Eltrombopag for treating chronic immune thrombocytopenia

4.7 The Committee considered the evidence on the clinical effectiveness of eltrombopag, noting that the evidence was derived mainly from the RAISE trial. It noted that the available evidence showed that, for people with chronic ITP for whom other treatments had failed, eltrombopag was clinically effective when compared with placebo in attaining the target platelet count and reducing the need for rescue therapy.

4.8 The Committee discussed the safety and tolerability of eltrombopag and noted that the adverse reactions other than bleeding were similar between people who took eltrombopag or placebo in the RAISE trial. The Committee noted the lack of long-term safety data for both eltrombopag and romiplostim; however, it acknowledged that both treatments had better safety profiles than most standard treatments for chronic ITP.

4.9 The Committee discussed whether the manufacturer's indirect comparison between eltrombopag and romiplostim was appropriate. The Committee was aware that both the manufacturer and the Evidence Review Group (ERG) had advised caution when interpreting the results of the indirect comparison because of differences in baseline patient characteristics between RAISE and the 2 Kuter et al. (2008) trials for: duration of ITP, the proportion of patients who had received more than 3 prior ITP therapies, and the proportion of patients receiving concomitant ITP medication at baseline. The Committee discussed the sources of heterogeneity between the trials, and heard from the clinical specialists that the romiplostim trials were conducted before the eltrombopag trials; the Committee recognised that this may have caused more patients with severe chronic ITP to be enrolled into the earlier romiplostim trials. The Committee agreed that the differences between the RAISE trial and the 2 Kuter et al. trials may have introduced bias in the indirect comparison, but it concluded that it would be appropriate to perform an indirect comparison between the 2 treatments.

4.10 The Committee considered the manufacturer's indirect comparison and the ERG's exploratory analysis between eltrombopag and romiplostim, noting the different statistical approaches used to estimate the results. The Committee noted that, for the outcomes durable response and overall response, the manufacturer's indirect comparison gave odds ratios lower than 1 (favouring romiplostim), but the confidence intervals around those odds ratios suggested that the differences between eltrombopag and romiplostim were not statistically significant; for example, the odds ratio for overall response was 0.22 (in favour of romiplostim) with an upper limit of the 95% confidence interval of 1.02. The Committee understood that the statistical approach used by the manufacturer did not account for the heterogeneity between the 2 Kuter et al. (2008) trials, whereas the ERG's exploratory indirect comparison, which used a Bayesian approach, treated the Kuter et al. trials separately. The Committee noted that, for overall response, the ERG's indirect comparison suggested a statistically significant difference in favour of romiplostim (odds ratio 0.15). For durable response, the Committee noted that the ERG found non-significant differences similar to those found by the manufacturer. The Committee was aware that the ERG's results depended on the degree of heterogeneity it assumed. The Committee agreed that the point estimates in both the manufacturer's and the ERG's analyses were associated with considerable uncertainty.

4.11 The Committee considered the relative effectiveness of eltrombopag and romiplostim in light of the manufacturer's indirect comparison and the ERG's exploratory analysis. It noted that the manufacturer interpreted the effectiveness of the 2 drugs as the same (that is, not different) on the basis that its indirect comparison did not show a statistically significant difference between them. The Committee heard from the clinical specialists that, while it is difficult to know whether 1 treatment is superior to the other, the use of eltrombopag and romiplostim in clinical practice is broadly interchangeable. The Committee accepted that the manufacturer's indirect comparison may have underestimated the clinical effectiveness of romiplostim given that the romiplostim trials preceded those for eltrombopag, and so the clinical trials for romiplostim may have enrolled patients whose condition was relatively more severe. The Committee noted that more patients in the romiplostim trials had received multiple previous therapies, which suggests that they better reflected patients whose condition had not responded than those in the eltrombopag trials. The Committee agreed that the available evidence suggested that romiplostim was likely to be more effective than eltrombopag rather than equally effective, and so it did not agree with the manufacturer's assumption used for the modelling that the treatments were equally effective. The Committee concluded that the most plausible odds ratio for overall response for eltrombopag compared with romiplostim would be less than 1.00 but, given the uncertainty around the point estimates obtained from the indirect comparison, it could not determine the likely value of this ratio.

4.12 The Committee considered the clinical effectiveness of eltrombopag compared with the pathway of standard care alone. The Committee understood that there was no direct evidence comparing eltrombopag with standard care, and so discussed the manufacturer's indirect comparison (section 3.16). The Committee noted that the evidence for treatments used in standard care was derived mainly from non-randomised, highly heterogeneous trials. It also noted that the manufacturer had altered its inclusion criteria after performing the literature review, and pooled response estimates using a simple weighted average of treatment groups. The Committee agreed that the indirect comparison lacked methodological rigor, and concluded that the results of the indirect comparison were not sufficiently robust to compare eltrombopag with the pathway of standard care alone.

4.13 The Committee considered the manufacturer's cost-effectiveness analyses, and the ERG's critique of the analyses. The Committee agreed that, of the 3 economic evaluations (the base case, the alternative evaluation and the scenario analysis), the alternative evaluation represented the most valid analysis because the modelling applied data derived directly from the pivotal trials of eltrombopag and the manufacturer's own systematic review.

4.14 The Committee considered the cost effectiveness of eltrombopag compared with the pathway of standard care alone in the alternative evaluation, that is, for people with ITP for whom romiplostim is not recommended. It noted that the manufacturer had estimated response rates and time on treatment for drugs used in standard care from its indirect comparison of treatments used in standard care (section 3.22), and used the estimates in the alternative evaluation. The Committee, however, agreed that the indirect comparison lacked methodological rigor (section 4.12), and that there was no sufficiently robust cost-effectiveness evidence to make a recommendation for eltrombopag compared with the pathway of standard care alone. The Committee concluded that it could not recommend eltrombopag for patients who do not have severe disease and a high risk of bleeding.

4.15 The Committee then considered the cost effectiveness of eltrombopag compared with the pathway of standard care plus romiplostim in the alternative evaluation. It noted that the results of this comparison would apply only to people with severe chronic ITP and a persistent high risk of bleeding (that is, people for whom romiplostim is recommended in NICE technology appraisal guidance 221 for romiplostim). The Committee considered the sensitivity analyses in which romiplostim was more effective than eltrombopag, and noted that neither the manufacturer nor the ERG had provided incremental cost-effectiveness ratios (ICERs) for eltrombopag compared with romiplostim. However, from the costs and QALYs presented by the ERG from its exploratory sensitivity analyses on the alternative evaluation (section 3.49), the Committee initially estimated that the ICERs would be more than £400,000 saved per QALY lost for patients who had or had not had a splenectomy. The Committee noted that a comment received during consultation on the first appraisal consultation document indicated that it would be more appropriate to include in the sensitivity analyses the odds ratio for overall response of 0.15 estimated from the ERG's indirect comparison (section 3.40). The Committee heard from the ERG that, in response to this comment, it had carried out sensitivity analyses within the alternative evaluation, varying the parameter for overall response rate in the model (section 3.50). The Committee noted that, when the ERG applied an odds ratio of 0.22 to derive overall response rates, the ICERs for eltrombopag compared with romiplostim were savings of £689,000 and £373,000 per QALY lost for patients who had and had not had a splenectomy respectively; when the ERG applied an odds ratio of 0.15, these ICERs decreased to £638,000 and £351,000 per QALY lost respectively. The Committee was aware that the clinical specialists felt that eltrombopag and romiplostim were broadly interchangeable (section 4.11), and it concluded that, if the odds ratio for overall response moved towards 1.0 (as implied by the clinical specialists' willingness to substitute 1 treatment for another), the ICERs would further increase leading to further savings per QALY lost.

4.16 The Committee noted that the ERG questioned the source of the data on health-related quality of life used in the manufacturer's model because the manufacturer did not use the SF‑36 health-related quality-of-life data collected from RAISE and EXTEND that it had mapped on to the SF‑6D. The Committee noted that the manufacturer applied the SF‑6D utility data in a sensitivity analysis within the base case, but not within the alternative evaluation favoured by the Committee. The Committee noted comments received during consultation on the second appraisal consultation document suggesting that it would be more appropriate to use, in sensitivity analyses within the alternative evaluation, EQ‑5D utility data obtained either from mapping SF‑36 data on to the EQ‑5D or from other studies from the literature identified by the manufacturer (such as Szende et al. 2010). The Committee heard that mathematical algorithms exist to map from SF‑36 (used in RAISE and EXTEND) on to EQ‑5D, and that, because no single algorithm is considered more valid than others, the EQ‑5D data obtained from using a particular algorithm would be associated with further uncertainty. It also heard from the ERG that the Szende et al. study used by the manufacturer in its base-case analysis did not report EQ‑5D utility values. The Committee was aware that the reference case outlined in NICE's Guide to the methods of technology appraisal states that EQ‑5D is the preferred measure of health-related quality of life in adults. However, in the absence of EQ‑5D data, the Committee concluded that, of the utility data available, the SF‑6D data provided by the manufacturer were the most appropriate to use within the alternative evaluation.

4.17 The Committee discussed the ERG's concern about the dosing of romiplostim in the analyses (section 3.46). It noted that the doses of romiplostim used in the manufacturer's model did not depend on whether or not the patient's condition had responded, whereas in the Kuter et al. (2008) trials, the median dose of romiplostim in patients whose condition had responded was 40–60% lower than that across the trial as a whole. The Committee noted a comment received during consultation on the second appraisal consultation document suggesting to set the dose of eltrombopag after 24 weeks of treatment equal to the average dose from week 12 to week 23 in the RAISE trial (which was 7–12% higher than the originally modelled doses) to link the doses of eltrombopag with the modelled response rates from RAISE. The Committee discussed the impact of this suggestion on the cost effectiveness of eltrombopag compared with romiplostim, and heard from the ERG that the proposed dosing for eltrombopag would only minimally affect the relative cost effectiveness of the 2 treatments. In addition, the Committee was aware that patient experts indicated that some patients take a tablet of eltrombopag only once every 3 days rather than daily (section 4.4), which implies that the dose of eltrombopag in clinical practice might in fact be lower than that observed in the RAISE trial. The Committee concluded that it was appropriate to use a median dose of romiplostim that is 40–60% lower than that used in the Kuter et al. trials and that the dose of eltrombopag used in the model was appropriate.

4.18 The Committee noted that comments received during consultation on the first appraisal consultation document raised concerns about some of the parameters used in the model, namely that the duration of treatment for romiplostim is longer than that for eltrombopag; that doses of romiplostim should be calculated in line with the approach used for NICE technology appraisal guidance 221 (that is, 1.54 vials for patients who had had a splenectomy and 1.10 vials for patients who had not had a splenectomy); that time to response for eltrombopag should be equal to that for romiplostim; that an in-hospital cost of £11.75 per administration for romiplostim should be used; and that anti‑D should be excluded as a rescue therapy for patients who had not had a splenectomy.

4.19 The Committee considered these comments as follows:

4.20 The Committee noted the comment received during consultation on the second appraisal consultation document, which suggested that it would be more appropriate to model higher rates of bleed and rescue therapy from the Kuter et al. (2008) trial to reflect the severely affected population for which romiplostim is recommended. The Committee recognised that rates of bleeding and use of rescue therapies are important parameters in the model. It noted that both rates were higher in the romiplostim trials (Kuter et al.) than in the eltrombopag trials (RAISE and EXTEND). It heard from the ERG that the SF‑6D utility data (favoured by the Committee, section 4.16) for bleeding events were aligned with the definition of bleeds in RAISE, but not with the definition in the Kuter et al. trials. The Committee was aware that neither the manufacturer nor the ERG applied the bleed and rescue rates from Kuter et al. in sensitivity analyses within the alternative evaluation favoured by the Committee. The Committee noted that there was no information available on bleeding and rescue rates in clinical practice, but it heard from the ERG that including the higher rates from Kuter et al. would not lower the cost effectiveness of eltrombopag compared with romiplostim below £30,000 saved per QALY lost. It accepted that, if the higher rates of bleeds and rescue therapy from Kuter et al. were more pertinent to the population with severe ITP for which romiplostim is recommended and for which eltrombopag was being considered, this would have an impact on the ICERs in favour of romiplostim, but it would be extremely unlikely to affect the relative cost effectiveness of eltrombopag and romiplostim to a degree where the Committee would change its recommendations. The Committee concluded that it did not need to consider further sensitivity analyses incorporating the higher bleed and rescue rates from the Kuter et al. trials.

4.21 The Committee discussed the most plausible ICERs for eltrombopag compared with romiplostim. The Committee noted that the ERG did not initially report ICERs for eltrombopag compared with romiplostim from its exploratory sensitivity analyses within the alternative evaluation. It was also aware that no ICERs were available that incorporated all the parameter inputs favoured by the Committee within a single analysis. Therefore, the Committee considered the additional sensitivity analyses carried out by the ERG in response to comments on the first appraisal consultation document (section 4.18). The Committee agreed that romiplostim is likely to be more clinically effective than eltrombopag; that it was appropriate to use the SF‑6D utility data collected from RAISE and EXTEND, a lower romiplostim dose and a lower administration cost for romiplostim, and to exclude anti‑D. The Committee considered the analysis that mirrored this, and noted that the resulting ICERs for eltrombopag compared with romiplostim were £389,000 saved per QALY lost for patients who had had a splenectomy and £271,000 saved per QALY lost for patients who had not had a splenectomy. The Committee acknowledged that these ICERs are associated with considerable uncertainty. It accepted that the ICERs would be higher (in favour of eltrombopag) when accounting for a romiplostim administration cost in hospital of more than £11.50, or if romiplostim relative to eltrombopag was less effective (that is, if the odds ratio for overall response was greater than the 0.15 used in the ERG's analyses). The Committee also accepted that the ICER would be lower (in favour of romiplostim) if the rates of bleeding and rescue therapy in clinical practice were higher than those applied in the cost-effectiveness analysis. The Committee accepted that there was a degree of uncertainty surrounding the ICER for eltrombopag compared with romiplostim. However, it was satisfied that, based on the evidence it had seen and the comments received during consultation on 2 appraisal consultation documents, eltrombopag can be considered a cost-effective use of NHS resources. The Committee noted that, in situations in which an ICER is derived from a technology that is less effective and less costly than its comparator, the commonly assumed decision rule of accepting ICERs below a given threshold is reversed, and so the higher the ICER, the more cost effective a treatment becomes. The Committee concluded that eltrombopag should be recommended as specified in its marketing authorisation (that is, in adults who have had a splenectomy and whose condition is refractory to other treatments, or as a second-line treatment in adults who have not had a splenectomy because surgery is contraindicated) as an option for treating adults with chronic ITP, but only if their condition is refractory to standard active treatments and rescue therapies, or they have severe disease and a high risk of bleeding that needs frequent courses of rescue therapies, and the manufacturer provides eltrombopag with the discount agreed in the patient access scheme.

4.22 The Committee noted that comments received in response to consultation on the appraisal consultation documents suggested that a specialist haematologist should supervise treatment with eltrombopag. The Committee agreed that, because the summary of product characteristics stipulates that eltrombopag treatment should remain under the supervision of a physician who is experienced in the treatment of haematological diseases (section 2.2), it did not need to repeat this in its recommendations.

4.23 The Committee was aware of comments received in response to the consultation on the appraisal consultation documents expressing concerns about the wording of the preliminary recommendation for eltrombopag, which, unlike the recommendation for romiplostim in NICE technology appraisal guidance 221, included reference to whether a person had or had not had a splenectomy. While the Committee appreciated the concerns raised, it was aware that the wording of the marketing authorisations for eltrombopag and romiplostim stipulates that both of these treatments should only be used if a person has had a splenectomy or has not had a splenectomy because such surgery is contraindicated. The Committee acknowledged that NICE cannot recommend a treatment outside its marketing authorisation, and agreed that the wording of the recommendation in section 1 should reflect the wording of the marketing authorisation for eltrombopag.

4.24 The Committee discussed the limited published data on the long-term safety and effectiveness for both eltrombopag and romiplostim (sections 4.8 and 4.11), and on observed rates of bleeding and rescue therapy in clinical practice (section 4.20). The Committee considered that the collection of more data on the clinical effectiveness of both eltrombopag and romiplostim would be useful for future appraisals of treatments for chronic ITP because it would enable a more robust estimate of the clinical and cost effectiveness of the treatments. Given the difficulties of conducting randomised controlled trials and in generalising their results to clinical practice, the Committee supported generating and analysing observational data including, but not limited to, the existing UK ITP Registry, which collects data on the long-term outcomes of patients treated with eltrombopag and romiplostim.

4.25 The Committee discussed the differences in the wording of the guidance developed in this appraisal and the wording in the recommendations for romiplostim in NICE technology appraisal guidance 221, the latter of which were developed some time ago. The Committee concluded that it is necessary to ensure clarity around the fact that the recommendations for eltrombopag and romiplostim are for exactly the same patient population.