Edoxaban for treating and for preventing deep vein thrombosis and pulmonary embolism

The committee considered the Hokusai‑VTE trial. It noted that it had an unconventional design in a number of ways; for example, patients could change dosage during the trial, there was a flexible treatment duration, and the primary efficacy outcome was measured at 12 months irrespective of the time when treatment stopped (which could have been as early as 3 months). The committee noted that the trial therefore differed from VTE trials for other anticoagulants. The clinical expert agreed that it was unusual that the primary outcome was measured in patients who had potentially been untreated for up to 9 months (although the committee noted that 60% of patients in the trial had treatment for at least 6 months). However, the company stated that this approach accurately reflected real‑world clinical practice, and led to a more conservative estimate of efficacy, with which the clinical expert agreed. The committee was aware that the design of the trial was not strictly comparable with other recent trials for newer anticoagulants, and that the value of assessing efficacy several months after some patients had stopped treatment was questionable. However, the trial did allow a choice of dosage, and did not predetermine the length of treatment at the start, both of which mirrored clinical practice. The committee concluded that Hokusai‑VTE was well designed and suitable for evaluating the clinical effectiveness of edoxaban.

The committee considered the baseline characteristics of the patients in the trial. It noted the ERG's concerns about generalisability (section 3.14). However, it heard from the clinical and patient experts that they had no concerns about the generalisability of the trial, including the age of participants, with 1 expert explaining that people can experience VTE at any age. The committee also noted that the patient population was comparable to that in other trials for newer anticoagulants. The committee concluded that the results of the clinical trial were generalisable to people with VTE in the NHS and were appropriate for decision making on the clinical effectiveness of edoxaban.

The committee discussed the clinical efficacy results from Hokusai‑VTE. It noted that the trial had shown a higher rate of VTE recurrence in the edoxaban group compared with warfarin in the first 30 days, but agreed that caution was needed in interpreting this difference because the numbers were small and the difference was not statistically significant. The committee concluded that overall the trial had demonstrated that edoxaban was statistically non‑inferior to warfarin for VTE recurrence.

The committee considered the network meta‑analysis presented by the company. The committee noted the wide credible intervals and the non‑inferiority design of the trials in the network. It further noted that the unconventional design of Hokusai‑VTE compared with the other trials in the network had led to heterogeneity. The committee concluded that no clear differences between treatments for any outcome had been demonstrated, but that the comparative evidence was weak (because of the lack of direct evidence comparing the newer anticoagulants, and the issues noted with the network meta‑analysis) and therefore the results needed to be interpreted with caution.

The committee noted that a small number of people with cancer had been included in the trial but that no subgroup analysis results had been presented for these patients. The committee heard from the clinical expert that the standard treatment for VTE in people with cancer is low‑molecular‑weight heparin. Hokusai‑VTE did not include this as a comparator group, so there was no comparison of edoxaban with the current standard of care for these patients. The committee concluded that the trial did not provide relevant data for people with cancer who experienced VTE, and it was unable to make any specific recommendation for this subgroup of patients.

The committee discussed the adverse events associated with edoxaban. It noted that there were 18 primary intracranial haemorrhages in the warfarin group (of which 6 were fatal) and 5 in the edoxaban group, none of which were fatal (section 3.9). The committee was aware that intracranial haemorrhage is considered to be the single most serious complication of anticoagulation treatment. It heard from the clinical expert that the reduced risk of intracranial haemorrhage is recognised as a benefit of newer oral anticoagulants in general and there is also the suggestion that, when intracranial haemorrhage does occur, the bleed may be less extensive than with warfarin. However, this benefit needs to be balanced against an increased risk of gastrointestinal bleeding as suggested in the trial; there were 9 more gastrointestinal bleeds in the edoxaban group than in the warfarin group (section 3.9). The committee heard concerns from the ERG that although Hokusai‑VTE demonstrated a statistically significant reduction in bleeding (major and clinically relevant non‑major bleeding), the reduction for major bleeding alone was not statistically significant. The committee concluded that there was no statistically significant difference for major bleeding for edoxaban compared with warfarin, but that a reduction in intracranial haemorrhage with edoxaban was a potential substantial benefit.

The committee considered the structure of the company's health economic model and the assumptions used. The ERG had raised a number of parameter and structural concerns about the model. The committee generally agreed with these concerns, particularly:

• The assumption that some adverse events such as chronic thromboembolic pulmonary hypertension and post‑thrombotic syndrome were treatment‑related rather than disease‑related.

• The exclusion of treatment‑switching after VTE recurrence (the committee heard from the clinical expert that returning to a treatment that had failed to prevent recurrence was not clinically plausible).

• The inclusion of both ischaemic and haemorrhagic stroke within the stroke health state. Ischaemic stroke is not treatment‑related whereas intracranial haemorrhage is; putting them together was an oversimplification and not clinically plausible. It also resulted in potential double‑counting of intracranial haemorrhage because this was also included in the major bleeding health state.
  
  Despite these and other concerns about the reliability of the model, the committee heard from the ERG that flaws in the model, although important methodologically, did not have a substantial impact on the cost‑effectiveness results, because they affected both arms equally, or because the probabilities of the associated adverse events were low. The committee concluded that some of the assumptions in the model and model structure lacked clinical plausibility but, taking into account the ERG's comments and analyses, these flaws were not key drivers of cost‑effectiveness.

The committee considered the clinical‑effectiveness estimates that had been used in the company's economic model, noting that the estimates were from a network meta‑analysis which had limited capacity to demonstrate statistically significant differences between treatments. The committee concluded that the efficacy data from the network meta‑analysis contributed to uncertainty in the cost effectiveness results.

The committee considered the warfarin monitoring‑cost assumptions used in the company model. The committee noted that the company had assumed an annual cost of warfarin monitoring of approximately £630. The committee discussed whether this cost was reasonable. It heard from the ERG that according to their clinical experts the company had overestimated the costs of monitoring – the frequency was too high, and the monitoring would usually be done by a nurse rather than a consultant as assumed in the company base case. After reducing the frequency of visits and assuming monitoring would be nurse‑led, the ERG had used a monitoring cost of £342 in the first year and £190 after that. The committee heard from the company that it had access to unpublished preliminary registry data suggesting that monitoring costs in clinical practice were substantially higher than the estimates used by the ERG. The committee was aware that there is considerable variation in how warfarin monitoring is provided in the NHS, which would affect the costs, and that there is also individual variation (some patients have INR levels which are more unstable than others and need more frequent monitoring). The exact frequency and cost of warfarin monitoring is therefore unknown. The committee referred to previous appraisals in its consideration of the very different estimates from the company and ERG. It noted that the cost assumed by the company was substantially higher than the range considered plausible in previous appraisals for VTE (£304 to £379). The committee concluded that the company estimates were higher than had previously been accepted as plausible and the ERG estimates for the first year were closer to those previously accepted. However, the precise costs of warfarin monitoring remained uncertain.

The committee considered the cost‑effectiveness estimates generated by the company and ERG. It noted that the warfarin monitoring cost was the main driver of the cost‑effectiveness estimates. The committee agreed that the cost of monitoring in the first year in the ERG base case was more consistent with previous appraisals than the company's estimate. Therefore, it considered that the ICER relative to warfarin was likely to be closer to the ERG estimate of £26,000 per QALY gained than the company estimate of approximately £2,500 per QALY gained. Nevertheless, it considered that both ICERs were subject to high levels of uncertainty because of the previously discussed flaws in the company model on which they were based, and the lack of definitive warfarin monitoring costs in the NHS. Noting these uncertainties, the committee further considered the cost effectiveness of edoxaban compared with other anticoagulants that had been considered in previous appraisals for the treatment of VTE. The committee accepted that the clinical effectiveness of edoxaban had been adequately demonstrated by the clinical trial. It also noted that the price of edoxaban was similar to one of the other agents, rivaroxaban. Taking into account the lack of any clear trial evidence that edoxaban was substantially different from the other newer oral anticoagulants, and the testimony of the experts, the committee concluded that the most plausible ICER was likely to be in line with that of the other oral anticoagulants already recommended in previous NICE guidance for the treatment of VTE. The committee therefore concluded that edoxaban could be recommended as a cost‑effective use of NHS resources.

The committee considered whether it should take into account the consequences of the Pharmaceutical Price Regulation Scheme (PPRS) 2014, and in particular the PPRS Payment Mechanism, when appraising edoxaban. The committee noted NICE's position statement in this regard, and 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 edoxaban. It therefore concluded that the PPRS Payment Mechanism was not relevant for its consideration of the cost effectiveness of edoxaban.