of white family background, the likelihood of finding a suitably matched donor is around 80%, which reduces to around 30% to 40% in people with a minority ethnic family background. Stakeholders and clinical experts highlighted that oral azacitidine should therefore be available to all people who are not able to have a transplant, including those with a minority ethnic family background who do not have a suitable donor. The committee noted that this issue had been reiterated in comments received in response to the appraisal consultation document. The committee acknowledged that unequal access to transplants because of ethnicity was a relevant consideration and it was mindful of its obligations in relation to the Equality Act 2010. It noted that the company had also highlighted that there may be geographical barriers to accessing a stem cell transplant based on how far away a person lives from an allograft transplant centre. The committee considered that issues around healthcare implementation cannot be addressed in a technology appraisal. It concluded that because it had recommended oral azacitidine for people with AML that this may help to reduce some of the potential equality issues raised during the appraisal.

3.17 The company and stakeholders considered oral azacitidine to be innovative because it improves survival, is well tolerated and because of its oral formulation. The committee recalled how treatment with oral azacitidine would offer benefits because people would be able to have treatment at home (see section 3.1). It heard how this would reduce time spent in hospital and allow people to spend more time with their family and friends. The committee agreed that these are important benefits and noted that oral azacitidine is the first maintenance treatment licensed for all people with AML. However, it concluded that it had not been presented with evidence of any additional benefits that could not be captured in the QALY calculations.