Bortezomib monotherapy for relapsed multiple myeloma

  • Technology appraisal guidance
  • TA129
  • Published: 
Download guidance (PDF)

3 The manufacturer's submission

The Appraisal Committee (appendix A) considered evidence submitted by the manufacturer of bortezomib and a review of this submission by the Evidence Review Group (ERG; appendix B).

3.1 The manufacturer's submission approached the decision problem by comparing the clinical effectiveness of bortezomib with that of high-dose dexamethasone (HDD), based on the results of the APEX (Assessment of Proteasome Inhibition for Extending Remissions) randomised controlled trial (RCT). The population considered was people with multiple myeloma at first or subsequent relapse; however, the manufacturer's submission placed emphasis on patients at first relapse. The manufacturer considered HDD to be the most appropriate comparator because it is an effective monotherapy for relapsed multiple myeloma that is commonly used in clinical practice in the UK, and its use at first relapse is within its licensed indications. In addition, HDD was the comparator agreed as the basis for regulatory approval of the APEX RCT.

3.2 In an interim analysis of the APEX trial (median follow-up of 8.3 months), it was found that people receiving bortezomib had a statistically significantly longer median time to disease progression compared with people receiving HDD (6.2 months compared with 3.5 months, hazard ratio 0.55, 95% confidence interval 0.44 to 0.69; p < 0.001). They also had a significantly improved overall survival (hazard ratio 0.57, 95% confidence interval 0.40 to 0.81; p = 0.001) and a significantly higher overall (complete or partial) response rate (38% compared with 18%; p < 0.001). As a result of the interim analysis and the recommendation of the data monitoring committee, all patients in the dexamethasone group were offered bortezomib. Updated analyses were performed at 15.8 months and 22 months of follow‑up. At 22 months follow-up, the median overall length of survival in the intention to treat population was 29.8 months in the bortezomib arm compared with 23.7 months in the HDD arm.

3.3 The manufacturer's submission provided cost-effectiveness evidence using a semi-Markov state-transition model to compare bortezomib with HDD. The manufacturer indicated that it did not include other comparators in the model because there is currently no UK consensus on best practice for the treatment of multiple myeloma at first relapse, because there are no other treatments available that hold a UK marketing authorisation for use at first relapse, and because of limitations in the available evidence. Because a high percentage of patients in the HDD arm of the APEX study were allowed to cross over to receive bortezomib, the manufacturer emphasised that the true difference in overall survival between the bortezomib and HDD arms was greater than in the reported results of the APEX study. Therefore, data from the Mayo Observational Study, which included some patients receiving a dexamethasone-containing regimen (a combination of vincristine, adriamycin and dexamethasone [VAD], of which dexamethasone is expected to be the most active ingredient), were used in addition to data from the APEX study in the modelling . The base case included people at first relapse only, resulting in a point estimate of the incremental cost-effectiveness ratio (ICER) of £31,000 per life year gained.

3.4 One-way sensitivity analyses of the key parameters identified in the manufacturer's model resulted in a range of ICERs from £28,000 to £31,000 per life year gained and showed that the duration of treatment effect was the most influential parameter. Three scenario analyses were also presented.

  • An analysis in which a rule was used by which patients whose disease had not responded to treatment (defined as not reaching complete or partial response using the European Blood and Marrow Transplant [EBMT] criteria) after three cycles would not continue treatment. Reductions in both bortezomib costs and survival benefit (resulting from discontinuing treatment) were included in the model. The reduction in survival benefit was calculated from the number of patients who responded within three cycles as a percentage of all those who responded.

  • An analysis in which the proportion of patients entering the model at first and second or subsequent relapse was assessed.

  • An analysis in which the use of bortezomib in combination with HDD was assessed.

3.5 In an additional analysis provided by the manufacturer in response to questions raised in the evidence-review phase, the base-case cost per life year gained of £31,000 was estimated to translate to £38,000 per quality-adjusted life year (QALY). The corresponding figure for a scenario with a three-cycle stopping rule with an ICER of £28,000 per life year gained was £33,500 per QALY gained. The QALYs were derived using utility values of 0.81 for the pre‑progression state, and 0.64 after progression, based on a published study of patients with previously untreated multiple myeloma. The manufacturer requested that due consideration be given to the view that it is more appropriate to measure cost effectiveness in terms of cost per life year gained in patients with multiple myeloma. The manufacturer argued that survival gain is the single most important outcome for people with relapsed multiple myeloma, that there is a lack of robust utility data to compute QALYs for people with relapsed multiple myeloma, and that the EuroQoL-5D (EQ-5D) quality of life (QoL) measure is not sensitive to some important facets of multiple myeloma.

3.6 The ERG raised a number of key issues about the manufacturer's submission.

  • It raised concerns about the uncertainty in the cost-effectiveness analysis resulting from using data from the Mayo Observational Study. For example:

    • HDD was not one of the reported regimens for the observational study

    • the data used were a subset of the Mayo patient data

    • of the 335 participants in the Mayo study who had received one prior therapy and were on their second regimen, only 33 received a dexamethasone-containing regimen

    • the Mayo Observational Study reported data collected in the United States over a 13-year period, so patients may not have benefited from the latest treatment protocols

    • the observational data were not specific about which patients had what treatment and when

    • there were some differences between the patient profiles in the APEX RCT and the Mayo Observational Study; for example, patients in the APEX RCT were diagnosed approximately 5 years earlier than the Mayo patients

    • the data used may have predicted a more severe disease progression profile (that is, a shorter time to progression and higher mortality) than would be expected in a hypothetical cohort of patients treated with HDD in the context relevant for this appraisal.

  • The model submitted by the manufacturer may have overestimated the true treatment effect of bortezomib because of the way in which data from the Mayo Observational Study were used to address the crossover in the APEX study.

  • Adverse effects were not included in the economic model, in terms of either reduction in QoL or increased use of resources.

  • The ERG's review of sensitivity analyses indicated a greater variability in cost-effectiveness estimates than was presented in the manufacturer's submission. The ERG found that the most influential parameters were the hazard ratio for time to disease progression and the cost of bortezomib.

  • The ERG stated that, if patients are treated at a later stage of multiple myeloma, the cost per life year gained increases significantly. The ERG found that when all patients were treated at second relapse, the ICER was £77,000 per life year gained; when all patients were treated at third relapse, the ICER was £107,000 per life year gained.

3.7 The manufacturer's response to the issues raised included clarification on the APEX study and a revised economic report that included additional scenarios involving vial sharing. The costs of grade 3 or 4 adverse events were included, based on the frequency reported in the APEX RCT, resulting in an additional average cost of £1463 for bortezomib and £703 for HDD. There were four categories of adverse events: anaemia, thrombocytopenia, neutropenia, and all other grade 3 or 4 adverse events. The manufacturer also provided clarification about the impact of using data from the Mayo Observational Study, and stated that this affected only the modelling of post-progression survival and that the survival gain predicted by the model was realistic, or even conservative.

3.8 The manufacturer was further requested to provide details of a response-based rebate scheme that had been proposed to the Department of Health. In the manufacturer's proposed rebate scheme, the Velcade Response Scheme, 'responders' were defined as patients after first relapse whose disease reached at least a minimal response (measured as at least a 25% reduction in the first serum M protein response seen ['initial M protein']) after up to four cycles of bortezomib treatment. For people with myeloma who do not have measurable serum M protein levels (approximately 10 to 15% of patients) response would be assessed in terms of reduction in urinary free light-chain (Bence-Jones protein) excretion. The manufacturer also was asked to clarify the way in which the modelling of such a scheme differed from the model previously reviewed by the Committee. The manufacturer stated that the only difference in the modelling of costs between the stopping rule without rebate and with rebate is that the bortezomib drug costs are removed for non-responders up to the point at which they cease treatment. Additional costs to the NHS of administering the scheme were not included. However, the manufacturer explained that these would be minimal because the rebate could be claimed using a simple form to be faxed to the manufacturer, and that the manufacturer would bear the cost of distributing replacement stock. As an alternative to replacement stock, the manufacturer would provide a credit note or cash refund.

3.9 The manufacturer provided the Committee with:

  • Analyses based on a four-cycle stopping rule in addition to the three-cycle stopping rule originally proposed (see sections 3.4 and 3.5), with and without rebate for non-responders

  • Analyses of response based on EBMT criteria as well as on the initial M protein response, with and without rebate for non-responders

  • Analyses that had been adjusted to reflect the rebate of bortezomib costs that the manufacturer would pay to the NHS for those patients whose disease does not meet the required response criteria. This was in addition to the reduction in bortezomib costs and in survival benefit that had been used to calculate the ICER of £33,500 per QALY gained for a three-cycle stopping rule without rebate (see sections 3.4 and 3.5).

  • Analyses in which minimal response was included in the definition of 'responder' used for the stopping rule and rebate scheme. This was in addition to the original analysis in which only complete and partial responders were defined as 'responders'. However, the manufacturer stated that because its model was based on time to progression and overall survival taken from the entire cohort in the bortezomib arm, the model did not allow separate estimation of these two outcomes for the minimal responder group. The manufacturer also stated that the minimal responder group of first-relapse patients in the APEX trail was too small to allow any meaningful analysis of time to progression and overall survival for this group alone. The model had not been constructed as a responder model, and specifically reflecting the expected health outcomes of minimal response patients would require a differently structured model.

3.10 The manufacturer provided data on the median time to progression separately for non-responders, complete responders, partial responders and minimal responders according to the levels of response at the fourth cycle of treatment. The details of this information were designated by the manufacturer to be commercial in confidence.

3.11 Data supplied by the manufacturer showed that, if no rebate scheme is applied, and EBMT criteria are used to assess response, the ICERs for bortezomib compared with HDD range from £33,500 per QALY gained for a three-cycle stopping rule for complete and partial responders only to £35,600 per QALY gained for a four-cycle stopping rule for complete, partial and minimal responders. If initial M protein is used to assess response and complete, partial and minimal responders are included, the ICERs for bortezomib compared with HDD range from £32,000 per QALY gained for a three-cycle stopping rule to £32,300 per QALY gained for a four‑cycle stopping rule. The manufacturer did not provide ICERs for a scenario in which initial M protein is used to measure response and in which only complete and partial responders are included (as in the original modelling). These ICERs could be established from the manufacturer's revised model, and were £26,500 and £29,000 per QALY gained for a three- and four-cycle stopping rule, respectively. However, the associated changes in cost and QALYs in the model showed that including minimal responders in the model resulted in higher costs but no further gain in QALYs.

3.12 Data supplied by the manufacturer showed that, if the manufacturer rebates the cost of treatment for patients whose disease does not meet the specified response criteria, and EBMT criteria are used to assess response, the ICERs for bortezomib compared with HDD ranged from £25,300 per QALY gained for a four-cycle stopping rule for complete and partial responders only to £28,100 per QALY gained for a three-cycle stopping rule for complete, partial and minimal responders. If initial M protein is used to assess response, the ICER for bortezomib compared with HDD ranged from £28,200 per QALY gained for a three-cycle stopping rule to £27,400 per QALY gained for a four-cycle stopping rule for complete, partial and minimal responders. The manufacturer did not provide ICERs for a scenario in which initial M protein is used to measure response and in which only complete and partial responders are included (as in the original modelling). However, these ICERs could be established from the manufacturer's revised model. This showed ICERs of £20,700 and £20,900 per QALY gained for a four- and three-cycle stopping rule, respectively. The model showed that compared with the scenario resulting in an ICER of £20,700, all other scenarios resulted in higher costs but no further gain in QALYs.

3.13 Full details of all the evidence are in the manufacturer's submission, the ERG report and the evaluation report.