This is a critical abstract of an economic evaluation that meets the criteria for inclusion on NHS EED. Each abstract contains a brief summary of the methods, the results and conclusions followed by a detailed critical assessment on the reliability of the study and the conclusions drawn.

Patients suffering from chronic myeloid leukaemia in the accelerated phase or blast crisis received 600 mg/day imatinib at home. On failing imatinib (progressing from the accelerated phase to blast crisis or losing their response), the patients switched to conventional therapies (combination chemotherapy or palliative care). Combination chemotherapy (DAT) comprised one course of daunorubicin, cytarabine arabinoside and 6-tioguanine, given in hospital on an inpatient basis. Palliative care could be provided at the hospital or in the patient's home. In the comparator cohorts, patients received conventional therapy from the outset.

Treatment.

Cost-utility analysis.

The study population comprised patients who were either newly diagnosed with CML in the accelerated phase or blast crisis, or who had failed conventional treatment in the chronic phase by progressing to the advanced disease stage. The analysis was undertaken for four patient cohorts:

patients presenting in the accelerated phase and being treated with imatinib;

patients presenting in the accelerated phase and being treated with conventional therapies;

patients presenting in blast crisis and being treated with imatinib; and

patients presenting in blast crisis and being treated with conventional therapies.

The practice setting was secondary care. The economic study was carried out in the UK.

The effectiveness data were derived from studies dating from 1996 to 2002. The authors did not state the dates to which the resource use data and prices related.

The evidence for the final outcomes was derived from a review of published studies, supplemented by a clinician panel.

A Markov model was developed to simulate the transitions of four hypothetical cohorts of 1,000 CML patients from the point at which they presented for treatment through a series of health states to death. Patients in the accelerated phase or blast crisis could respond to imatinib by having either:

a complete haematological response accompanied by a major cytogenetic response (CytR), that is, a return to normal white blood cell (WBC) counts and a reduction in Philadelphia-positive (Ph+) metaphases; or

complete haematological response without major cytogenetic response (CHR), that is, a return to normal WBC counts without a major reduction in Ph+ metaphases, or partial haematological response (PHR), where the WBC count falls but is still above the normal level.

The patients remained in response states until progressing to the accelerated phase or blast crisis. Unresponsive patients progressed from the accelerated phase to blast crisis or death, and from blast crisis to death. In all health states, the patients could die from other disease-unrelated causes.

The outcomes assessed in the review were:

the transition probabilities of moving from the accelerated phase of CML to various health states (CytR, CHR, PHR, the blast crisis phase, and death) when treated with imatinib;

the transition probability of moving from the accelerated-CytR/CHR/PHR phase to the accelerated or blast crisis phase when treated with imatinib;

the transition probability of moving from the accelerated phase of CML to the blast crisis phase when treated with conventional therapies;

the transition probabilities of moving from the blast crisis phase of CML to various health states (CytR, CHR, PHR, and death) when treated with imatinib;

the transition probability of moving from the blast crisis-CytR/CHR/PHR phase to the accelerated or blast crisis phase when treated with imatinib;

the transition probability of moving from the blast crisis phase of CML to death when treated with conventional therapies; and

the probability of death from disease-unrelated factors at any health state.

The transition probabilities for the accelerated phase and blast crisis study cohorts were obtained from the international Phase II imatinib clinical trials.

Not reported.

Not reported.

The probabilities for the accelerated phase cohort were derived from the clinical trial where 158 patients received 600 mg/day imatinib (Capdeville and Gathmann, see Other Publications of Related Interest). The probabilities for the blast crisis phase cohort were derived from the same clinical trial, but 165 patients received 600 mg/day imatinib.

Four studies in total were included in the review. The transition probabilities for the accelerated phase and blast crisis study cohorts were derived from a single study (Capdeville and Gathmann, see Other Publications of Related Interest). The transition probabilities for the comparator cohorts were derived from two other studies (Kolibaba and Druker, and Reese et al., see Other Publications of Related Interest). The probability of dying from disease-unrelated factors was derived from another single study (Kattan et al., see Other Publications of Related Interest).

Not reported.

Not reported.

The transition probabilities of moving from the accelerated phase of CML to the following health states when treated with imatinib were: CytR, 28.6%; CHR, 12.2%; PHR, 34.0%; the blast crisis phase, 10.9%; and death, 22.0% (95% confidence interval, CI: 15.0 - 29.0).

The transition probability of moving from the accelerated-CytR/CHR/PHR phase to the accelerated or blast crisis phase when treated with imatinib was 22.5% (95% CI: 13.6 - 31.5).

The transition probability of moving from the accelerated phase of CML to the blast crisis phase when treated with conventional therapies was 50% at 6 months.

The transition probabilities of moving from the blast crisis phase of CML to the following health states when treated with imatinib were: CytR, 15.7%; CHR, 0.0%; PHR, 20.9%; and death, 67.9% (95% CI: 56.8 - 79.0).

The transition probability of moving from the blast crisis-CytR/CHR/PHR phase to the accelerated or blast crisis phase when treated with imatinib was 46.1% (95% CI: 30.9 - 61.3).

The transition probability of moving from the blast crisis phase of CML to death when treated with conventional therapies was 50% at 4.5 months.

The probability of death from disease-unrelated factors at any health state was 0.04%.

A clinician panel was used to estimate the proportions of newly diagnosed patients in the UK assigned to each comparator therapy (i.e. DAT, hospital palliative care or home palliative care). The clinician panel also estimated the proportions for accelerated phase patients progressing to blast crisis for each treatment arm.

Due to the absence of utility data associated with DAT, imatinib or palliative care, a clinician panel was also used to estimate the quality of life for an average patient in each health state and treatment sub-group. The means of the clinician responses (six respondents) were used.

Ninety per cent of newly diagnosed patients in the accelerated phase received home palliative care and 10% received DAT.

Sixty per cent of newly diagnosed patients in the blast crisis phase received DAT, 30% received hospital palliative care and 10% received home palliative care.

Ten per cent of the patients in the accelerated phase who failed on imatinib received DAT and 90% received home palliative care.

Fifty per cent of the patients in the blast crisis phase who failed on imatinib received DAT, 10% received hospital palliative care and 40% received home palliative care.

Thirty per cent of the patients in the blast crisis phase who had been treated in the accelerated phase with DAT were treated again with combination chemotherapy, 20% received hospital palliative care and 50% received home palliative care.

Fifty per cent of the patients in the blast crisis phase who received hospital palliative care when in the accelerated phase were treated with DAT, 10% received hospital palliative care and 40% received home palliative care.

Forty-five per cent of the patients in the blast crisis phase who received home palliative care when in the accelerated phase were treated with DAT, 21% received hospital palliative care, and 34% received home palliative care.

The mean utility value derived for health states CytR, CHR and PHR was 0.91 (range: 0.73 - 1.00) when the patient was treated with imatinib.

In the accelerated phase health state, the mean utility value was 0.58 (range: 0.15 - 1) for treatment with imatinib, 0.01 (range: -0.33 - 0.52) for treatment with DAT, 0.07 (range: -0.33 - 0.52) for hospital palliative care, and 0.34 (range: 0.08 - 0.52) for home palliative care.

In the blast crisis phase health state, the mean utility value was 0.38 (range: 0.02 - 0.69) for treatment with imatinib, -0.09 (range: -0.33 - 0.52) for treatment with DAT, -0.18 (range: -0.33 - 0.02) for hospital palliative care, and 0.04 (range: -0.17 - 0.20) for home palliative care.

The mean utility value derived for the health state of death was 0.

The benefit measure used in the economic analysis was the quality-adjusted life-years (QALYs). The patient's quality of life for each treatment and health state was assessed using the "self-reported description" component of the EuroQol (EQ-5D) instrument. This instrument assesses five different dimensions (mobility, self-care, performance of usual activities, pain/discomfort and anxiety/depression). Each dimension then has three possible responses ("no problem", "some difficulties/moderate problem" or "unable/extreme problem").

The costs and resource use were reported separately. The direct costs incurred by the NHS were included in the analysis. These were for the drugs (imatinib and DAT), hospitalisation for patients receiving DAT, hospital palliative care, blood transfusions, diagnostic tests (chest X-ray, bone marrow test and computed tomography scan), outpatient clinic visits, and general practitioner and nurse visits. The cost data were obtained from published literature, the Chartered Institute of Public Finance and Accountancy, the Department of Health, and from personal communication with six NHS Trusts. Discounting was relevant since the costs were incurred during 5 years. The costs were discounted at an annual rate of 6%, in line with recommendations made by NICE. The study reported both the average costs and the incremental costs per patient. The authors did not state the dates to which the price data referred. Consumption taxes (VAT) were excluded, as they are a transfer payment from one sector in the economy to another with no net cost or gain to the government.

The resource use and unit costs were treated as point estimates (i.e. the data were deterministic).

The indirect costs were not included in the analysis.

UK pounds sterling ().

The parameters where there was the greatest uncertainty, and to which imatinib's cost-effectiveness was expected to be particularly sensitive, were varied within plausible ranges. These parameters included the price of imatinib, survival benefits, utility values, and the discount rates for costs and benefits. The incremental cost-utility ratios (ICURs) were also estimated for best- and worse-case scenarios. The parameters were set at the most optimistic or pessimistic limit in their range.

For the accelerated phase study cohort, the authors estimated median survivals of 39.5 months (cohort receiving imatinib) and 13.1 months (comparator cohort). This translated into 2.04 discounted QALYs per patient in the imatinib cohort and -0.04 discounted QALYs per patient in the comparator cohort.

For the blast crisis study cohort, the authors estimated median survivals of 8.7 months (imatinib cohort) and 4.5 months (comparator cohort). This translated into 0.53 discounted QALYs per patient in the imatinib cohort and -0.05 discounted QALYs per patient in the comparator cohort.

The discounted costs for patients in the accelerated phase were 78,593 per patient in the imatinib cohort, and 17,325 per patient in the comparator cohort. Thus, the incremental cost of imatinib was 61,268.

The discounted costs for patients in the blast crisis phase were 35,781 per patient in the imatinib cohort, and 11,085 per patient in the comparator cohort. Thus, the incremental cost of imatinib was 42,239.

The costs and benefits were combined by calculating an ICUR (additional cost required per QALY). The ICUR was 29,344/QALY for patients in the accelerated phase and 42,239/QALY for patients in the blast crisis phase.

Results from the sensitivity analysis showed that the cost-effectiveness of imatinib was sensitive to its price. For all discounted costs within the range tested, treatment with imatinib incurred additional costs in comparison with conventional treatments. If the price of imatinib was to be lowered by 50% the ICURs would be 54% lower. The cost-effectiveness of imatinib was insensitive to duration of response in the accelerated phase. In blast crisis, the ICUR was up to 7% higher or 8% lower, due to the wider CIs. Imatinib's cost-effectiveness was also sensitive to patient utility values. Under the best-case scenario, the ICURs were 69% lower (accelerated phase) and 43% lower (blast crisis). Under a worst-case scenario, the costs per QALY were substantially greater than at baseline (108% higher in accelerated phase; 189% higher in blast crisis).

In the treatment of chronic myeloid leukaemia (CML), the use of imatinib conferred considerably greater survival and quality of life than conventional treatments. However, these benefits came at a high cost.

Although no explicit justification was given for the comparator used, it would appear to represent current practice in the authors' setting. You should decide if the comparator represents current practice in your own setting.

The authors did not state that a systematic review of the literature was undertaken to identify relevant research and minimise biases. However, the authors appear to have used relevant and valid research from the international Phase II imatinib clinical trials 0109 and 0102 to derive their transition probabilities. When multiple studies were used to derive the transition probabilities for the comparator cohorts, the authors failed to state how results from these two studies were combined, if at all, and how differences between the studies were investigated. Hence, there is the possibility that transition probabilities for the comparator cohorts were derived by a selective interpretation of the literature. Further, these transition probabilities were not varied in the sensitivity analysis.

The authors also supplemented data from the literature review with a clinician panel to determine the proportion of patients receiving each type of conventional therapy. The authors did not report how these clinicians were selected, if they were a representative sample of their field, or how differences in responses were combined (e.g. by consensus or by taking an average). Further, these values were not varied in the sensitivity analysis. It was also unclear whether this was the same clinician panel as that used to derive the utility values.

The measure of health benefit was obtained through modelling parameters. It was not possible for the authors to elicit controlled utility values due to the nature of the Phase II trials used in the review. Hence, the authors had to resort to using a clinician panel to elicit these values. Even though the clinicians' responses may accurately depict experiences of CML patients (due to their expertise), the authors pointed out that they may not be a substitute for the actual patients' responses. However, to minimise this limitation the authors varied the elicited utility values in the sensitivity analysis. Following NICE recommendations, the QALYs were discounted at an annual rate of 1.5%.

For the cost perspective adopted (i.e. the NHS), all the relevant categories of cost were included in the analysis. With the exception of VAT, all the relevant costs for each category were included in the analysis. The authors excluded consumption taxes as they considered these to be a transfer within the economy, with no net gain or loss for the government. The costs and the quantities were reported separately, thus enhancing the generalisability of the authors' results. Resource use seems to have been derived on the basis of opinion and published sources. The authors did not undertake a sensitivity analysis of the quantities, which may limit the interpretation of the study findings. The unit costs were obtained from published sources and a sensitivity analysis of the prices was undertaken. Since all the costs were incurred during 5 years, the costs were discounted at an annual rate of 6% following NICE recommendations. The dates to which the prices related were not specified, which will hamper any possible reflation exercises.

The authors did not compare their findings with those from other studies, as no studies on the cost-effectiveness of treatments in the advanced stages of CML had been published. The issue of generalisability to other settings was partially addressed through the sensitivity analysis. The authors do not appear to have presented their results selectively and their conclusions reflected the scope of the analysis. The authors reported a number of further limitations to their study.

The authors assumed that a full month of imatinib treatment was given in each cycle. The authors believed that this would not substantially overstate the duration of imatinib treatment since a lag between disease progression (treatment failure) and clinical confirmation (cessation of treatment) seemed likely. As imatinib is more expensive than the comparator, this bias would favour the comparator.

The authors excluded movement between response rates representing best responses due to limitations in data availability. This will limit the accuracy of the overall estimates and may underestimate the effectiveness of imatinib.

The authors acknowledged that the duration of follow-up in the clinical trials was too short to assess long-term survival. Hence, the 5-year survival rates with imatinib may be substantially lower than those estimated by assuming a constant monthly probability of death.

The effectiveness data used for this study was obtained from the manufacturer of imatinib (Novartis), which might raise questions over the reliability of the effectiveness data. However, the same data were also reported in other published studies and reviewed by NICE. Hence, the authors considered that any potential bias should be minimal.

The authors included estimates of the likely net budget impact of introducing imatinib. For the estimated 1,000 advanced stage CML patients in the UK, the average cost to each of the 28 strategic health authorities would be 0.6 million pounds, having a relatively small impact on their budgets.

Funded by Novartis Pharmaceutical UK Ltd.

Capdeville R, Gathmann I. Integrated summary of efficacy - Update (cut-off 31 Jan 01). Basel: Novartis Pharma AG; 2001.

Kattan MW, Inoue Y, Giles FJ, et al. Cost-effectiveness of interferon-alpha and conventional chemotherapy in chronic myelogenous leukemia. Annals of Internal Medicine 1996;125:541-8.

Kolibaba KS, Druker BJ. Current status of treatment for chronic myelogenous leukaemia. Medscape Hematology-Oncology e-Journal 2000;3. http://www.medscape.com/viewarticle/408451

Lee SJ, Anasetti C, Kuntz KM, et al. The costs and cost-effectiveness of unrelated donor bone marrow transplantation for chronic phase chronic myelogenous leukemia. Blood 1998;92:4047-52.

Liberato NL, Quaglini S, Barosi G. Cost-effectiveness of interferon alfa in chronic myelogenous leukemia. Journal of Clinical Oncology 1997;15:2673-82.

Messori A. Cost-effectiveness of interferon in chronic myeloid leukaemia: analysis of four clinical studies. Annals of Oncology 2001;9:389-96.

Reese F, Wehrle E, Gathmann I. A phase III study of ST1571 Versus Interferon-alpha combined with cytarabine in patients with newly diagnosed previously untreated Philadelphia chromosone positive chronic myelogenous leukaemia in chronic phase (protocol). Basel: Novartis AG; 2000.