4 Evidence and interpretation

The Appraisal Committee (Appendix A) considered evidence from a number of sources (Appendix B).

4.1 Clinical effectiveness

4.1.1 Haematological response (HR) and cytogenetic response (CR) have been used as surrogate measures of efficacy in imatinib studies. HR refers to the normalisation of blood cell counts, whereas CR refers to the reduction (partial CR) or elimination (complete CR) from the bone marrow of white blood cell precursors with the Philadelphia chromosome.

4.1.2 One RCT and three case series were identified in published literature; all were sponsored by the manufacturer of imatinib. All three case series were described as open-label, non-randomised, phase II studies. Of these, one investigated the efficacy of imatinib as a second-line treatment (after failure of IFN-α) in 454 people with chronic-phase CML; one investigated the efficacy of imatinib in 181 people in the accelerated phase; and the third investigated imatinib in 229 people in blast crisis. These three case series were reviewed by the Institute when producing the previous NICE guidance on imatinib (NICE Technology Appraisal Guidance No. 50).

4.1.3 The RCT (known as the IRIS trial) compared imatinib with the combination of IFN-α and cytarabine (Ara-C) in 1106 people with newly diagnosed chronic-phase CML. Full details of the IRIS trial and a summary of the updated results from the case series were made available to the Institute as part of the manufacturer's submission for this appraisal. The major outcome measures in these studies were time to progression, CR, HR and survival. The IRIS trial also reported on measures of quality of life (QoL).

Chronic phase (first line)

4.1.4 In the IRIS trial, intention-to-treat analyses of the results at 18 months indicated that the disease had not progressed in an estimated 92% of people in the imatinib group (n = 553) compared with 74% of people in the IFN-α and Ara-C combination group (n = 553). Additionally at 18 months, complete CR was achieved more often in people in the imatinib group compared with those in the IFN-α and Ara-C combination group (76% versus 15% of people, respectively).

4.1.5 The difference in overall survival between groups was not statistically significant at 18 months: 97% in the imatinib group versus 95% in the IFN-α and Ara-C combination group (p = 0.16).

4.1.6 Withdrawal because of side effects was 2.2% in the imatinib group compared with 6.0% in the IFN-α and Ara-C combination group at 18 months. Crossover as a result of intolerance was much higher in the IFN-α and Ara-C combination group (25%) compared with the imatinib group (less than 1%).

4.1.7 QoL was reported to be better in the imatinib group compared with the IFN-α and Ara-C combination group when assessed at 1, 3 and 6 months using the 'Functional Assessment of Cancer Therapy – Biological Response Modifier' (FACT-BRM) instrument. However, interpretation of the results may be difficult because of different completion rates of the groups (80% in the imatinib group versus 59% in the IFN-α and Ara-C group at 12 months) and the high withdrawal rates in the IFN-α and Ara-C combination group. In addition, only three of the five subscales of the FACT-BRM were used in this trial.

4.1.8 In the IRIS trial, the overall crossover rate (including crossover as a result of treatment failure) between groups was high, in particular for those who were initially treated with the IFN-α and Ara-C combination. At 18 months, 11 out of 553 (2%) people in the imatinib group and 318 out of 553 (58%) in the IFN-α and Ara-C group had crossed over treatments.

Chronic phase (after failure of or intolerance to IFN-α)

4.1.9 In the case series that investigated the efficacy of imatinib in 454 people with late chronic-phase CML, for whom previous therapy with IFN-α had failed, complete HR was achieved in 95% of people, major CR (complete or partial) in 60% of people, and complete CR in 41% of people, at 18 months. Overall survival was 95% at 18 months. Additional analysis of data reported in this study suggested that progression-free survival at 18 months was significantly longer in those who demonstrated major CR at 3 months compared with those who did not (97% versus 88%, p = 0.005 by the log-rank test).

4.1.10 At the 31-month follow-up, 74% of the original study population remained on imatinib and 48% of these still demonstrated complete CR. The progression-free survival was estimated to be 87% at 24 months, with overall survival of 92% at 24 months.

Accelerated phase

4.1.11 In the published accelerated-phase study, the results were reported at median follow-up times of 9.9 (400-mg group, n = 77) and 11 months (600-mg group, n = 158). Combining both groups and time points, 53% achieved complete HR and 19% returned to the chronic phase. Major CR was reported in 24% of the study population, whereas 17% had complete CR. Estimated overall survival rates at 12 months were 65% (95% confidence interval [CI], 53% to 77%) for the 400-mg group, and 78% (95% CI, 68% to 81%) for the 600-mg group. Median survival had not been reached in either group at the time the study results were published. The progression-free survival rates at 12 months were 44% (95% CI, 31% to 56%) for the 400-mg group, and 67% (95% CI, 59% to 76%) for the 600-mg group.

4.1.12 At the 36-month follow-up in the 600-mg group, the overall survival rate was 66%, with an expected time to progression of 23 months.

Blast crisis

4.1.13 In the published blast-crisis study, sustained HR (lasting at least 4 weeks) was reported for 31% of people (n = 229); 8% of people had complete HR and 18% had returned to the chronic phase. The median duration of treatment was 4.0 months (600 mg). At this time point, 16% of people achieved major CR, with 7% having complete CR. Among those who achieved a sustained response, the median duration of HR was estimated to be 10 months. The overall 12-month survival rate was estimated to be 32%.

4.1.14 At the 36-month follow-up, 8% of the study population remained on imatinib, and the overall median survival was 6.9 months.

4.2 Cost effectiveness

4.2.1 Only one published abstract concerned with economic evaluation of second-line imatinib therapy (after IFN-α had failed) was identified in the literature. In addition, the manufacturer's submission presented an economic model, and the Assessment Group developed an independent economic model.

4.2.2 The published economic evaluation (abstract only) did not provide full details of methodology or sensitivity analyses. This study reported the incremental cost-effectiveness ratio (ICER) of imatinib as a second-line treatment over HU in the chronic phase to be £35,000 per quality-adjusted life year (QALY). The ICER for imatinib compared with combination chemotherapy or palliative care in the accelerated phase was around £22,000 per QALY and in the blast-crisis phase £43,500 per QALY. The year of costs was not stated but the abstract was presented in 2002.

4.2.3 The manufacturer's submission included an economic evaluation based on a new Markov model that compared the costs and QALYs in a hypothetical cohort of 1000 newly diagnosed people receiving imatinib as a first-line treatment with a similar cohort of 1000 people receiving IFN-α. The model runs for 30 years, using 1-month cycles. The key effectiveness data were based on the IRIS study. Using two different techniques to estimate the survival benefit, the manufacturer's model estimated that the ICERs for imatinib treatment when compared with IFN-α were £19,000 and £27,000 per QALY.

4.2.4 An independent economic model was developed by the Assessment Group to determine the ICER of imatinib compared with HU and IFN-α, and of IFN-α compared with HU in terms of cost per QALY. This is a Markov model that follows a cohort of 1000 people with CML from the start of treatment until death, or for a maximum of 20 years. The cycle length for the model is 3 months and costs are calculated based on an NHS perspective. Key effectiveness data comes from published literature.

4.2.5 The independent model estimated the ICER of imatinib compared with IFN-α to be around £26,000 per QALY gained (ranging from £13,500 to £52,000). Results were relatively robust when subjected to a number of sensitivity analyses. The highest ICER estimate was obtained when higher doses of imatinib were assumed (that is, 600 mg for the chronic and accelerated phases and 800 mg for the blast-crisis phase). Imatinib was less cost effective when compared with HU, with an ICER of £87,000 per QALY. The ICER of IFN-α when compared with HU was considerably higher – in excess of £1 million per QALY.

4.2.6 In the Institute's previous guidance (NICE Technology Appraisal Guidance No. 50), the ICER for imatinib treatment when compared with HU was estimated to be between £36,000 and £38,000 per QALY as a second-line treatment in chronic-phase CML, between £21,800 and £56,000 per QALY in the accelerated phase, and between £33,275 and £64,750 per QALY in the blast-crisis phase.

4.3 Consideration of the evidence

4.3.1 The Committee reviewed the data available on the clinical and cost effectiveness of imatinib for CML, having considered evidence on the nature of the condition and the value placed on the benefits of imatinib from people with CML, those who represent them, and clinical experts. It was also mindful of the need to take account of the efficient use of NHS resources.

4.3.2 The Committee was mindful of the current licensed indications for the use of imatinib and the previous guidance produced by the Institute regarding the use of imatinib in the circumstance of intolerance or resistance to first-line IFN-α treatment, which was based on evidence primarily from case series. Conversely, the new licensed indication for imatinib as first-line therapy is supported by a single RCT. However, the Committee's deliberations were hampered by the absence of long-term survival data. Thus, the published supportive evidence from the RCT relied principally on the surrogate measures of efficacy such as the achievement of an HR and/or a CR.

4.3.3 The extent to which CR (particularly) and HR, as surrogate measures, predict survival is central to the judgment about the clinical and cost effectiveness of imatinib. After reviewing the available data, the Committee considered it likely – based on current evidence of the significance of CR and HR in CML, and knowledge of the effect of imatinib – that the relationship between CR and survival is sufficiently strong to support the use of CR, in particular, as a surrogate measure of survival in people with chronic-phase CML.

4.3.4 The Committee was therefore persuaded, by the current evidence and by that presented by the experts for this appraisal, that the RCT reviewed for the first-line use of imatinib indicated that there was likely to be a significant survival advantage for imatinib over IFN-α in this clinical situation. However, the Committee did not believe it was possible, based on the current evidence, to determine precisely the absolute survival gain that would result from switching from IFN-α to imatinib as first-line treatment.

4.3.5 The Committee considered the implications of high crossover rates in the IRIS trial, and the appropriateness of using intention-to-treat and per protocol analyses. The Committee thought that the degree of the benefit of imatinib treatment would have been underestimated because of the high crossover rate in the IFN-α and Ara-C combination group.

4.3.6 The Committee additionally reviewed the clinical effectiveness evidence from the Assessment Report for other treatment options in chronic-phase CML, in particular IFN-α and HU. It discussed in detail the appropriateness of these comparator treatments in the context of first-line treatment with imatinib for chronic-phase CML. The Committee concluded that, although it was reasonable to regard HU as a comparator treatment in this context, current clinical practice (prior to the licensing of imatinib) uniformly considered IFN-α as the principal treatment of choice for people in the chronic phase of CML, provided it can be tolerated.

4.3.7 The Committee also carefully considered the cost effectiveness of imatinib treatment compared with alternatives, including both IFN-α and HU. Compared with IFN-α, the Committee considered that imatinib was a cost-effective option. The results from the independent model suggested, however, that the cost effectiveness of imatinib when compared with HU was not acceptable, with an ICER of around £87,000 per QALY. The ICER of IFN-α compared with HU was very much higher, in excess of £1 million per QALY.

4.3.8 In line with the considerations outlined in Section 4.3.5, the Committee asked the Assessment Team to test the impact of using per protocol values instead of intention-to-treat values on the cost-effectiveness results. The additional analysis using new assumptions, including the use of per protocol values, resulted in slightly improved ICERs for imatinib, to around £60,000 when compared with HU.

4.3.9 The Committee further discussed the issues of comparing imatinib with IFN-α versus HU. If IFN-α would not be considered a cost-effective treatment compared with HU, then how should the ICER of imatinib compared with IFN-α be viewed by the Committee? The Committee was, however, persuaded that, because IFN-α is currently accepted as a standard first-line treatment for people with CML (although it might not be considered cost-effective), it was appropriate to compare imatinib with IFN-α in terms of its ICER.

4.3.10 The Committee considered the implications of this decision on the overall cost effectiveness of first-line treatment of CML. Although imatinib treatment was not cost effective when compared with HU, the introduction of imatinib as a first-line therapy for chronic-phase CML would displace the use of IFN-α for this purpose, except in people currently satisfactorily treated with IFN-α. The Committee therefore considered that this may result in a better use of NHS resources for CML.

4.3.11 The Committee considered the continuing use of imatinib in chronic-phase CML for people who had not achieved the principal endpoints of treatment as identified in the IRIS trial. The 13-month data from the IRIS trial suggested that only 6.5% in the imatinib arm fell into this category and thus were treated by dose escalation as per protocol. The clinical experts suggested that this group may be larger in clinical practice than was reflected in the trial. The effect of dose escalation on the cost effectiveness of imatinib in this situation was discussed by the Committee. They also reviewed the sensitivity analysis provided in the Assessment Team's economic model relating to differing doses of imatinib. The Committee was persuaded that, for the majority of people receiving imatinib, dose escalation would not be required and that for those that did, further information on the effectiveness of this strategy was required in the longer term to inform the appraisal review.

4.3.12 The Committee considered the use of imatinib in people who were diagnosed in chronic-phase CML but then progressed to the accelerated phase or blast crisis without a trial of imatinib treatment. The Committee was persuaded that imatinib may provide a clinically and cost-effective treatment option for these people.

4.3.13 The Committee considered the situation in which individuals progressed from the chronic phase to the accelerated phase or blast crisis while they were receiving imatinib. The Committee was aware that in this circumstance there is currently no clinical consensus of the correct therapeutic approach. In addition, there is no evidence on the clinical and cost effectiveness of continuing imatinib treatment by escalating the dose after it has failed to prevent disease progression in the chronic phase. The Committee was also fully aware of the implications regarding the significantly increased cost-effectiveness ratios where dosages of imatinib greater than 400 mg/day were used. Therefore, the Committee concluded that, for individuals who progress to the accelerated or blast-crisis phase whilst taking imatinib, the continuing use of the drug could only be recommended on the basis of collection of prospective data to inform the clinical and cost effectiveness of dose-escalation strategies. The information gathered will enable the Institute to review its guidance on the use of imatinib in 2006.

4.3.14 For people currently receiving IFN-α treatment for chronic-phase CML, the decision about whether to change to imatinib should be dictated by the current response of the disease to treatment and by the tolerance of the person to IFN-α. This decision should be made after informed discussion between the person with CML and the responsible clinician, taking full account of the evidence on the effects of imatinib and the wishes of the person.