Cost-utility analysis of life-long prophylaxis with recombinant factor VIIIFc vs recombinant factor VIII for the management of severe hemophilia A in Sweden

Abstract

Aims: Prophylaxis with recombinant factor VIII (rFVIII) is the standard of care for severe hemophilia A in Sweden. The need for frequent injections with existing rFVIII products may, however, result in poor adherence to prophylaxis, leading to increased bleeding and long-term joint damage. Recombinant FVIIIFc (rFVIIIFc) is an extended half-life fusion protein which can offer prolonged protection and reduced dosing frequency. The objective of this study was to evaluate the cost-utility of prophylaxis with rFVIIIFc in severe hemophilia A from the perspective of the Swedish health system.

Methods: A Markov model was built to estimate lifetime costs and benefits of prophylaxis with rFVIIIFc vs rFVIII products. Clinical outcomes were represented by annualized bleeding rate (ABR) and quality of life via disutility applied to bleeding events and injection frequency. Costs included the cost of FVIII for routine prophylaxis and bleed resolution. The pooled comparator was costed by weighting the cost of individual products by their market share.

Results: In the base case, rFVIIIFc was dominant vs the pooled comparator. Savings of SEK 9.0 million per patient resulted from lower factor consumption for prophylaxis and bleed resolution. Fewer bleeds and reduced injection frequency yielded an estimated 0.59 quality-adjusted life years (QALYs). Results were sensitive to drug dosage and robust to variation in other parameters. Probabilistic sensitivity analysis suggested a greater than 85% probability of rFVIIIFc being cost-effective at a willingness-to-pay threshold of 500,000 SEK/QALY.

Limitations: Due to unavailibilty of patient-level data, treatment benefit was based on a non-adjusted indirect comparison. Dosing and treatment outcomes were assumed to persist over the model duration in the absence of long-term outcome data.

Conclusion: The results suggest that rFVIIIFc may be a cost-effective option for hemophilia A prophylaxis, generating greater quality of life and reduced costs for the Swedish payer compared to more frequently administered rFVIII alternatives.

Keywords:

• Cost-effectiveness analysis
• cost-utility
• hemophilia A
• prophylaxis
• factor VIII
• annualized bleeding rate
• extended half-life

Introduction

Routine prophylactic factor VIII (FVIII) replacement therapy is the recommended standard of care for patients with severe hemophilia A¹. The severe phenotype, which accounts for ∼60% of all diagnosed hemophilia A cases, occurs in patients with circulating plasma FVIII levels <1% of normal.

The management of hemophilia A has undergone a paradigm shift with the introduction of routine prophylaxis as a means of increasing plasma FVIII levels to prevent spontaneous bleeding and resulting joint damage². Compared with on-demand treatment, prophylaxis significantly reduces bleeding rates, attenuates joint deterioration, and improves quality of life in affected patients^3–7.

Despite major advances in the management of hemophilia A, regular replacement therapy with existing FVIII products requires frequent intravenous injections and significant time commitment, which may limit participation in social life^8,9. All conventional recombinant FVIII (rFVIII) products have nearly indistinguishable pharmacokinetic profiles, with a relatively short half-life of ∼12 h¹⁰, which necessitates frequent factor infusions to maintain FVIII trough levels at ≥1% of normal required for prevention of spontaneous bleeding¹¹. Such a treatment burden may have a considerable impact on patient adherence and long-term outcomes of hemophilia A¹².

The licencing of rFVIIIFc has expanded the opportunities for improvement in hemophilia care. It is an extended half-life rFVIIIFc fusion protein approved for use in hemophilia A prophylaxis as well as the treatment of bleeding episodes and perioperative hemostatic management¹³. The safety, efficacy, and extended half-life of rFVIIIFc have been demonstrated within studies in adults and children^14,15. A study in adults has demonstrated an ∼1.5-fold increase in the half-life of rFVIIIFc compared to rFVIII (Advate^TM, recombinant anti-hemophilic factor – protein-free medium)¹⁴. Similarly, rFVIIIFc can achieve comparable factor VIII activity levels with a lower dose or less frequent dosing than conventional FVIII products¹⁶. As such, rFVIIIFc offers the potential for reduced frequency of dosing and improved protection from spontaneous bleeding.

This study explores the cost-utility of prophylaxis with extended half-life rFVIIIFc in comparison to routine prophylaxis regimens with conventional rFVIII products, which constitute the current standard of practice in Sweden. A cost-utility model was developed which enables assessment of lifetime costs and outcomes of prophylaxis in severe hemophilia A, as well as the impact of a reduction in dosing frequency on the health-related quality of life (HRQoL) over an extended period of time.

Methods

A two-state Markov model was developed in Microsoft Excel (Microsoft Inc., Redmond, WA), as shown in Figure 1. All patients enter the model within the alive state. Within this state, patients are treated prophylactically and experience a defined number of bleeds, represented by the annualised bleed rate (ABR), as determined by the clinical efficacy of each respective treatment. The incidence of ABR was also linked to its impact on HRQoL by applying a utility decrement to each occurrence of a bleed. A disutility associated with frequency of prophylactic infusions was included in the model, while morbidities of hemophilia A such as joint deterioration, associated surgery, and intracranial hemorrhages were not due to lack of evidence linking their incidence with ABRs, which was the primary efficacy end-point of clinical studies in hemophilia.

Figure 1. Structure of the Markov model used in cost-utility analysis of prophylaxis with extended half-life rFVIIIFc vs rFVIII.

Patients could also move to the dead state. In the absence of robust population-specific data on mortality in hemophilia patients, the probability of death reflected the general, male age-adjusted mortality only.

A lifetime horizon was used (70 years) and the cycle length was set to 1 year. A discount rate of 3% was applied to costs and effects in the base case, while discount rates of 0% and 5% were explored in sensitivity analysis, as recommended by the Swedish Dental and Pharmaceutical Benefits Agency (Tandvårds- och läkemedelsförmånsverket; TLV)¹⁷.

The base case analysis was undertaken to explore the economic and health consequences of primary prophylaxis, which is typically initiated before or after the first joint bleed and prior to the onset of clinical arthropathy. The base case population comprised subjects with severe hemophilia A (endogenous factor VIII activity level <1% of normal) whose prophylactic treatment was initiated at the age of 1 year, consistent with the minimum age of subjects in the Kids A-LONG study¹⁵. To address the uncertainty introduced in the model by means of a non-adjusted indirect cost- minimization (IDC) of treatment efficacy, we also performed a cost minimization analysis by assuming equal efficacy (ABRs) and equal HRQoL of conventional rFVIII and the long-acting rFVIIIFc treatment. Three additional scenarios assessed the impact of alternative assumptions on model results, including a comparison of pharmacokinetic-tailored prophylaxis with rFVIIIFc vs rFVIII, based on the clinical studies in which the treatment goal was to maintain FVIII trough levels at or above 1 IU/dL^14,18 required for the prevention of spontaneous bleeding, a scenario utilizing mean rather than median weekly factor consumption data, and a scenario considering long-term outcomes with rFVIIIFc¹⁹.

One-way sensitivity analysis (OWSA) was used to evaluate uncertainty in key model inputs on the outputs of the analysis. The plausible parameter ranges for the OWSA were obtained from the literature wherever available, or, alternatively, a variation of 20% on default parameter values was applied (Supplementary Table S2).

To assess the second order uncertainty in the model and evaluate overall robustness of results, probabilistic sensitivity analysis (PSA) was run with 2,000 iterations by simultaneously sampling all relevant model parameters from probability distributions assigned based on the characteristics of these variables (Supplementary Table S2).

Model inputs

Efficacy

Treatment efficacy was measured in the form of ABRs. The efficacy inputs for rFVIIIFc were based on the pivotal phase III study A-LONG, an open-label, multi-center, partially randomized study of rFVIIIFc in previously treated patients with severe hemophilia A aged ≥12 years¹⁴. In the absence of head-to-head comparisons and robust real world data, the mean ABR and weekly factor consumption during continuous prophylaxis with rFVIIIFc as reported in the A-LONG¹⁴ were indirectly compared with those reported in studies of conventional rFVIII products identified in a systematic literature review, as described by Iorio et al.²⁰. To ensure comparability with A-LONG subjects, eligible rFVIII studies were required to have a majority of subjects between 12 and 65 years of age with severe or moderately severe hemophilia A (endogenous FVIII level ≤2% of normal). Eligible studies were also required to include subjects with a similar number of bleeds in the 12 months prior to the study enrolment as the A-LONG subjects in the individualized prophylaxis arm (arm 1). The weekly prophylaxis with rFVIIIFc (arm 2) is not currently licenced and was not utilized in the model, as it was designed to provide data on prior on-demand patients unable to perform multiple infusions in a week, rather than on patients aiming to maintain a pre-specified trough FVIII activity.

In total, six studies of rFVIII products met the eligibility criteria for comparison with the A-LONG trial. Their characteristics and efficacy readouts are summarized in Table 1. Given the small number of published rFVIII studies and limited data on patient characteristics reported in them, an adjusted indirect comparison was not feasible. The indirect comparison of mean ABRs was based on the estimated mean ABR for the set of included rFVIII studies pooled via meta analysis²⁰.

Table 1. Study characteristics and prophylaxis ABRs reported in RCTs included in the indirect comparison.

Study	Product	Prophylaxis regimen	Number of subjects	ABR Mean ± SD^a	Median weekly factor consumption (IU/kg/week)	Median dose per bleeding episode (IU/kg)	Mean weekly factor consumption (IU/kg/week)	Mean dose per bleeding episode (IU/kg)
Mahlangu et al.^14	rFVIIIFc (Elocta)	1.4–2.4 times weekly; 25–65 IU/kg	117	2.9 ± 3.9	77.90	28.23	88.8^a	36.33
Tarantino et al.^36	rFVIII (Advate)	3–4 times weekly; 25–40 IU/kg	107	6.3 ± 8.5^b	107.45	34.50	NR	52.15
Valentino et al.^18	rFVIII (Advate)	2.3 times weekly; 20–80 IU/kg	34	4.3 ± 5.1	110.55^a	43.91^c	109.75^a	NR
		3.5 times weekly; 20–40 IU/kg	32	3.3 ± 5.8	99.62^a		109.75^a
Recht et al.^37	rFVIII (Xyntha)	3 times weekly; 30–45 IU/kg	94	3.9 ± 6.5	90.60	46.15^d	NR	NR
Lentz et al.^38	rFVIII (Novoeight)	3–4 times weekly; 20–50 IU/kg	150	6.5 ± 8.3^b	NR	NR	71.06	45.60
Lissitchkov et al.^39	rFVIII (Nuwiq)	3–4 times weekly; 30–40 IU/kg	32	2.3 ± 3.7	107.79^e	33.30	107.2^e	60.4
Pollmann et al.^40	rFVIII (REFACTO)	Post-marketing surveillance study; prophylactic regimen defined as ≥2 prophylactic administrations per week for ≥70% of weeks documented during a year	81	6.4 ± 7.1	87.66	29.0	104.3^a	NR

Abbreviation. NR, not reported.

^a These doses were reported as including factor consumption associated with both prophylaxis and the treatment of breakthrough bleeds.

^b Standard deviation (SD) of the ABR for Tarantino et al.³⁶ was not reported, and was estimated assuming Poisson distributions and adjusted for over-dispersion. The adjustment factor used was 3.08. The SD of the ABR for Lentz et al.³⁸ was calculated from the reported 95% confidence interval.

^c Derived as a product of median dose per infusion (30.7 IU/kg) and mean number of infusions per bleeding episode (1.43).

^d Derived as a product of median dose per infusion (30.6 IU/kg) and mean number of infusions per bleeding episode (1.51).

^e Lissitchkov et al.³⁹ have not reported whether Nuwiq doses include the treatment of breakthorough bleeds.

The mean ABR in the A-LONG study of rFVIIIFc was 2.91, whereas the pooled mean ABR estimate for the rFVIII comparator was 4.86. The pooled mean ABR for rFVIII products was significantly higher compared with rFVIIIFc (difference = 2.0; p = .007)²⁰. Clinical studies of rFVIII products Kogenate and Helixate, LIP-LONG and SPINART^4,21, were not included in the indirect comparison due to milder severity of hemophilia A in their study populations. The efficacy of these treatments was assumed to be adequately represented by the pooled mean ABR of other rFVIII products.

The efficacy and factor consumption in severe pediatric patients on prophylaxis with rFVIIIFc was taken from the Kids A-LONG study¹⁵. For children on prophylaxis with conventional rFVIII, ABRs and dosages were derived from published studies^22–24. The mean ABR in the Kids A-LONG was 2.62. The mean ABR for the pooled rFVIII comparator was calculated as the average of the reported values weighted by the study sample size to be 4.55. These data were applied in the model to the cohort aged 1–11 years.

In the base case, the ABR and factor consumption per unit of body weight (IU/kg) associated with each treatment were assumed to remain constant over the duration of the model, as validated by expert opinion in the absence of data on longer-term outcomes of prophylaxis with standard half-life FVIII products. However, longer-term data on bleeds and factor consumption in patients on individualized rFVIIIFc prophylaxis from the ASPIRE study²⁵ were incorporated in a scenario analysis to capture the impact of longer-term prophylaxis with extended half-life FVIII product.

Dosage

The weekly factor consumption (IU/kg/week) observed with each prophylactic treatment was derived from the publications included in the indirect comparison by multiplying the reported treatment frequency per week by the median or mean dose per administration, respectively (Table 1). Compared with most rFVIII studies, the reported weekly factor consumption was lower with rFVIIIFc [mean differences = 15.5–21.8 IU/kg/week (17–26%), median differences = 12.7–29.8 IU/kg/week (16–37%)]²⁰. In addition to prophylaxis, the factor consumption required for resolving bleeding episodes per kilogram of body weight were obtained from the aforementioned studies in pediatric, adolescent, and adult patients. Median doses for prophylaxis and bleed resolution were applied in the base case, since this metric was more commonly reported in the included studies (Table 1). Mean factor consumption data was applied in one of the scenario analyses. In the absence of relevant RCTs of the rFVIII products Kogenate and Helixate, their factor consumption was imputed as the arithmetic average of factor consumption reported for other rFVIII products. While the factor consumption per unit of body weight remained constant for each treatment throughout the duration of the model, the per patient rFVIII utilization varied as the patient progressed through the model in accordance with the age-specific average male weight in Sweden²⁶.

Cost

Cost inputs included the cost of factor needed for prophylaxis and resolution of spontaneous bleeds within prophylaxis. Pharmacy purchase prices (PPP) per unit of each included product were derived from the TLV via Apotekens Service AB²⁷. The analysis uses the approved and reimbursed PPP of 6.15 SEK/IU for rFVIIIFc and 5.69 SEK/IU for each rFVIII product. The unit cost of rFVIIIFc and the pooled rFVIII comparator were multiplied by the median or mean weekly dose and the number of weeks in a year (52.18) to derive the annual cost of factor consumption per kilogram of body weight. The annual cost of the pooled comparator was calculated by weighting the annual cost of each rFVIII by its market share in Sweden, obtained from the IMS MIDAS database. The costs of factor consumption in adolescent and adult patients is shown in Table 2, and those of pediatric patients in Supplementary Table S1.

Table 2. The cost of treating bleeding events and the cost of prophylaxis with recombinant factor VIII products, based on median doses (patients ≥12 year of age).

Product and source	Cost of treating a bleeding episode, per kg (SEK)	Cost of prophylaxis per kg per year (SEK)	Market share based on volume (%)
rFVIIIFc (Elocta)^14	173.61	24,997.97	NA
rFVIII (Advate)*^36	196.31	31,901.48	15.76
rFVIII (Advate)*^18	249.83	32,821.06	15.76
rFVIII (Advate)*^18		29,575.48	15.76
rFVIII (Xyntha)^37	262.57	26,898.78	14.72
rFVIII (Novoeight)^38	212.64	29,870.71	2.94
rFVIII (Nuwiq)^39	189.48	32,002.84	6.17
rFVIII (REFACTO) (calculated)	262.57	26,026.06	14.72
rFVIII (Kogenate) (calculated)	212.64	29,870.71	10.57
rFVIII (Helixate) (calculated)	212.64	29,870.71	3.60
Pooled rFVIII comparator	216.48	29,736.98	100

Abbreviation. NA, not applicable.

* Note; data on Advate administered in three slightly different prophylaxis regimens were available, and all were included for completeness.

To generate the full per patient cost of factor consumption in each cycle, the cost per kilogram per year was multiplied by age-specific male weight obtained from official Swedish statistics²⁶. The cost of resolving bleeding episodes was included by multiplying the median or mean dose per bleeding episode per kilogram (Table 1) by the cost per unit and patient weight to derive the total cost of treating bleeds (Table 2).

Utility

A utility decrement was assigned to bleeding events in either treatment arm, based on the reported utility associated with bleed (0.66) vs non-bleed (0.82) days in hemophilia patients with inhibitors²⁸. In the absence of alternative data, the utility decrement of a bleed was assumed to be applicable to patients without inhibitors. The difference in utility between the bleed and non-bleed states was used to calculate a proportional reduction in baseline utility elicited from a bleeding event. In the absence of published evidence, a bleeding event was assumed to induce a quality of life impairment lasting 7 days on average. In the base case, the utility decrement per bleed was, therefore, multiplied by 7/365 before applying it to the baseline age-specific health-related utility of the Swedish population²⁹. Due to a lack of published data, baseline utility for patients aged 18–24 years were assumed to apply to younger patients.

In addition, utility associated with reduced frequency of prophylactic infusions with rFVIIIFc vs rFVIII was included in the model based on a study of patient and population valuation of treatment attributes for prophylactic regimens in severe hemophilia A³⁰. The web-based survey involved 1,657 participants from the Swedish patient association (FBIS) and the general Swedish population. Respondents assessed four treatment scenarios using the time-trade off methodology to elicit utilities associated with treatment attributes such as factor injection frequency, participation in physical activity, and protection from bleeding³⁰.

Extending the dosing interval from every other day to every fifth day, while maintaining the same ABR and physical activity participation, resulted in a utility increase of 0.04. This utility increment was applied to the proportion of the A-LONG cohort who achieved the 5 day dosing interval within the time frame of the study (39 of 117 subjects, i.e. 33.3%) and was assumed to be constant over the time horizon of the model.

The impact of time on treatment and joint deterioration as measured by Pettersson score on HRQoL was not modeled due to the lack of published evidence.

Mortality

In the absence of hemophilia-specific data, the transition probability between alive and dead states was approximated with the age- and sex-adjusted mortality of the general population in Sweden, derived from official sources³¹. Factor replacement therapy has led to a dramatic decrease in mortality from hemophilia and most patients in developed countries now have near normal life expectancy³²; hence, this was considered a reasonable approximation.

Results

Base case analysis

The base case results of the cost-utility analysis (CUA) are shown in Table 3. With the described base case inputs, rFVIIIFc was found to be a dominant treatment option for life-long prophylaxis, generating fewer costs and greater QALYs than the pooled rFVIII comparator. Over the 70 year time horizon, per patient cost savings of SEK 9.0 million resulted from reduced treatment costs due to lower prophylactic and bleed resolution dosing with rFVIIIFc. An estimated gain of 0.59 QALYs resulted from the lower incidence of bleeding with rFVIIIFc as well as the reduced frequency of treatment administration.

Table 3. Base case results of the cost-utility analysis comparing costs and benefits of routine prophylaxis with rFVIIIFc vs rFVIII in patients with severe hemophilia A (≥1 year of age).

Outcome	rFVIIIFc	rFVIII (pooled)	Incremental
Total cost of bleeding episodes	SEK 975,469	SEK 1,892,270	–SEK 916,801
Total cost of prophylaxis	SEK 44,977,021	SEK 53,042,764	–SEK 8,065,743
Total cost	SEK 45,952,490	SEK 54,935,034	–SEK 8,982,544
Total QALYs	26.72	26.13	0.59
ICER (ΔCost/ΔQALY)			Dominant
NMB			SEK 9,276,441

Abbreviation. QALY, quality-adjusted life year; ICER, incremental cost-effectiveness ratio; NMB, net monetary benefit; SEK, Swedish krona.

Scenario analysis

Results of the cost-minimization analysis suggest that, even with the conservative assumption of no difference in annual bleed rates and no health utility associated with extended injection intervals, routine prophylaxis with rFVIIIFc yields per patient cost-savings of SEK 8.2 million over the 70 year time horizon (Table 4). This reduction in the total cost was due to lower dosing required for routine prophylaxis with rFVIIIFc compared to conventional rFVIII products.

Table 4. Cost-minimization analysis comparing total costs of routine prophylaxis with rFVIIIFc vs rFVIII in patients with severe hemophilia A who are ≥1 year of age. The analysis assumes equal efficacy (as determined by annualized bleed rate) and equal HRQoL associated with the conventional and long-acting rFVIII treatments.

Outcome	rFVIIIFc	rFVIII (pooled)	Incremental
Total cost of bleeding episodes	SEK 975,469	SEK 1,127,149	–SEK 151,680
Total cost of prophylaxis	SEK 44,977,021	SEK 53,042,764	–SEK 8,065,743
Total cost	SEK 45,952,490	SEK 54,169,913	–SEK 8,217,423
Total QALYs	26.32	26.32	0.00
ICER (ΔCost/ΔQALY)			Not applicable
NMB			SEK 8,217,423

Abbreviation. QALY, quality-adjusted life year; ICER, incremental cost-effectiveness ratio; NMB, net monetary benefit; SEK, Swedish krona.

An additional scenario was analyzed to compare the costs and benefits of prophylaxis with rFVIIIFc vs rFVIII based on dosing regimens designed to maintain plasma FVIII at ≥1 IU/dL. Results shown in Supplementary Table S3 suggest that pharmacokinetic-driven prophylaxis with rFVIIIFc vs rFVIII results in greater cost savings but smaller QALY gains than the base case. Since pharmacokinetic prophylaxis was individually adjusted to maintain the trough plasma FVIII above levels associated with increased frequency of breakthrough bleeding (>1 IU/dL), the reduction in QALY gains was as expected. The reduced cost of individualized prophylaxis with FVIIIFc is consistent with the longer time to 1 IU/dL FVIII trough level for rFVIIIFc than for the identical dose of rFVIII¹⁴.

Results of the remaining two scenarios which considered mean instead of median factor consumption values and long-term outcomes of prophylaxis with rFVIIIFc are shown in Supplementary Tables S4 and S5, respectively. In both cases, rFVIIIFc was found to be a dominant treatment option. The scenario based on mean factor consumption data yielded smaller cost savings than the base case, due to the smaller differential in mean factor consumption of rFVIIIFc and rFVIII products compared to their median dosages. The scenario incorporating long-term efficacy and rFVIIIFc consumption from the ASPIRE study generated greater QALY gains but smaller cost savings than the base case. The increase in QALY gains was due to a reduction in ABR observed after ∼2 years of prophylaxis with rFVIIIFc. No such long-term data was available with conventional products; hence, the efficacy and factor consumption derived from the IDC were assumed to remain constant over the 70 year time horizon.

Sensitivity analyses

Due to difficulties associated with the interpretation of negative incremental cost-effectiveness ratios (ICERs), OWSA results were presented using net monetary benefit (NMB) via a tornado diagram (Figure 2). Assuming a willingness-to-pay (WTP) threshold of SEK 500,000, the base case NMB was SEK 9.3 million, as shown at the baseline of the tornado diagram. OWSA results demonstrate that model outputs were most sensitive to variation in the median weekly dose of rFVIIIFc and rFVIII comparators, discount rate, and time horizon. Nevertheless, variation in any of the tested parameters outlined in Supplementary Table S2 did not generate a scenario in which rFVIIIFc was not cost-effective (with a negative net monetary benefit), suggesting that base case cost-effectiveness estimates were robust to changes in key inputs.

Figure 2. Tornado diagram rFVIIIFc vs rFVIII. The diagram depicts the variation of net monetary benefit (NMB) from the base case value of SEK 9.3 million, at lower and upper estimate of the value for 10 most influential parameters. The middle of the tornado diagram corresponds to the base case NMB, the blue bars represent the cost-effectiveness at low estimate of each parameter, and the orange bars at high estimate.

The cost-effectiveness acceptability curve generated from the PSA shows a greater than 85% likelihood of routine prophylaxis with rFVIIIFc being cost-effective over a wide range of WTP thresholds (Figure 3). This is particularly relevant in Sweden, which has no formal WTP threshold³³.

Figure 3. Cost-effectiveness acceptability curve for rFVIIIFc vs rFVIII.

Discussion

This CUA investigates the cost-effectiveness of routine prophylaxis with rFVIIIFc vs conventional rFVIII by evaluating long-term costs and health outcomes in patients with severe hemophilia A. In addition to drug efficacy, our analysis incorporates patients’ perception of their HRQoL through utility associated with extending the dosing interval while maintaining low bleeding rate³⁰. Base case results indicate that life-long prophylaxis with rFVIIIFc dominates the pooled rFVIII comparator due to lower factor consumption for routine prophylaxis and resolution of bleeds, and greater quality-of-life resulting from reduced incidence of bleeds and lower treatment burden.

Due to the small number of published rFVIII studies and unavailability of patient level data, the estimated differences in efficacy and weekly factor consumption with rFVIIIFc vs conventional rFVIII were based on a non-adjusted indirect comparison, and could have, therefore, been biased. In addition, evidence from a recent retrospective study of the Malmö hemophilia register suggested that a disparity may exist between real world bleeding rates in Swedish patients on rFVIII prophylaxis and those observed in the RCTs, with real world ABRs being as low as 1.9³⁴. Although low real world ABRs may have been attributable to under-reporting, we addressed the uncertainty in the estimates of efficacy by performing a cost-minimization analysis which assumed equal efficacy between treatment arms and no HRQoL impact associated with injection frequency. Even under these conservative assumptions, rFVIIIFc remained a cost-saving treatment option as a result of median (as well as mean) rFVIIIFc consumption in A-LONG study being lower than that of rFVIIIs observed in the included trials. The assumption of lower factor consumption with rFVIIIFc is supported by the findings from a recent retrospective study of factor utilization in Canadian patients switching from standard half-life rFVIII to rFVIIIFc. In the 6 months post-switching, the median factor utilization was found to be reduced by 19% compared to 6 months pre-switch³⁵, lending support to our assumption that the improved pharmacokinetic profile of rFVIIIFc leads to lower weekly consumption of this product vs the standard half-life rFVIII.

In the scenario comparing pharmacokinetic-tailored prophylaxis with rFVIIIFc and rFVIII designed to maintain plasma FVIII trough levels at or above 1 IU/dL, rFVIIIFc was also found to be a dominant treatment option, generating a slightly higher NMB than the base case, due to greater savings in factor consumption with rFVIIIFc vs rFVIII. The scenario incorporating long-term efficacy and rFVIIIFc consumption from the ASPIRE study preserved the dominance of rFVIIIFc, as did the analysis based on mean factor consumption data.

The results of OWSA and PSA suggested that base case outputs were robust to parameter variation.

One-way variation of 29 different variables from their low to high estimates in no instance resulted in cost-effectiveness not being achieved. The PSA indicated a greater than 85% probabilitiy of rFVIIIFc being cost-effective at tested WTP thresholds between SEK 0 and SEK 1,000,000.

Our study has several limitations. In the absence of long-term data, treatment benefits and factor consumption were assumed to be constant over the model duration, an approach validated by expert opinion. The impact of reduced ABR on HRQoL was captured only through a disutility associated with bleeding, while any effect on attenuation of joint deterioration was omitted due to lack of quantitative data linking ABRs to joint deterioration. The model outputs are, therefore, likely to underestimate the QALY gains associated with lower ABR achievable with rFVIIIFc. Future studies establishing a quantitative relationship between ABR and clinical events for which an effect on quality of life is clearly demonstrated will enhance the robustness of economic models. Furthermore, the disutility of a bleeding event was considered independent of its timing or type. In practice, the impact of a bleed on HRQoL is likely to vary with the type of bleed, as well as by the number of preceding bleeds. However, data capturing bleed-associated quality of life at this level of granularity are not currently available.

The costs associated with joint surgery and specialist consultations were not included due to the difficulty in attributing their incidence to cumulative bleeds or clinical measures of joint deterioration. Cost savings attributable to prophylaxis with rFVIIIFc may, therefore, have been underestimated.

The only costs assigned to bleeding events were those of factor consumption. In practice, a proportion of bleeding events, which range in severity, are likely to require specialist consultations. No additional costs were assigned to the treatment with rFVIII to account for more frequent administration, despite the fact that in current clinical practice a proportion of infusions are administered by a nurse at a hemophilia treatment center. Furthermore, the absence of data linking lost work days with ABR and clinical measures of joint damage precluded the inclusion of indirect costs in the model.

Conclusion

This CUA is the first to compare two prophylactic therapies for hemophilia whilst incorporating a novel aspect of HRQoL associated with the frequency of prophylactic infusions. Our results suggest that extended half-life rFVIIIFc may be a cost-effective life-long treatment option in severe hemophilia A patients, which lessens the treatment burden without compromising outcomes. Although additional comparative research is needed to gain an accurate understanding of its efficacy and factor utilization, our study reaffirms the role of extended half-life rFVIIIFc in the clinical practice.

Transparency

Declaration of funding

This research was funded by Swedish Orphan Biovitrum AB (publ) (Sobi).

Declaration of financial/other relationships

NH, JJ, and MS are consultants of IQVIA, a consulting company that has received funding from Sobi for the current study. PK and KW are employees of Sobi, a company involved in the development and commercialization of rFVIIIFc (Elocta). K-J M is an employee of Sobi and a Sobi share-holder.

Previous presentations

This work was presented in poster format at the 10th Annual Congress of the European Association of Haemophilia and Allied Disorders in Paris, February 1–3, 2017.