Edoxaban for preventing stroke and systemic embolism in people with non-valvular atrial fibrillation

4.2 The Committee considered the clinical‑effectiveness data from ENGAGE AF‑TIMI 48, that compared edoxaban with warfarin. It considered that this trial was of good quality and discussed whether the results were generalisable to people with atrial fibrillation in the UK. The Committee noted that ENGAGE AF‑TIMI 48, like other trials of newer anticoagulants, used CHADS₂ to assess the risk of stroke rather than CHADS₂‑VASc, which is now used in clinical practice, as recommended in the NICE guideline on managing atrial fibrillation. The Committee understood from the clinical expert that the CHADS₂‑VASc scoring system was developed to better define those who would benefit from anticoagulation because a number people with a CHADS₂ score of 1 would still benefit. It also heard that although these people were not included in ENGAGE AF‑TIMI 48, a lower baseline risk of stroke would not be expected to reduce the relative efficacy of the treatment. In clinical practice, edoxaban is expected to be offered in the same place in the treatment pathway as other anticoagulants (that is, to women with a CHADS₂‑VASc score of 2 and above, and to men with a score of 1 or above), while taking bleeding risk into account. The Committee concluded that the trial was well designed and generalisable to clinical practice.

4.3 The Committee considered the results of ENGAGE AF‑TIMI 48. It noted that the primary efficacy outcome was a composite of stroke (both ischaemic and haemorrhagic) and systemic embolism. However, ischaemic stroke and systemic embolism could be considered direct treatment effects, whereas haemorrhagic stroke was a bleeding outcome and therefore an adverse event. The Committee noted that for the composite primary outcome, edoxaban was non‑inferior to, but not superior to, well‑controlled warfarin (which was defined in the trial as a median time in therapeutic range [TTR] of 68.4%). The Committee noted that when the individual components of the primary outcome were considered separately, there was only a statistically significant reduction in haemorrhagic stroke with edoxaban compared with warfarin. The Committee concluded that edoxaban was as clinically effective as warfarin for the primary efficacy outcome of reducing stroke (ischaemic and haemorrhagic) and systemic embolism, and had nearly half the rate of haemorrhagic stroke events compared to warfarin.

4.4 The Committee considered the results of the company's subgroup analyses, which used data from ENGAGE AF‑TIMI 48. It noted that the company presented data for subgroups based on international normalised ratio (INR) control, that compared the efficacy of edoxaban and warfarin in relation to the median TTR for the study centre. One of the analyses showed that the relative benefits of edoxaban compared with warfarin were greater in centres where the centre‑level TTR was less than 60%. The Committee noted comments from the company and the Evidence Review Group (ERG) that this was not consistent across all analysis sets. The Committee concluded that there was insufficient evidence to consider different treatment effects according to centre‑level TTR.

4.5 The Committee noted that the company's subgroup analyses for risk of stroke (as defined by CHADS₂ score) showed that the hazard ratio for edoxaban compared with warfarin was stable and non‑inferior across CHADS₂ scores of 2 to 6. The Committee concluded that there was no biologically plausible reason to indicate that the relative treatment effect would be dependent on the baseline risk of stroke.

4.6 The Committee discussed the subgroup analysis based on renal function, which used 3 categories of creatinine clearance (normal renal function, and mild or moderate impairment). It noted that the results of this analysis suggested a trend towards decreasing efficacy of edoxaban with increasing creatinine clearance (see section 3.6). The Committee heard from a clinical expert that this was likely to be because with better renal function edoxaban is removed by the kidneys more quickly, leading to a reduction in treatment effect. It also heard that this may apply to all newer oral anticoagulants, but data need to be re‑evaluated to confirm this. It heard from the clinical experts that the proportion of people with good renal function (measured by creatinine clearance) who would be eligible for treatment with edoxaban was in the region of 5% to 10%, and that these are often younger people. The Committee noted the company's rationale that the results of this sub‑group analysis should be interpreted with caution (see section 3.6). It also noted the summary of product characteristics which states that, in people with non‑valvular atrial fibrillation and high creatinine clearance, edoxaban should only be used after careful evaluation of a person's thromboembolic and bleeding risk. The Committee concluded that if edoxaban is used in accordance with the summary of product characteristics, there is no reason to make differential recommendations based on creatinine clearance.

4.7 The Committee considered the adverse events reported in ENGAGE AF‑TIMI 48. It noted that for the primary safety outcome of major bleeding, edoxaban resulted in statistically significantly fewer bleeds than warfarin. Edoxaban also had statistically significantly fewer other bleeding events including fatal, intracranial and clinically relevant non‑major bleeds. The Committee recognised the particular importance of the reduction in intracranial bleeding compared with warfarin. It also noted the statistically significantly higher numbers of gastrointestinal bleeds in people treated with edoxaban compared with warfarin. The Committee was aware that this is not unique to edoxaban, and that clinicians are now more experienced in using the newer oral anticoagulants and in managing the adverse events. It also heard from the clinical experts that administration of 4‑factor prothrombin complex concentrate has been shown to reverse the effects of edoxaban. The Committee concluded that the risk–benefit profile of edoxaban was acceptable.

4.8 The Committee discussed the data for edoxaban compared with rivaroxaban, apixaban, dabigatran etexilate (110 mg twice daily and 150 mg twice daily) and rivaroxaban, that were used in the company's network meta‑analysis. The Committee noted that the trials included in the network meta‑analysis were not directly comparable; for example, they had different baseline risks of stroke (with different CHADS₂ inclusion criteria and mean CHADS₂ scores) and differences in time in the therapeutic range in the warfarin groups. The Committee also noted the ERG's concerns about the violation of the proportional hazards assumption in data from ENGAGE AF‑TIMI 48, from the trials of the other 3 newer oral anticoagulants, and in the warfarin groups of the 4 trials included in the network meta‑analysis. It understood from the ERG that this meant that the hazard ratios produced by the network meta‑analysis were not sufficiently robust to compare the relative clinical effectiveness of the newer oral anticoagulants. The Committee considered the results of the network meta‑analysis in the light of the methodological issues and noted that all the newer oral anticoagulants appeared to have comparable efficacy for the composite primary and bleeding outcomes. The Committee concluded that the network meta‑analysis results should be interpreted with caution, but edoxaban is unlikely to be different from rivaroxaban, apixaban and dabigatran etexilate in clinical practice.

4.9 The Committee considered the company's economic model. It noted that the economic analysis was largely based on the model used in NICE's technology appraisal guidance on apixaban for preventing stroke and systemic embolism in people with non-valvular atrial fibrillation, which captured the main efficacy and adverse events of treatment. The Committee agreed that the model structure, perspective and time horizon were appropriate, although it questioned the relevance of the inclusion of myocardial infarction. It concluded that the analysis was consistent with the NICE reference case.

4.10 The Committee considered the clinical‑effectiveness estimates used in the company's model. It noted that the comparison of edoxaban with warfarin used direct evidence from ENGAGE AF‑TIMI 48 to inform the company's economic model. The Committee was aware of the ERG's concern that the assumption of proportional hazards for edoxaban and warfarin for haemorrhagic stroke (one of the components of the primary outcome) appeared to be violated in ENGAGE AF‑TIMI 48. However, the Committee considered that the general modelling approach and the pairwise comparison with warfarin were appropriate. The Committee noted that for the comparison of edoxaban with the other newer oral anticoagulants, hazard ratios obtained from the network meta‑analysis were used in the economic model and that these estimates were considered unreliable by the ERG (see section 4.8). The Committee concluded that data from ENGAGE AF‑TIMI 48 were appropriate for calculating the cost effectiveness of edoxaban compared with warfarin, but the estimates of the cost effectiveness of edoxaban compared with dabigatran etexilate, apixaban and rivaroxaban were based on data that were associated with a high degree of uncertainty.

4.11 The Committee heard from the ERG that there were differences in the utility values used in the economic model, compared with other NICE technology appraisals for atrial fibrillation (apixaban for preventing stroke and systemic embolism in people with non-valvular atrial fibrillation; rivaroxaban for preventing stroke and systemic embolism in people with non-valvular atrial fibrillation; dabigatran for preventing stroke and systemic embolism in people with non-valvular atrial fibrillation). It noted that even though EQ‑5D data were collected at baseline in ENGAGE AF‑TIMI 48 , the baseline utility value for stable atrial fibrillation used in the model was from another small UK study. The Committee noted that the ERG had identified a number of inconsistencies and had raised concerns about some of the sources of data used in the company model. However, the Committee noted that when the ERG's suggested revisions (alternative utility estimates for systemic embolism, myocardial infarction, and transient ischaemic attack) were applied, together with an amended age‑adjusted utility decrement per year of -0.00646 instead of -0.00029 (see sections 3.40 and 3.41), they had only a minor impact on the incremental cost‑effectiveness ratio (ICER). The Committee concluded that the utility values used in the model, although open to debate, were not key drivers of the cost effectiveness.

4.12 The Committee considered the costs used in the company's model. It noted that costs for ischaemic stroke, haemorrhagic stroke, and systemic embolism were based on the Oxford Vascular Study (a cohort study of a UK population) and the costs were similar to those used in other NICE technology appraisals for atrial fibrillation (apixaban for preventing stroke and systemic embolism in people with non-valvular atrial fibrillation; rivaroxaban for preventing stroke and systemic embolism in people with non-valvular atrial fibrillation; dabigatran for preventing stroke and systemic embolism in people with non-valvular atrial fibrillation). The Committee also noted that an INR monitoring cost of £265 was used by the company, and that this fell within a range previously accepted in NICE technology appraisals. The Committee concluded that the costs used in the model were appropriate.

4.13 The Committee considered the cost effectiveness of edoxaban compared with warfarin. It noted that the company's base‑case deterministic and probabilistic ICERs for edoxaban compared with warfarin were £12,900 and £16,900 per QALY gained respectively. The Committee noted that the ERG considered the economic model to be robust to all of the company's sensitivity analyses, and to most of those done by the ERG. The Committee further considered the ERG's exploratory analyses. It noted that the change which had the largest single impact on the ICER was applying the hazard ratio from ENGAGE AF‑TIMI 48 for haemorrhagic stroke (which increased the ICER to £17,100 per QALY gained). The Committee noted that the inclusion of all the ERG's preferred values in the model (see sections 3.40 and 3.41) resulted in a deterministic ICER of £16,000 per QALY gained and a probabilistic ICER of £22,100 per QALY gained. The Committee concluded that taking all of the analyses into account, edoxaban was cost effective compared with warfarin and could be recommended as an alternative to warfarin for preventing stroke and systemic embolism in people with non‑valvular atrial fibrillation who have 1 or more risk factors for stroke.

4.14 The Committee noted that the cost effectiveness of edoxaban compared with other newer oral anticoagulants was calculated using hazard ratios from the network meta‑analysis, which the Committee considered to lack robustness (see section 4.8). In the full incremental analysis edoxaban, dabigatran etexilate 110 mg, apixaban and rivaroxaban were strictly dominated by dabigatran etexilate 150 mg, which had an ICER of £7645 per additional QALY gained compared to warfarin. However, there were very small differences in QALYs and costs between the newer oral anticoagulants. The Committee concluded that there was insufficient evidence to distinguish between the clinical and cost effectiveness of edoxaban and the newer oral anticoagulants recommended in previous appraisals (apixaban, dabigatran etexilate and rivaroxaban). Therefore, edoxaban could be recommended as a cost‑effective treatment for non‑valvular atrial fibrillation in people who have 1 or more risk factors for stroke.

4.15 The Committee concluded that the decision about whether to start treatment with edoxaban should be made after an informed discussion between the clinician and the person about the risks and benefits of edoxaban compared with warfarin, apixaban, dabigatran etexilate and rivaroxaban. For people considering switching from warfarin to edoxaban, the potential risks and benefits of edoxaban should be considered in the light of their level of international normalised ratio (INR) control.

4.16 The Committee was aware of NICE's position statement with regard to the Pharmaceutical Price Regulation Scheme (PPRS) 2014, and in particular the PPRS payment mechanism, when appraising edoxaban. It accepted the conclusion 'that the 2014 PPRS payment mechanism should not, as a matter of course, be regarded as a relevant consideration in its assessment of the cost effectiveness of branded medicines'. The Committee heard nothing to suggest that there is any basis for taking a different view with regard to the relevance of the PPRS to this appraisal of everolimus. It therefore concluded that the PPRS payment mechanism was irrelevant for the consideration of the cost effectiveness of edoxaban.