Daratumumab with bortezomib and dexamethasone for previously treated multiple myeloma

3.1 This evaluation reviews the evidence for daratumumab plus bortezomib and dexamethasone for relapsed multiple myeloma, which was approved for use in the Cancer Drugs Fund in NICE technology appraisal guidance 573. It also reviews new data collected as part of the managed access agreement. The company presented evidence for daratumumab plus bortezomib and dexamethasone as a second-line treatment only, in line with the original company submission and recommendation in technology appraisal guidance 573. The committee recognised that limiting this treatment to second line was narrower than its marketing authorisation. But concluded that it would appraise daratumumab plus bortezomib and dexamethasone after only 1 previous treatment, having been presented only with evidence for its use as a second-line treatment.

3.2 Multiple myeloma is a chronic condition that affects how long people live and their quality of life. A clinical expert described it as a relapsing and remitting disease with a complex and evolving pathway. The Cancer Drugs Fund clinical lead explained that each appraisal is a snapshot in time. They explained that since daratumumab plus bortezomib and dexamethasone was available through the Cancer Drugs Fund several new multiple myeloma drugs have been recommended for routine commissioning in the NHS. This has changed what would be offered to people with a new diagnosis, and what subsequent treatment is offered to those who could have daratumumab plus bortezomib and dexamethasone. The committee acknowledged that this makes interpreting clinical trial evidence for this appraisal challenging, because the trial (see section 3.5) started several years ago and may not reflect the current multiple myeloma pathway. The committee noted that the company submission also included new data collected as part of the managed access agreement. It was aware that this data may have limitations because the managed access period mostly took place during the COVID‑19 pandemic. The pandemic may have affected the data collected in 2 ways. First, by affecting the care pathway, because some other treatment options were made available for an interim period. Second, because there may be excess mortality associated with COVID‑19 in people who had daratumumab plus bortezomib and dexamethasone. The committee agreed that it would consider these limitations in its decision making.

3.3 Patient experts stated that in their experience, daratumumab plus bortezomib and dexamethasone had very few side effects. This was echoed in a patient survey done by Myeloma UK, which said that 95% of respondents would recommend the treatment, despite some people having side effects. One patient expert noted that having daratumumab plus bortezomib and dexamethasone had dramatically increased their quality of life. They said it helped with maintaining day to day routines, and meant that they were likely to be well enough for more options later in the treatment pathway. They explained that this was important because the condition becomes more resistant to treatment with each relapse. The committee recognised the need for effective, well-tolerated treatment options for people with multiple myeloma who have had a previous treatment.

3.4 Treatment options for people with newly diagnosed multiple myeloma depend on if a stem cell transplant is a suitable treatment option. Once the disease progresses, treatment options depend on what treatments people have had before, response to these treatments, and patient preference. The committee noted that progression through the pathway is slow, and each remission and progression may span several years. It noted that treatment choice at each line may differ in clinical practice depending on when a person entered the treatment pathway, and what treatments have been available to them before. For someone who has had 1 previous line of treatment, currently available options at second line are:

3.5 Clinical evidence for daratumumab plus bortezomib and dexamethasone compared with bortezomib plus dexamethasone came from the CASTOR trial. CASTOR is an ongoing, randomised, open-label, multicentre, phase 3 trial. The population included adults with relapsed or refractory multiple myeloma. Because the company chose to focus on daratumumab plus bortezomib and dexamethasone as a second-line treatment (see section 3.1), it presented data from the trial for people who had only had 1 previous treatment. In the original submission for technology appraisal guidance 573, the CASTOR trial had a median follow up of 26.9 months. After the period of managed access, this was 50.2 months for progression-free survival and 72.6 months for overall survival. The trial evidence showed that daratumumab plus bortezomib and dexamethasone reduced the risk of disease progression or death by 79% (hazard ratio [HR] 0.21, 95% confidence interval [CI] 0.15 to 0.31) compared with bortezomib plus dexamethasone. Daratumumab plus bortezomib and dexamethasone also reduced the risk of death by 44% (HR 0.56, 95% CI 0.39 to 0.80) compared with bortezomib plus dexamethasone. The committee questioned if previous treatment with bortezomib or daratumumab (both available at first line) would affect the results in the NHS. The clinical expert and Cancer Drugs Fund clinical lead explained that having bortezomib or daratumumab before is not expected to affect the effectiveness of daratumumab plus bortezomib and dexamethasone, if these treatments are used for a finite time rather than until disease progression. They noted that there are often several years between remissions. The committee concluded that the CASTOR trial showed that daratumumab plus bortezomib and dexamethasone is clinically effective compared with bortezomib plus dexamethasone. It further concluded that CASTOR was the most suitable source for establishing the relative effect of these 2 treatments.

3.6 The CASTOR trial was a global trial, which meant not all subsequent treatments used in the trial (at third line and beyond) are available in the NHS. Because alternative treatments may affect survival results, the company presented an inverse probability of censoring weights (IPCW) analysis to adjust the survival estimates. The adjusted HR showed a small but important difference in the clinical trial results. In the CASTOR trial some people had daratumumab as a subsequent treatment, but not at fourth line (where it is recommended in the NHS, see NICE's technology appraisal guidance on daratumumab monotherapy). If a person has daratumumab at an earlier line, they would not have fourth-line daratumumab monotherapy. At the committee meeting the company explained that the adjusted analyses from the trial estimated that only a proportion of people who had second-line bortezomib plus dexamethasone would have had had fourth-line daratumumab monotherapy (the exact figure is academic in confidence). The clinical expert noted that in the NHS most people who have not had a CD38 targeted therapy previously (for example, daratumamab or isatuximab) would have fourth-line daratumumab monotherapy. The committee agreed it is appropriate to adjust the analyses for subsequent treatments not available in the NHS, but that the company's adjusted analyses did not reflect current practice. It agreed that most people progressing to fourth-line treatment who had not had a CD38 targeted therapy before would have fourth-line daratumumab monotherapy. This is because no other CD38 targeted therapies were routinely commissioned at the time of the first committee meeting. The committee was aware that daratumumab monotherapy was expected to have a survival benefit at fourth line. The committee noted that the benefit appeared to have been underestimated in the adjusted analysis for people who had bortezomib plus dexamethasone. So, the hazard ratio estimated by the IPCW analysis is likely to be biased in favour of daratumumab plus bortezomib and dexamethasone. The committee concluded that the adjusted and unadjusted HRs were associated with uncertainty and reflected the higher and lower bounds of clinical effectiveness. The true effect of daratumumab plus bortezomib and dexamethasone compared with bortezomib plus dexamethasone was likely to lie between the adjusted and unadjusted HRs.

3.7 There was no trial directly comparing daratumumab plus bortezomib and dexamethasone with carfilzomib plus dexamethasone. So, the company did a network meta-analysis using data from the second-line subgroup of CASTOR and ENDEAVOR (which compared carfilzomib plus dexamethasone with bortezomib plus dexamethasone). Evidence from the network meta-analysis showed that daratumumab plus bortezomib and dexamethasone improves overall survival and progression-free survival compared with carfilzomib plus dexamethasone. The committee concluded that the network meta-analysis was appropriate for decision making.

3.8 The CASTOR trial took place across 16 countries, with no study centres in England. Interpreting the available evidence is difficult because the clinical trial happened several years ago and the current multiple myeloma pathway in England is significantly different to when the trial started (see section 3.6). The systemic anti-cancer therapy (SACT) dataset provides real world evidence for people having daratumumab plus bortezomib and dexamethasone in the NHS in England, starting in March 2019. During the managed access period data was collected on overall survival and median treatment duration. A naive comparison shows that overall survival reported in the SACT dataset is lower than that reported in the CASTOR trial. The population in the SACT dataset was on average older than in CASTOR, but limited data was available on comorbidities and any increased risk of mortality. To address the differences in patient population between CASTOR and the SACT data, the company did a matching-adjusted indirect comparison. This adjusted for differences in various baseline characteristics, including age. But the results showed that the prognostic factors explored did not explain the differences between the datasets. The committee agreed that the SACT dataset is more likely to reflect the true experience of people having daratumumab plus bortezomib and dexamethasone in England, but that it has several limitations. It mostly took place during the COVID‑19 pandemic, had a shorter follow-up time than the CASTOR trial and data was missing for key prognostic variables (such as ECOG performance status and international staging system). The committee noted that SACT data might include excess mortality because of COVID‑19. The committee would have liked to have seen the survival outcomes for people who entered the SACT dataset before March 2020 to see if this projected a different survival curve than the complete managed access period. The patient experts explained that many people with multiple myeloma shielded through the pandemic, which reduced the risk of being infected. But they noted that people who had daratumumab plus bortezomib and dexamethasone were still attending hospital appointments at least once a month. One patient expert explained that if people with multiple myeloma were infected with SARS‑CoV‑2 (the virus that causes COVID‑19) they were likely to have poorer outcomes than the general population in England. The clinical expert explained that the pandemic is also likely to have changed treatment decisions, because oral ixazomib with lenalidomide and dexamethasone was made available at second line. This aimed to reduce frequent visits to hospital for treatment injections. The Cancer Drugs Fund clinical lead said that the use of ixazomib with lenalidomide and dexamethasone peaked at 15% of second-line treatments. Because of the availability of alternative treatment options at second-line, some people had daratumumab plus bortezomib and dexamethasone at third line. But it was clarified that this proportion was very small and would be unlikely to affect the results. The committee was aware of the limitations of the SACT data but noted that it included a larger sample size than the CASTOR trial. The committee concluded that the SACT data appeared to be a better source to estimate absolute (baseline) event rates for overall survival. This is because it better represented the population in NHS clinical practice than the CASTOR trial. But it further concluded that the impact of COVID‑19 on survival outcomes from the SACT data was uncertain. The committee considered this during decision making.

3.9 Although the SACT data provided evidence for people having daratumumab plus bortezomib and dexamethasone in the NHS, there was no data for comparator treatments. The company used data from 3 real-world cohorts of people who did not have daratumumab plus bortezomib and dexamethasone, to compare overall survival with SACT data in a naive comparison. The committee noted that the comparisons were likely to be at high risk of bias because the populations included in the studies differed. The company also presented a scenario analysis where it simulated a bortezomib plus dexamethasone survival curve using the absolute (baseline) event rates from the SACT data and applying the relative treatment effect observed in CASTOR. The real-world cohorts were then used to validate this comparison. The EAG noted that although the naive comparison has limitations, the cohorts show a similar survival trajectory to the simulated bortezomib plus dexamethasone curve. The committee concluded the scenarios were associated with uncertainty but suggest that the relative effects seen in CASTOR would hold in clinical practice.

3.10 The company chose a partitioned survival model to estimate the cost effectiveness of daratumumab plus bortezomib and dexamethasone. The model included 3 health states: progression-free, progressed disease and death. The probability of being in a given health state was calculated using the overall survival and progression-free survival curves. The model time horizon was 30 years. The model had the same structure as the original appraisal but included data from CASTOR and the indirect comparison in the network meta-analysis. Also, baseline characteristics for age and gender were updated to reflect those seen in the SACT dataset. The company explained that this increased the age-related mortality in the model because the starting age was higher. The committee concluded that the model structure is acceptable.

3.11 The company's base case used updated data from CASTOR to simulate time to stopping treatment, progression-free survival, and overall survival. The company fitted parametric curves to the trial data for daratumumab plus bortezomib and dexamethasone and bortezomib plus dexamethasone to extrapolate the observed data beyond the period of follow up. After technical engagement, the company changed the parametric function used to extrapolate the data for the bortezomib plus dexamethasone arm. It selected the Weibull curve based on it having good visual and statistical fit, and clinical plausibility, estimating a small proportion of people would be alive at 10 years. To simulate the survival path of people who have carfilzomib plus dexamethasone, the company applied the HRs from the network meta-analysis to the daratumumab plus bortezomib and dexamethasone arm. The committee recalled it was appropriate to use the SACT data for estimating survival for people who had daratumumab plus bortezomib and dexamethasone. NICE's technical support document 13 recommends using registry data to estimate absolute baseline event rates, and that randomised evidence should be used to quantify relative differences. The committee also agreed that the company's approach for the comparison of daratumumab plus bortezomib and dexamethasone with bortezomib plus dexamethasone allowed the 2 curves to diverge over time. This may overestimate the benefit of daratumumab plus bortezomib and dexamethasone.

3.12 The company presented 2 exploratory scenario analyses using the SACT data. In both cases, the model estimated overall survival with daratumumab plus bortezomib and dexamethasone using a Weibull curve fitted to the SACT data to extrapolate to the time horizon of the model. The company then simulated a bortezomib plus dexamethasone arm by applying the either the adjusted or unadjusted HRs for overall survival from CASTOR (that is, the relative effect between treatments estimated from the randomised data) to the extrapolated daratumumab plus bortezomib and dexamethasone data from SACT. The company made several assumptions to estimate progression-free survival in this analysis. These included using time to treatment stopping data from the SACT dataset and applying an HR from CASTOR to account for people stopping treatment for reasons other than disease progression. The committee noted this was a reasonable approximation because progression-free survival data was not captured in the SACT dataset. The company then applied the progression-free survival relative treatment effect HR from CASTOR (either adjusted or unadjusted) to estimate progression-free survival with bortezomib plus dexamethasone. Analyses that used the additional follow-up data available in CASTOR could have been useful to reduce some of the uncertainty. The committee noted that this could be done in several ways. For example, by applying the appropriate shape and scale parameters (using a Weibull distribution) to the CASTOR data and adjusting them so that they match the survival in the SACT data. Or, the CASTOR data could be used to extrapolate beyond the SACT follow-up period but with appropriate survival constraints. The committee concluded that although these additional scenarios would be informative, they would need additional assumptions. So the simulations based on SACT data extrapolations using the relative treatment effect from the trial were preferred to model survival for daratumumab plus bortezomib and dexamethasone, and bortezomib plus dexamethasone. The committee noted that using the fixed HR from the trial and applying it to the SACT data reduced the risk of overestimating the benefit of daratumumab plus bortezomib and dexamethasone. It agreed that both the adjusted and unadjusted hazard ratio should be considered for decision making because the IPCW analysis did not accurately reflect subsequent treatment use in the NHS (see sections 3.4 and 3.5, and section 3.13). The committee agreed that the scenarios were associated with uncertainty because of the unknown impact of COVID‑19 on the survival estimates, but that it would consider this in its decision making.

3.13 The company modelled the costs of subsequent treatments using a basket approach, applying one-off costs based on the CASTOR trial. The committee recalled that the trial happened several years ago and subsequent treatments did not reflect current NHS practice. The model only included 1 subsequent line of treatment applied for the proportion of people still alive at the point of disease progression. But in practice people who have multiple myeloma are likely to have several subsequent treatments. The committee agreed this simplification prevented full exploration of subsequent treatments and their effects on the cost-effectiveness modelling. It noted that lenalidomide plus dexamethasone as a subsequent line of treatment is likely to have been overestimated and was inconsistent with the assumption that most people have lenalidomide at first line (see section 3.3). The committee agreed that it would have preferred to see estimates where the proportion was close to zero for those who have daratumumab plus bortezomib and dexamethasone at second line. This is because this would reflect the current treatment pathway. The committee also noted that the costs and clinical estimates for subsequent daratumumab treatment were not aligned in the modelling. It noted that a higher proportion of people who had bortezomib plus dexamethasone had costs for fourth-line daratumumab monotherapy than the company stated had been included in the clinical estimates using the adjusted effectiveness data. This would bias the company's cost-effectiveness estimates in favour of daratumumab plus bortezomib and dexamethasone. The committee agreed that the basket costs applied for subsequent treatments are not likely to reflect current practice. It noted that this was likely to reflect the evolving treatment pathway and the simplistic application in the model, but the impact of subsequent treatments on the cost-effectiveness results is uncertain.

3.14 The company used utility values in the model based on EQ-5D data collected in the ENDEAVOR trial, which was preferred during the original evaluation of daratumumab plus bortezomib and dexamethasone. Disutilities were applied based on the rate of grade 3 and 4 adverse events in CASTOR trial. The clinical and patient experts explained that grade 1 or 2 adverse events would not lead to people stopping treatment but would likely affect quality of life. One patient expert explained the positive experience they had had while having daratumumab, with limited side effects. The committee noted that grade 1 and 2 events were not included in the modelling but noted that these were more frequent in the bortezomib plus dexamethasone arm of the trial. It concluded that there is likely a small underestimate on how daratumumab plus bortezomib and dexamethasone affects quality of life.

3.15 NICE's manual on health technology evaluation notes that above a most plausible incremental cost-effectiveness ratio (ICER) of £20,000 per quality-adjusted life year (QALY) gained, decisions about the acceptability of a technology as an effective use of NHS resources will take into account the degree of certainty around the ICER. The committee will be more cautious about recommending a technology if it is less certain about the evidence presented but will also take into account other aspects including uncaptured health benefits. The committee agreed that CASTOR trial showed that daratumumab plus bortezomib and dexamethasone is a clinically effective treatment. It noted that the benefit was likely to remain long term and the relative effect is likely to hold when used outside the clinical trial setting. It also heard from patient and clinical experts about the relative ease and lack of side effects associated with taking this treatment, and the importance of having the most effective treatments possible available at second line. The committee recognised that there was uncaptured health benefit from not including low-level side effects of treatment, and the ease of administration (including using subcutaneous administration instead of intravenous). It agreed that it would accept an ICER at the upper end of the acceptable range if this was based on more conservative modelling assumptions. This is because it would allow the committee to have more confidence that residual uncertainties would not result in the cost effectiveness estimates being above what NICE considers an acceptable use of NHS resources.

3.16 Because of confidential commercial arrangements for daratumumab, bortezomib and post-progression treatments, the cost-effectiveness results are confidential and cannot be reported here. After technical engagement the company and EAG had the same base case, which used updated data from CASTOR to simulate time to stopping treatment, progression-free survival, and overall survival (see section 3.11). The committee agreed that its preferred scenarios for the comparison with bortezomib plus dexamethasone used the extrapolated daratumumab plus bortezomib and dexamethasone data from SACT, and applied the relative treatment effect from CASTOR to estimate outcomes in the bortezomib plus dexamethasone arm (see section 3.11). The preferred scenarios for the comparison with carfilzomib plus dexamethasone applied the relative treatment effect from the network meta-analysis to the same analysis. The committee agreed that uncertainty remained in 3 areas of the cost-effectiveness estimates: