`D01 10.1007/s40263 014 0207 x
`
`ORI GINA L RESEARC H ARTIC L E
`
`Cost Effectiveness of Fingolimod, Teriftunomide,
`Dimethyl Fumarate and Intramuscular Interferon-P1a
`in Relapsing-Remitting Multiple Sclerosis
`
`Xinke Zhang· Joel W. Hay· Xiaoli Niu
`
`Published online: 19 October 2014
`© Springer international Publishing Switz.erland 2014
`
`Abstract
`Objective The aim of the study was to compare the cost
`teriflunomide, dimethyl
`effectiveness of fingolimod,
`fumarate, and intramuscular (IM) interferon (IFN)- ~1 a as
`in the treatment of patients with
`first-line therapies
`relapsing-remitting multiple sclerosis (RRMS).
`Methods A Markov model was developed to evaluate the
`cost effectiveness of disease-modifying drugs (DMDs) from
`a US societal perspective. The time horizon in the base case
`was 5 years. The primary outcome was incremental net
`monetary benefit (INMB), and the secondary outcome was
`incremental cost-effectiveness ratio (ICER). The base case
`INMB willingness-to-pay (WTP) threshold was assumed to
`be US$150,000 per quality-adjusted life year (QALY), and
`the costs were in 2012 US dollars. One-way sensitivity
`analyses and probabilistic sensitivity analysis were con(cid:173)
`ducted to test the robustness of the model results.
`Results Dimethy 1 fumarate dominated all other therapies
`over the range of WTPs, from US$0 to US$180,000.
`Compared with IM IFN-~1 a, at a WTP of US$150,000,
`INMBs were estimated at US$36,567, US$49,780, and
`US$80,611 for fingolimod, teriflunomide, and dimethyl
`fumarate, respectively. The ICER of fingolimod versus
`teriflunomide was US$3,201,672. One-way sensitivity
`analyses demonstrated the model results were sensitive to
`the acquisition costs of DMDs and the time horizon, but in
`most scenarios, cost-effectiveness rankings remained sta(cid:173)
`ble. Probabilistic sensitivity analysis showed that for more
`
`X. Zhang · J. W. Hay (121) · X. Niu
`Department of Clinical Pharmacy and Pharmaceutical
`Economics and Policy, Leonard D. Schaeffer Center for Health
`Policy and Economics, University of Southern California,
`University Park Campus, VPD 214 L, Los Angeles,
`CA 90089 3333, USA
`e mail: jhay@usc.edu
`
`than 90 % of the simulations, dimethyl fumarate was the
`optimal therapy across all WTP values.
`Conclusion The three oral therapies were favored in the cost(cid:173)
`effecti veness analysis. Of the four DMDs, dimethyl fumarate
`was a dominant therapy to manage RRMS. Apart from dime(cid:173)
`thyl fumarate, teriflunomide was the most cost-effective ther(cid:173)
`apy compared with IM IFN- ~1 a, with an ICER ofUS$7,115.
`
`Key Points
`
`This is the first cost-effectiveness analysis in
`relapsing-remitting multiple sclerosis to (1) make
`comprehensive comparisons between the three new oral
`disease-modifying drugs and the established therapy
`intramuscular (IM) interferon (IFN)-~1 a, (2) incorporate
`second-line therapy in the model, and (3) presentresults
`in terms of incremental net monetary benefit (INMB)
`
`Dimethyl fumarate dominated all other therapies
`over the range of willingness-to-pay (WTP) values,
`from US$0 to US$180,000. Compared with IM IFN(cid:173)
`~1a, at a WTP of US$150,000, INMBs were
`estimated at US$36,567, US$49,780, and US$80,611
`for fingolimod, teriflunomide, and dimethyl
`fumarate, respectively. The three oral therapies were
`favored in the cost-effectiveness analysis
`
`After dimethyl fumarat.e, teriflunomide was the most
`cost-effective therapy compared with IM IFN- ~1 a,
`with an incremental cost-effectiveness ratio of
`US$7,115. When the monthly cost is below
`US$5,132, fingolimod is cost effective compared
`with IM IFN- ~ia· However, fingolimod is not cost
`effective compared with teriflunornide
`
`L\Adis
`
`MYLAN PHARMS. INC. EXHIBIT 1100 PAGE 1
`
`
`
`72
`
`1 Introduction
`
`Before the introduction of oral fingolimod (GilenyaTM,
`Novartis, East Hanover, NJ, USA), over half of the patients
`with relapsing-remitting multiple sclerosis (RRMS) who
`were treated with disease-modifying drugs (DMDs) were
`using injected interferons (IFNs) [1], and intramuscular
`Ò
`(IM) IFN-b1a (Avonex
`, Biogen Idec, Weston, MA, USA)
`had the largest market share in 2010 [2]. However, IM
`IFN-b1a and other traditional DMDs require long-term
`parenteral administration, which imposes a burden on
`patients and may have a significant impact on medication
`adherence. Over the past few years, three new oral DMDs,
`Ò
`namely fingolimod,
`teriflunomide
`(Aubagio
`, Sanofi
`Aventis, Cambridge, MA, USA), and dimethyl fumarate
`Ò
`(Tecfidera
`, Biogen Idec, Weston, MA, USA), were
`approved by the FDA in 2010, 2012, and 2013, respec-
`tively. Fingolimod was the first oral therapy approved, and
`the Trial Assessing Injectable Interferon versus FTY720
`Oral in Relapsing-Remitting Multiple Sclerosis (TRANS-
`FORMS) showed that fingolimod appeared to be more
`effective than IM IFN-b1a in reducing the frequency of
`relapses [3]. The large-scale phase III clinical trials the
`Teriflunomide Multiple Sclerosis Oral (TEMSO) trial and
`the Determination of the Efficacy and Safety of Oral
`Fumarate in Relapsing-Remitting MS (DEFINE) trial also
`demonstrated that teriflunomide and dimethyl fumarate,
`respectively, significantly reduced annualized relapse rates,
`slowed disability progression, and reduced the number of
`lesions on magnetic resonance imaging [4, 5]. Although
`these new oral therapies were thought to contribute to the
`growth of the total costs of multiple sclerosis (MS), so far
`there is no comprehensive evidence on either the cost
`effectiveness of the new oral DMDs compared with the
`established treatment IM IFN-b1a, or incremental cost
`effectiveness among the oral therapies. For these reasons,
`this paper compares the cost effectiveness of fingolimod,
`teriflunomide, dimethyl fumarate, and IM IFN-b1a as first-
`line therapies in the treatment of patients diagnosed with
`RRMS.
`
`X. Zhang et al.
`
`2 Materials and Methods
`
`2.1 Model Overview
`
`The cost-effectiveness analysis was conducted from a US
`societal perspective over a 5-year time horizon. We chose
`5-year rather than 10-year or life time as the time horizon
`because (1) extrapolating a 1- or 2-year randomized con-
`trolled trial (RCT) over long time horizons requires more
`unreliable assumptions on model extrapolations [6] and (2)
`high discontinuation rates imply that a large proportion of
`patients will discontinue or develop secondary-progressive
`multiple sclerosis (SPMS) over time [3 5, 7]. Costs were
`reported in 2012 US dollars, and both costs and outcomes
`were discounted at a 3 % annual rate in the base case
`scenario. The primary outcome was incremental net mon-
`etary benefit (INMB), and the secondary outcome was
`incremental cost-effectiveness ratios (ICERs). INMB was
`chosen as the primary outcome since, when comparing
`multiple treatment options,
`it more clearly delineates
`treatments with dominance or extended dominance [8, 9].
`The willingness-to-pay (WTP) threshold was assumed to
`be US$150,000 per quality-adjusted life year (QALY),
`which is three times the 2012 US gross domestic product
`(GDP) per capita, as recommended by the World Health
`Organization [10, 11]. The choice of US$150,000 as the
`WTP threshold rather than the antiquated US$50,000 value
`in the US context is also supported by the study of Brai-
`thwaite et al. [12] and is used in numerous previous studies
`[13 16].
`Ò
`A Markov model was developed in Microsoft
`Excel to
`simulate the disease progression of patients with RRMS
`(Fig. 1). The cycle is 1 month. The comparators included
`oral fingolimod at a daily dose of 0.5 mg, oral terifluno-
`mide 14 mg once daily, oral dimethyl fumarate 120 mg
`twice a day for the first 7 days and 240 mg twice a day
`after 7 days, and IM IFN-b1a at a weekly dose of 30 lg
`[17]. The disease progression was modeled by the Expan-
`ded Disability Status Scale (EDSS), which is most widely
`used in the measurement of MS [18]. Specifically, health
`
`Fig. 1 Markov model for the
`disease progression of multiple
`sclerosis. EDSS Expanded
`Disability Status Scale
`
`EDSS 0.0-2.5
`Relapse
`
`EDSS 3.0-5.5
`Relapse
`
`EDSS 0.0-2.5
`
`EDSS 3.0-5.5
`
`EDSS 6.0-7.5
`
`EDSS 8.0-9.5
`
`EDSS 10.0
`(Death)
`
`All health states may
`progress to death
`
`MYLAN PHARMS. INC. EXHIBIT 1100 PAGE 2
`
`
`
`Cost Effectiveness of Oral Disease Modifying Drugs Versus IM IFN b1a
`
`73
`
`states were divided as EDSS 0.0 2.5 (no or mild disabil-
`ity), EDSS 3.0 5.5 (moderate disability, ambulatory with-
`out aid), EDSS 6.0 7.5 (waking aid required), EDSS
`8.0 9.5 (restricted to bed), EDSS 10.0 (death) and another
`two relapse states. Since MS is a progressive disease,
`patients were assumed to only progress to a more severe
`health state or a relapse state.
`A cohort of 1,000 patients was assumed to enter the
`model. Consistent with the clinical trials, all patients were
`initially distributed to EDSS 0.0 2.5 and 3.0 5.5 states and
`treated with first-line DMDs [3 5]. The EDSS distribution
`ratio between the two states was estimated from a national
`cross-sectional survey [19]. In any cycle during the simu-
`lation, patients could discontinue the drug and then tran-
`sition to a second-line treatment, natalizumab, or to the
`symptom management
`(SM) arm without active drug
`therapy. Moreover, patients could also discontinue natal-
`izumab due to insufficient response or adverse events and
`then switch to SM treatment.
`The decision to choose natalizumab as the second-line
`therapy was based on the fact that (1) natalizumab was
`specifically indicated for use when previous DMDs failed,
`as recommended by American Academy of Neurology
`[20]; (2) a retrospective cohort study found that approxi-
`mately 10 % of patients who were initially treated with
`IFN-b or glatiramer acetate (GA) experienced break-
`through disease and either switched to natalizumab or an
`immunosuppressant (e.g., mitoxantrone) or declined new
`therapy [21] (however, according to another study, which
`followed a cohort from 2000 to 2008, only 1 % of the first-
`line and second-line DMD populations used mitoxantrone
`[22]); and (3) other first-line drugs are often used as sec-
`ond-line therapies, despite not being indicated after failure
`of a previous DMD, and they are actually similar in
`
`efficacy; however, there is evidence that switching to na-
`talizumab is more effective than switching to other first-
`line drugs [23]. Therefore, patients were assumed to
`receive natalizumab as second-line therapy.
`Patients in EDSS 0.0 2.5 and 3.0 5.5 states would
`likely transition to a temporary state of relapse and stay for
`a cycle (1 month). Following a relapse, patients could
`transition back to the previous state or progress to a next
`more severe health state. According to a recent natural
`history study of SPMS, for patients initially diagnosed with
`RRMS, 80.0 % reached SPMS at or before EDSS 6.0 and
`99.5 % at or before EDSS 8.0 [24]. That is to say, for those
`transitioned to EDSS 6.0, at least 80 % of the patients
`would have already reached SPMS, and so would almost
`all of the patients who progressed to EDSS 8.0. Therefore,
`it was assumed that patients in EDSS 6.0 7.5 and EDSS
`8.0 9.5 had developed SPMS and thus were not associated
`with further relapses. Since these DMDs are indicated for
`relapse forms of MS, patients transitioned to EDSS 6.0 7.5
`and EDSS 8.0 9.5 would stop DMD treatment and then be
`treated with SM.
`The model design in this paper was consistent with
`previous cost-effectiveness studies comparing DMDs in
`that the same health states were classified and the same
`disease progression path was defined [2, 25 27]. The health
`states were decided in a way that the transition points
`(EDSS 3.0, 6.0, 8.0, and 10) reflected key disability levels
`in the natural history of MS and are critical in defining
`clinical course [7, 28 30]. In our model, we also allowed
`the patients to switch to second-line DMD treatment when
`they discontinued the first-line therapy, to better reflect
`clinical practice [20, 31]. In addition, we had each author
`verify the model equations and computations indepen-
`dently to ensure the internal validity [32].
`
`Table 1 Baseline characteristics of the patients
`
`Variable
`
`FREEDOMS [33]
`
`TRANSFORMS [3]
`
`TEMSO [4]
`
`DEFINE [5]
`
`Placebo
`
`FIN
`
`FIN
`
`IM IFN b1a
`
`TER
`
`DF
`
`Demographic characteristics
`
`Age, years
`
`Mean ± SD
`
`Median (range)
`
`Female sex, %
`
`White race, %
`
`Clinical characteristics
`
`Relapses, No.
`
`37.2 ± 8.6
`
`36.6 ± 8.8
`
`36.7 ± 8.8
`
`36.0 ± 8.3
`
`37.8 ± 8.2
`
`38.1 ± 9.1
`
`37.0 (18 55)
`
`36.0 (18 55)
`
`37 (18 55)
`
`36 (18 55)
`
`71.30
`
`69.60
`
`65.40
`
`93.70
`
`67.80
`
`93.80
`
`71
`
`96.90
`
`72
`
`78
`
`In previous year, mean ± SD
`
`1.4 ± 0.7
`
`In previous 2 years, mean ± SD
`
`2.2 ± 1.2
`
`EDSS score, mean ± SD
`
`2.5 ± 1.3
`
`1.5 ± 0.8
`
`2.1 ± 1.1
`
`2.3 ± 1.3
`
`1.5 ± 1.2
`
`2.3 ± 2.2
`
`1.5 ± 0.8
`
`2.3 ± 1.2
`
`1.3 ± 0.7
`
`2.2 ± 1.0
`
`1.3 ± 0.7
`
`2.24 ± 1.33
`
`2.19 ± 1.26
`
`2.67 ± 1.24
`
`2.40 ± 1.29
`
`DF dimethyl fumarate, EDSS Expanded Disability Status Scale, FIN fingolimod, IFN interferon, IM intramuscular, TER teriflunomide
`
`MYLAN PHARMS. INC. EXHIBIT 1100 PAGE 3
`
`
`
`74
`
`2.2 Patient Characteristics
`
`The baseline characteristics of the modeled patients were
`very similar in the phase III clinical trials across the four
`DMDs (Table 1)
`[3 5, 33]. Generally, patients were
`between 18 and 55 years old, had a diagnosis of RRMS,
`had had at least two relapses during the previous 2 years or
`at least one relapse during the previous year before ran-
`domization, and had an EDSS score of 0 5.5. Based on a
`national survey study, the initial proportions of patients
`distributed in EDSS 0.0 2.5 and EDSS 3.0 5.5 were esti-
`mated at 41.3 and 58.7 %, respectively [19].
`
`2.3 Transition Probabilities
`
`Transition probabilities for disease progression, relapses,
`and discontinuation were obtained from the literature and
`modeled using the DEALE method (Table 2) [3 5, 33
`36]. For patients in SM, the EDSS progression proba-
`bilities were estimated from the London Ontario natural
`history study of MS [7]. The London Ontario data were
`used because, unlike in other studies, the patients in the
`study did not receive disease-modifying therapies and the
`database was subjected to a rigorous quality review in
`2009 [37]. There were 806 RRMS-onset patients in the
`database, and the shortest follow-up was 16 years. Since
`the patients were similar in demographics and clinical
`characteristics, for patients treated with fingolimod, teri-
`flunomide, dimethyl
`fumarate, and natalizumab,
`the
`hazard ratios of disease progression for DMDs compared
`with placebo reported in phase III placebo-controlled
`trials were used to derive the 1-month transition proba-
`bilities for each DMD. For the IFN-b1a arm, the hazard
`ratio from the head-to-head trial TRANSFORMS between
`fingolimod and IFN-b1a was also used to estimate tran-
`sition probabilities [3].
`The transition probabilities of relapses for patients in
`SM were obtained from the placebo group in the FTY720
`Research Evaluating Effects of Daily Oral Therapy in
`Multiple Sclerosis (FREEDOMS) trial [33]. Hazard ratios
`of
`relapses between DMDs
`(teriflunomide, dimethyl
`fumarate, and natalizumab) and placebo were used to
`derive the transition probabilities to relapse state for the
`DMDs. For patients treated with fingolimod and IFN-b1a,
`relapse probabilities were estimated by using the data in the
`TRANSFORMS trial [3]. All discontinuation rates were
`extracted from the corresponding phase III clinical trials.
`After discontinuation of the first-line therapy, the assign-
`ment ratio between natalizumab and SM was assumed to be
`equal in the base case scenario, and extreme cases were
`tested in sensitivity analyses. Since the mortality rate due
`to MS is very low, survival probabilities were based on the
`mortality rates of the general population [38]. The age-
`
`X. Zhang et al.
`
`specific mortality rates were estimated from the life
`expectancy data in national vital statistics reports using the
`DEALE method [35, 36, 39].
`
`2.4 Utilities
`
`Since utilities were not available in the pivotal RCTs, we
`reviewed the literature and identified the best available
`evidence to support the utility estimates. The utilities for
`each health state from EDSS 0.0 to EDSS 9.5 and the
`disutility for IFN-b1a were obtained from the Prosser et al.
`[40] quality-of-life study. The study used the standard-
`gamble method to measure patient and community pref-
`erences for treatments and health states in patients with
`RRMS. The Prosser et al. [41] study was used because
`standard-gamble was thought to be the gold standard in
`preference elicitation since it is the only method that esti-
`mates Von-Neumann Morgenstern utility (preference
`measured under uncertainty) [41]. Also, since this study
`was performed from a societal perspective, use of com-
`munity preferences was more appropriate as it reflected the
`society’s preference on the resource allocation [42]. Dis-
`utility for relapses was based on the Kobelt et al. study
`[19]. For the effects of fingolimod and natalizumab, though
`there was evidence that fingolimod and natalizumab could
`improve the quality of life of MS patients significantly [43
`45], no study on utility impacts was available. Therefore, to
`be conservative, the disutility for fingolimod and natal-
`izumab was assumed to be 0 in the base case scenario.
`Changes in assumed base case utility were explored in
`sensitivity analyses. For
`teriflunomide, one study has
`demonstrated that there was no disutility associated with
`administration of teriflunomide, so the impact of teriflun-
`omide on utility was assumed to be 0 in the base case
`analysis [46]. Dimethyl fumarate has been reported to have
`significant improvements in physical and mental aspects of
`health and functioning, where the change in EQ-5D value
`was 0.01 [47, 48]. The base case utilities and the effects of
`DMDs on utilities are shown in Table 2.
`
`2.5 Costs
`
`Costs in the model were mainly obtained from the cost-of-
`illness study by Kobelt et al. [19] and inflated to 2012
`dollars (Table 2). We applied the results from the Kobelt
`et al. [19] study because the costs were reported on the
`basis of stratified EDSS score, which corresponded to each
`health state in our model. The costs included costs of
`hospital
`inpatient care, ambulatory care,
`tests, drugs
`(DMDs and other drugs), services, adaptations and costs of
`informal care. The productivity losses were not included,
`because the costs associated with productivity were cap-
`tured in the QALYs [49]. All drug costs were obtained
`
`MYLAN PHARMS. INC. EXHIBIT 1100 PAGE 4
`
`
`
`Cost Effectiveness of Oral Disease Modifying Drugs Versus IM IFN b1a
`
`75
`
`Table 2 Parameters and range
`in one way sensitivity analysis
`
`Parameters
`
`Base case
`
`One-way SA rangea
`
`Sources
`
`Monthly probability of disease progression (SM)
`
`EDSS 0.0 2.5
`
`EDSS 3.0 5.5
`
`EDSS 6.0 7.5
`
`Monthly probability of progressing to death
`
`EDSS 0.0 2.5
`
`EDSS 3.0 5.5
`
`EDSS 6.0 7.5
`
`EDSS 8.0 9.5
`
`Annual relapse rate for SM
`
`Annual relapse rate for FIN
`Annual relapse rate for IM IFN-b1a
`HR of disease progression
`
`0.005760
`
`0.007194
`
`0.005760
`
`0.001684
`
`0.002348
`
`0.003121
`
`0.004457
`
`0.400
`
`0.160
`
`0.330
`
`0.700
`
`1.353
`
`N/A
`
`N/A
`
`N/A
`
`0.120
`
`0.248
`
`0.525
`
`1.015
`
`0.200
`
`0.413
`
`0.875
`
`1.692
`
`[7]
`
`[7]
`
`[7]
`
`[39]
`
`[39]
`
`[39]
`
`[39]
`
`[39]
`
`[3]
`
`[3]
`
`[33]
`
`[3]
`
`[4]
`
`FIN vs. SM
`IM IFN-b1a vs. FIN
`TER vs. SM
`
`DF vs. SM
`
`NAT vs. SM
`
`HR of annual relapse rate
`
`TER vs. SM
`
`DF vs. SM
`
`NAT vs. SM
`
`Annual discontinuation rate for FIN
`Annual discontinuation rate for IM IFN-b1a
`Discontinuation rate for TER, 2 year
`
`Discontinuation rate for DF, 2 year
`
`Discontinuation rate for NAT, 2 year
`
`0.700
`
`0.620
`
`0.580
`
`0.720
`
`0.510
`
`0.410
`
`0.103
`
`0.118
`
`0.265
`
`0.310
`
`0.083
`
`0.5:0.5
`
`0.525
`
`0.465
`
`N/A
`
`0.540
`
`0.383
`
`N/A
`
`0.077
`
`0.089
`
`0.199
`
`0.233
`
`N/A
`
`0:1
`
`0.875
`
`0.775
`
`0.900
`
`0.638
`
`0.128
`
`0.148
`
`0.332
`
`0.388
`
`1:0
`
`[5]
`
`[34]
`
`[4]
`
`[5]
`
`[34]
`
`[3]
`
`[3]
`
`[4]
`
`[5]
`
`[34]
`
`Assignment ratio between NAT and SM
`
`Utilities estimates
`
`Utility EDSS 0.0 0 2.5
`
`Utility EDSS 3.0 0 5.5
`
`Utility EDSS 6.0 0 7.5
`
`Utility EDSS 8.0 0 9.5
`
`Disutility for relapse
`Disutility for IM IFN-b1a
`Impact of FIN on utility
`
`Impact of TER on utility
`
`Impact of DF on utility
`
`Impact of NAT on utility
`
`Monthly costs, 2012 US dollars
`
`0.899
`
`0.821
`
`0.769
`
`0.491
`
`0.094
`
`0.115
`
`0
`
`0
`
`0.01
`
`0
`
`0.674
`
`0.616
`
`0.577
`
`0.368
`
`0.071
`
`0.086
`
`0.03
`
`0.03
`
`0.03
`
`N/A
`
`1
`
`1
`
`0.961
`
`0.614
`
`0.118
`
`0.144
`
`0.03
`
`0.03
`
`0.03
`
`[40]
`
`[40]
`
`[40]
`
`[40]
`
`[19]
`
`[40]
`
`[43, 44]
`
`[46]
`
`[47, 48]
`
`[45]
`
`$4,164
`
`$3,123
`
`$5,204
`
`[50]
`
`WAC for FIN
`WAC for IM IFN-b1a
`WAC for NAT
`
`WAC for TER
`
`WAC for DF
`
`Cost of EDSS 0.0 2.5
`
`Cost of EDSS 3.0 5.5
`
`Cost of EDSS 6.0 7.5
`
`Cost of EDSS 8.0 9.5
`
`Cost of relapse
`
`Discount rate
`
`Time horizon
`
`DF dimethyl fumarate, EDSS
`Expanded Disability Status Scale,
`FIN fingolimod, HR hazard ratio,
`IFN interferon, IM intramuscular,
`NAT natalizumab, SA sensitivity
`analysis, SM symptom
`management, TER teriflunomide,
`WAC wholesale average cost
`a ±25 % unless indicated
`
`$3,835
`
`$3,320
`
`$3,704
`
`$3,346
`
`$615
`
`$1,289
`
`$3,198
`
`$6,369
`
`$5,008
`
`0.03
`
`$2,876
`
`$2,490
`
`$2,778
`
`$2,509
`
`$1,298
`
`$2,768
`
`$4,047
`
`$8,093
`
`$3,756
`
`0
`
`5 years
`
`2 years
`
`[50]
`
`[50]
`
`[50]
`
`[50]
`
`[19]
`
`[19]
`
`[19]
`
`[19]
`
`[51]
`
`$4,794
`
`$4,150
`
`$4,630
`
`$4,182
`
`$2,163
`
`$4,614
`
`$6,744
`
`$13,489
`
`$6,259
`
`0.05
`
`10 years
`
`MYLAN PHARMS. INC. EXHIBIT 1100 PAGE 5
`
`
`
`76
`
`X. Zhang et al.
`
`from the Federal Supply Schedule drug prices [50]. Costs
`of relapses were estimated from a cross-sectional, web-
`based survey that investigated the impacts of relapses on
`costs and quality of life for patients with RRMS in the USA
`[51].
`
`2.6 Sensitivity Analysis
`
`One-way sensitivity analyses were conducted to test the
`robustness of the pairwise comparisons of three oral ther-
`apies versus IM IFN-b1a and the robustness of the optimal
`therapy selection by using INMB as the outcome. The base
`case inputs of the parameters were varied by 25 % in both
`positive and negative directions, unless indicated otherwise
`(Table 2). We varied the parameters by 25 % so that the
`upper and lower bound of the sensitivity analysis range
`differed markedly from the base case inputs. The 25 %
`variation ranges of the parameters were also comparable to
`their corresponding 95 % confidence intervals where
`available. Key parameters that may affect
`the disease
`progression, utilities and costs were included in the ana-
`lysis. In addition, a threshold analysis was conducted if a
`
`parameter variation resulted in a major change in conclu-
`sion [49]. Tornado diagrams were plotted in the order from
`most sensitive parameter to least sensitive. Moreover,
`sensitivity to time horizon was specifically tested by
`varying the time horizon from 2 to 30 years under both
`discounted and non-discounted scenarios.
`The robustness of the base case results was also tested
`by probabilistic sensitivity analysis based on a second
`order Monte Carlo simulation (1,000 times). Choice of
`the distribution for the model inputs was based on the
`recommendations
`and reflected how the
`confidence
`interval of each parameter was estimated [52]. The dis-
`tributions of hazard ratios and annual relapse rates were
`assumed to be log-normal [53]. Utilities were assumed to
`follow a beta distribution which is confined between 0
`and 1 [53]. Health care costs for each health state and
`drug acquisition costs were assumed to follow a gamma
`distribution [53]. The result of the probabilistic sensitivity
`analysis was reported as the probability of each drug
`maximizing the net monetary benefits (NMBs) over the
`range of WTPs [54]. That is the probability of each drug
`being the optimal therapy.
`
`Table 3 Results for the base case scenario (WTP
`
`US$150,000)
`
`DMDs
`
`Cost
`
`QALY
`
`NMB
`
`INMB vs. IM IFN b1a
`
`ICER vs. IM IFN b1a
`
`ICER: FIN vs. TER
`
`IM IFN b1a
`TER
`
`FIN
`
`DF
`
`$223,606
`
`$226,085
`
`$239,947
`
`$200,145
`
`3.34
`
`3.68
`
`3.69
`
`3.72
`
`$276,745
`
`$326,525
`
`$313,312
`
`$357,356
`
`$49,780
`
`$36,567
`
`$80,611
`
`$7,115
`
`$46,328
`
`Dominant
`
`$3,201,672
`
`DF dimethyl fumarate, DMDs disease modifying drugs, FIN fingolimod, ICER incremental cost effectiveness ratio, IFN interferon, IM intra
`muscular, INMB incremental net monetary benefit, NMB net monetary benefit, QALY quality adjusted life year, TER teriflunomide, WTP
`willingness to pay
`
`0
`
`20000
`
`40000
`
`60000
`
`80000
`
`100000
`
`120000
`
`140000
`
`160000
`
`180000
`
`Dimethyl fumarate
`
`Teriflunomide
`
`Fingolimod
`
`IM IFN β 1a
`
`Willingness-to-pay
`
`500000
`
`400000
`
`300000
`
`200000
`
`100000
`
`0
`
`100000
`
`200000
`
`300000
`
`Net monetary benefits
`
`Fig. 2 Net monetary benefits of
`the disease modifying drugs.
`IFN interferon,
`IM intramuscular
`
`MYLAN PHARMS. INC. EXHIBIT 1100 PAGE 6
`
`
`
`Cost Effectiveness of Oral Disease Modifying Drugs Versus IM IFN b1a
`
`77
`
`Fig. 3 Results for one way
`sensitivity analysis. DF
`dimethyl fumarate, EDSS
`Expanded Disability Status
`Scale, FIN fingolimod, IFN
`interferon, IM intramuscular,
`NAT natalizumab, SM symptom
`management, TER
`teriflunomide, WAC wholesale
`average cost, yr years
`
`(a) Fingolimod vs. IM IFN β-1a
`Monthly WAC for FIN
`Monthly WAC for IM IFN β 1a
`Time horizon
`Impact of FIN on utility
`Disutility for IM IFN β 1a
`Annual discontinuation rate for IM IFN β 1a
`Annual discontinuation rate for FIN
`Utility EDSS 6.0 0 7.5
`Discount rate
`Assignment ratio between NAT and SM
`$10,000
`
`(b) Teriflunomide VS. IM IFN β-1a
`Monthly WAC for TER
`Monthly WAC for IM IFN β 1a
`Time horizon
`Impact of TER on utility
`Disutility for IM IFN β 1a
`Annual discontinuation rate for IM IFN β 1a
`Discontinuation rate for TER, 2yr
`Assignment ratio between NAT and SM
`Utility EDSS 6.0 0 7.5
`Discount rate
`
`$70,000
`$50,000
`$30,000
`$10,000
`Incremental net monetary benefit
`
`$90,000
`
`$0
`
`$80,000
`$60,000
`$40,000
`$20,000
`Incremental net monetary benefit
`
`$100,000
`
`(c) Dimethyl fumarate VS. IM IFN β-1a
`Time horizon
`Monthly WAC for IM IFN β 1a
`Monthly WAC for DF
`Disutility for IM IFN β 1a
`Impact of DF on utility
`Annual discontinuation rate for IM IFN β 1a
`Assignment ratio between NAT and SM
`Discontinuation rate for DF, 2yr
`Discount rate
`Utility EDSS 6.0 0 7.5
`$30,000
`
`$110,000
`$90,000
`$70,000
`$50,000
`Incremental net monetary benefit
`
`$130,000
`
`3 Results
`
`3.1 Base-Case Scenario
`
`Over 5 years, the total costs per patient were estimated at
`US$223,606, US$239,947, US$226,085, and US$200,145
`for IM IFN-b1a, fingolimod, teriflunomide, and dimethyl
`fumarate, respectively (Table 3). The accumulated QALYs
`were 3.34, 3.69, 3.68, and 3.72 for IM IFN-b1a, fingolimod,
`teriflunomide,
`and
`dimethyl
`fumarate,
`respectively.
`Assuming a WTP at US$150,000,
`the NMBs were at
`US$276,745, US$313,312, US$326,525, and US$357,356
`for each of the DMDs above. Having the lowest costs and
`
`highest QALYs, dimethyl fumarate dominated all other
`drugs.
`Compared with IM IFN-b1a,
`INMBs were
`the
`US$36,567, US$49,780, and US$80,611 for fingolimod,
`teriflunomide,
`and
`dimethyl
`fumarate,
`respectively
`(Table 3). The NMBs of the four DMDs over a range of
`WTPs are shown in Fig. 2. As long as the WTP was greater
`than US$100,000, NMBs of all the drugs would be greater
`than zero. Dimethyl fumarate dominated all other drugs
`across the range of WTPs.
`The ICERs for fingolimod and teriflunomide compared
`with IM IFN-b1a were US$46,328 and US$7,115, respec-
`tively, so both fingolimod and teriflunomide are cost
`
`MYLAN PHARMS. INC. EXHIBIT 1100 PAGE 7
`
`
`
`78
`
`X. Zhang et al.
`
`effective compared with IM IFN- ~1 a. In addition, the ICER
`of fingolimod compared against
`teriflunornide was
`US$3,201,672 (Table 3). Therefore, apart from dimethyl
`fumarate, teriflunornide was the most cost effective therapy
`at a WTP threshold of US$150,000.
`
`3.2 One-Way Sensitivity Analysis
`
`The ten most sensitive parameters are shown in Fig. 3.
`Generally, the results for base case analysis were stable to
`the change in most parameters. For the comparison
`between fingolimod and IM IFN- ~1 a (Fig. 3a), the monthly
`costs for fingolimod and IM IFN- ~1 a were the most sen(cid:173)
`sitive parameters. When the monthly cost for fingolimod
`exceeded US$5,132, fingolimod was no longer cost effec(cid:173)
`tive. A decrease in the cost of IM IFN- ~1 a substantially
`reduced the INMB, but fingolimod still remained cost
`effective. For the sensitivity analysis of teriflunornide and
`dimethyl fumarate compared with IM IFN- ~1 a (Fig. 3b, c),
`the INMBs were consistently greater than 0. For all of the
`three oral therapies versus IM IFN-~1 ,., lNMB increased as
`the time horiwn became longer under both discounted and
`non-discounted cases (Fig. 4), indicating oral therapies are
`associated with greater benefits for long-term care.
`For the sensitivity analysis of therapy selection, dime(cid:173)
`thyl fumarate remained the optimal therapy in almost all of
`the cases. The result was only sensitive to the monthly cost
`for teriflunornide. If the monthly cost for teriflunornide was
`lower than US$2,908, teriflunornide would be the most
`cost-effective therapy.
`
`3.3 Probabilistic Sensitivity Analysis
`
`Figure 5 shows the probability that each drug maximiz.ed
`the NMBs. Over the range of WTPs, the probability that
`dimethyl fumarate was the highest value therapy always
`exceeded 90 %. As a result, both teriflunornide and fin(cid:173)
`golimod had a less than 10 % chance of having the highest
`value. However, the probability for teriflunornide being
`preferred was constantly higher than that for fingolimod.
`Finally, IM IFN- ~1 a had a negligible probability of being
`the highest value treatment.
`
`4 Discussion
`
`This paper evaluated the cost effectiveness of three oral
`therapies, fingolimod , teriflunornide, and dimethyl fuma(cid:173)
`rate, compared with IM IFN- ~1 a as first-line therapies in
`the treatment of RRMS patients. A Markov model based on
`EDSS disability level was developed to simulate disease
`progression over a 5-year time horiwn.
`To our knowledge, this is the first paper to compare
`the cost effectiveness of the new oral DMDs compre(cid:173)
`the
`incorporate second-line therapy in
`hensively and
`model. Model results favored oral therapies in economic
`and health benefits compared with IM IFN- ~ia· Over a
`range of time horizons and WTPs, dimethyl fumarate was
`always the dominant strategy because of high QALY
`gained and low total costs. Leaving aside the dominance
`of dimethyl
`fumarate, given a WTP
`threshold of
`
`Fig. 4 incremental net
`monetary benefit vs.
`intramuscular inteiferon ~1 •
`
`250000
`
`200000
`-
`~
`ii
`~ s
`~ 150000
`0
`E
`~
`~ 100000
`Qi
`E
`~ c=
`
`50000
`
`---
`
`--
`
`---
`
`---
`...
`....
`---------------
`.... ········· ..
`
`a
`
`5 years
`
`10years
`
`~ Dimethyl fumarate
`(Non discounted)
`- Dimethyl fumarate
`(Discounted)
`
`-
`
`20 years
`15 years
`Time horizon
`Teriflunomide
`(Non discounted)
`-- - - - Teriflunomide
`(Discounted)
`
`-
`
`25years
`
`30 years
`
`~ Fingolimod
`(Non discounted)
`· · · · · · · · Fingol imod
`(Discounted)
`
`L\Adis
`
`MYLAN PHARMS. INC. EXHIBIT 1100 PAGE 8
`
`
`
`Cost Effectiveness of Oral Disease Modifying Drugs Versus IM IFN b1a
`
`79
`
`80000
`
`100000
`
`120000
`
`160000
`140000
`Willingness-to-pay
`
`180000
`
`200000
`
`Dimethyl fumarate
`
`Teriflunomide
`
`Fingolimod
`
`IM IFN β 1a
`
`1
`
`0.9
`
`0.8
`
`0.7
`
`0.6
`
`0.5
`
`0.4
`
`0.3
`
`0.2
`
`0.1
`
`0
`
`0.1
`
`Probabty of maxmzng NMB
`
`Fig. 5 Probability that each
`therapy maximizes NMB. IFN
`interferon, IM intramuscular,
`NMB net monetary benefit
`
`US$150,000, fingolimod and teriflunomide were cost
`effective compared with IM IFN-b1a, with ICERs of
`US$46,328 and US$7,115, respectively. However, fingo-
`limod was not cost effective compared with terifluno-
`mide, and thus teriflunomide was the most cost-effective
`therapy after dimethyl
`fumarate. One-way sensitivity
`analyses indicated that model results were most sensitive
`to acquisition costs for DMDs and time horizon, yet the
`results were rarely reversed and the decision-making
`rankings based on the results were robust
`in one-way
`sensitivity analyses. Probabilistic sensitivity analysis also
`showed dimethyl fumarate was the optimal therapy most
`of the time, whereas teriflunomide was the second best
`choice.
`In prior studies, fingolimod was the only oral therapy
`that was evaluated using cost-effectiveness analysis. Lee
`et al. [2] compared fingolimod with IM IFN-b1a by build-
`ing a Markov model on a 10-year time horizon. They
`estimated the ICER for fingolimod at US$73,975 compared
`with US$46,328 in this paper. However, given the annual
`discontinuation rate of about 10 %, a 10-year time horizon
`would imply that essentially all patients would have dis-
`continued by the end of the simulation. Moreover, since
`Lee et al. [2] ignored second-line treatments, their results
`may not be realistic. O’Day et al. [55] compared fingoli-
`mod with natalizumab in the treatment of RRMS, with
`incremental cost per relapse avoided as the outcome. The
`study found that natalizumab dominated fingolimod since it
`was less costly and more effective in reducing relapses.
`Natalizumab was not included as a comparator in this
`study, because it is recommended for use after alternative
`therapies failed, while fing