Azacitidine for the treatment of myelodysplastic syndromes, chronic myelomonocytic leukaemia and acute myeloid leukaemia

The Committee considered the clinical-effectiveness evidence from AZA-001 presented by the manufacturer. The Committee understood that patients were preselected for treatment with 1 of the conventional care regimens before randomisation, and this was based on age, ECOG performance status, the presence of comorbidities and patient preference. It also understood that patients randomised to treatment with azacitidine were compared with patients in their respective pre-randomisation regimen. The Committee also heard from the clinical specialists that the proportion of patients in each preselection group in AZA-001 broadly represented the treatment patients with these conditions receive in the UK (that is, treatment with chemotherapy plus best supportive care is appropriate for considerably fewer patients than treatment with best supportive care alone).

The Committee noted that the median overall survival for patients receiving azacitidine was longer than for patients receiving the conventional care regimens. The Committee further noted that median time to transformation to acute myeloid leukaemia was longer for patients receiving azacitidine and the percentage of patients becoming independent of blood transfusions was higher for patients receiving azacitidine than for patients receiving the conventional care regimens. The Committee noted that when outcomes were analysed separately for each conventional care regimen, the difference in overall survival between the treatment arms in the standard-dose chemotherapy group was not statistically significant, and the differences between the treatment arms in the estimates of time to transformation to acute myeloid leukaemia for either the low-dose or standard-dose chemotherapy groups were not statistically significant. The Committee was aware that the small patient numbers limited the precision and certainty of the outcome estimates in these groups, but concluded that the estimates of total overall survival compared with conventional care appeared robust. The Committee noted that the problems relating to loss of patients to follow-up, as described by the ERG (see section 3.13), may have introduced bias into estimates of relative effectiveness, but concluded that this effect was likely to be minimal.

The Committee considered the role of patient preference in the design and analysis of AZA-001. The Committee heard from the DSU and the manufacturer that the term 'patient preference trial' is used to describe trial designs that take account of a patient's preference to receive either the study treatment (for example, azacitidine) or the comparator (for example, conventional care). It noted that in AZA-001, patient preference informed preselection to 1 of the conventional care regimens before randomisation, but did not inform randomisation to either azacitidine or conventional care. The Committee concluded that the role of patient preference in AZA-001 did not affect the way in which the trial results or subsequent analyses should be considered.

The Committee considered the potential adverse effects associated with treatments for myelodysplastic syndromes and the impact of these effects on quality of life. The Committee heard from the clinical specialists that common adverse effects of treatment with azacitidine include peripheral blood cytopenias, myelosuppression, nausea, vomiting and injection site reactions. The patient experts and clinical specialists agreed that these adverse effects are generally well tolerated. The Committee heard from the patient experts that compared with other treatment options, azacitidine was associated with relief from fatigue, fewer infection-related hospitalisations, a decreased need for blood and platelet transfusion, and increased ability to perform day-to-day activities. The Committee noted that no quality-of-life data were collected in AZA-001, although EORTC data collected in CALGB 9221 suggested improvements in overall health with azacitidine.

The Committee concluded on the basis of the clinical-effectiveness evidence and the evidence from the clinical specialists and patient experts that azacitidine is a clinically effective treatment for myelodysplastic syndromes, chronic myelomonocytic leukaemia and acute myeloid leukaemia.

The Committee considered evidence on the cost effectiveness of azacitidine for the treatment of myelodysplastic syndromes, chronic myelomonocytic leukaemia and acute myeloid leukaemia. It discussed the likelihood of vial sharing, noting that because of the small number of patients, it may be difficult to implement a vial-sharing scheme. It concluded that the manufacturer's estimate of 49% of patients being able to receive treatment at the same time (and therefore share vials) seemed optimistic. The Committee concluded that the analyses incorporating estimated vial sharing did not produce plausible results and therefore would not form the basis for its decision on the use of azacitidine in the NHS.

The Committee noted the ERG's concerns about the manufacturer's initial model, mainly relating to the selection of the parametric function to model overall survival. The ERG stated that the most important influence on the model's outputs was overall survival, and that the choice of parametric distribution used to extrapolate estimates of overall survival from AZA-001 greatly influenced the results. The Committee noted that the manufacturer's initial base case used the log-normal function to extrapolate overall survival from the trial data, which the manufacturer justified with supporting data from a Düsseldorf myelodysplastic syndromes registry. The Committee understood that the use of the log-normal distribution modelled survival in such a way that some patients were predicted to live to an unrealistic age given the nature of the condition (see section 3.12), that is the use of log-normal distribution led to an overestimation of survival. The Committee concluded that the Weibull distribution generally provided the best overall fit to the Düsseldorf registry long-term survival data, and that modelling that incorporated the Weibull distribution should be used to inform a decision on the use of azacitidine in the NHS.

The Committee considered the estimates of quality of life included in the model (see also sections 4.12 and 4.13). The Committee first considered the derivation of the utility values. It was aware of the ERG's concerns about the mapping of EORTC values to the EQ-5D (see section 3.14). The Committee concluded that because the algorithm had been developed using data from patients with oesophageal cancer, the values would be associated with greater uncertainty than if a validated algorithm based on patients with myelodysplastic syndromes, chronic myelomonocytic leukaemia and acute myeloid leukaemia had been used.

The Committee then considered the face validity of the quality-of-life gains attributed to azacitidine in the model. The patient experts and clinical specialists stated that treatment with azacitidine reduces symptoms (such as fatigue) and the need for blood transfusions. The Committee agreed that the associated utility gains of these should ideally be reflected in the model. The Committee noted that greater independence from blood transfusions was not explicitly included in the utility value estimate. It was aware that the manufacturer estimated the utility value by mapping to the EQ-5D. The Committee understood that the EQ-5D does not include fatigue as a dimension, although it would capture some of the effects of fatigue on the patient's ability to undertake day-to-day activities. The Committee also understood that the EORTC measure includes a domain that captures the impact of treatment on quality of life, and that dependence on transfusion could be expected to affect this. It considered that reduced fatigue after azacitidine treatment may not have been completely captured in the modelled utility values. The Committee noted the alternative utility values provided by MDS UK after the appeal, which were meant to better reflect the utility values associated with dependence on or independence from blood transfusions. However, the Committee noted the DSU's concern that the alternative utility values did not separate dependence or independence from transfusion from associated symptoms (see section 3.22) and it did not accept these in preference to the manufacturer's previous estimates. The Committee also noted that the model applied constant utility values within each health state, therefore assuming quality-of-life gains from the first day of treatment, which it considered unrealistic. Taking all these points into account, the Committee concluded that it was uncertain whether the utility values used in the model under- or over-estimated the true utility values associated with myelodysplastic syndromes.

The Committee considered the uncertainty around the quality-of-life estimates included in the model (see sections 4.11 and 4.12). The Committee noted that sensitivity analyses carried out by the manufacturer showed that variations in the utility values had relatively little impact on the ICERs. It concluded that because the ICER estimate was largely driven by the incremental life years gained and the acquisition cost of azacitidine, and was only minimally affected by the changes in health-related quality of life, the impact of any over- or underestimation of quality-of-life gains was likely to be small.

The Committee considered the inclusion of costs in the economic model. The Committee noted the use of the NHS 2009 to 2010 tariff. It considered that using the NHS 2009 to 2010 tariff was appropriate because it could provide a more precise estimate of hospital costs by breaking down costs attributable to adverse events. The Committee also noted the assumed increase in the cost for treatment received at the weekend. The Committee concluded that the modelled increased costs of weekend administration were reasonable, aware that the associated impact on the ICERs was relatively small. On balance, the Committee concluded that costs included in the model were acceptable.

The Committee considered the estimated cost effectiveness of azacitidine. The Committee noted that the manufacturer had submitted 2 approaches to estimating the ICERs for azacitidine:

• 1 analysis comparing azacitidine separately with each of the conventional care regimens specified by the Appeal Panel (that is, compared with best supportive care alone and compared with low-dose chemotherapy plus best supportive care) and

• 3 analyses comparing azacitidine with usual care (that is, a single estimate representing a weighted average of all the conventional care regimens together), with each analysis using different proportions of conventional care to form the weighted average.
  
  The Committee noted that no cost-effectiveness evidence was presented for the subgroup of patients with the -7/del(7q) chromosomal abnormality.

The Committee considered the 2 approaches to estimating the cost effectiveness of azacitidine. The Committee first considered the separate conventional care regimen analyses. The Committee noted that patients randomised to receive azacitidine in the group preselected to receive low-dose chemotherapy plus best supportive care incurred higher total costs (£101,100) than those randomised to azacitidine who had been preselected to receive best supportive care alone (£91,800). It further noted that the number of total QALYs (that is, not the incremental QALY gain) was greater in those randomised to azacitidine who had been preselected to receive low-dose chemotherapy (2.44) than in those randomised to azacitidine who had been preselected to receive best supportive care alone (2.04). The Committee also noted that in the analyses presented before the appeal comparing azacitidine with standard-dose chemotherapy, the number of QALYs associated with those randomised to azacitidine who had been preselected to receive standard-dose chemotherapy in the azacitidine arm was 1.91. The Committee concluded that there appeared to be no reason, other than differences in baseline patient characteristics, why those who were randomised to azacitidine but were preselected to receive low-dose chemotherapy should have gained greater benefit than those who were randomised to azacitidine but were preselected to receive standard-dose chemotherapy plus best supportive care or best supportive care alone. The Committee agreed that this uncertainty should be noted when considering the most appropriate analyses on which to base its recommendations.

The Committee then considered the analyses of azacitidine compared with a weighted average of usual care (see sections 3.24 to 3.26). The Committee understood that the weighted ICERs had each been calculated by combining the individual ICERs for the respective conventional care regimens, each weighted by the proportion of patients receiving these regimens in AZA-001 and in the HMRN registry respectively. The Committee understood the significant methodological limitations associated with analyses involving the use of such a weighted average, in particular the need for equivalent patient characteristics (such as age or disease severity) at baseline among any of the groups being combined (see section 4.16). It also acknowledged the importance of ensuring that the full range of appropriate comparators was considered within these groups. It noted that because of differences in the baseline patient characteristics of the conventional care regimen subgroups (see section 3.1) the patient populations and associated population-specific results may not necessarily be appropriate for statistical analysis, which combines estimates across the groups using a simple weighted average.

The Committee then considered which of the 2 analytical approaches to considering cost effectiveness provided the most appropriate basis for its decision. It agreed that the most important consideration in whether a decision should be based on the separate comparisons with the different conventional care regimens was the need for a clear definition of the groups of patients eligible to receive each of the conventional care regimens. The Committee agreed that because a set of clear and objective clinical characteristics defining the eligibility of patients to receive chemotherapy had not been agreed among haematologists, it could not make recommendations based on any of the separate conventional care regimen groups (see section 4.2).

The Committee was aware of the significant methodological limitations with using weighted averages. The Committee understood that weighted averages can mask differences in the incremental costs and/or QALYs of the technologies being combined. The Committee was aware that the NICE 'Guide to the methods of technology appraisal' states that best practice should be considered as a comparator for appraisal of health technologies. If best practice is defined as a cost-effective treatment option, the standard approach to assessing the cost effectiveness of azacitidine in this context would be to consider all treatment options (that is, each of the conventional care regimens) in an iterative incremental analysis (that is, assessing the ratio of the additional cost and benefit of each technology compared with the next best alternative) to identify the most cost-effective strategy. The Committee understood that the weighted average approach assumes that each conventional care regimen is the most cost-effective treatment option available for the patient group for whom it is used. It heard from the DSU that the necessary evidence to test this assumption is not available, given the remit of this appraisal, and therefore the use of a weighted average would result in some uncertainty in the ICER produced. It further heard from the DSU that if the cost effectiveness of each of the individual conventional care regimens was not established, the magnitude and direction of uncertainty in the weighted average ICER is unknown. The Committee also understood that because the populations eligible for each of the conventional care regimens cannot be clearly defined (see section 4.2), an incremental analysis (as preferred by the DSU) is not possible in this case. Taking into account the limitations of the available evidence and in the absence of a satisfactory alternative, the Committee hesitantly concluded that any decision on the cost effectiveness of azacitidine would need to be made using the weighted average.

The Committee considered the proportions of each conventional care regimen used to calculate the weighted average. It understood that the manufacturer's base-case estimate was based on the proportions of people receiving each conventional care regimen in AZA-001, but that alternative analyses used the HMRN registry data submitted by the manufacturer (see section 3.25). The Committee heard from the clinical specialists that the HMRN registry, a UK database of over 600 patients, provided a representative estimate of the management of myelodysplastic syndromes in the UK. It noted that the HMRN registry includes a wider range of patients than those covered by the marketing authorisation for azacitidine but acknowledged clinical advice that within the HMRN registry, the subset of patients classified as having IPSS intermediate-2 or high-risk myelodysplastic syndromes provided the best available estimate of the proportion of patients receiving each of the conventional care regimens in the patient population for whom azacitidine is licensed. The Committee concluded that a weighted average of conventional care regimens should be calculated using the HMRN registry data (for the subset of patients having IPSS intermediate-2 or high-risk myelodysplastic syndromes) rather than the AZA-001 data.

The Committee considered the ICERs calculated using a weighted average of the proportions receiving conventional care regimens in the HMRN registry for patients classified as IPSS intermediate-2 or high risk. It understood that in this patient population approximately 59% of patients received best supportive care alone, 12% received low-dose chemotherapy plus best supportive care and 29% received standard-dose chemotherapy plus best supportive care. The Committee noted that the manufacturer's deterministic ICER using these proportions of patients was £49,800 per QALY gained. The Committee heard from the DSU that the probabilistic estimate of the ICER associated with these proportions was approximately £47,200 per QALY gained (see section 3.27). The Committee considered that the probabilistic ICER was a more valid estimate than the deterministic estimate because it takes account of joint parameter uncertainty. The Committee noted that because the incremental cost-effectiveness estimates of each respective conventional care regimen that comprise the weighted average were not known (see section 4.19), uncertainty about the true ICER for azacitidine compared with usual care remained. Taking into account the limitations associated with the use of a weighted average and the uncertainty associated with the cost effectiveness of each individual conventional care regimen, the Committee concluded that £47,200 per QALY gained represented the best available estimate of the cost effectiveness of azacitidine.

The Committee considered supplementary advice from NICE that should be taken into account when appraising treatments that may extend the life of patients with a 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.

• The treatment is indicated for patients with a short life expectancy, normally less than 24 months.

• There is sufficient evidence to indicate that the treatment offers an extension to life, normally of at least an additional 3 months, compared with current NHS treatment.

• The treatment is licensed or otherwise indicated for small patient populations.
  
  In addition, when taking these criteria into account, the Committee must be persuaded that the estimates of the extension to life are robust and that the assumptions used in the reference case economic modelling are plausible, objective and robust.

The Committee discussed whether the benefit provided by azacitidine for the treatment of myelodysplastic syndromes, chronic myelomonocytic leukaemia and acute myeloid leukaemia fulfilled the criteria for consideration as a life-extending, end-of-life treatment. The Committee understood that there are approximately 700 patients with IPSS intermediate-2 and high-risk myelodysplastic syndromes in England and Wales. The Committee considered that life expectancy with best supportive care alone was likely to be approximately 11.5 months. It considered the evidence from AZA-001 and noted that the median overall survival for patients treated with azacitidine in the best supportive care preselection group was 21.1 months. The Committee agreed that azacitidine would improve the treatment of myelodysplastic syndromes, chronic myelomonocytic leukaemia and acute myeloid leukaemia and that it was likely that azacitidine would increase overall survival by approximately 9.6 months. The Committee agreed that the estimates of clinical effectiveness informing the best available estimate of the ICER were sufficiently robust and concluded that azacitidine meets the criteria for being a life-extending, end-of-life treatment.

The Committee then considered the ICER taking into account the end-of-life considerations. It considered that the best available estimate of the base-case ICER was approximately £47,200 per QALY gained (see section 4.21). The Committee accepted the uncertainty associated with this estimate and acknowledged the difficulty of assessing the impact of uncertainty on the best estimate of the ICER (see section 4.19). The Committee was aware that other technologies had been recommended using the end-of-life criteria with ICERs as high as the 1 in this appraisal, but was conscious of the increasing cost pressures in the NHS and the opportunity costs that would result from a recommendation to fund a technology with an ICER of this magnitude. However, the Committee recognised that azacitidine represents an important change in the treatment of patients with myelodysplastic syndromes, noting the substantial benefits associated with its use. The Committee considered that on balance, the additional weight that would need to be assigned to the QALY benefits for the cost effectiveness of azacitidine to fall within the current threshold range was acceptable on this occasion. Therefore, the Committee considered that azacitidine when provided by the manufacturer with the discount agreed in the revised patient access scheme agreed by the Department of Health in January 2011 was a cost-effective use of NHS resources as a treatment option for people with myelodysplastic syndromes, as stated in the marketing authorisation. This includes adults who are not eligible for haematopoietic stem cell transplantation with intermediate-2 and high-risk myelodysplastic syndromes according to the International Prognostic Scoring System (IPSS), chronic myelomonocytic leukaemia with 10% to 29% marrow blasts without myeloproliferative disorder or acute myeloid leukaemia with 20% to 30% blasts and multilineage dysplasia, according to World Health Organization classification.

The Committee considered whether NICE's duties under the equalities legislation required it to alter or to add to its recommendations in any way. At an earlier meeting, the Committee noted that azacitidine may be of specific benefit to those who, for clinical or religious reasons, are unable to receive blood transfusions, because patients treated with azacitidine require fewer blood transfusions than patients treated with best supportive care. However, the Committee noted that no representations had been made or evidence received about the pathway of care for this particular group of patients, or about the effectiveness of azacitidine in this patient population. Therefore, the Committee agreed that it would not be appropriate to make recommendations for a subgroup of patients unable to receive blood transfusions. Because the final recommendations (see section 4.24) do not restrict access to azacitidine for any particular group of patients, the Committee concluded that there was now no need to alter or add to its recommendations in any case.