Cost-effectiveness analysis of febuxostat in patients with gout in Spain

Abstract

Objectives Cost-effectiveness of febuxostat compared with allopurinol in the treatment of hyperuricemia in patients with gout.

Methods Costs, clinical outcomes, and QALYs were estimated using a Markov model. Febuxostat 80 mg and 120 mg sequentially, used as first line and second line therapy, was compared with allopurinol 300 mg. Patients switched to the next treatment in the sequence according to a dichotomous response vs no response (target serum urate level < 6 mg/dl outcome) after 3 months of active treatment. A 3% discount rate and 5-year time horizon were applied. Perspective: National Health System.

Results The addition of febuxostat to any therapeutic strategy was an efficient option, with incremental cost-effectiveness ratios (ICER) compared with allopurinol 300 mg ranging from €5268–€9737.

Conclusions Febuxostat is a cost-effective treatment in Spain for the management of hyperuricemia in gout patients, with ICERs far below accepted Spanish efficiency thresholds (30 000€/QALY).

Keywords:

• Cost-effectiveness
• Gout
• Febuxostat
• Allopurinol
• Cost
• Spain

Introduction

Gout is a disease caused by the nucleation and deposition of monosodium urate crystals (MSUCs) in the tissues, mainly in the musculoskeletal and subcutaneous. MSUCs deposition is conditioned by sustained hyperuricemia (considered as the solubility threshold of serum urate (sUA) over 6.8 mg/dl), which leads to super saturation¹.

As any storage disease, it is conceptually a chronic process, although clinical manifestations vary from acute flares, asymptomatic periods, and persistent complaints derived from chronic synovitis and joint structural damage related to tophi. In addition, gout is associated with a wide range of comorbidities including cardiovascular disease, renal disease, and the metabolic syndrome^2–7.

Hyperuricemia and gout are frequent conditions in the adult population⁸. The prevalence of gout in Western countries ranges between 0.5–1%, and rises with age to exceed 3% in persons aged ≥65 years⁹. In Spain, it has been estimated to be 0.2% in males aged 18–54 years and 0.5% in males aged ≥65 years, suggesting a figure of more than 200 000 people affected¹⁰. Despite of this, data on the economic impact in Spain are limited, and proper control of hyperuricemia in patients with gout did not reach target in over 50% of patients previously to the release of febuxostat¹¹.

Febuxostat is a non-purine selective xanthine oxidase inhibitor, which is approved by the European Medicines Agency (EMA), the Food and Drug Administration (FDA), and the Pharmaceuticals and Medical Devices Agency (PMDA) of Japan for the treatment of hyperuricemia in situations when urate deposition has already occurred (including a history of or the presence of tophus and/or gouty arthritis).

The cost-effectiveness of febuxostat vs allopurinol in the treatment of hyperuricemia was evaluated by the National Institute for Health and Clinical Excellence (NICE)¹² and the Scottish Medicines Consortium (SMC)¹³ in 2008 and 2010, respectively. Recently, Beard et al.¹⁴ have reported on the cost-effectiveness model presented to the SMC and concluded that febuxostat 80 mg followed by 120 mg is highly cost-effective in reducing sUA levels, especially when used as second line treatment after allopurinol.

The aim of this study was to determine the cost-effectiveness of different treatment sequences for the treatment of hyperuricemia in gout patients in the Spanish setting, not including indirect costs, and focused on febuxostat vs allopurinol 300 mg alone as first line treatment, thereby complementing the study by Beard et al.¹⁴ in which no-urate lowering therapies (ULT) was used as a reference comparator.

Methods

This cost-effectiveness analysis adapted a pharmacoeconomic model in which the effects and costs over time of different treatment sequences for hyperuricemia were compared in the Spanish setting.

Study population and comparators

The characteristics of the cohort of patients considered in the base case were intended to reflect the population of the APEX¹⁵ and FACT¹⁶ studies: adult patients with gout and sUA ≥8 mg/dl to be treated with allopurinol 300 mg/day or febuxostat 80 or 120 mg/day. Supplementary analyses were made for patients with mild-to-moderate renal function impairment (in whom it is recommended to be treated with lower doses of allopurinol), and patients in whom allopurinol was not an appropriate treatment, due either to intolerance or lack of response. Benzbromarone was not included in the analysis, since its use in Spain is restricted to severe gout in patients with failure or intolerance to allopurinol, and is only prescribed by rheumatology and nephrology specialists.

Different treatment sequences, in addition to febuxostat vs allopurinol as first line treatment, were compared (Table 1). Treatment sequences including second line therapies were included in order to reflect standard practice of treatment in Spain, where patients may be switched to an alternative option once the current treatment no longer provides clinical benefit. Thus, treatment sequences consist of an initial dose of febuxostat 80 mg once daily, which was increased to 120 mg once daily if required, allopurinol 300 mg once daily (with dose reduction in the supplementary analysis in patients with mild or moderate renal failure [30–89 ml/min]), and no treatment (NT) for patients intolerant to allopurinol or who have discontinued active treatment (last step of the therapeutic sequence).

Table 1. Types of analyses and strategies compared.

Primary/Supplementary analyses	Strategies compared
Patients with hyperuricemia and gout (primary)	0. Alo 300 → NT
	1. Feb 80 → Feb120 → NT
	2. Alo 300 → Feb 80 → Feb 120 → NT
	3. Feb 80 → Feb 120 → Alo 300 → NT
Patients with mild or moderate renal failure (supplementary)	0. Alo 100-200 → NT
	1. Feb 80 → Feb 120 → NT
	2. Alo 100-200 → Feb 80 → Feb 120 → NT
	3. Feb 80 → Feb 120→ Alo 100-200 → NT
Patients intolerant of or unresponsive to allopurinol (supplementary)	0. NT
	1. Feb 80 → Feb 120 → NT

Alo, Allopurinol; Feb, Febuxostat; NT, No treatment.

Type of analysis

The results were expressed as incremental cost-effectiveness ratios (ICER): where Cost_S1-S2-S3 and Cost_S0 are the costs associated with the different strategies, S0 is the reference, and Effectiveness_S1-S2-S3 and Effectiveness_S0 are the clinical consequences, expressed in quality-adjusted life years (QALY), of the different strategies, with S0 as the reference. Thus, the ICER expresses the incremental cost per QALY of each strategy (S1–S3) relative to S0.

The time horizon of the base case was 5 years, and a discount rate of 3% was applied to account for future costs and effects occurring after the first year, as recommended in Spanish health technology evaluation guidelines^17,18. The analysis was conducted from the perspective of the Spanish National Health System (NHS) and, therefore, only direct medical costs were considered.

Model description

The Markov model, which was adapted to the Spanish setting, was based on cycles of 3 months, reflecting the period in which the therapeutic response to the initial treatments and whether to continue with the same ULT or switch to a new one should be assessed. Figure 1 shows the structure of the model, distinguishing between the initial treatment (first 3 months with each ULT of the sequence) and maintenance therapy (long-term, represented by subsequent Markov cycles).

Figure 1. Diagram of model (adapted from Beard et al.¹⁴). Tx, treatment.

The cohort of patients started the model receiving the first ULT of each sequence (initial treatment) and were assigned, according to the response to such treatment, to the following health states (based on sUA levels): responders (sUA1: ≤6 mg/dL), non-responders (sUA2: >6 mg/dL and ≤8 mg/dL, sUA3: >8 mg/dL and ≤10 mg/dL, sUA4: >10 mg/dL).

Patients who responded (sUA1), were assumed to remain on the same health state unless failure and treatment discontinuation, defined as treatment drop-out. When patients did not achieve an adequate response based on sUA levels (non-responders: sUA2, sUA3, or sUA4), or did achieve a response (responders: sUA1), but suffered treatment drop-out, they moved to the next treatment in the sequence (Sw), with the response (initial treatment) being determined again according to sUA levels for the new ULT, as described above.

Patients at the end of the treatment sequence were distributed across the sUA levels associated with the no-treatment health state. In each cycle of the model, the probability of experiencing acute gout flares at any sUA level (assuming a duration of 1 week) was considered. Acute gout flares did not involve a change of ULT, but did have consequences in terms of cost and reduced quality-of-life.

The analysis included general mortality in order to calculate the long-term results in terms of QALYs. The direct risk of mortality related to gout was not included in the model, assuming, conservatively, that hyperuricemia had no impact on patient survival, despite increased risk of mortality has been recently reported as associated to both severity and the highest levels of sUA in patients with gout compared to the general population¹⁹.

Clinical data

Febuxostat has demonstrated its efficacy and safety in the treatment of hyperuricemia in patients with gout in the randomized, controlled, double-blind phase 3 APEX¹⁵, FACT¹⁶, and CONFIRMS²⁰ trials, and the long-term FOCUS²¹ and EXCEL²² studies (5 and 3 years, respectively).

The clinical efficacy of febuxostat 80 mg/120 mg and allopurinol 300 mg once daily, defined as the reduction in sUA levels achieved, was obtained from aggregated data from the APEX¹⁵ and FACT¹⁶ controlled clinical trials (Table 2). Efficacy data for the supplementary analysis in patients with chronic kidney disease (CKD) were obtained from the CONFIRMS controlled clinical trial²⁰, which included a sub-set of patients with mild-to-moderate renal impairment (creatinine clearance 60–89 ml/min and 30–59 ml/min, respectively) who received a lower dose of allopurinol.

Table 2. Efficacy of treatments for hyperuricemia.

sUA level	Allopurinol 300 mg	Febuxostat 80 mg	Febuxostat 120 mg	No treatment
Responders*
<6.0 mg/dL	37.6%	73.3%	79.3%	0.8%
Non-responders	62.4%	26.7%	20.7%	99.2%
>6 and ≤8 mg/dL	79.00%	74.10%	67.20%	0%
>8 and ≤10 mg/dL	17.50%	21.30%	29.50%	63.40%
>10 mg/dL	3.50%	4.60%	3.30%	36.60%

sUA, serum uric acid.

*Defined as proportion of patients with SUA ≤6 mg/dl at the final visit.

Drop-outs from treatment with either allopurinol or febuxostat by any cause, defined in the clinical trials as ‘premature discontinuation’, were included in the model. Drop-out rates were obtained from the pooled data of the APEX and FACT studies^15,16 for the first year of ULT and from the EXCEL long-term, open-label extension study²² for subsequent maintenance treatment. Table 3 shows the drop-out probabilities for the first year and subsequent years for each treatment.

Table 3. Discontinuation rates.

Treatment	First year*			Successive years** Annual
	0–3 months	4–6 months	7–12 months
Allopurinol 300 mg	11.9%	8.8%	10.2%	28.9%
Febuxostat 80 mg	17.4%	13.9%	14.5%	12.2%
Febuxostat 120 mg	17.7%	11.9%	11.8%	18.2%

*Pool APEX and FACT.

**Annual rate calculated from years 1, 2 and 3 of the EXCEL study.

The rate of acute gout flares was obtained from the APEX and FACT studies^15,16 for the initial treatment (first cycle of 3 months). In clinical practice, the initiation of ULT is commonly associated with increases in gout flares, for which prophylaxis for 6 months is nowadays recommended^2,23. Since the prophylaxis period was only 8 weeks in the APEX and FACT studies^15,16, an adjustment was made in the flare rate observed in the APEX and FACT studies, according to the reduction in the risk of flare with and without prophylaxis observed at 12 weeks in the study by Borstad et al.²⁴ to estimate the rate of acute gout flares when prophylaxis was extended. As a result, the adjusted probability of suffering a gout flare in the first 3 months of treatment was 0.51 for allopurinol 300 mg, 0.63 and 0.87 for febuxostat 80 mg and 120 mg, respectively, and 0.25 for no treatment.

For cycles subsequent to the first 3 months of treatment, the probability of having a gout flare depended on sUA levels, which were obtained from a study by Institute of Medical Science (IMS) Europe of 140 patients in the UK, Germany, and France²⁵. For each health state in the model, the probabilities of experiencing flare were 0.65 for sUA1, 0.73 for sUA2, 0.80 for sUA3, and 0.86 for sUA4.

Utilities and costs

Utility values were used to represent the impact on the quality-of-life associated with a specific health state on a scale from 0 (death) to 1 (perfect health). Quality-of-life data were obtained from the EQ-5D questionnaire administered in the IMS Europe study²⁵ which found a mean value for patients with gout of 0.710 (95% CI = 0.638–0.736). Utility values were assigned to each health state using a multivariate analysis that evaluated the impact of sUA levels on the EQ-5D scores. The utility values assigned to each health state were 0.7463 for sUA1, 0.7120 for sUA2, 0.6777 for sUA3, and 0.6435 for sUA4. A reduction in utility (disutility) of 0.0097 was assigned to acute gout flares, assuming a 7-day quality-of-life effect²⁵.

The drug costs of ULT were expressed as the retail price plus value added tax (RP + VAT)²⁶, taking into account the corresponding deductions according to Royal Decree-Law 8/2010²⁷. The resulting daily cost for each ULT is shown in Table 4. The cost of an acute gout flare was estimated from the use of resources reported in the observational study by Sicras-Mainar et al.²⁸, to which reported Spanish tariff unit costs were assigned²⁹ (Table 5).

Table 4. Drug costs.

	100–200 mg allopurinol	Allopurinol 300 mg	Febuxostat 80 mg and 120 mg
Daily cost	€0.05*	€0.09	€1.33

*Mean doses of 100 mg and 200 mg.

Table 5. Cost of an acute gout flare.

Gout flare cost
Cost type	% Use (3130 patients)*	Unit cost**	Total cost
Primary care
Medical visits	100%	€65.80	€65.80
Laboratory tests	96.10%	€74.90	€71.98
Conventional radiology	83.90%	€17.77	€14.91
Other tests	47%	€28.81	€13.54
Specialist care
Medical visits	13.20%	€139.36	€18.40
Hospitalization days	3.10%	€2856.70	€88.56
Emergency	11.20%	€183.70	€20.57
Total acute gout flare = €293.76

*Taken from Sicras-Mainar et al.²⁸.

**eHealth Data Base²⁹.

Sensitivity analysis

To evaluate the uncertainty associated with some model parameters, and to further determine the robustness of the results, deterministic and probabilistic sensitivity analyses were developed.

In the deterministic sensitivity analyses, variables were modified individually with respect to the base case (univariate), in order to identify variables with the greatest impact on the results of the model. The results at different time horizons (1, 2, 10 years and lifetime) were analyzed, the response threshold to the ULT was modified (≤5 and 7 mg/dL), the discount rate for costs and effects was modified (0% and 5%), and the baseline value of the utilities and the cost of gout flares was modified by ±20%.

In the probabilistic sensitivity analysis, 1000 simulations were performed using a Monte-Carlo second-order simulation³⁰, simultaneously changing the parameters of reductions in sUA, treatment discontinuation, and annual rate (after one year) of gout flares (beta distribution), the flare rates during the first cycle (log-normal distribution), and the healthcare cost (gamma distribution).

Results

Primary analysis (base case)

Table 6 shows the clinical results (expressed as QALYs, the number of acute gout flares, and the days with sUA ≤6 mg/dL) and economic results for each of the sequences (strategies S0–S3). Table 6 also shows the incremental output (in QALYs and costs) of the S1–S3 strategies with respect to the S0 strategy. All strategies that included the use of febuxostat 80 mg/120 mg were cost-effective with respect to the use of first-line allopurinol 300 mg alone, using the cost-effectiveness threshold of €30 000 per QALY commonly accepted in Spain³¹.

Table 6. Results of the primary analysis.

	Strategy 0 A300/NT	Strategy 1 F80/F120/NT	Strategy 2 A300/F80/F120/NT	Strategy 3 F80/F120/A300/NT
Effectiveness
Days sUA ≤6 mg/dL	366	871	1048	1060
Number of initial flares (treatment related)	0.98	1.03	1.26	1.38
Number of later flares (sUA related)	5.64	5.27	4.84	4.84
QALYs	3.13	3.24	3.28	3.28
Costs
Drug cost	€37	€1193	€987	€1213
Cost of gout flares	€1840	€1752	€1697	€1752
Total cost	€1877	€2944	€2684	€2965
ICER
Additional QALYs vs S0	—	0.11	0.15	0.15
Additional cost vs S0	—	€1067	€806	€1088
Cost per QALY gained vs S0	—	€9737	€5268	€7129

sUA, Serum uric acid; QALY, Quality-adjusted life year; ICER, Incremental cost-effectiveness ratio.

Supplementary analyses

The results of the analyses made in patients with CKD and those intolerant to allopurinol are in line with those obtained in the primary analysis (Table 6), showing that febuxostat was a cost-effective option compared with first-line allopurinol. In a cohort of patients with mild-to-moderate CKD, typically treated with lower doses of allopurinol (100–200 mg/day), the ICER resulting from the comparison of the first-line strategies (S1 vs S0) was €12,106 per QALY gained. For sequential strategies vs first-line allopurinol 300 mg alone (S2 and S3 vs S0) the ICERs were €5358 and €7708 per QALY gained, respectively.

In a cohort of patients in whom allopurinol was not a suitable treatment due to intolerance or inadequate response, the ICER for febuxostat 80/120 mg compared with no treatment was €6348 per QALY gained.

Sensitivity analysis

Table 7 shows the results of the deterministic sensitivity analysis.

Table 7. Results of univariate deterministic sensitivity analysis.

		ICER (€per QALY)
		S1 vs SO	S2 vs SO	S3 vs SO
Base case		€9737	€5268	€7129
Time horizon	1 year	€14,237	€6329	€11,902
	2 years	€12,330	€5697	€9338
	10 years	€8292	€5121	€6159
	Lifetime	€7477	€5053	€5630
Threshold of response  to treatment	<5 mg/dL	€6955	€5395	€6366
	<4 mg/dL	€7169	€6202	€7320
Utility values	+20%	€8150	€4412	€5960
	−20%	€12,091	€6534	€8869
Cost of gout	+20%	€9575	€5080	€7013
	−20%	€9 899	€5455	€7244
Discount rate	0%	€9632	€5250	€7040
	5%	€9 806	€5279	€7187

QALY, Quality-adjusted life year; ICER, Incremental cost-effectiveness ratio.

Figure 2 shows the results of the probabilistic sensitivity analysis using an incremental cost-effectiveness plane to compare the S0 and S1 strategies. The ordinate axis represents the incremental cost of S1 compared with S0, and the horizontal axis the incremental effect of S1 compared with S0 expressed in QALYs. All simulations are located in the first quadrant, representing greater effectivity and a greater cost of S1 compared with S0. Using the above-mentioned threshold of €30,000/QALY³¹, 96.2% of simulations showed S1 as a cost-effective option compared with S0.

Figure 2. Probabilistic sensitivity analysis. Incremental cost-effectiveness plane.

Figure 3 shows the probabilistic sensitivity analysis results in the form of an acceptability curve for S1 compared with S0 at different cost-effectiveness thresholds. For the above-mentioned threshold of €30,000/QALY, the probability of S1 being cost-effective compared with S0 was 100%: this fell to 92.8% and 63.3% when the acceptability threshold was lowered to €20,000 and €10,000 per QALY gained, respectively.

Figure 3. Probabilistic sensitivity analysis. Acceptability curve.

Discussion

Gout has a substantial economic impact in terms of both direct healthcare costs (medical visits, drugs, etc.) and indirect costs such as lost productivity^32,33. A Spanish population-based study of 3130 patients with gout that aimed to determine the use of resources and the economic impact of patients with gout found that the total cost of the patients included in the study amounted to 7 million euros, of which 96.9% were direct health costs and 3.1% non-healthcare costs (lost productivity)²⁸. In addition, gout has a high impact on the quality-of-life of patients, which may be due both to direct musculoskeletal manifestations and to associated comorbidities^2,34.

The results obtained in this analysis, the first to determine the cost-effectiveness of febuxostat in Spain, show that initiating treatment for hyperuricemia with febuxostat instead of allopurinol is a cost-effective option for the NHS.

Several economic evaluations of the treatment of hyperuricemia in gout with febuxostat and allopurinol have been published^14,35–40, including the study by Beard et al.¹⁴ describing the Markov model presented to the SMC, which has been adapted to the Spanish setting in the present analysis. These economic evaluations differ in the time horizons used, but in models using longer horizons^{14,35,36,38,40}, even despite the higher acquisition cost of febuxostat, it was a cost-effective option for both first-line^35,36,40 and second-line treatment^14,38 when costs and effects are modeled in the long-term.

The model has limitations, some of which are inherent to this type of pharmacoeconomic models, which have structural rigidity that could make correct representation of clinical reality difficult.

For example, this type of model usually uses long-term extrapolations of data from studies of limited duration. The 28 and 52 weeks duration of the APEX and FACT studies, respectively, provide significant information about reductions in sUA levels, but not on the long-term clinical outcomes, reflected in reductions of gout flares that are associated to improvement in quality-of-life⁴¹. Therefore, in the model, the reduction in sUA levels occurs at the beginning of each ULT (based on the evidence of the APEX and FACT studies), assuming that this reduction is maintained over time if the patient continues treatment correctly.

For the drop-out rate beyond the first year (for which the pooled results of the APEX and FACT studies were used), it was also necessary to extrapolate from the results of a study of limited duration. Using the results of the 3-years study EXCEL, the annual probability of discontinuation was estimated and, therefore, applied constant from the second year onwards. In this respect, discontinuation was over-estimated, since the EXCEL study found that treatment abandonment fell over time and, in our model, the annual probability calculated was the same for each year. The CONFIRMS study found slightly-lower rates of drop-out than the pooled results of the APEX and FACT studies, and also favorable to febuxostat 80 mg (the study did not consider febuxostat 120 mg) over allopurinol. Therefore, the analysis is conservative in considering higher drop-out rates during the first year of treatment. Neither were the costs associated with the adverse events of febuxostat and allopurinol included in the model, since clinical evidence suggests no significant difference in the adverse events rate between the two treatments. However, the treatment discontinuation due to adverse effects itself is recorded within the variable ‘drop-out’.

Another limitation was due to the adaptation to the Spanish setting: the lack of a national cost database. This made it necessary to estimate the cost associated with an acute gout flare according to the use of resources in primary and specialist care reported in the observational study by Sicras-Mainar et al.²⁸. Importantly from the clinical practice point of view, other costs associated with the management of hyperuricemia and gout were not included in the analysis, such as follow-up visits or laboratory tests for escalating doses of allopurinol as recommended²³. Since no evidence is available on possible differences in the use of these resources according to sUA levels, it is not possible to establish differences in the cost of disease management for each health state (sUA1, sUA2, sUA3, and sUA4). Therefore, if the non-drug management cost is the same for all health states, this is irrelevant in incremental terms, given that the clinical benefits of febuxostat are due to a reduction in sUA levels. As the analysis was made from the Spanish National Health System perspective, indirect costs due to the lost productivity associated with hyperuricemia were not included. The indirect costs associated with gout flares are considerable. The study by Brook et al.³² estimated the indirect costs of the active population suffering gout as €2883, while a recent study by Spaetgens et al.⁴² estimated the cost of work absenteeism due to gout as €4982. The study by Sicras-Mainar et al.²⁸ reported lower indirect costs associated with gout flares, this difference might be due to their conservative information source for the unit cost (the minimum wage instead of average wage cost) or the registration of job loss by the insurers (temporary job loss might not be registered). In any case, given the reduction in sUA levels observed in patients treated with febuxostat, which lead to a reduction in gout flares in the medium- and long-term, if indirect costs had been included in the analysis, the ICER of febuxostat vs allopurinol as first-line treatment would have been lower.

To overcome the uncertainty associated with all these limitations, the assumptions made and the values finally considered in the analysis were validated by a clinical expert, in addition to the deterministic and probabilistic sensitivity analyses developed.

In conclusion, and, even though it only considers direct costs, this analysis confirms and supplements the results of the study by Beard et al.¹⁴, and shows that febuxostat 80 mg to 120 mg daily is a cost-effective option in the treatment of hyperuricemia of gout in the Spanish setting, not only when both ULT are used sequentially, but also when febuxostat is considered as first-line treatment.

Further economic evaluations should be developed in order to include indirect costs, or if relevant price changes befall.

Transparency

Declaration of funding

The present study was supported by Menarini Group, who market Febuxostat (Adenuric®) in Spain.

Declaration of financial/other relationships

FP received fees as advisor/speaker from AstraZeneca, Menarini, and Cymabay. DC received financing from Menarini for the pharmacoeconomic analysis and drafting the manuscript. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.