4 Consideration of the evidence
4.1 The Appraisal Committee reviewed the data available on the clinical and cost effectiveness of febuxostat for the management of hyperuricaemia in people with gout, having considered evidence on the nature of the condition and the value placed on the benefits of febuxostat by people with chronic hyperuricaemia in conditions where uric acid deposition has already occurred (including a history or the presence of tophi and/or gouty arthritis), those who represent them, and clinical specialists. It was also mindful of the need to take account of the effective use of NHS resources.
4.2 The Committee discussed the decision problem presented in the manufacturer's submission. It noted that the manufacturer had initially presented evidence for the clinical and cost effectiveness of febuxostat only as a first-line therapy, whereas the marketing authorisation does not make this restriction. It heard from the clinical specialists that in current clinical practice allopurinol is used as standard first-line therapy, and that febuxostat was a plausible improvement on current second-line options. These options are considered where allopurinol is not appropriate (for example because of intolerance or lack of response), and include benzbromarone, sulphinpyrazone and probenecid, all of which have limitations such as limited availability, adverse effects and poor effectiveness. The Committee was, however, persuaded that these options need not be considered as routine or best-practice comparators.
4.3 The Committee discussed the possible use of febuxostat as an alternative to allopurinol as first-line therapy; that is, where allopurinol is tolerated and not contraindicated. It identified that it has been established in clinical practice (as set out in BSR and EULAR guidelines) that allopurinol is most effective when given in a titrated regimen depending on the serum uric acid concentration. The Committee heard that even though many GPs limit their prescribing of allopurinol to a fixed dose of 300 mg/day, this is not best practice, particularly in the subset of people with gout for whom febuxostat might be considered. The Committee concluded that up-titrated allopurinol (to a maximum of 900 mg/day) should be considered as the best-practice comparator for febuxostat.
4.4 The Committee also discussed the outcome measures used. In particular, it discussed the surrogate outcome of serum uric acid concentration. It heard from the clinical specialists that a significant proportion of the population have a high serum uric acid concentration, but that comparatively few people present with clinical symptoms related to gout. However, although it noted that there remained some uncertainty about the relationship between serum uric acid concentration and clinical benefits (such as gout flare control, renal impairment and reduction in tophi size and number), the Committee was persuaded that a reduction in the serum uric acid concentration below the 'saturation point' (approximately 6 mg/100 ml) was necessary to avoid precipitation of uric acid crystals in tissues in the long term. Additionally, although the association between serum uric acid concentration and symptoms of gout is complex and not completely understood, the Committee concluded that it was reasonable to assume that a relationship with symptoms is likely above a serum uric acid concentration of 6 mg/100 ml.
4.5 The Committee discussed the evidence for the clinical effectiveness of febuxostat, and specifically the RCTs and extension studies in which febuxostat was compared with fixed-dose allopurinol. It noted the ERG critique of the efficacy results from the pooled analysis of data from the APEX and FACT trials and agreed that the approach was methodologically inappropriate, in that it failed to preserve randomisation in the RCT evidence. The Committee noted that the manufacturer was asked to conduct a meta-analysis of the RCT data but this was not provided. The Committee noted, however, that an appropriately conducted meta-analysis from the ERG showed very similar results, demonstrating that febuxostat is more effective than allopurinol in lowering serum uric acid concentration.
4.6 The Committee noted that in the main comparative trials, fixed-dose allopurinol was found to effectively reduce the serum uric acid concentration below the saturation point of 6 mg/100 ml in a substantial proportion of patients. The Committee accepted that febuxostat is more effective at reducing serum uric acid concentration than fixed-dose allopurinol. The Committee also agreed that it was plausible that a reduction in serum uric acid concentration was associated with a reduction in gout symptoms, However, it concluded that the benefits of febuxostat compared with allopurinol (using a fully titrated dosing schedule) in improving clinical outcomes, such as gout flare control, reduction in tophi size and number, and avoidance of joint and organ damage as a result of urate deposition in the longer term, had not been clearly demonstrated.
4.7 The Committee discussed the observation that the proportion of people with recurrent gout flares needing treatment was higher in the febuxostat arm than in the allopurinol arm of the APEX and FACT RCTs, although the number of gout flares declined over time. It heard from the clinical specialists that this phenomenon is seen with all uric acid-lowering therapies. Although the mechanism is not completely understood, it is believed to relate to the rate of change in serum uric acid concentration, with treatments that reduce serum uric acid concentration more effectively and rapidly having a more pronounced effect. The Committee accepted this explanation and noted that an initial rise in clinical gout flares can be mitigated with the use of prophylaxis with an NSAID or colchicine. Nevertheless, it concluded that, where they remain, these initial flares are of clinical importance.
4.8 The Committee considered the adverse effects associated with febuxostat in comparison with fixed-dose allopurinol. In particular, it noted that there were a higher number of cardiovascular events and deaths across the febuxostat arms of the APEX, FACT and EXCEL studies. It noted that the manufacturer had reported such differences as not being statistically significant, but that the ERG had found a lack of completeness in reporting, despite requests for clarification. However, it was noted that no dose–response relationship between these events and increasing doses of febuxostat had been observed in RCTs. The Committee also noted that the marketing authorisation for febuxostat carries special warnings and precautions that preclude its use in patients with ischaemic heart disease or congestive heart failure.
4.9 The Committee discussed the economic model presented by the manufacturer, initially covering only the use of febuxostat as an alternative to first-line allopurinol therapy, and the ERG's subsequent critique. It had a number of concerns about the model structure and parameter assumptions in the model. The Committee was concerned that the evidence base was incomplete because the comparison presented by the manufacturer was limited to a suboptimal (that is, fixed-dose) regimen of allopurinol. The manufacturer did not consider that it could include up-titrated allopurinol as a comparator in the current model. The Committee concluded that the manufacturer's ICER calculation for febuxostat compared with fixed-dose allopurinol, having failed to consider up-titrated allopurinol, was a substantial underestimate.
4.10 The Committee heard from clinical specialists that new and relatively unfamiliar drugs are often used, initially at least, when current drugs are inappropriate or have failed to achieve a response. It understood that this was the case for febuxostat, and that its place in the pathway of care was preferentially as second-line treatment after allopurinol. It was mindful that the ERG had requested modelling of sequential use when patients progress to the need for alternative treatments (following lack of response to allopurinol treatment) or no-treatment options, and that the manufacturer had declined the request on the basis of lack of evidence. The Committee noted that these shortcomings of the manufacturer's model were too great for the ERG to be able to rectify and offer an alternative ICER. It also noted that the manufacturer had subsequently offered to model sequential treatment strategies, but that it estimated that this would take between 4 and 6 months.
4.11 The Committee discussed the relationship assumed in the manufacturer's model between serum uric acid concentration, frequency of gout flares and how this may translate into improvements in long-term health-related quality of life. The Committee noted clarification from the manufacturer that the estimated relationship was non-linear. The Committee considered the ERG's concerns about the validity of the 'multivariate analysis' from which this assumption was derived. It noted the ERG's critique that this was based on a dataset from which some data points had been selectively excluded. The Committee also considered the statements from the clinical specialists that although a relationship between serum uric acid concentration above the saturation point and both frequency of gout flares and long-term adverse outcomes of gout is plausible, the strength and nature of the relationship is not clearly understood. The Committee concluded that there remained some uncertainty about the relationship between absolute serum uric acid concentration and gout symptoms in general, and that this was an additional source of uncertainty in the estimation of the incremental QALYs gained.
4.12 The Committee discussed the exploratory analysis by the ERG of the incremental QALY gain associated with the effect of lowering the serum uric acid concentration. The overall incremental QALY gain (0.032) included both the incremental QALY gain from the avoidance of gout flares and the 'chronic utility gain' from the prevention of gout-related symptoms. This is much higher than the overall incremental QALY gain (0.006) obtained from including the avoidance of gout flares alone. The impact of this difference on the final ICER was proportionately substantial. The Committee noted that removing the component of incremental QALY gain associated with the 'chronic utility gain' from lowering serum uric acid concentration increased the base-case ICER from £15,000 to £81,000 per QALY gained over a 2-year time horizon. It considered, however, that uncertainty about the strength and nature of the relationship between serum uric acid concentration, gout flares and utility gain added to the uncertainties surrounding the manufacturer's base case. Although the Committee was persuaded that removal of the 'chronic utility gain' would lead to an underestimation of the long-term clinical benefits of febuxostat treatment, it considered that the true base-case ICER, even when compared with fixed-dose allopurinol, would be within a wide range of between £15,000 and £81,000 per QALY gained.
4.13 The Committee then considered how improvements in health-related quality of life obtained with febuxostat had been estimated. The Committee discussed clarifications received from the manufacturer that estimates reported in the submission were based on utility decrements associated with the acute effects of gout flares and on a 'chronic utility gain' associated with the benefits of a lower serum uric acid concentration and of the avoidance of gout-related symptoms such as tophi, joint erosion and renal complications. It noted that this had been reported in the manufacturer's submission on the basis of a separate analysis in which a direct and continuous relationship between utility and serum uric acid concentration had been assumed. The Committee remained concerned (as described in section 4.11) that although this relationship was plausible, evidence supporting it was uncertain and not clearly established.
4.14 The Committee also considered that the difference in quality of life between experiencing a gout flare or not might have been overestimated because of bias in the research used to inform this parameter, since individuals had been asked to recall or imagine states they were not currently in. While it understood that the pain of flares can be extreme, it thought that the reported difference in EQ-5D scores for experiencing a gout flare of 1 week's duration was an overestimate. On the other hand, it heard from a patient expert that the assumption that gout flares last for 1 week might be a significant underestimate. Overall, the Committee thought that there was considerable uncertainty over the estimation of disutility associated with gout flares. The Committee noted additionally that published research studies have reported a number of anomalous responses, such as better overall health when experiencing a flare than when not, and 'perfect' health during a flare.
4.15 The Committee discussed the time horizons used in the manufacturer's model. It thought that for this chronic illness, which requires life-long therapy, a lifetime time horizon would have been appropriate. The Committee further took into account evidence from the APEX and FACT trials showing that a higher percentage of patients receiving febuxostat discontinued treatment compared with those receiving allopurinol. The Committee believed that the assumption over the 2-year time horizon that patients would remain on treatment, even if only a partial treatment effect was achieved, potentially biased the results of the economic analysis in favour of febuxostat. It was, however, mindful of conflicting evidence from the EXCEL trial that patients on febuxostat were more likely to continue treatment. It also noted that there could potentially be long-term benefits associated with febuxostat, such as avoidance of adverse renal events in comparison with other therapies. The Committee considered the univariate sensitivity analysis, which showed that an extension of the manufacturer's economic model to 5 years had a marginal impact on the base-case ICER. However, it noted that this simply reflects the assumption that patients continue to accrue treatment benefits beyond the 2-year time horizon.
4.16 The Committee discussed cost assumptions in the model. It considered it plausible that the main drivers of incremental cost would be the drug cost (noting the 13-fold difference in cost between febuxostat and allopurinol), and the costs saved by avoiding treatment costs associated with any gout flares. It concluded that the total cost avoided by reduction in flares may have been overstated because the difference in expected number of flares may have been overestimated, as discussed above (although this may be offset by the cost savings from the long-term benefits of alleviating gout symptoms).
4.17 The Committee discussed the results of a second economic analysis presented by the manufacturer showing that a comparison between febuxostat and a no-treatment (placebo) option gave an ICER of £3727 per QALY gained over a 2-year time horizon. It considered that this estimate was also subject to substantial uncertainty, given the concerns discussed above regarding the approach taken by the manufacturer in the economic analysis. The Committee considered, however, that given that the option of titrated allopurinol was obviously not available for the treatment of people who are intolerant of allopurinol or for whom allopurinol is contraindicated, the ICER of febuxostat in these groups of patients was very likely to be at an acceptable level, despite the degree of uncertainty because of lack of evidence.
4.18 The Committee further discussed the use of febuxostat in cases where treatment goals were not achieved with allopurinol. The Committee considered that there is no evidence, and no basic pharmacodynamic reason, to suppose that febuxostat would confer any benefit for patients in whom up-titrated allopurinol has failed to lower the serum uric acid concentration below the 'saturation point'. Thus the Committee considered that febuxostat could be recommended as a treatment option in people with chronic hyperuricaemia and symptomatic gout who are intolerant of allopurinol (defined as adverse effects that are sufficiently severe to warrant its discontinuation, or to prevent full dose escalation for optimal effectiveness as appropriate within its marketing authorisation), or for whom allopurinol is contraindicated (for instance in cases of severe renal impairment or interactions with other medications).
4.19 Overall, the Committee concluded that although febuxostat had been shown to be more effective than fixed-dose allopurinol in lowering serum uric acid concentration, it had not been shown to be clinically or cost effective compared with the more appropriate comparator of allopurinol up-titrated in accordance with established best clinical practice. However, it concluded that febuxostat should be recommended as an option for the management of chronic hyperuricaemia in gout for people who are intolerant of allopurinol (defined as adverse effects that are sufficiently severe to warrant its discontinuation, or to prevent full dose escalation for optimal effectiveness as appropriate within its marketing authorisation), or for whom allopurinol is contraindicated.