3 The manufacturer's submission
3.1 The manufacturer produced an analysis of the clinical and cost effectiveness of lenalidomide for the treatment of multiple myeloma in people who had received at least one prior therapy. This included people at first and subsequent relapse and people who had progressive disease after two or more cycles of anti-myeloma treatment. The trial population was divided into five subgroups for the economic analysis. For people who had received only one prior therapy the main comparator was bortezomib monotherapy, which is currently recommended as a treatment option in 'Bortezomib monotherapy for relapsed multiple myeloma' (NICE technology appraisal guidance 129). For people in whom bortezomib was contraindicated, people who had received two or more prior therapies and people who had received prior thalidomide (only one prior therapy or two or more prior therapies), the comparator was dexamethasone.
3.2 Two randomised controlled trials (RCTs), of identical design but differing in their locations (MM-009 and MM-010), compared treatment with lenalidomide plus dexamethasone (len/dex) with dexamethasone alone for patients with multiple myeloma who had received at least one prior therapy. In both arms, the regimen of dexamethasone was pulsed high-dose dexamethasone in 28-day cycles. The trials enrolled 353 and 351 patients, respectively (n = 704). Patients were stratified according to their serum concentration of β2-microglobulin, previous stem-cell transplantation and number of previous anti-myeloma therapies. Treatment was continued until the disease progressed or unacceptable adverse effects occurred. The primary outcome was time to progression (TTP). Secondary outcomes were overall survival, response rates, adverse effects and time to decrease in performance status. Response was assessed using the European Group for Blood and Marrow Transplantation criteria, and six response categories were defined: complete response, near-complete response, partial response, stable disease, disease progression and 'response not evaluable'. A number of post-hoc subgroups from the pooled populations were investigated, including patients with pre-existing peripheral neuropathy and patients who had received prior thalidomide or bortezomib therapy. At disease progression or unblinding, patients in the dexamethasone monotherapy group were allowed to receive lenalidomide.
3.3 The median TTP at unblinding from the pooled trials was 48.3 weeks (95% confidence interval [CI] 41.1 to 60.1 weeks) for the len/dex arms and 20.1 weeks (95% CI 19.9 to 20.7 weeks) for the dexamethasone arms. The pooled hazard ratio for TTP was 0.35 (95% CI 0.29 to 0.43; log-rank p < 0.001). The median overall survival in one trial, analysed 3 years and 3 months after study initiation, was 29.6 months in the len/dex arm and 20.2 months in the dexamethasone arm (hazard ratio 0.44; 95% CI 0.30 to 0.65; p < 0.001). In the second trial, the median overall survival was analysed 2 years and 8 months after study initiation; it could not be estimated in the len/dex arm (because of the number of patients still alive), and was 20.6 months in the dexamethasone arm (hazard ratio 0.66; 95% CI 0.45 to 0.96; p = 0.03). For the pooled trials, the subgroup of patients who had received one prior therapy had a median survival of 169.1 weeks in the len/dex arm compared with 145.4 weeks in the dexamethasone arm. For the subgroup of patients who had received two or more prior therapies, the median survival was 144.0 weeks in the len/dex arm compared with 118.0 weeks in the dexamethasone arm. A complete, near-complete or partial response was obtained in 60.6% of patients in the len/dex arms and 21.9% of patients in the dexamethasone arms. The remaining 39.4% of patients in the len/dex arms and 78.1% of patients in the dexamethasone arms had stable or progressive disease, or were not evaluable. The odds ratio for this dichotomised response (complete, near-complete or partial response versus stable disease, progressive disease or response not evaluable) was 5.48 (95% CI 3.94 to 7.63; p < 0.001). Over the course of the first 23 cycles in the trial, about 70% of the treatment days for all patients in the trial were at the full dose of lenalidomide. The dose of lenalidomide was reduced on about 25% of the treatment days, and treatment was interrupted on about 5% of the days.
3.4 The results for overall survival were affected by crossover of patients at unblinding: 170 of 351 patients in the dexamethasone arm opted to receive lenalidomide at disease progression or unblinding. However, these patients were analysed as remaining in the dexamethasone arm. The TTP was also affected by the crossover, but to a lesser degree because most patients (over 75%) had shown disease progression at unblinding.
3.5 In both trials, the differences in TTP and response rates (in favour of len/dex) were observed in all of the prespecified subgroups. The post-hoc subgroups in the trial showed that the efficacy of len/dex relative to dexamethasone alone remained statistically significant in subgroups that had received prior treatment with thalidomide or bortezomib and in subgroups specified by the number of previous therapies for multiple myeloma.
3.6 A meta-analysis was also performed to combine the results of the trials and to confirm the results obtained by the pooling of trials. This resulted in a median difference in TTP of 28.24 weeks (95% CI 18.39 to 38.08 weeks) and an odds ratio for overall survival of 1.44 (95% CI 1.34 to 1.56). The hazard ratio was not calculated. There was no evidence of heterogeneity between the trials.
3.7 An indirect comparison was undertaken to compare len/dex with bortezomib monotherapy because there were no head-to-head trials. The results of the trials for len/dex were compared with the results of the Assessment of Proteasome Inhibition for Extending Remissions (APEX) RCT. The APEX study compared bortezomib with high-dose dexamethasone. For median TTP, len/dex had a 34-week advantage over bortezomib for people who had received one prior therapy only, and there were no statistically significant differences for the secondary outcomes of complete response, partial response and progressive disease. However, this analysis is limited by the small number of data points. In addition, the common comparator (high-dose dexamethasone) was an active treatment and was not used in the same dose across the trials, and the definition of response differed between the trials.
3.8 The economic evaluation in the manufacturer's submission used a discrete-event simulation model. This model used two separate prediction equations to calculate TTP and post-progression survival values, which were then added together to give overall survival. A cohort was created by randomly sampling (with replacement) patients from the pooled trial populations. For subgroups within the model, the cohort was created from the relevant population. The model attempted to capture the variability between individuals in the trial and to allow correlation between observed parameters to be retained.
3.9 The model divided patients from both arms of both trials into four groups according to their level of response. In building a cohort for the study population or any of the subgroups, the model ensured that the proportion of patients achieving a particular response in the trial was replicated in the cohort. To calculate TTP, the model assumed a Weibull distribution. For bortezomib, the response rates were taken from the APEX trial and the equation for TTP was calibrated such that the median TTP was the same as that within the trial.
3.10 The equation for post-progression survival was assumed to take an exponential form. However, the trial results were affected by the crossover of patients at unblinding from the dexamethasone arm to receive lenalidomide. Therefore the equation included an adjustment factor that calibrated the modelled median overall survival in the dexamethasone group after progression to be equal to that observed in the UK Medical Research Council (MRC) multiple myeloma trials. This assumed that survival of people with multiple myeloma in this cohort was the same when treated with dexamethasone as with all other regimens used in the MRC trials. The patient profiles from the RCTs in the manufacturer's submission were applied to the predictors in the survival equations derived from the MRC trial data to predict survival in the dexamethasone arm if crossover had not occurred.
3.11 The model considered subgroups of patients who had received one prior therapy (with this group divided further into those who did and did not have peripheral neuropathy), patients who had received two or more prior therapies, and patients who had received thalidomide (divided further into those who had received only one prior therapy and those who had received two or more prior therapies). For patients who had received only one prior therapy, len/dex was compared with bortezomib monotherapy. For patients with peripheral neuropathy and for those who had received two or more prior therapies, the comparator was dexamethasone alone.
3.12 The utility values were based on a study that evaluated intensive chemotherapy followed by myeloablation and autologous stem-cell transplantation in people with multiple myeloma. For the complete response, partial response and stable disease states, a utility value of 0.81 was used. This value was based on the utility of the general public at an age (median 54 years) corresponding to that of the patients in the study. A utility value of 0.64 was applied to the progressive disease state. After 2 years, a utility value of 0.77 was applied to those patients whose disease had not progressed.
3.13 Only grade 3 and 4 adverse effects were included in the model. Utility decrements for adverse effects were not included. Resource-use data associated with, for example, adverse effects, routine follow-up and laboratory tests were collected to build up a profile of resource use, depending on disease state and treatment. Resource-use profiles were developed for people during relapse and/or on treatment, and for people in remission on maintenance therapy or off therapy. Resource use was estimated by interviewing 15 specialists across England and Wales who specialised in the management of multiple myeloma.
3.14 The economic analysis did not produce cost-effectiveness estimates for the whole trial population. In the base case, for the subgroup with one prior therapy, the model resulted in an incremental cost-effectiveness ratio (ICER) per quality-adjusted life year (QALY) gained for lenalidomide that the manufacturer stated was not cost effective compared with bortezomib. For the one prior therapy subgroup, the ICER was £46,865 per QALY gained for lenalidomide compared with dexamethasone. For patients who had received two or more prior therapies the ICER was £24,584 per QALY gained. In the subgroup of patients who had received prior thalidomide, the ICERs were £38,861 per QALY gained for patients with only one prior therapy and £22,589 per QALY gained for patients who had received two or more prior therapies.
3.15 The ERG explored the precision with which the fitted curve for len/dex matched the actual trial data used in the model. It observed that the fitted curve overestimated overall survival. The ERG noted that the overall survival curve for the dexamethasone arm in the cost-effectiveness model had been adjusted to predict the median overall survival predicted from the MRC trials. However, the ERG stated that it was more methodologically correct to adjust the dexamethasone overall survival calculation in the model to predict the mean overall survival predicted from the MRC trials, because the ICERs calculated from the model were a ratio of means and not medians.
3.16 The ERG conducted an exploratory analysis with an improved fit of the len/dex overall survival curve and also with the dexamethasone curve adjusted to the mean overall survival in the MRC trials. For the subgroup of patients who had received only one prior therapy where len/dex was compared with bortezomib, the ICER increased more than 30-fold from the manufacturer's base case. For the subgroup who had received one prior therapy where the comparator was dexamethasone, the ICER increased from £46,865 to £69,500 per QALY gained. For the subgroup of patients who had received two or more prior therapies, the ICER increased from £24,584 to £47,100 per QALY gained. For patients who had received prior thalidomide, the ICER increased from £38,861 to £56,500 per QALY gained if they had received only one prior therapy and from £22,589 to £43,600 per QALY gained if they had received two or more prior therapies.
3.17 In addition, the ERG noted that the costs associated with routine medical management (non-drug costs) assumed in the model were lower than the figures that were accepted in the appraisal of bortezomib and may therefore have been underestimated. It also noted that the model had no disutility attached to the occurrence of adverse effects. Finally, it noted that the cost of anti-thrombotic prophylaxis that was routinely used with lenalidomide was not included in the model. The ERG stated that the inclusion of the above considerations in the model would further increase the ICERs for all subgroups above the values obtained in the exploratory reanalyses quoted above.
3.18 For the subgroup of patients who had received only one prior therapy, the ERG repeated the indirect comparison of len/dex with bortezomib using methods that it considered to be more appropriate. This resulted in a hazard ratio of 0.557 (95% CI 0.337 to 0.912). The ERG pointed out that this comparison was with bortezomib as monotherapy and that bortezomib was commonly used in combination with dexamethasone in routine clinical practice. The economic analysis for the comparison of len/dex with bortezomib also assumed a maximum of eight cycles of bortezomib instead of the 11 allowed in the trial, and did not model the response-based rebate scheme recommended in NICE technology appraisal guidance 129. The ERG also suggested that the administration costs associated with bortezomib may have been overestimated in the manufacturer's model. In addition, the dose intensity for bortezomib was assumed to be 100%; that is, the analysis did not allow for dose reductions and treatment interruptions, which had been included for lenalidomide. All of the above issues would have had the effect of increasing the ICERs for the comparison of len/dex with bortezomib in the subgroup of patients who had received only one prior therapy.
3.19 After the first Appraisal Committee meeting, the manufacturer presented an updated cost-effectiveness analysis. The manufacturer stated that this updated analysis would only consider all patients who had received two or more prior therapies and the subgroup of patients who had received thalidomide as one of these prior therapies. The manufacturer accepted the ERG's approach to modelling the len/dex overall survival curve. However, the manufacturer did not agree with the calibration of the overall survival curve for dexamethasone alone in the cost-effectiveness model to the mean overall survival predicted from the MRC trials. The manufacturer stated that a curve calibrated to the mean rather than the median overall survival was less representative of the published curves, and that using the mean placed more emphasis on the tail of the distribution, where there were fewer patients and greater uncertainty. The manufacturer maintained that it was appropriate to calibrate the curve to the predicted median survival in the updated analysis. In addition, the updated analysis incorporated costs for outpatient visits into routine management costs, and inflated the routine management costs to 2008 values. The effects of adding costs for the prophylaxis of deep vein thrombosis and disutility for long-term adverse effects were also explored in the updated model through sensitivity analyses. The manufacturer also proposed a patient access scheme in which the cost of lenalidomide to the NHS for a person with multiple myeloma will be capped at 26 cycles of treatment (each of 28 days, so normally administered over 2 years). A cycle will still be considered as having been completed within the scheme even if there are dose reductions and treatment interruptions during the cycle. The drug cost of lenalidomide (excluding any related costs) for people who remain on treatment for more than 26 cycles will be met by the manufacturer. The Department of Health in England and the Department of Health and Social Services in Wales accepted the consideration of this scheme by NICE.
3.20 Taking all these factors into account in the model, for patients who had received two or more prior therapies the incremental life-year gain was 2.77, the incremental QALY gain was 1.86 and the ICER was £30,350 per QALY gained. For patients who had received thalidomide as one of the prior therapies the incremental life-year gain was 2.51, the incremental QALY gain was 1.7 and the ICER was £28,941 per QALY gained. These ICERs were relatively insensitive to the addition of costs for prophylaxis of deep vein thrombosis or disutility for long-term adverse effects.
3.21 The ERG considered the updated analyses from the manufacturer and agreed with the implementation of the stated changes. It noted that the model predicted that, for the group of patients who had received two or more prior therapies, the patient access scheme applied to 17% of the people in the model. The estimated average cost of treatment with lenalidomide to the NHS per person over a modelled lifetime (median overall survival approximately 2.7 years) decreased from £59,800 to £51,800 with the patient access scheme. For the subgroup of patients who had received thalidomide as one of their prior therapies the patient access scheme applied to 11% of people in the model, and the modelled lifetime cost of treatment with lenalidomide to the NHS per person decreased from £49,800 to £46,300 with the patient access scheme.
3.22 The ERG repeated the exploratory analysis on the updated model using their preferred approach with the dexamethasone overall survival curve calibrated to the mean survival predicted from the MRC trials and the patient access scheme implemented in the model. For the subgroup of patients who had received two or more prior therapies, the incremental life-year gain was 1.81 and the incremental QALY gain was 1.24 at an incremental cost of £54,291, giving an ICER of £43,800 per QALY gained. For patients who had received thalidomide as one of the prior therapies, the incremental life-year gain was 1.71 and the incremental QALY gain was 1.15 at an incremental cost of £47,531, giving an ICER of £41,300 per QALY gained.
3.23 Full details of all the evidence are in the manufacturer's submission and the ERG report.