Auspar attachment 3: Extract from the Supplementary Clinical Evaluation Report for Daclizumab

7.2.Supplementary material submitted to the EMA in relation to pivotal efficacy data

7.2.1.EMA Question 70

‘The indication claimed by the applicant is for the treatment of Relapsing Multiple Sclerosis. However in order to gain this unrestricted indication the applicant should further justify a positive effect on disability progression in relapsing forms of Secondary Progressive Multiple Sclerosis. The absence of such an effect may need to be reflected in the European Summary of Product Characteristics.’

This question relates to an issue already flagged in this SCER: the sponsor performed two pivotal studies with entry criteria that explicitly excluded SPMS, but nonetheless proposes an indication that applies to the excluded MS category. The EMA question is specifically framed in terms of the endpoint of disability progression, but similar concerns could be raised about the overall efficacy of DAC HYP in SPMS, including its effect on relapse rate.

In their response to this question, the sponsor attempted to defend their extension of the indication beyond the original target population. The evaluator was not convinced that the arguments raised by the Sponsor were valid, and strongly recommends that the indication in the PI should match the entry criteria of the pivotal studies. At a minimum, this would mean that DAC HYP was restricted to subjects with RRMS. A strict interpretation of the pivotal studies would require that DAC HYP was further restricted to subjects with RRMS and evidence of recent disease activity, as evidenced by at least 2 relapses in the previous 3 years (or a radiological substitute for clinical relapses).

7.2.1.1.Sponsor’s response

The sponsor began by pointing out that some subjects in the pivotal studies had SPMS, despite the exclusion criteria:

‘In the clinical development of DAC HYP in MS, the 2 pivotal trials were of sufficient duration and size that subjects included in these trials could be identified as having SPMS with superimposed relapses based on the observation of sustained disability progression that occurred independently of, or in the absence of, clinical relapses.’

The evaluator acknowledges that the pivotal studies were almost certainly contaminated with some subjects who had SPMS. Given that SPMS was explicitly listed as an exclusion criterion, this raises a number of difficulties in interpretation. If subjects had SPMS at baseline, then they were not entered into the study appropriately; conversely, if subjects developed SPMS during the study, then they were not prospectively randomised after entering the subgroup of interest.

For subjects who had SPMS at baseline, it might be possible to perform a post hoc subgroup analysis, if such subjects could be reliably identified, but their accidental inclusion does not constitute a clear prospective assessment of efficacy in SPMS, particularly if they only constituted a small proportion of the overall cohort. If they constituted a large proportion of the cohort, it implies widespread disregard for the entry criteria, and suggests the study was not methodologically sound.

For subjects who developed possible SPMS during the study, there are also difficulties with interpretation. Progression in the sense mentioned in the sponsor’s statement above, ‘sustained disability progression’, could be attributed to two quite different processes: experiencing a relapse and failing to recover fully (giving a stepwise increase in disability), or progressing slowly in the absence of a relapse (giving a gradual increase in disability). Only the latter counts as SPMS, though it is difficult to separate the two categories accurately in clinical practice.

A third category also needs to be excluded, spurious progression, in which subjects have not yet had time to exhibit recovery from their most recent relapse; it is important that the methodological requirement for assessing disability progression at a time distant from the relapse is not conflated with the issue of whether the increased disability was caused by a relapse. The sponsor’s expression ‘independently of, or in the absence of, relapses’ and much of their subsequent discussion, is ambiguous about whether an aetiological or a temporal independence is intended. Only those subjects with gradual disease progression during the study could count as having developed SPMS post-baseline, and the number of such subjects is uncertain. The mere presence of subjects with ‘confirmed disability progression’ does not prove that all or most of those subjects had gradual disease progression, and therefore had SPMS (much less that they had SPMS at baseline).

If subjects developed SPMS during the pivotal studies, then this could be seen as a treatment failure and it is not logical to conclude that a benefit demonstrated for the whole cohort applies equally well to those who experienced a treatment failure. Even if it were possible to identify a subgroup in which true SPMS developed during the studies (by excluding subjects in whom the deterioration was causally associated with a relapse), it would be difficult to draw any conclusions from a subgroup analysis of their subsequent course. Firstly, their time left in the study would have been shortened by the requirement that they develop SPMS prior to entering the subgroup of interest, reducing the time left available in which they could reach new clinical endpoints. Secondly, as a subgroup, they are unlikely to have been present in large enough numbers for their assessment to achieve statistical power. Thirdly, their non-random entry into the SPMS subgroup while already on treatment would raise major issues of interpretation because treatment allocation would not necessarily be random across all subjects with SPMS. Fourthly, the fact that they had progressed despite taking DAC HYP is likely to mean that they are a subgroup in which DAC HYP efficacy was limited, so such a subgroup analysis could produce disappointing results.

The only analysis that would properly address the efficacy of DAC HYP in SPMS would be one that took subjects with clinically overt SPMS and examined them prospectively, in a randomised study. When dedicated SPMS studies have been attempted with other disease-modifying agents, the results have often been disappointing, and it is widely recognised that SPMS is a more difficult entity to treat than early, non-progressive RRMS. Decades of experience with MS research has shown that immune-modifying agents have blunted efficacy in this disease category, so efficacy in SPMS subjects must be demonstrated directly, not inferred from the results in RRMS subjects, or from the accidental inclusion of SPMS subjects in studies of RRMS.

The sponsor’s response continued:

‘Furthermore, analysis of these subjects provided evidence that DAC HYP was more effective than IFN β-1a at preventing the progression of sustained disability progression that occurred independently of clinical relapses.’

This statement appears to be in direct contradiction of the fact that Study 205MS301 did not show a significant effect on sustained progression of disability for the cohort as a whole, or for any prospectively defined subgroup. A nominally significant result was shown for the tertiary endpoint of 24 week confirmed progression, but not for the key secondary endpoint of 12 week confirmed progression. In EMA Question 94 (discussed below), a subgroup was identified post hoc that a showed nominally significant benefit for DAC HYP on progession, but this subgroup analysis was not based on the presence of SPMS. From the quoted sentence above the expression ‘analysis of these subjects’ was used in the context of a discussion of SPMS. This implies that a subgroup analysis of SPMS subjects was undertaken and that this subgroup had reduced progression on DAC HYP. No prospective subgroup analysis addressed this issue, because SPMS subjects were not supposed to be enrolled. The statement therefore appears to be unjustified.

Both pivotal studies included some subjects with elevated EDSS at baseline, which could arguably serve as an imprecise surrogate marker for the presence or absence of SPMS. One section of the sponsor’s response to EMA Question 70 proposed that EDSS ≥ 3.5 could be interpreted this way. In the same section, the sponsor also appeared to suggest additional post hoc methods of identifying SPMS subjects, as follows:

‘In order to explore the issue of DAC HYP’s effect on patients with SPMS with superimposed relapses, the following analysis was performed for Study 301:

1. 
   Identification of subjects with higher levels of baseline disability: EDSS ≥3.5, when the transition to SPMS is common.
   
2. 
   Definition of a gradual worsening of neurologic function using standard criteria to measure disability progression in SPMS clinical trials that is based on a composite measure of 6-month confirmed worsening on either:
   
   1. 
      EDSS,
      
   2. 
      20% worsening in gait as measured by the T25FW, or
      
   3. 
      20% worsening on upper extremity function as measured by the 9HPT. Progression is confirmed at a visit at least 6 months later and also at the last study visit.
      
3. 
   Requirement that both the initiation of the neurologic worsening and the 6-month confirmation of the neurologic worsening occur in the absence of relapse within the prior month of the evaluation. As a further sensitivity analysis to ensure that new clinical relapses were not the cause of the worsening, an additional analysis of 6-month confirmed worsening was performed that was restricted to subjects who were relapse free during the entirety of Study 205MS301.’
   

Each of these proposed methods for identifying SPMS subjects is potentially flawed, mostly because relapses in RRMS subjects can cause deficits in all of the domains studied: elevations of EDSS, worsening of gait, and worsening of upper limb function. Excluding a relapse in the previous month reduces the risk that a current relapse will be misinterpreted as disease progression, but does not prevent a previous relapse with incomplete recovery from being misidentified as a marker of SPMS. Note that the sponsor’s original definition of confirmed progression required a 12-week delay to minimise the inclusion of cases with spurious progression, not the one-month delay proposed for this analysis. Also, note that the category of sustained progression is not limited to sustained gradual progression: sustained EDSS progression could include a stepwise deterioration due to a relapse, and would not identify a patient as having SPMS.

Overall, then, the sponsor’s proposed subgroup analyses do not clearly identify a subgroup of subjects with SPMS. Even if they did, they would constitute a special subset of occult SPMS subjects – those who, at baseline, were explicitly thought by their clinicians not to have SPMS, but had it or developed it anyway. Even if efficacy were confirmed in such a subgroup, this would not apply to the broader disease category of overt SPMS. Further discussion of this important issue is found in the conclusions of this section.

Prior to submitting this supplementary material, the Sponsor had already performed a basic subgroup analysis based on EDSS. In Study 205MS201, subgroup analysis of subjects with higher baseline EDSS (> 2.5 versus ≤ 2.5) showed that, even in subjects with higher EDSS, DAC HYP has favourable effects on the primary endpoint of annualized relapse rate, compared to placebo. A similar subgroup analysis for the minor endpoint of disease progression appears not to have been performed in Study 205MS201, which is understandable as this was a tertiary endpoint and the study was too brief for a robust assessment of progression.

In Study 205MS301, subgroup analyses based on EDSS showed a benefit of DAC HYP over IFN β1a on ARR in subjects with moderately elevated EDSS (> 2.5 versus ≤ 2.5) but the benefit was reduced, and not statistically significant, in subjects with even higher EDSS (≥ 3.5). For the secondary endpoint of disease progression, favourable trends were observed across a range of EDSS categories. The treatment effect for this endpoint was not statistically significant for the cohort as a whole, so it is not surprising that EDSS subgroups also failed to show a significant reduction in disease progression.

Given that the higher EDSS subjects did not have overt SPMS at baseline (according to the protocol), the favourable trends for reduced ARR and the weaker trends for reduced progression cannot be generalised to other subjects with high EDSS and overt SPMS. The high-EDSS subgroups (> 2.5 or ≥ 3.5) would have included many subjects in whom baseline EDSS was elevated because of previous relapses with incomplete recovery, rather than because of SPMS. Indeed, according to the protocol, this was the only permissible reason for having a high baseline EDSS.

In the sponsor’s response to EMA Question 70, the main efficacy measure of interest was disease progression, which was subjected to a new analysis, with alternate definitions of progression based on a composite of EDSS, timed walk and 9-hole peg test.

The sponsor’s presentation of these supplementary analyses was somewhat unclear. Their discussion was primarily focussed on the extent to which DAC HYP was effective in subjects with SPMS that were inadvertently included in the pivotal studies or who developed SPMS during the studies. The proposed supplementary definitions of progression could be interpreted as an alternative means of identifying SPMS for potential subgroup analysis, and they were apparently introduced in that context. Further inspection of the tables suggests that these additional progression measures were used as alternative progression endpoints in the overall cohort, not as a means of identifying an SPMS subgroup. The sponsor was asked to confirm the nature of this analysis, with the following question:

‘In the sponsor’s response to EMA Question 70, two tables were supplied intended to address the efficacy of DAC HYP in subjects with Secondary Progressive MS (SPMS). From the sponsor’s brief presentation of this data, it is somewhat unclear what analysis has actually been performed. Is the proportion with confirmed progression listed in each column of the table simply the proportion of the total EDSS-specific cohort that were considered to have progressed according to each of the listed criteria?

For instance, in the first table, does the ‘0.241’ listed opposite the Timed 25-Foot Walk under IFN β-1a simply mean that, of the 291 subjects at risk in this EDSS category, approximately 24.1% showed progression identified on the 25-Foot Walk? (Or similarlybut more precisely that the Proportional Hazards Model predicts that 24.1% would progress?)

If so, is this exercise primarily an analysis of the risk of developing SPMS (identified by a range of markers), given a particular baseline EDSS? Was any analysis done of the risk of progression after being identified as having SPMS by any of these tests? If not, this exercise appears to consist of no more than a new subgroup analysis of EDSS categories, using a non-standard definition of progression instead of the protocol-specified definition as a 12-week sustained EDSS worsening. The analysis does not appear to provide any prospective assessment of subjects with SPMS.

The core part of their answer is as follows:

‘The values listed in the table are the estimated proportion of subjects who experienced disability progression by the listed criteria based on the Kaplan Meier product limit method (for example, from Table 1, 24.1% of IFN β-1a-treated subjects with baseline EDSS ≥ 3.5 are estimated to have 6-month confirmed 20% worsening on the timed 25-foot walk that did not start and was not confirmed at a visit within 29 days of an MS relapse).’

This response confirms that the scores on the 9HPT and 25FW have been used as efficacy outcome variables, not as a means of identifying SPMS subjects for subsequent subgroup analysis. The relevance of this analysis to subjects with SPMS is indirect.

The results of the sponsor’s reanalysis of progression are shown in the two tables below (Tables 14 and 15). The use of the term ‘Independent of Relapse’ in the title of the first table is potentially misleading, because it merely indicates that no relapse had occurred in the month prior to the documentation of progression, not that the progression was causally independent of a relapse. In the second table, the analysis is restricted to subjects without any on-study relapses, and it appears more likely that, in such subjects, the observed progression does actually identify these subjects as having probable SPMS. Note that the number of such subjects was low: only 163 and 154 in the IFN β-1a and DAC HYP groups, respectively, had EDSS ≥ 3.5 and were free of relapses, and less than a quarter of these (23.4% in the IFN β-1a group) could be considered to have had SPMS by the ‘Composite’ progression measure at the end of the study. At baseline, none of them were thought to have SPMS by the enrolling clinician, and even if we accept that all such subjects had SPMS, it is unclear how many of them had SPMS at baseline and how many of them developed it during the study. In nearly all of the subgroups identified in this manner, and over a range of different definitions of progression, there are favourable trends for DAC HYP but the treatment benefit relative to IFN β-1a is not significant. Indeed, of 24 analyses across the two tables, only one had a 95% CI that excluded unity in favour of DAC HYP. Contrary to the sponsor’ s conclusions, this analysis suggests that the number of SPMS subjects in the pivotal studies was very low, that the overall results cannot be extended to the SPMS population, and that the benefit of DAC HYP over IFN β-1a is not robust for the endpoint of progression.

Table 14. Summary of confirmed progression independent of relapse in Study 205MS301

Table 15. Summary of confirmed progression in relapse-free population in Study 205MS301

7.2.1.2.Evaluator’s conclusion

Overall, despite the sponsor’s arguments to the contrary, there has been no clear assessment of the efficacy of DAC HYP in subjects with SPMS, given that the pivotal studies explicitly excluded this disease category, efficacy in this population is unknown.

Subjects with a moderately high baseline EDSS (attributed by their clinician to relapses rather than to SPMS) appeared to respond to DAC HYP with a reduced relapse rate in both pivotal studies. This implies that the drug has efficacy in subjects with high EDSS when it is due to incomplete recovery from previous relapses, but it does not allow a generalisation to subjects with a high EDSS due to SPMS, which was explicitly not intended to be assessed in the pivotal studies.

The data for progression are less favourable than those for ARR. According to prospective analysis methods in Study 205MS201, the study did not show a significantly reduced rate of progression relative to placebo. The proposed DAC HYP dose of 150 mg achieved nominal statistical significance for progression in this study, but only if issues with multiplicity are ignored. (Relative to placebo, the hazard ratio for disability progression was 0.43 (95% CI: 0.21 to 0.88) in the DAC HYP 150 mg group and 0.57 (95% CI: 0.30 to 1.09) in the DAC HYP 300 mg group.) No subgroup analysis of progression data in SPMS subjects from Study 205MS201 was presented.

In Study 205MS301, high-EDSS subjects did not have a significantly reduced rate of progression relative to IFN β-1a, using the major prospective definition of progression (12 week sustained EDSS worsening) in keeping with the lack of a significant effect of DAC HYP on progression in the cohort as a whole.

The sponsor’s supplementary analyses of Study 205MS301, shown in the tables above, appear to identify some subjects who had probable SPMS at the end of the study, but it remains completely unknown how many of these subjects had SPMS at the time of study entry. The group of subjects with SPMS at baseline is likely to have been too small to allow any robust inferences to be drawn, even if such subjects could be identified. At best, the provided tables indicate that DAC HYP may reduce the development of SPMS in high EDSS subjects, as suggested by favourable trends in this post hoc analysis, but this is not a statistically robust finding and would require confirmation in a prospective study. The tables do not constitute an analysis of DAC HYP efficacy in subjects with SPMS.

Even if it were known that DAC HYP reduced the development of SPMS, this would not establish that it has efficacy in SPMS, and it is unfortunate that the Sponsor’s discussion frequently conflated these issues. If we translate the issue to a different domain, it becomes clear that the ability to prevent a disease is not the same as efficacy in treating the disease (showing that a vaccine could prevent influenza would not logically imply that it also had efficacy in the treatment of influenza.) For SPMS, the situation is more complex, because there is a spectrum of disease between MS dominated by relapses and MS dominated by progression. There are a priori reasons to suspect that mechanisms of action that prevented relapses in early disease might still be beneficial when MS has reached a secondary progressive phase. Nonetheless, there is also substantial evidence from decades of MS research suggesting that immune therapies have less efficacy in SPMS, and this means that efficacy in SPMS must be demonstrated explicitly. If this were not the case, the sponsor would not have listed SPMS as an exclusion criterion in the first place.

To some extent, the sponsor’s entire discussion in their response to EMA Question 70 is based on the assumption that accidental violations of inclusion criteria justify subsequent broadening of the target population. This reasoning is rejected for three main reasons:

1. 
   It encourages a second round of ‘bracket creep’ in the attempt to identify the target population. Many of the standard MS categories lack clear boundaries, including the category of SPMS. It is therefore inevitable that a large MS study would inadvertently include some subjects in whom the categorisation of their MS was open to debate and potentially subject to redefinition. Importantly, though, the same is also true of clinical characterisation of MS when prescribing a drug in an MS patient. Some degree of blurring of the boundaries is inevitable in both trial recruitment and clinical prescribing and in the absence of further evidence we should assume that the blurring is about the same in both situations. (It is performed by the same clinicians treating the same disease). If some subjects in the pivotal studies had a clinical appearance suggestive of RRMS but actually had occult SPMS and if those subjects appeared to benefit from DAC HYP, then the same is likely to be true of subjects in clinical practice that appear to have RRMS and actually have occult SPMS. Such subjects would end up treated anyway, using an indication that matched the entry criteria. By matching the target population to the study entry criteria, the blurring of the diagnostic categories in both situations would be expected to include the same subjects, and produce similar benefits. Conversely, if the imprecision of the diagnostic categories in a study is used to justify a more inclusive indication, then the imprecision of the clinical categorisation at the time of prescribing adds a second round of blurring, leading to inclusion of subjects not represented in the original study. In this particular context, if the indication for DAC HYP were worded as ‘relapsing forms of MS’ subjects with overt SPMS who were not actually experiencing ongoing true relapses could be included because:
   
   1. 
      their MS had begun with relapses, and therefore they could be considered to have a relapsing form of MS;
      
   2. 
      the tendency of MS-related pre-existing deficits to flare in the context of inter-current illnesses (‘pseudo-relapses’) could be construed as ongoing relapses;
      
   3. 
      the tendency of gradual functional decline to cross non-gradual milestones could produce the appearance of a stepwise decline and be construed as a relapse, such as first need of a walking aid, or first use of a wheelchair;
      
   4. 
      intercurrent soft-tissue injuries, which are common complications of MS-related motor disability, could produce temporary deteriorations and be classified as relapses;
      
   5. 
      non-MS-related deficits, such as complicated migraines or middle ear infections, could be labelled as relapses;
      
   6. 
      clinicians or patients could misrepresent the underlying clinical symptoms deliberately or subconsciously because of their hope that ‘doing something’ for the MS is better than giving up.
      

For all of these reasons, some spread of the target population could be expected at the time of prescribing, which should not be added to over-inclusiveness in the wording of the indication.

The accidental inclusion of some subjects with occult SPMS in the pivotal studies does not prove that the drug works in this category. Studies only provide clear evidence of efficacy if they are prospective tests of well-defined hypotheses, and accidental inclusion of subjects with a different condition constitutes a methodological flaw. If some subjects with small-vessel cerebrovascular disease had been misdiagnosed as having MS and been accidentally recruited, their inclusion would clearly not support the claim that DAC HYP worked in small-vessel disease.

The sponsor’s original study design with its explicit entry criteria was a tacit admission that these subjects are more difficult to treat than RRMS subjects, or at least different to RRMS subjects. The sponsor is yet to account for the inherent contradiction in their approach: their study design explicitly excluded these difficult-to-treat subjects, but they have nonetheless proposed including them in the indication.

As there is a spectrum of MS disease types from pure RRMS to pure non-relapsing SPMS, it is likely that some efficacy is achievable in subjects with SPMS, at least for those subjects still experiencing relapses, but the onus is on the sponsor to asses that potential for efficacy, rather than inferring it from studies that explicitly excluded such subjects. The indication in the PI should therefore be changed to match the entry criteria of the pivotal studies. There is, ultimately, no logical rationale for performing a study with one set of entry criteria and then proposing that clinicians prescribe on the basis of more inclusive criteria.

7.2.2.EMA Question 94

‘The applicant is invited to redefine ‘highly active MS’ as per Tecfidera and Aubagio SmPC definitions, and should perform a comparison of safety data (for) high versus low disease activity, as this will have impact in the benefit/risk discussion.’

This question, as summarised in the material provided to the clinical evaluator, did not explicitly ask for an efficacy analysis using the new definition, but this would appear to the main point of the redefinition, and it is implied that the safety analysis was additional. As discussed below, the estimation of the efficacy of DAC HYP was potentially affected by the redefinition, but the safety assessment was unaffected. Discussion of safety results in this subgroup analysis is included in the Safety section of this report.

Table 16. Original and new sponsor definition of ‘highly active MS’

Original definition Study 205MS201	Original definition Study 205MS301	New definition
As a post-hoc analysis, the efficacy of DAC HYP was also evaluated in subjects with high disease activity at baseline, defined as: ≥ 2 relapses in the year prior to randomisation and; ≥ 1 Gd-enhancing lesion at baseline as well in subjects with and without prior MS treatment experience (excluding steroids). Post hoc	High disease activity was defined as: ≥ 2 relapses in the year prior to randomisation and; ≥1 Gd+ lesion on the baseline MRI. Prospective	Definition 1: Subjects with ≥ 2 relapses in 1 year and; ≥ 1 Gd-enhanching lesions on brain MRI OR Definition 2: Subjects who failed to respond to a full and adequate course (≥ 1 year) of β-IFN, having had ≥ 1 relapse in the previous year while on therapy, and; ≥ 9 T2 lesions in cranial MRI or ≥ 1 Gd-enhancing lesion, or having an unchanged or increased relapse rate in the prior year as compared to the previous 2 years Post hoc

It should be noted that, for each pivotal study, the sponsor had already performed their own subgroup analysis of subjects with high disease activity.

In Study 205MS201, the original analysis suggested that efficacy was similar in the high-disease-activity and low-disease-activity subgroups for most major endpoints including the primary endpoint, ARR, as well as radiological endpoints and the tertiary endpoint of disease progression. Unfortunately, given that this was a post hoc analysis, it lacks formal statistical validity. The EMA definition is also post hoc, but because this definition has been used for other products, it could potentially be considered at least partly independent of the results actually obtained with DAC HYP.

In Study 205MS301, the original definition of high disease activity appeared to be prospective. The original subgroup analysis of high disease activity and low disease activity suggested that, for the primary endpoint of ARR, superiority of DAC HYP over IFN β-1a was statistically significant in both subgroups. Similarly, for new T2 lesions, a similar and statistically significant benefit of DAC HYP relative to IFN β-1a was observed in both subgroups. For disease progression, a significant benefit with DAC HYP relative to IFN β-1a was not demonstrated for the cohort as a whole, or for any subgroup, including subgroups defined on the basis of disease activity.

The main difference between the original definition of high disease activity and the new definition proposed by the EMA is that the new definition allows subjects to be defined as having high disease activity on the basis of having failed treatment with beta interferons. Thus, it does not simply identify a group with high disease activity, but high activity and/or interferon resistance. If the results of subgroup analyses are different with this definition, then the difference is likely to be due to the addition of interferon-resistant subjects to the subgroup. The revised definition and the subsequent selective enrichment of this subgroup with interferon-resistant subjects would be expected to improve the apparent efficacy of DAC HYP relative to an interferon control therapy. As shown in the sponsor’s response, below, this is what was observed. Conversely, subjects with low disease activity did not show a significant benefit with DAC HYP relative to IFN β-1a.

The fact that the proposed definition was post hoc further undermines the statistical validity of the analysis. The p-values cited have been produced by the application of statistical tests that assume, incorrectly in this instance, that they will be applied as isolated tests of a well-defined, prospective hypothesis.

The sponsor’s response consisted of two and half pages of text, followed by several pages of tables. The text is reproduced in its entirety below, followed by a discussion that is limited to the underlined and bolded efficacy endpoints. The safety endpoints mentioned in the response are discussed separately, in the Safety section of this report but there was no evidence of a substantially different safety profile in subjects with high or low disease activity.

7.2.2.1.Sponsor’s response

The applicant redefined high disease activity as per the Tecfidera and Aubagio SmPCs. This modified definition added a second criterion to the definition used in the applicant’s primary analysis as shown below:

1. 
   Subjects with 2 or more relapses in 1 year, and with 1 or more Gd-enhancing lesions on brain MRI, or
   
2. 
   Subjects who failed to respond to a full and adequate course (at least 1 year of treatment) of beta-interferon, having had at least 1 relapse in the previous year while on therapy, and at least 9 T2-hyperintense lesions in cranial MRI or at least 1 Gd-enhancing lesion, or having an unchanged or increased relapse rate in the prior year as compared to the previous 2 years
   

Although the definition above matches the definitions in the Tecfidera SmPC and Aubagio SmPC, the analysis in the Aubagio SmPC was based on the first criterion only as no data were available for the second criterion.

Subjects who did not meet the criteria for high disease activity were classified in our analyses as having low/unknown disease activity.

To facilitate the assessment of benefit/risk based on this new definition of high disease activity, analyses were performed on the data from Study 205MS201 and Study 205MS301 for the following endpoints by baseline disease activity level:

Summary of AEs

Incidence of maximum values in liver function tests (Study 205MS301 only)

Annualised relapse rate (using INEC confirmed relapses)

Number of new or newly enlarging T2 lesions

6-month sustained disability progression

Study 205MS201

In Study 205MS201, the overall AE profile was similar for the subjects with high and low/unknown disease activity at baseline. The incidence of AEs and SAEs reported were also similar among subjects with high disease activity and low/unknown disease activity. Notably the incidence of AEs in the high and low disease activity subgroups of the total DAC HYP group was similar for events in the Infections and Infestations SOC (53% and 52%, respectively) and the Skin and Subcutaneous Tissue Disorders SOC (16% and 21%, respectively). The results of the analyses of annualised relapse rate and new or newly enlarging T2 lesions by baseline disease activity demonstrate the superiority of DAC HYP over placebo for both the high and low/unknown disease activity subgroups. The reductions in the annualised relapse rate in the DAC HYP 150 mg group relative to placebo were similar, with a 52% reduction (p = 0.0493) in the high disease activity group and a 54% reduction (p = 0.0003) in the low/unknown disease activity (see Table 17, below) In the analysis of new or newly enlarging T2 lesions (see Table 18, below), the reduction relative to placebo was greater in the high disease activity group (78%, p < 0.0001) than in the low/unknown disease activity group (66%, p < 0.0001).

In the analyses of disability progression (see Table 19, below), treatment with DAC HYP 150 mg was associated with a markedly lower rate of 6-month sustained progression compared to placebo in both the high disease activity group (hazard ratio = 0.23, p = 0.2034) and the low/unknown disease activity group (hazard ratio = 0.24, p = 0.0093).

Study 205MS301

As was the case in Study 205MS201, there were no notable imbalances in the safety data between the high and low/unknown disease activity groups in Study 205MS301. The incidence of SAEs was greater in subjects with high disease activity as compared to subjects with low disease activity in both treatment groups, suggesting the differences were associated with baseline disease severity and were not indicative of treatment-related differences. In the DAC HYP arm, the incidence of AEs was slightly higher in the high disease activity subgroup as compared to the low disease activity subgroup for the Infections and Infestations SOC (70% versus 62%) and the Skin and Subcutaneous Disorders SOC (41% versus 62%). However, a similar trend was also seen in the IFN β-1a group, which suggests the differences are primarily a function of greater disease severity in these subjects.

Maximum values for liver function tests were also similar in the high and low/unknown disease activity groups of Study 205MS301. Most subjects in both subgroups had maximum values that were between ≤ 3 x ULN. The incidence of maximum values ≥ 5 x ULN was low and similar between the disease activity subgroups and the DAC HYP and IFN β-1a arms. The results of the analyses of annualized relapse rate and new or newly enlarging T2 lesions by baseline disease activity demonstrate the superiority of DAC HYP over IFN β-1a for both the high and low/unknown disease activity subgroups, with highly significant p values (< 0.0001). For annualised relapse rate (see Table 20, below), the effect relative to IFN β-1a was greater in the high disease activity group (rate ratio 0.497: 95% CI 0.397 to 0.621) than in the low/unknown disease activity group (rate ratio = 0.614: 95% CI 0.490 to 0.770). For new or newly enlarging T2 lesions (see Table 21, below), the results by baseline activity were comparable (reductions of 53.7% and 52.3%, respectively, for high and low/unknown disease activity).

In Study 205MS301, there was a 43% reduction in 6-month sustained disability progression with DAC HYP compared to IFN β-1a in the high disease activity subgroup (hazard ratio = 0.57, p = 0.0102). No significant difference was evident between treatment groups in the low/unknown disease activity group (hazard ratio = 0.89, p = 0.5662) (see Table 22, below). The stronger treatment effect in the high disease activity subgroup may be due to a higher rate of disease progression in the IFN β-1a group, which provides more power to detect a treatment benefit. Conversely, the low rate of disease progression in the IFN β-1a arm provides less power to detect a treatment effect in the low disease activity subgroup. A similar pattern has been seen in other MS development programs in which a significant treatment benefit over IFN β-1a has been difficult to establish when there is a low progression rate. Nevertheless, the clearly superior findings of efficacy against disability progression compared to placebo in the low disease activity subgroup of Study 205MS201 provide evidence that DAC HYP does have a beneficial effect on disability progression in these subjects.

The results of these analyses demonstrate that the benefit/risk profile of DAC HYP remains favourable when high disease activity is redefined based on the Tecfidera/Aubagio SmPCs. The overall safety profile of DAC HYP is consistent in subjects with low and high disease activity at baseline in both studies. Likewise, DAC HYP provides a meaningful and consistent efficacy benefit over placebo and IFN β-1a whether measured in terms of relapses (annualized relapse rate), number of new/newly enlarging T2 lesions, or disability progression in subjects with both high and low disease activity at baseline. The differences between subgroups for some of the safety and efficacy results in both studies were generally observed in both the DAC HYP and control groups and were consistent with the greater level of disease activity at baseline.

The key tables mentioned in this response that are relevant to efficacy are reproduced below (with different numbering than in the sponsor’s original text). The evaluator’s interpretation of the significance of these results is included below the tables.

Table 17. Annualised relapse rate by disease activity and treatment, Study 205MS201

Table 18. New or newly enlarged T2 lesions by disease activity and treatment, Study 205MS201

Table 19. Time to sustained progression by disease activity and treatment, Study 205MS201

Table 20. Annualised relapse rate by disease activity and treatment, Study 205MS301

Table 21. New or newly enlarged T2 lesions by disease activity and treatment, Study 201MS301

Table 22. Time to sustained progression by disease activity and treatment, Study 205MS301

7.2.2.2.Evaluator’s conclusions

Most of the results in this supplementary analysis are concordant with the original analysis of the each study, and therefore do not provide any major new insights. To the extent that the results differ may be attributed to post hoc changes in the definition of progression (switching from 12 week confirmed progression to 6 month confirmed progression) as well as a post hoc revision of the definition of high disease activity (including subjects with proven poor response to the active comparator of Study 205MS301).

For Study 205MS201, a significant benefit of DAC HYP over placebo was demonstrated for ARR in the 150 mg dose group for subjects with high and low disease activity, but for the 300 mg dose group a significant benefit was only demonstrated for low disease activity. The hazard ratios were similar for subjects with high disease activity administered DAC HYP 300 mg, but the analysis was underpowered. For T2 lesions, all four combinations of dose and disease activity produced significant superiority of DAC HYP over placebo. For disease progression, a significant benefit was only observed for one of the four combinations (DAC HYP 150 mg and low disease activity), but the general trends were favourable across doses and subgroups. A pvalue was not calculated for the combination of DAC HYP 300 mg and high disease activity, because none of these subjects progressed. In general, the number of subjects with high disease activity was too low for robust statistical analysis., but the trends suggested that the relative benefit of DAC HYP over IFN β-1a was broadly similar across subgroups.

For Study 205MS301, this new subgroup analysis showed broadly similar benefits in both subgroups, when ARR and T2 lesion load were analysed (p < 0.0001 for each endpoint for each subgroup). This is broadly consistent with the original subgroup analysis. In both the original analysis and the supplementary analysis, the rate ratio for ARR was lower (more favourable) in the high-activity subgroup and higher (but still significantly in favour of DAC HYP) in the low-activity group.

Table 23. Table excerpt of ARR by baseline disease activity, Study 205MS301

The main discordant result in the new analysis was that a nominally significant treatment effect on disease progression was observed in high-activity/interferon-resistant subjects receiving DAC HYP, compared to those receiving IFN β-1a. Recall that, in the original presentation of the progression results in this study by the prospectively specified analysis method, there was no significant difference between the treatment groups: the hazard ratio for confirmed progression was 0.84 (DAC HYP/IFN β-1a), but the 95% CI included the possibility that progression was increased with DAC HYP (95% CI: 0.66 to 1.07). Furthermore, the hazard ratios in subgroups defined on the basis of high or low disease activity were very similar (0.80 and 0.83), as shown in the table excerpt below.

Table 24. Table excerpt of 3-month disability progression (increase in EDSS) by baseline disease activity, Study 205MS301

The post hoc shift of interferon-resistant subjects into the high-disease-activity subgroup improved the hazard ratio in this subgroup but worsened the hazard ratio for the low-disease-activity subgroup. This is not a particularly surprising result. To some extent, the hazard ratio was improved because a higher proportion of high-activity IFN β-1a subjects progressed when the subgroup was redefined to include a history of interferon resistance (24.0% progressed instead of 20.1%).

This discordant result has a weaker claim to validity than the original subgroup analysis, because it is post-hoc. After suitable adjustments for multiplicity, and with adequate appreciation of the post hoc nature of this analysis, these discordant results cannot be considered statistically robust despite nominally significant estimates for the unadjusted p-values. Furthermore, even if the finding were accepted as significant, this analysis relies on a specific definition of high activity that includes a poor response to interferon, and therefore merely suggests that DAC HYP may be more useful than beta interferon in subjects with demonstrated resistance to beta interferon. At best, this would be a weak result, and would represent a biased approach to a head-to-head comparison, focussing on subjects known to be resistant to the active comparator.

This supplementary analysis does not substantially alter the efficacy conclusions. The benefit of DAC HYP over once-weekly IFN β-1a is primarily evident in relapse rate and radiological endpoints. The evidence that DAC HYP reduces relapses in both high- and low-activity subgroups is robust, and suggests a clinically worthwhile effect. The endpoint of progression showed a trend in favour of DAC HYP in both studies, and it is possible to strengthen this finding with a number of post hoc approaches, including the enrichment of one subgroup with subjects known to be resistant to the active comparator. The placebo-controlled data failed to show a statistically significant benefit on progression, but a nominally significant p-value was obtained (if issues of multiplicity of dose groups are ignored). The cumulative effect of multiple marginal results suggests that DAC HYP is likely to have a clinically worthwhile benefit on progression, but this is yet to be confirmed in a rigorous manner. This new post hoc analysis does not substantially alter that conclusion, but merely suggests that efficacy relative to IFN β-1a may be better if subjects have already demonstrated that they respond poorly to IFN β-1a. If confirmed in a suitable prospective study, this would in turn suggest that DAC HYP could be useful as second-line or rescue therapy, which is likely to be the therapeutic role it occupies anyway.

7.2.3.EMA question 95

‘The statistical and clinical significance of patient reported outcomes (PRO) and INEC confirmed disability progression are not robust. Please compare to current available treatments, considering both patients with high versus low disease activity.’

This is an important question, asking the sponsor to place the progression results in the broader context of current available treatments, but it should be noted that the subdivision of patients into ‘high versus low disease activity’ is not a simple exercise. As discussed in the discussion of EMA Question 94 above, one of the definitions of high disease activity included in the supplementary data was based on beta interferon treatment failure (implying interferon resistance) in addition to other measures of disease activity.

7.2.3.1.Sponsor’s response

The sponsor begins their response with a statement that does not appear fully supported by the evidence:

‘All measures of disease progression assessed in the DAC HYP clinical trials provide clinically meaningful evidence that DAC HYP-treated subjects experience less disease progression than subjects treated with placebo or IFN β-1a. Although some of the trends do not reach statistical significance, the consistency of the results provides reassurance that they are not due to chance.’

A reduction in disease progression can only be ‘clinically meaningful’ if it actually exists, and in Study 205MS301 the 95% CI for the difference in disease progression was consistent with their being zero difference between DAC HYP and IFN β-1a, or even a slight inferiority of DAC HYP versus IFN β-1a. It is not the case that a significant benefit was shown but a few trends failed to reach significance, as implied in the sentences quoted above: no significant benefit was demonstrated, but a few analysis methods create nominally significant p-values if issues of multiplicity are ignored.

The sponsor’s appeal to ‘consistency of the results’ has partial merit. A series of independent trends can sometimes, in aggregate, provide strong clues of a non-random effect, even though they are individually not significant. Furthermore, there was a broad consistency between the pivotal studies, with DAC HYP showing a strong trend to superiority over placebo at both doses (with nominally significant p-values at one dose) and a weaker trend over IFN β-1a.

Despite these consistencies across studies, the sponsor’s expression ‘reassurance that they are not due to chance’ overstates the strength of the evidence. The question of whether chance alone could give rise to an observation can be partially quantified, and indeed this is the whole point of statistical hypothesis testing.

The sponsor’s response then mentions the MSIS-29 scale, and proceeds to provide evidence of benefit with DAC HYP on MSIS-29 scores. The MSIS-29 scale is not a measure of disability progression and is not directly relevant to the question being addressed but the sponsor argues that it correlates with disease progression:

‘The endpoints of the MSIS-29 physical scale (as a patient-reported outcome (PRO)) and disability progression by EDSS were assessed by subjects’ baseline level of disease activity and support the overall conclusion that DAC HYP 150 mg offers a consistent advantage over both placebo and active comparators… The MSIS-29 physical scale has been shown to be a valid and sensitive PRO measure, which correlates highly with the EDSS (r = 0.704) and the multiple sclerosis functional composite (MSFC) (r = 0.577).’

The link between MSIS-29 scores and disability progression is not sufficiently tight that one measure can be used as a surrogate for another and the fact that the two measures have been shown to be correlated in subjects not on DAC HYP does not allow robust inferences to be made about whether DAC HYP might differentially modify each measure.

The MSIS-29 results in Study 205MS301 have already been discussed in the description of that study, where it was noted that a significant benefit was achieved relative to IFN β-1a, but it was unclear to what extent this reflected tolerability issues with IFN β-1a. The sponsor’s new comments on the MSIS-29 scores in their response to EMA Question 95 are reproduced below. Overall, there was a benefit demonstrated for this measure, and it appears plausible that this was, in part, due to benefits in disease progression. If the results for disease progression had been positive in Study 205MS301, the MSIS-29 results would have given the positive progression results some extra plausibility; they do not, however, overturn the fact that negative results were obtained in the direct analysis of disease progression.

‘In Study 205MS201, DAC HYP 150 mg resulted in a significant decrease in the proportion of subjects with a clinically meaningful worsening (defined as an increase of ≥ 7.5 points) on the MSIS-29 physical scale compared to placebo (odds ratio = 0.56 (95% CI: 0.35 to 0.88; p = 0.0125]). Among the subjects with high disease activity at baseline in Study 201, the proportion of subjects who experienced a clinically meaningful decline on the MSIS-29 physical scale was higher in the placebo group (38%) than in the DAC HYP 150 mg group (22%, p = 0.1104). Amongst subjects with low disease activity at baseline, a similar benefit was seen for the DAC HYP 150 mg arm: 30% of subjects in the placebo group experienced worsening compared with 20% of subjects in the DAC HYP 150 mg group (p = 0.0462) (Table 25, below).

Similarly, in Study 205MS301, DAC HYP 150 mg prevented clinically meaningful (≥ 7.5 point increase) decline on the MSIS-29 physical score, a predefined secondary study endpoint, compared with IFN β-1a (odds ratio = 0.76 (95% CI: 0.60 to 0.95; p = 0.0176)). Among subjects with high disease activity at baseline, a greater proportion of subjects in the IFN β-1a group experienced worsening as compared with the DAC HYP group (25% versus 19%, odds ratio = 0.69, p = 0.0409). The trend in subjects with low disease activity was consistent with that of the high disease activity subgroup, but the difference between the treatments was less pronounced (22% versus 19%; odds ratio = 0.80; p = 0.1534) (Table 26, below)

The remainder of the sponsor’s response addressed the EMA’s question more directly, by considering the progression results of the two pivotal studies alongside studies with other disease-modifying MS treatments:

‘Slowing of disability progression remains one of the most important goals of MS therapy. Although numerous agents have been shown to have a clinically meaningful and statistically significant improvement over placebo in controlled trials (Tables 25 and 26, below) no approved agent has demonstrated a consistent benefit over an active comparator agent (Tables 27 and 28, below).’

This is a fair statement and it describes the current literature reasonably well. The tables provided with the response are reproduced below (an error in the first table, in which placebo and DAC HYP results were transposed, has been fixed by the evaluator). The placebo-controlled results for DAC HYP compare favourably with other active treatments in their respective placebo-controlled studies: DAC HYP at the proposed dose was associated with the most favourable hazard ratio in the table (0.43), and the nominal p-value (p = 0.0211) compares favourably with most of the other disease-modifying agents. The placebo-controlled results for ‘6-Month’ (24-Week) confirmed progression were even stronger, with a low hazard ratio (0.24), and the nominal p-value appears to show a high degree of significance (p = 0.0037). Against this, it should be noted that the 150 mg dose group was the secondary dose group in Study 205MS201, and a hierarchical testing procedure was used to control for multiplicity. By this procedure, no significance was demonstrated for the 150 mg dose group, and the cited pvalues should be considered spurious. On the other hand, if the nominal p-values for 150 mg were doubled, to approximate the effects of assessing two doses in the same study, they would remain nominally significant. The results for 24-week confirmed progression have not been corrected for multiplicity, but doubling to account for multiplicity of doses (150 mg and 300 mg) and doubling again to account for multiplicity of confirmation times (12 and 24 weeks) would produce a nominal p-value of 0.015, well below the traditional cut-off of 0.05.

The results for the active-controlled studies show that a significant benefit versus an active comparator has not often been achieved. Fingolimod failed to show a significant benefit for 3month confirmed disability against IFN β-1a, and alemtuzumab showed a significant benefit over IFN β-1a (44mcg three times a week, not 30mcg weekly) in one study but not another. In this respect, the results for DAC HYP are no worse than for other agents against active comparators: DAC HYP failed to show a significant benefit for the main prospective endpoint of 12 week confirmed progression, but it did show a significant benefit for the supportive, tertiary endpoint of 24 week confirmed progression.

In their response to this question, the sponsor also included results based on the subgroup analysis already discussed in relation to EMA Question 94, in which high disease activity was redefined to involve, not just high disease activity, but treatment failure on beta interferon. This post hoc analysis produces nominal hazard ratios and p-values that compare even more favourably to other disease-modifying agents, but the validity of this post hoc approach is questionable for the reasons discussed earlier.

Table 25. Impact on 3-month (12-week) confirmed disability progression of MS therapies in placebo controlled trials

Note active and placebo values were switched in Sponsor’s original table for DAC HYP.

Table 26. Impact on 6-month confirmed disability progression of MS therapies in placebo controlled trials

Table 27. Impact on 3-month confirmed disability progression of MS therapies in active controlled trials

Table 28. Impact on 6-month confirmed disability progression of MS therapies in active controlled trials

7.2.3.2.Evaluator’s conclusion

The sponsor’s response to this question confirms that it has generally been difficult to show that disease-modifying agents have a substantial benefit in reducing disease progression, relative to active controls. The evidence that DAC HYP reduced progression is inconclusive, from a statistical perspective, but the absolute magnitude of the observed trends is favourable when compared to other new agents.

7.3.Other efficacy studies

The original CER described a number of minor studies, including extension studies (Studies 205MS202, 205MS203, 205MS303). Only Study 205MS202 was blinded. These studies shared a problem with most extension studies, in that they are subject to a number of biases due to prior exposure to study medication, non-random entry into the extension cohort, and incomplete follow-up.

Overall, these minor studies were broadly consistent with the pivotal studies, but lacked the statistical power and methodological rigour required to modify conclusions drawn directly from the pivotal studies alone. The studies were broadly reassuring in that there was no obvious waning of efficacy with continued treatment. The original CER should be consulted for details.

7.4.Efficacy comparisons with other disease-modifying agents

The submitted evidence suggests that DAC HYP has superior efficacy in comparison to once-weekly IFN β-1a, and broadly similar efficacy to other new disease-modifying agents.

In the pivotal placebo-controlled study, Study 205MS201, the observed reduction in relapse rate (approximately 50 to 54%, depending on which dose group is considered) resembled the reported reductions in ARR observed in other recent pivotal placebo-controlled studies of different disease-modifying agents, including fingolimod and dimethyl fumarate. For the pivotal placebo-controlled fingolimod trial, the ARR was 0.18 in the active group, compared to 0.40 in the placebo group, a relative reduction 55% (p < 0.001, approved PI for fingolimod). In the pivotal study of dimethyl fumarate, the reduction in ARR was also similar: ‘The annualised relapse rate at 2 years was 0.17 in the twice-daily BG-12 group and 0.19 in the thrice-daily BG12 group, as compared with 0.36 in the placebo group, representing relative reductions of 53% and 48% with the two BG-12 regimens, respectively (p < 0.001 for the comparison of each BG-12 regimen with placebo).’⁵

In a direct comparison of DAC HYP with once-weekly IFN β-1a, in Study 205MS301, the adjusted annualized relapse rate in the IFN β-1a group was 0.393 relapses/year, compared to a rate of 0.216 relapses/year in the DAC HYP group (95% CIs: 0.353 to 0.438 in the IFN β-1a treatment group and 0.191 to 0.244 in the DAC HYP treatment group). These results correspond to a relative reduction of 45% in ARR (p < 0.0001) with DAC HYP, compared to IFN β-1a.

DAC HYP was not shown to have a significant effect on disease progression in either of the pivotal studies, but there were favourable trends in both studies and the observed hazard ratios were broadly consistent with other disease modifying agents, as discussed.

7.5.Evaluator’s overall conclusions on clinical efficacy

The sponsor provided a summary table of the key results of the two pivotal studies, and this table is reproduced below. It should be noted that the p-values flagged in the table as ‘nominal’ should not be considered statistically significant, indeed by a strict application of the closed testing procedure, these values should not even have been calculated or reported. Also, the cited p-values do not include any correction for multiplicity. In particular, despite the nominal pvalue of 0.0211 cited for sustained disability progression in Study 205MS201, a significant benefit on progression cannot be inferred.

The table also uses relative risk estimates that have been inflated by the practice of using instantaneous hazard ratios to estimate relative risk, as discussed previously in this report (see Sections: Results of Efficacy Outcomes for both pivotal studies).

Finally, the benefits of DAC HYP have only been demonstrated in subjects with RRMS who satisfied the entry criteria for the two pivotal studies. Extrapolation to a broader population is not warranted.

With these limitations in mind, the evaluator concludes that the following efficacy benefits are supported by the evidence:

DAC HYP at a dose of 150 mg or 300 mg SC 4-weekly reduced annualised relapse rate by 50 to 54%, relative to placebo (p ≤ 0.0002)

DAC HYP at the proposed dose of 150 mg SC 4-weekly reduced relapse rate by 45%, relative to once-weekly IFN β-1a (p < 0.0001)

DAC HYP at the proposed dose reduced the proportion of subjects relapsing by 44 to 47%, relative to placebo, and by 34 to 35%, relative to IFN β-1a, depending on the duration of follow-up. (Note: this is less benefit than claimed by the sponsor, for calculations see Sections: Results of Efficacy Outcomes for both pivotal studies)

DAC HYP at the proposed dose reduced the number of new Gd-enhancing lesions by 69 to 78%, relative to placebo (p < 0.0001)

DAC HYP at the proposed dose reduced the number of new or newly enlarging T2 lesions by 70 to 79% relative to placebo (p < 0.0001), and by 54% relative to IFN β-1a (p < 0.0001)

Compared to placebo, DAC HYP showed a trend to benefits in quality of life at the proposed dose, as estimated by the MSIS-29 physical impact score, but by the closed testing procedure failed to achieve significance, and trends were inconsistent across dose groups

Compared to beta interferon, DAC HYP showed significant superiority in quality of life, as estimated by the MSIS-29 physical impact score

DAC HYP is associated with a strong trend to reduced disability progression

DAC HYP produced a broadly similar benefit across all major subgroups in the study population

DAC HYP at the proposed dose has better efficacy, relative to IFN β-1a, in a population enriched for subjects with proven resistance to IFN β-1a, and in this population a nominally significant post hoc p-value can be obtained for the endpoint of progression

DAC HYP at the proposed dose has a broadly similar efficacy to other new disease-modifying agents

DAC HYP has not been studied in subjects with overt SPMS, and its efficacy in this population is unknown

Despite the fact that the supplementary evaluator and the sponsor have drawn different conclusions about the statistical robustness of the progression data, these efficacy results are considered satisfactory. The supplementary evaluator does not believe that a clear benefit on progression endpoints should be an absolute requirement for a new disease-modifying agent in MS. In subjects with RRMS, a large proportion of disability progression is due to damage sustained during relapses, and preventing relapses is a worthwhile achievement in its own right, provided that there is at least no adverse effect on progression. Although the data do not provide robust confirmation of a benefit for progression endpoints, there is a consistency across multiple different analyses that, in aggregate, strongly suggest that DAC HYP has a favourable effect on progression, and at least DAC HYP appears highly unlikely to have an adverse effect on progression. Coupled with strong evidence of a reduced relapse rate, this is sufficient to support the claim of efficacy in RRMS.

The efficacy of DAC HYP in subjects with SPMS has not been characterised, and there is currently no basis for recommending this treatment in subjects with SPMS.

The lowest dose of DAC HYP capable of producing a substantial reduction in relapse rate has not been established.

Table 29. Primary, secondary, and selected tertiary efficacy endpoints in the DAC HYP pivotal studies 205MS201 and 205MS301, DAC HYP 150 mg

Note: the evaluator does not agree with all of the figures of this table – see text for details)