3 Committee discussion
The appraisal committee considered evidence submitted by Biogen, a review of this submission by the evidence review group (ERG), and the technical report developed through engagement with stakeholders. See the committee papers for full details of the evidence.
The committee was aware that several issues were resolved during the technical engagement stage. It agreed that the company had adequately justified some discrepancies in the company submission (both between the clinical- and cost-effectiveness sections, and when compared with previous appraisals).
The committee recognised that there were remaining areas of uncertainty associated with the analyses presented (see technical report, table 2, page 36), and took these into account in its decision making. It discussed the following issues, which were outstanding after the technical engagement stage:
issue 1: the population and appropriateness of comparators
issue 2: the minimum clinically significant reduction in outcome measures
issue 3: the generalisability of the ADVANCE trial population
issue 5: treatment waning
issue 6: stopping treatment for any reason
issue 7: the utility values.
Any relapse can affect people with relapsing–remitting multiple sclerosis and is clinically significant
3.1 Relapsing–remitting multiple sclerosis is a chronic, disabling, lifelong neurological condition. People with the condition can have episodes of relapse, in which the symptoms worsen, followed by remission. However, over time, relapses result in worsening disability. The condition is associated with signs and symptoms such as pain, disturbance to muscle tone, chronic fatigue, unsteady gait, speech problems, incontinence, visual disturbance and cognitive impairment. The committee noted that reducing the frequency of relapses was a key outcome for people with the condition. It considered whether there was a clinically meaningful reduction in the number of relapses. The patient and clinical experts explained that even 1 mild relapse can be devastating and can affect a person's family life and career. The committee concluded that any reduction in relapse was clinically significant.
Peginterferon beta-1a would most likely be offered first line or as an alternative first-line treatment when other options are not tolerated
3.2 Peginterferon beta‑1a has been commissioned by NHS England since 2015 for people with relapsing–remitting multiple sclerosis as a first-line treatment or as an alternative when other first-line treatments are not tolerated. The committee noted that the licence is broader than this because it does not restrict peginterferon beta‑1a to a particular line of therapy. It considered whether peginterferon beta‑1a would also be of value in the more severe subgroups of 'highly active' and 'rapidly evolving severe' multiple sclerosis. The patient and clinical experts explained that peginterferon beta‑1a would continue to be used primarily as a first-line treatment or as an alternative when other first-line treatments are not tolerated. They said that it would not be their first choice of treatment for people with more severe forms of relapsing–remitting multiple sclerosis. However, they stated that, in clinical practice, multiple sclerosis does not lend itself to such clear categorisation. They also noted that patient choice is an important aspect of the treatment pathway, so they would not want to exclude use of peginterferon beta‑1a later in the treatment pathway. For example, some people with more severe disease might prefer peginterferon beta‑1a than other options if they were better able to tolerate it. The committee concluded that, in clinical practice, peginterferon beta‑1a would likely continue to be used primarily as a first-line treatment or as an alternative when other first-line treatments are not tolerated. However, given the importance of patient choice, it agreed that any recommendation should not explicitly comment on its use later in the treatment pathway.
3.3 At the time of the committee discussion (November 2019), peginterferon beta‑1a was already commissioned by NHS England as a first-line treatment option along with other beta interferons, dimethyl fumarate, glatiramer acetate, teriflunomide, alemtuzumab and ocrelizumab. The clinical experts explained that, of these, other beta interferons and glatiramer acetate were the most relevant comparators. Alemtuzumab and ocrelizumab are more active immunosuppressant treatments with a higher chance of more serious adverse effects, so would typically be used later in the treatment pathway. Also, as of November 2019, the Committee for Medicinal Products for Human Use recommended that the use of alemtuzumab should be restricted to 'relapsing–remitting multiple sclerosis that is highly active despite adequate treatment with at least one disease-modifying therapy or if the disease is worsening rapidly with at least two disabling relapses in a year and brain-imaging showing new damage'. The committee concluded that alemtuzumab and ocrelizumab were not relevant comparators at the place in the pathway where people were most likely to use peginterferon beta‑1a.
3.4 The committee considered how people would choose between the several beta interferons available. The patient and clinical experts explained that, if a treatment with 1 beta interferon had failed, another would not typically be offered. They stated that peginterferon beta‑1a is not considered to be more effective than other beta interferons but has a less frequent dosing schedule. It is administered twice monthly, whereas other beta interferons may need to be administered as often as 3 times a week. The clinical experts stated that people were less likely to develop neutralising antibodies with peginterferon beta‑1a than with other interferons. However, the committee was not shown data to support this and questioned its relevance. One clinical expert explained that, when offering a choice of treatments, clinicians generally remain neutral to allow people to choose the best treatment for their lifestyle. For example, some people prefer the reduced dosing schedule of peginterferon beta‑1a, whereas others, such as those with cognitive issues affecting memory, might prefer a more frequent, regular schedule. The committee concluded that the availability of peginterferon beta‑1a increased patient choice.
3.5 The company presented the evidence for peginterferon beta‑1a from:
ADVANCE, a phase 3 double-blind randomised placebo-controlled trial and
ATTAIN, a 2‑year blinded follow-up study (of peginterferon beta‑1a only; all people in the placebo arm of ADVANCE switched to peginterferon beta‑1a after 1 year).
The ADVANCE trial recruited patients mainly from the eastern European region, and included only 8% from western Europe (of which 14 patients were from the UK). The ERG explained that there were some numerical differences in effectiveness across regions, with region 1 (which included the UK) performing the worst. However, the company stated that it did not find statistical interaction by region. The patient and clinical experts explained that any patients from the UK in the trial were likely not typical of those seen in NHS clinical practice. This was because, given the many active treatments available on the NHS, there would be little incentive for people with multiple sclerosis to join a placebo-controlled trial. Also, there may be differences in clinical practice, treatments and standard of care across regions. The committee agreed that all these factors raised questions around how representative the trial would be of a UK population. However, it noted that most patients in the trial were treatment naive (the expected place in the treatment pathway for peginterferon beta‑1a). It also noted that, if there were any differences in the population, this was unlikely to have had an impact on the treatment effect of peginterferon beta‑1a compared with placebo. The committee concluded that it had minor concerns about the generalisability of the ADVANCE and ATTAIN trials. However, overall, it considered that they were appropriate for decision making.
It is disappointing that there is no evidence to help address gaps in evidence on areas such as generalisability, treatment waning and stopping treatment
3.6 Given its concerns about generalisability, the committee considered whether there was any pharmacoepidemiologic evidence available from current use of peginterferon beta‑1a in the NHS. The company explained that, of around 12,000 people in the UK currently using injectable treatments for relapsing–remitting multiple sclerosis, about 1,400 use peginterferon beta‑1a. The committee stated that, if the company had provided data on these patients, it would have helped inform evidence gaps in areas such as generalisability (see section 3.5), stopping treatment (see section 3.13) and treatment waning (see section 3.14). It was disappointed that it had not been presented with this evidence, which could have helped to address these evidence gaps.
3.7 In ADVANCE, there were statistically significant (p<0.05) improvements with peginterferon beta‑1a compared with placebo for the primary outcome (annualised relapse rate) and several important secondary outcomes, including disability progression sustained for 3 or 6 months:
annualised relapse rate: rate ratio 0.64 (0.50 to 0.83; p=0.0007)
confirmed disability progression at 3 months: hazard ratio 0.62 (0.40 to 0.97; p=0.04)
confirmed disability progression at 6 months: hazard ratio 0.46 (0.26 to 0.81; p=0.007).
The committee agreed that there had been improvements in important outcomes such as relapse frequency when compared with placebo. It also noted its previous conclusion that any reduction in the frequency of relapses was clinically significant (see section 3.1). The committee concluded that peginterferon beta‑1a was both a clinically and statistically significantly effective treatment when compared with placebo.
Peginterferon beta-1a does not appear to be more effective than its main comparators in an indirect comparison with active treatments
3.8 Because there was no direct trial evidence comparing peginterferon beta‑1a with active comparators, the company also conducted indirect analyses comparing it with the other beta interferons, dimethyl fumarate, glatiramer acetate, teriflunomide, alemtuzumab and ocrelizumab. Several of the results are academic in confidence and cannot be presented here. However, overall, the committee concluded that there were no statistically significant differences for peginterferon beta‑1a compared with its main comparators (that is, other beta interferons and glatiramer acetate).
3.9 The patient and clinical experts explained that the most common adverse events were injection-site reactions and flu-like symptoms. Although these could sometimes be severe, they were generally mild to moderate and easily treatable with common analgesics such as paracetamol and ibuprofen. The experts were not aware of any differences in adverse effects between the beta interferons. The committee concluded that peginterferon beta‑1a is an established treatment with a well-known side effect profile.
3.10 The company estimated disease progression in the model using 21 health states. It defined these using Expanded Disability Status Scale (EDSS) scores ranging from 0 to 9.5 (with a higher score indicating worse disease) and either relapsing–remitting or secondary progressive multiple sclerosis, plus a death state. In each cycle of the model, a patient with relapsing–remitting multiple sclerosis could move to a higher or lower EDSS state (that is, their disability could worsen or improve) or remain in the same state. The disease could also advance from relapsing–remitting multiple sclerosis to secondary progressive multiple sclerosis but could not then move back to relapsing–remitting disease. The clinical experts explained that it was not realistic to assume that patients could not improve from secondary progressive to relapsing–remitting multiple sclerosis. Also, the company assumed that, once treatment with peginterferon beta‑1a stopped, a patient's condition followed the untreated natural history of the disease. However, in clinical practice, people would have several other treatments to choose from. The company stated it had chosen not to include any treatment sequencing because it could have biased the model, making it unclear whether any observed treatment effect was coming from the main intervention or the next line of treatment. The committee agreed with this approach. It noted that the overall structure of the company's model was similar to models used in previous NICE technology appraisals for the first-line treatment of relapsing–remitting multiple sclerosis. The committee concluded that the structure of the company's economic model was appropriate for decision making.
3.11 The company used the characteristics of patients in the ADVANCE trial to reflect the baseline population in the model. The committee discussed whether ADVANCE was the most appropriate source, given its earlier minor concerns that there were issues with generalisability in the trial (see section 3.5). It was aware that models in some previous appraisals for this disease area had used the multiple sclerosis risk sharing scheme (RSS) as the source for its baseline population. The RSS collected data for more than 5,000 patients in the NHS for over 10 years. The committee agreed that this population in the RSS was more generalisable to the NHS than the population in ADVANCE. It concluded that it would have preferred that the model baseline characteristics had been based on the RSS.
3.12 The company used utility values from Orme et al. (2007) for consistency with previous technology appraisals. It also provided utility values from a more recent study, Thompson et al. (2017). The ERG provided a scenario analysis using Thompson et al. (2017). The committee noted that the utility values in both Thompson et al. and Orme et al. were very similar. However, it noted that there were substantial differences in the quality-adjusted life years generated by the deterministic and probabilistic models. This was unusual and could suggest that there may have been an issue with the underlying utility values used in the model. The committee discussed whether it could identify any underlying problems with the utility values. The company did not adjust utility values to worsen with age, which the committee agreed was unrealistic. It also noted that older people were more likely to be in a higher EDSS score group in the model (in which higher EDSS scores were associated with a worse utility value), and that the Orme et al. and Thompson et al. studies differed with respect to age, which could have led to different model predictions. The committee agreed that, because the utility values seemed to be a key driver of cost effectiveness in the model, the company should have explored this more thoroughly, for example, by using values derived from a meta-analysis. The committee concluded that the utility values were a source of uncertainty in the cost-effectiveness results.
3.13 The company applied a probability of stopping treatment for any reason in its base-case analysis. It used the annualised stopping rates from 18 trials. It weighted these based on sample size to derive the risk of stopping treatment for any reason for each disease-modifying therapy. The ERG considered that deriving the stopping risk weighted by person time would be more appropriate than using the trial sample size. This was because an annualised stopping rate would not capture changes over time. The company accepted this correction. However, the ERG stated that it would have preferred the use of a 5% stopping rule across all treatments, which was derived from the multiple sclerosis RSS. The committee agreed that it was more likely that different treatments would have different stopping rates. It also noted that data on treatment-specific stopping rates in the UK should be available because they are recorded electronically on NHS payment systems. It was disappointed that these had not been presented. The committee concluded that, in the absence of the actual data, it was more plausible to assume a treatment-specific rate than a flat rate.
3.14 The company assumed in the model that the treatment effect of all treatments waned over time, and all at the same rate:
years 1 and 2: no waning
years 3 to 5: 75% of full treatment effect
year 6 onwards: 50% of full treatment effect.
The clinical experts stated that most treatments for multiple sclerosis become less effective over time. This is either because the person's immune system develops neutralising antibodies, or because the condition worsens and becomes resistant to treatment. It was unclear whether the treatment effect waning with peginterferon beta‑1a would differ from other treatments. The committee noted that peginterferon beta‑1a is an established treatment and that data on treatment effect waning could have been collated. However, the company had not attempted to analyse whether the treatment effect of peginterferon beta‑1a changed over time, which left an important gap in the evidence. The committee concluded that it was appropriate to include some treatment effect waning in the model, and that it would have been more plausible to assume treatment-specific rates rather than a constant rate. However, the committee noted that varying this assumption (that is, using a constant rate for all treatments, or varying the rate by treatment) did not substantially affect cost-effectiveness results. Because of this, the committee accepted the company's base-case assumption.
3.15 The ERG made several changes to the company's base case. It stated that the most important of these were:
changing the treatment-specific stopping rates to a flat rate of 5% from the multiple sclerosis RSS (because it considered these data more informative than trial data for the outcome of stopping treatment) and
the source of decrements in the utility of caregivers (because its preferred source assumed utility decrements rose with worsening health state).
The committee agreed with some, but not all, of the ERG's changes. For example, it agreed that caregiver utility decrements should rise with worsening EDSS health state, and that the company should have used the baseline population from the RSS rather than the trial. However, it disagreed that it was more plausible to assume a constant stopping rate for all treatments than a treatment-specific rate. The committee concluded it would take into account the various company and ERG base cases and scenarios in its decision making.
3.16 The committee agreed that enough evidence had been presented to consider the cost effectiveness of peginterferon beta‑1a. However, given the number of years that peginterferon beta‑1a has been an established treatment in the NHS, it was disappointed that the company had not provided data reflecting the experience with peginterferon beta‑1a in NHS practice. These data would have helped to address uncertainty in the model, including for stopping rates and treatment waning. The ERG and company presented a number of exploratory analyses that, in part, helped to reduce some of these uncertainties. Results of all these analyses suggested that the incremental cost-effectiveness results of peginterferon beta‑1a were in the range that NICE normally considers a cost-effective use of NHS resources. The results are commercial in confidence because of confidential comparator patient access schemes, so cannot be reported here.