Dimethyl fumarate for treating relapsing‑remitting multiple sclerosis

Clinical effectiveness

4.6 The Committee discussed the clinical‑effectiveness evidence from the DEFINE and CONFIRM trials. It heard from the clinical specialists that the trial populations broadly represent patients who would be offered beta interferon or glatiramer acetate in the UK, in line with the Association of British Neurologists' guidelines. The Committee noted that the trial populations had more severe relapsing‑remitting multiple sclerosis than the population covered by the marketing authorisation. The Committee noted that the 2 trials included different primary endpoints for measuring relapse, that is, the proportion of patients with relapse at 2 years in the DEFINE trial and the annualised relapse rate in the CONFIRM trial, and heard from the manufacturer that this was because the European Medicines Agency and the Food and Drug Administration preferred different approaches to measuring relapse. The Committee heard from the clinical specialists that the 2 endpoints have the same influence on clinical decisions. In addition, the Committee heard from the clinical specialists and patient experts that it is difficult to define a relapse because each relapse varies in nature and severity, and that it is the disability that follows, rather than the relapse itself, that has the greater impact on the patient's health‑related quality‑of‑life. The Committee acknowledged that confirming a relapse may include a degree of subjectivity. The Committee noted that the results presented from the manufacturer's trials and meta‑analysis showed that dimethyl fumarate statistically significantly reduced both the rate of relapses and the proportion of patients experiencing a relapse compared with placebo. The Committee discussed the manufacturer's approaches to analysing the efficacy outcomes for its trials. The Committee was concerned about a few aspects of the analysis: in its original submission, the manufacturer had adjusted the analysis for a number of factors, including region. However, the manufacturer clarified that the statistical analyses of data by region had been documented in the statistical analysis plan prior to data base lock and as such had constituted a pre‑specified analysis. The Committee also noted that patients in the DEFINE and CONFIRM trials taking dimethyl fumarate experienced more flushing than patients taking placebo and this may have led to functional unblinding of the treatment arms. However, the ERG confirmed that protocols were put in place to avoid functional unblinding. The Committee concluded that, overall, the evidence suggested that dimethyl fumarate reduces relapses in people with relapsing‑remitting multiple sclerosis compared with placebo.

4.7 The Committee was aware of another factor potentially affecting the magnitude of the treatment effect of dimethyl fumarate compared with placebo, in that patients in the DEFINE and CONFIRM trials were eligible to switch to alternative active therapies for multiple sclerosis if they had 1 or more relapse or confirmed progression of disability for 3 months. The Committee acknowledged that a higher proportion of patients randomised to placebo (13%) switched to active treatment than patients randomised to dimethyl fumarate (6%). The Committee heard from the manufacturer that in its base‑case efficacy analysis, it included only outcomes measured before patients switched treatment, but conducted a sensitivity analysis that included outcomes after patients switched treatment. However, the estimated treatment effect for dimethyl fumarate compared with placebo did not differ between the base‑case analysis and this sensitivity analysis. The Committee concluded that it was satisfied that switching to alternative treatments in the DEFINE and CONFIRM trials did not affect the estimated treatment effect of dimethyl fumarate compared with placebo as measured by the primary efficacy end points.

4.8 The Committee discussed the results of the manufacturer's mixed treatment comparison for disability progression. It understood that in response to clarification requests from the Evidence Review Group (ERG), the manufacturer revised its estimates for the sustained disability progression outcomes, presenting the effect measure as hazard ratios rather than risk ratios as in its original submission. The Committee heard from the ERG that it preferred hazard ratios because they represent the instantaneous risk over the study period whereas risk ratios measure the cumulative risk over the entire study. The Committee concluded that it was more appropriate to measure outcomes measuring sustained disability progression using hazard ratios.

4.9 The Committee discussed the trials' outcome measure of sustained disability progression. The Committee noted that the manufacturer's mixed treatment comparison suggested that compared with placebo, dimethyl fumarate statistically significantly reduced confirmed disability progression sustained for 3 months in the 2 years of the trials, but the reduction for disability progression sustained for 6 months at 2 years was not statistically significant. The Committee heard from the clinical specialists that patients may not have permanent disability progression after a relapse and that recovery may take up to 12 months, but on average people will recover within 3 or 4 months. The clinical specialists stated that sustained disability progression lasting for 6 months is a more appropriate outcome measure than disability progression lasting for 3 months, and it was also preferred by the European Medicines Agency in its draft guideline for the clinical investigation of medicinal products for the treatment of multiple sclerosis. The Committee heard from the ERG that most trials of relapsing‑remitting multiple sclerosis measure sustained disability progression lasting for 3 months, and the Committee agreed that it would consider this in its decision‑making. However, the Committee concluded that sustained disability progression confirmed for 6 months provides a more robust indication of the treatment effect given that patients may recover from relapse.

4.10 The Committee was aware that the diagnostic criteria, clinical management and prognosis of multiple sclerosis have changed since the year 2000. The Committee noted that the manufacturer included trials that were published before the year 2000 in its mixed treatment comparison, and that the ERG observed differences among the baseline relapse rates of the trials of relapsing‑remitting multiple sclerosis. The Committee heard from the ERG that these differences were likely to be clinically meaningful, in that there is potential for more heterogeneity when using an unadjusted model (rather than a model adjusted for baseline relapse as a covariate). The Committee understood that, in response to consultation, the manufacturer had presented additional analyses of relapse rate and disability progression from its trials and the mixed treatment comparison adjusted for baseline relapse rate only. The Committee noted that adjusting the trial outcomes only for baseline relapse rate (rather than for baseline age, EDSS, relapse rate and geographical region) and the mixed treatment comparison only for baseline relapse rate (rather than unadjusted) did not change the results. The Committee questioned why the statistics reflecting model fit performed better for the unadjusted model than for the adjusted model. The Committee heard from the manufacturer that it had not tested either the adjusted or unadjusted mixed treatment comparison for heterogeneity. The Committee considered that estimating heterogeneity for each approach would better indicate the most appropriate approach. The Committee concluded that there remains some uncertainty about whether the manufacturer had appropriately modelled the adjustment for baseline relapse rate in its mixed treatment comparison, but that in this case it preferred the results of the unadjusted mixed treatment comparison because it provided the better statistical fit.

4.11 The Committee discussed the results of the mixed treatment comparisons and agreed that they showed that dimethyl fumarate is more effective than beta interferons and glatiramer acetate in reducing relapses. However, a treatment effect on disability progression was less clear in that the hazard ratios for disability progression indicated an effect of dimethyl fumarate compared with beta interferons and glatiramer acetate but the difference was not statistically significant. The Committee concluded that, compared with beta interferons and glatiramer acetate, dimethyl fumarate is more effective in reducing relapse rates and as effective for disability progression.

4.12 The Committee noted that, in response to consultation, the manufacturer presented evidence of clinical effectiveness for dimethyl fumarate compared with placebo for the highly active relapsing‑remitting multiple sclerosis subgroup from its DEFINE and CONFIRM trials. The Committee commented that the number of patients with highly active relapsing‑remitting multiple sclerosis in the trials was low, and that the data suggested that dimethyl fumarate was beneficial in terms of reducing relapses (with a statistically significant rate ratio according to the confidence interval). The treatment effect on disability progression was less clear because of the small sample size (with a hazard ratio suggesting that dimethyl fumarate increased the hazard of disability progression, but with no statistical significance according to the confidence interval). The Committee also noted that no trials exist that directly compare dimethyl fumarate with either fingolimod or natalizumab, and that the manufacturer had not submitted a mixed treatment comparison for patients with highly active relapsing‑remitting multiple sclerosis or with rapidly evolving severe relapsing‑remitting multiple sclerosis respectively. Therefore, the Committee agreed that it could not draw any conclusions about the clinical effectiveness of dimethyl fumarate compared with natalizumab or with fingolimod in the respective subgroups. The Committee concluded that it had insufficient evidence from the manufacturer to recommend dimethyl fumarate in these subgroups.

4.13 The Committee considered the safety data from the DEFINE and CONFIRM trials, which showed that patients taking dimethyl fumarate experienced more gastrointestinal events and flushing and skin reactions, particularly in the first months of treatment, than patients not taking dimethyl fumarate. The Committee heard from the manufacturer that most of these episodes were mild to moderate in severity and that approximately 4% of patients taking dimethyl fumarate discontinued the study drug because of flushing. It was also aware that episodes of progressive multifocal leukoencephalopathy reported in patients taking Fumaderm or a compound formulation of dimethyl fumarate and copper monomethyl fumarate are unlikely to be relevant here, and that no episodes of progressive multifocal leukoencephalopathy had been reported in patients taking dimethyl fumarate. The Committee concluded that, although dimethyl fumarate can lead to several different adverse reactions, it is generally well tolerated.

Cost effectiveness

4.14 The Committee commented that the manufacturer had submitted a model structurally similarly to models used in previous NICE technology appraisals. The Committee concluded that it could consider only the ICERs for dimethyl fumarate compared with beta interferons and glatiramer acetate because of the lack of data for the subgroups for whom natalizumab and fingolimod have been recommended (see section 4.12).

4.15 The Committee discussed how the manufacturer modelled the natural history of multiple sclerosis. It commented that it was appropriate to allow modelled patients with relapsing‑remitting multiple sclerosis to move to lower as well as to higher EDSS states, that is, to allow for the condition to improve and to get worse, which is in line with what is seen in clinical practice for patients in the lower EDSS states. The Committee noted the inherent limitations associated with using the London Ontario dataset to model the natural history of disease, namely, that it allowed only for movement to higher EDSS states, and that it reflected a patient population from the 1970s and 1980s. However, the Committee understood that the manufacturer had used the London Ontario data set to model the natural history of disease only for EDSS scores of 7 or more in patients with relapsing‑remitting multiple sclerosis, for rates of progression to secondary progressive multiple sclerosis, and in patients with secondary progressive multiple sclerosis. The Committee also heard from the clinical specialists that once patients are in a higher EDSS state they are less likely to relapse, and therefore the possibility of moving to lower EDSS states is less plausible. The Committee recognised that the manufacturer had used the DEFINE and CONFIRM trial data to model the natural history of relapsing‑remitting multiple sclerosis at lower EDSS states, that the model allowed patients to move to lower EDSS states, and that the trial population more closely reflected the population in UK clinical practice than did the population in the older London Ontario data set, especially considering that the prognosis for people with multiple sclerosis has improved in the last 20 years. The Committee concluded that by using its trial data, the manufacturer had appropriately modelled the natural history of disease.

4.16 The Committee discussed the mortality data included in the manufacturer's economic model. It was aware that the manufacturer had used mortality multipliers by EDSS score from Pokorski et al. (1997) in a Danish population diagnosed with multiple sclerosis from 1948 onwards. The Committee heard from the clinical specialists that they would anticipate that the relative risk of mortality in people with multiple sclerosis compared with the general population is lower than reported in this publication because the life expectancy of people with multiple sclerosis has improved. However, the Committee was aware that the manufacturer provided scenario analyses around mortality that showed that this had little impact on the ICERs and therefore concluded that it did not need to pursue this issue any further.

4.17 The Committee noted that because the trials lasted 2 years, but the manufacturer assumed that patients would take dimethyl fumarate indefinitely, the manufacturer modelled a waning of treatment effect because of the uncertain longer‑term benefits of dimethyl fumarate. The Committee heard from the clinical specialists that the manufacturer's assumption seemed reasonable but, given the uncertainty, they could not comment on the degree to which dimethyl fumarate's effect might wane. The Committee also recognised that it may be possible that the effect of dimethyl fumarate might wane at a different rate than other treatments, but this was uncertain. Therefore, the Committee accepted the manufacturer's approach using the same rate of waning of effect for each treatment. The Committee noted that, in response to consultation, the manufacturer presented data from the open‑label ENDORSE extension study, which suggested that dimethyl fumarate maintains its effect over 4 years. The Committee noted that the treatment effect in the economic model waned to 75% after 2 years, and therefore the manufacturer's model may have overestimated waning in the short term. However, the Committee recognised that the manufacturer had used a time horizon of 30 years in its economic model and therefore the longer‑term benefits of the treatment remained unknown. The Committee concluded that a cautious modelling approach was appropriate.

4.18 The Committee discussed the costs and resource use values included in the manufacturer's economic model. The Committee heard from the ERG that several publications were available that presented the annual costs of relapsing‑remitting multiple sclerosis by EDSS state, and that although they were also based on the UK Multiple Sclerosis Survey, the annual costs by EDSS state varied considerably. The ERG explained to the Committee that each publication used different unit costs and different cost items, and that some of the cost items were non‑medical (and therefore potentially not considered from the perspective of the NHS and personal social services), and so it was unclear whether these items met the NICE reference case, as detailed in NICE's Guide to the methods of technology appraisal). The Committee understood from the ERG that it was unable to judge the most appropriate data source for annual costs by EDSS state, and that an approach that removed non‑medical costs was more plausible (unless the manufacturer could prove that the non‑medical costs met the NICE reference case). The Committee was aware that, in response to consultation, the manufacturer explored the impact of including or excluding all non‑medical costs because it was not possible to identify in the data set the non‑medical costs relating to personal social services relevant to the NICE reference case. The Committee heard from the ERG that, for some EDSS states, the manufacturer had estimated higher costs when excluding rather than including non‑medical costs. However, the Committee noted that using either approach in the manufacturer's economic model had little impact on the ICERs. The Committee concluded that it preferred excluding non‑medical costs, but acknowledged that the ICERs were likely to be lower for dimethyl fumarate if the personal social services costs had been included.

4.19 The Committee was also aware that the manufacturer's chosen number and cost of visits to a neurologist needed by patients differed from those preferred by the ERG. The Committee considered that the number of visits included in the manufacturer's model was reasonable because patients taking dimethyl fumarate were unlikely to need more intensive monitoring than patients using other disease‑modifying treatments. However, the Committee agreed that the ERG's assumed cost (outpatient) for a visit to a neurologist was more plausible than the manufacturer's (day case). The Committee noted that the manufacturer had lowered the cost of relapse in its revised model in response to consultation, but not to the value preferred by the Committee (section 3.41). However, it heard from the ERG that either of these 2 lower values could be plausible and had little impact on the ICERs. The Committee was disappointed that all of the sources used by the manufacturer to estimate the cost of relapse were of low methodological quality, and encouraged further research to identify more robust data for future NICE technology appraisals. The Committee was satisfied that, given the current evidence, the manufacturer adequately addressed and explored all of the uncertainties associated with the costs in the economic model.

4.20 The Committee noted that the manufacturer had collected EQ‑5D utility data in its clinical trials of dimethyl fumarate for people with relapsing‑remitting multiple sclerosis without relapses, and adjusted these values for patients with secondary progressive multiple sclerosis and for patients experiencing a relapse using data from the UK Multiple Sclerosis Survey (see section 3.18). The Committee heard from the clinical specialists that the health‑related quality‑of‑life of people with multiple sclerosis was more closely related to their EDSS score than to the clinical form of their multiple sclerosis (that is, relapsing‑remitting or secondary progressive). The clinical specialists stated that it is difficult to clearly identify when a patient's disease becomes secondary progressive multiple sclerosis, and therefore it is also difficult to gauge the relative health‑related quality‑of‑life effects of the different clinical forms of multiple sclerosis. The Committee acknowledged that the ERG's exploratory analyses showed that using alternative utility values and alternative assumptions relating to the rate of conversion from relapsing‑remitting multiple sclerosis to secondary progressive multiple sclerosis had little impact on the ICERs. The Committee noted that the model included disutility to carers of people with relapsing‑remitting multiple sclerosis that increased with increasing disability of the patient. The Committee was aware that carer disutility had featured in Fingolimod for the treatment of highly active relapsing-remitting multiple sclerosis (NICE technology appraisal guidance 254) and Natalizumab for the treatment of adults with highly active relapsing-remitting multiple sclerosis (NICE technology appraisal guidance 127), and concluded that including these carer disutility values was appropriate. It noted that the ERG updated the disutility of flu‑like symptoms and flu in its exploratory analyses of the manufacturer's economic model but this had little impact on the estimated ICERs. The Committee concluded that the results of the ICERs were robust to changes in these parameters, and considered that the EQ‑5D utility values from the trial represented the best evidence available because they more closely reflected the population with relapsing‑remitting multiple sclerosis treated with disease‑modifying therapy in UK clinical practice.

4.21 The Committee discussed the assumption in the manufacturer's original economic model that people with relapsing‑remitting multiple sclerosis do not switch to another active treatment when their disease does not respond, or when they have adverse reactions. The Committee recognised that, whereas no specific sequence of disease‑modifying treatments defines standard practice in the NHS, it heard from the clinical specialists that people with relapsing‑remitting multiple sclerosis are likely to take another treatment in these circumstances, and confirmed that people would choose a treatment with a different side‑effect profile. The Committee noted that, in response to consultation, the manufacturer had provided cost‑effectiveness estimates for dimethyl fumarate when included in a treatment sequence. The Committee heard from the ERG that, when dimethyl fumarate is included in a treatment sequence, the resulting ICERs were slightly higher than those estimated in the manufacturer's original model, which excluded subsequent treatments. The Committee considered it important to explore the sensitivity of the ICERs from different treatment sequences, and agreed that, for future NICE multiple technology appraisals in multiple sclerosis, exploring several sequences would be useful. However, the Committee concluded that analysing individual drugs (without a sequence) was appropriate for its decision‑making in this appraisal because:

4.22 The Committee understood that the main drivers of the ICERs were the costs of treatment, how likely a patient was to experience disease progression, the probability of stopping treatment, and the magnitude of the treatment waning effect. The Committee heard from the clinical specialists that the rate of stopping treatment is likely to be lower in the longer term than that observed in the 2‑year trials because patients are more likely to have adverse reactions and discontinue treatment early in the treatment course. The Committee noted the ERG's observation that in the manufacturer's economic model, the sooner a patient stops treatment, the more cost effective the treatment appears (section 3.34). The Committee noted that the manufacturer, in response to consultation, carried out an external validation exercise to explore how similar the cost‑effectiveness estimates for current active treatments (all beta interferons and glatiramer acetate) compared with no treatment in its model were to those already established for the NHS risk‑sharing scheme for multiple sclerosis. It acknowledged that the manufacturer attempted to update its economic model to reflect the model used in the NHS risk‑sharing scheme as closely as possible, but noted that some differences remained between the models (for example, between the sources of evidences used, methods used to synthesise the evidence and structural assumptions). The Committee noted that the manufacturer and ERG were unable to explain any differences between the ICERs resulting from the manufacturer's model and from the model behind the risk‑sharing scheme, and highlighted that there was still uncertainty related to the validity of the manufacturer's model. The Committee acknowledged that showing close convergence between the previous and present analyses was challenging. The Committee concluded that it was satisfied that the manufacturer's economic model was sufficiently robust for decision‑making.

4.23 The Committee noted that the manufacturer's revised ICERs estimated from deterministic analyses were substantially lower than the ICERs estimated from probabilistic analyses. It heard from the manufacturer that this is because it included uncertainty around EDSS state transitions for which there is no evidence because they were not observed in the trial (for example, moving from the lower EDSS states of relapsing‑remitting multiple sclerosis to the higher EDSS states of relapsing‑remitting multiple sclerosis, such as from EDSS 1 to EDSS 6, 7, 8 or 9). The manufacturer highlighted to the Committee that it considered the probabilistic ICERs to be conservative estimates of cost effectiveness because the probabilistic ICERs were similar to the deterministic ICERs when the model did not include the uncertainty around the state transitions for which there was no evidence. The Committee considered it was appropriate to capture this uncertainty in the probabilistic analysis because some patients do experience these rare changes. The Committee was aware that the ERG also preferred probabilistic sensitivity analyses because of the non‑linear nature of the manufacturer's economic model (see section 3.31). The Committee concluded that it preferred the probabilistic ICERs.

4.24 The Committee discussed the innovative nature of dimethyl fumarate and whether the economic analysis had captured all changes in health‑related quality‑of‑life. In its submission, the manufacturer stated that dimethyl fumarate was innovative because it is taken orally, and because its mechanism of action targets the nuclear factor erythroid‑derived 2‑like 2 (Nrf2) pathway. The Committee recognised that a drug taken orally may give people with relapsing‑remitting multiple sclerosis a valuable alternative to current first‑line treatment options, but acknowledged comments from professional and patient groups that its twice‑daily administration schedule may lower adherence compared with once‑daily options. The benefit related to being an oral drug was not captured in the analysis because the manufacturer's economic model applied the same utility values to dimethyl fumarate as to beta interferons and glatiramer acetate. The Committee therefore acknowledged that dimethyl fumarate provides health‑related quality‑of‑life benefits other than those captured in the QALY calculation for patients currently taking beta interferons and glatiramer acetate, and that the ICER may decrease when the benefits of oral treatment were taken into consideration. The Committee heard from the clinical specialists that little is known about what causes multiple sclerosis and therefore it could not advise the Committee whether dimethyl fumarate's mechanism of action could be considered relevant to the pathophysiology of multiple sclerosis, and therefore innovative. The Committee also noted the comments received during consultation stating that dimethyl fumarate could be a preferred treatment option for women of child‑bearing age because of its short washout period compared with another oral treatment currently available. The Committee concluded that dimethyl fumarate was innovative, and that additional health‑related quality‑of‑life benefits associated with oral treatment and short washout duration may not have been fully captured within the manufacturer's economic modelling.

4.25 The Committee discussed the most plausible ICER for dimethyl fumarate for the group of people with relapsing‑remitting multiple sclerosis whose disease is eligible for active treatment under the Association of British Neurologists' guidelines (see section 4.3). The Committee acknowledged that the manufacturer had used the best available evidence to model the natural history of the disease, used EQ‑5D utility data as preferred by NICE in its Guide to the methods of technology appraisal and included waning of the treatment effect. The Committee agreed that the most plausible ICER should be based on:

Summary of Appraisal Committee's key conclusions