J Manag Care Pharm. 2007 Apr;13(3) 245-61

8/14/2019 J Manag Care Pharm. 2007 Apr;13(3) 245-61

1/17

www.amcp.org Vol. 13, No. 3 April 2007 JMCP Journal of Managed Care Pharmacy 245

Cost-e ectiveness o Four Immunomodulatory Therapiesor Relapsing-Remitting Multiple Sclerosis:

A Markov Model Based on Long-term Clinical DataChristopher Bell, MS; Jonathan Graham, MS; Stephanie Earnshaw, PhD;

MerriKay Oleen-Burkey, PhD; Jane Castelli-Haley, MBA; and Kenneth Johnson, MD

ABSTRACT

Authors

FORMULARY MANAGEMENT

Note: A d a ubj c a c a a ag 287-89 u .

Christopher Bell, Ms, d c , h a ec m c , JonAthAnGrAhAM, Ms, a a c m , a d stephAnie eArnshAw, p D,

g ba ad, U.s., h a ec m c , rti-h a s u , r a c t a gpa k, n Ca a; MerriKAy oleen-BUrKey, p D, d c , ou c mr a c a d M d ca o a , a d JAne CAstelli-hAley, MBA, ma ag ,h a ou c m r a c , t va n u c c , i c., Ka a C , M u ;Kenneth Johnson, MD, a u g a d d c , Ma a dC Mu sc , sc M d c , U v Ma a d, Ba m ,Ma a d.

AUthor CorresponDenCe: C B , Ms, c/ s a ea a ,p D, G ba h ad, U.s. h a ec m c , rti-h a s u , 200 pa k D .,po B x 12194, r a c t a g pa k, nC 27709. t : (919) 485-2730;Fax: (919) 541-7222; e-ma : a a @ . g

C g 2007, Acad m Ma ag d Ca p a mac . A g v d.

BACKGROUND: Be ore the introduction o the immunomodulatory therapies ormultiple sclerosis (MS), treatment options or MS consisted o symptomaticmanagement (physical therapy and pharmacological treatment or symptommanagement). Symptomatic management or MS has been supplementedin the past decade by 2 new classes o immunomodulatory therapies thathave been approved as rst-line treatments or relapsing-remitting multiplesclerosis (RRMS): subcutaneous glatiramer acetate (SC GA) and 3b-inter erons:

intramuscular inter eron b-1a (IM IFNb-1a), SC IFNb-1a, and SC IFNb-1b.OBJECTIVE: To estimate the economic outcomes o 5 treatment strategies:symptom management alone, symptom management combined with SC GA,IM IFNb1-a, SC IFNb1-a, or SC IFNb1-b in patients diagnosed with RRMS.

METHODS: A literature-based Markov model was developed to assess thecost-e ectiveness o 5 treatment strategies or managing a hypotheticalcohort o patients diagnosed with RRMS in the United States4 immuno-modulatory drug therapies and symptom management alone. Health stateswere based on the Kurtzke Expanded Disability Status Scale (EDSS), a widelyaccepted scale or assessing RRMS (higher EDSS scores = increased diseaseseverity). Baseline relapse and disease progression transition probabilities orsymptom management were obtained rom natural history studies. Treatmente ects o the immunomodulatory therapies were estimated by applying apercentage reduction to the symptom management transition probabilities

or relapse (27% reduction) and disease progression (30% reduction).Transition probabilities were subsequently adjusted to account or (1) thee ects o neutralizing antibodies, speci cally on relapse rates by assumingno additional therapy bene ts a ter the second year o continuous therapy,and (2) treatment discontinuation. Therapy-speci c data were obtained romclinical trials and long-term ollow-up observational studies. Transitionsamong health states occurred in 1-month cycles or the li etime o a patient.Costs (2005 US$) and outcomes were discounted at 3% annually.

RESULTS: The incremental cost per quality-adjusted li e-year or the 4 immuno-modulatory therapies is $258,465, $303,968, $416,301, and $310,691 or SCGA, IM IFNb-1a, SC IFNb-1a, and SC IFNb-1b, respectively, compared withsymptom management alone. Sensitivity analyses demonstrated that resultswere sensitive to changes in utilities, disease progression rates, time horizon,and immunomodulatory therapy cost.

CONCLUSIONS: The pharmacoeconomic model determined that SC GA was thebest strategy o the 4 immunomodulatory therapies used to manage MS andresulted in better outcomes than symptom management alone. Sensitivityanalyses indicated that the model was sensitive to changes in a numbero key parameters, and thus changes in these key parameters would likelyinfuence the estimated cost-e ectiveness results. Head-to-head randomizedclinical trials comparing the immunomodulatory therapies or the treatmento MS are necessary to validate the projections rom the pharmacoeconomicanalyses, particularly since the results available today rom the clinical trialsdo not account adequately or treatment dropouts.

KEYWORDS: Multiple sclerosis, Immunomodulatory therapy, Markov model,Cost-e ectiveness

J Manag Care Pharm. 2007;13(3):245-61

Multiple sclerosis (MS) is a chronic, neurodegenerativein ammatory disease o the central nervous system. Inthe United States, the prevalence rate o MS is estimatedto range between 85 and 177 cases per 100,000, translating toan overall prevalence o approximately 400,000. 1-3 In a majorityo patients (70%), the onset o MS occurs during early adulthood(age 15-45 years), making MS one o the most common causeso neurological disability in young and middle-aged adults. 4,5 Common symptoms o MS include atigue, walking di fcultiesleading to reduced mobility, bowel/bladder disturbances, opticneuritis and other visual changes, changes in cognitive unction,pain, sensory loss, and depression. 6 Three main types o MSare generally recognized: (1) relapsing-remitting MS (RRMS),(2) secondary progressive MS (SPMS), and (3) primaryprogressive/relapsing MS (PPMS or PRMS).7,8 At disease onset,RRMS is diagnosed in approximately 80% to 85% o MSpatients, with the remaining proportion o patients diagnosedwith the primary progressive orms o MS (PPMS or PRMS). 4,9,10

Among RRMS-diagnosed patients, a majority will progress toSPMS (50% o patients will experience a gradual progression o disability within 10 years o the initial attack and 90% o patientswill have progressive disease a ter 25 years). 11


2/17

246 Journal of Managed Care Pharmacy JMCP April 2007 Vol. 13, No. 3 www.amcp.org

Cost-effectiveness of Four Immunomodulatory Therapies for Relapsing-Remitting Multiple Sclerosis:A Markov Model Based on Long-term Clinical Data

C1MSSG(1995)

MSCRG(1996)

EIFNBDCS(2002)

PRISMS(1998)

IFNBMSSG(1995)

INCOMIN(2002)

EVIDENCE(2002)

Re erence Johnson12

Jacobs13

Clanet14

PRISMS15

IFNB MS StudyGroup 16,17 Durelli18

Panitch19

Intervention GA 20 mg

PLA

IFNb-1a 30 g

PLA

IFNb-1a 30 g

IFNb-1a 60 g

IFNb-1a 22 g

IFNb-1a 44 g

PLA

IFNb-1b 1.6 MIU

IFNb-1b 8 MIU

PLA

SC IFNb-1b 8 MIU

IM IFNb-1a 30g

SC IFNb-1a 44 g

IM IFNb-1a 30 g

Dosing SC once daily IM once weekly IM once weekly SC 3 timesweekly

SC every otherday

SC every otherday

IM once weekly

SC 3 timesweekly

IM once weekly

Study design RCT, DBUnited States24-months

RCT, DBUnited States24 months

RCT, DBEurope36 months

RCT, DB9 countries24 months

RCT, DBUnited States/ Canada24 months

RCTItaly24 months

RCTEurope/Canada/ United States48 weeks

Number o patients

Total: 251

GA: 125

PLA: 126

Total: 301

30 g: 158

PLA: 143

Total: 802

30 g: 402

60 g: 400

Total: 560

22 g: 189

44 g: 184

PLA: 187

Total: 372

1.6 MIU:125

8 MIU: 124

PLA:123

Total: 188

8 MIU: 96

30 g: 92

Total: 677

44 g: 339

30 g: 338

Patientcharacteristics

EDSS 0.0-5.5

Mean (SD)EDSS:

GA: 2.8 (1.2)

PLA: 2.4 (1.3)

2 relapses inprevious 2 years

EDSS 1.0-3.5

Mean (SE)EDSS:

30 g: 2.4 (0.05)

PLA: 2.3 (0.05)


EDSS 2.0-5.5

Mean (SD)EDSS:

30 g: 3.6 (1.0)

60 g: 3.6 (1.0)


EDSS 0.0-5.0

Mean (SD)EDSS:

22 g: 2.5 (1.2)

44 g: 2.5 (1.3)

PLA: 2.4 (1.2)


EDSS 0.0-5.5

Mean (SE)EDSS:

1.6 MIU: 2.9(0.1)

8 MIU: 3.0 (0.1)

PLA: 2.8 (0.1)


EDSS 1.0-3.5

Mean (SD)EDSS:

8 MIU: 1.9 (0.6)

30 g: 2.0 (0.7)


EDSS 0.0-5.5

Mean (SD)EDSS:

44 g: 2.3 (N/R)

30 g: 2.3 (N/R)


Number (%) o patients whocompletedthe study andcontributed toclinical outcomes

Total: 215/251(85.7%)

GA: 106/125(84.8%)

PLA: 109/126(86.5%)

Total: 129/301(57.1%)

30 g: 85/158(53.8%)

PLA: 87/143(60.8%)

Total: 559/802(69.7%)

30 g: 281/402(69.9%)

60 g: 273/400(69.5%)

Total: 526/560(93.9%)

22 g: 167/189(88.4%)

44 g: 165/184(89.7%)

PLA: 170/187(90.9%)

Total: 338/372(90.9%)

1.6 MIU:111/125(88.8%)

8 MIU: 115/124(92.7%)

PLA: 112/123(91.1%)

Total: 158/188(84.0%)

8 MIU: 85/96(88.5%)

30 g: 73/92(79.3%)

Total: 649/677(95.9%)

44 g: 325/339(95.9%)

30 g: 324/338(95.9%)

Primary endpoint Number o relapses

Time to confrmedEDSS progression(increase o 1.0point on EDSSsustained or 6months)

Sustaineddisabilityprogression

Relapse rate Annual relapserate and theproportion o relapse- reepatients

Number o relapse- reepatients andthe number o patients with nonew T2 lesionson MRI

Proportiono patientsrelapse- reeat 24 weeks

Summary o results Comparedwith PLA, GAis e ective inreducing relapserates.

Compared withPLA, 30g ise ective inprolonging thetime to disabilityand reducingrelapse rates;initial resultsbased on non-ITT principles.

No di erencein relapseor sustaineddisabilityprogressionbetweentreatmentgroups.

Comparedwith PLA, 22g and 44 gare e ective inreducing relapsesand delaying theprogression o disability.

Compared withPLA, 1.6 MIUand 8 MIUare e ectivein reducingrelapses and onmeasures o MRIactivity.

Compared with30 g, 8 MIUhad a greaterpercentage o patients whoremained relapse-ree and was moree ective in termso reducing MRIdisease activity.

Compared with30 g, 44 g ismore e ectivein preventingrelapses andreducing MRIdisease activity.

Table of Immunomodulatory Therapy Clinical Trials in Relapsing-Remitting Multiple Sclerosis (continued) TABLE 1

(continued on next page)


3/17



Relapse results6 Mean numbero relapses over2 years:

GA: 1.19

PLA: 1.68

RR (vs. PLA):

GA = 0.71

Annual relapserate (allpatients):

30 g: 0.67

PLA: 0.82

RR (vs. PLA):

30 g = 0.82

Annual relapserate (2-yearcompleters):

30 g: 0.61

PLA: 0.90

RR (vs. PLA):30 g = 0.68

Annualizedrelapse rate:

30 g: 0.77

60 g: 0.81

Relapse rate over2 years:

22 g: 1.82

44 g: 1.73g: 1.73g: 1.73

PLA: 2.56

RR (vs. PLA):

22 g= 0.71g=0.71g=0.71= 0.710.71

44 g: 0.68g: 0.68g: 0.68

Annual relapserate over 2 years:

1.6 MIU: 1.17

8 MIU: 084

PLA: 1.27

RR (vs. PLA):

1.6 MIU = 0.92

8 MIU = 0.66

Relapse- reepatients a ter2 years:

1.6 MIU: 23%

8 MIU: 36%PLA: 18%

Relapse- reepatients a ter2 years:

8 MIU: 51%

30 g: 36%

Relapse- reepatients a ter24 weeks:

44 g: 75%

30 g: 63%

Relapse- reepatients a ter48 weeks:

44 g: 62%

30 g: 52%

Diseaseprogressionresults6

Patients withincrease o 1.0point on EDSS(unsustained):

GA: 20.8%

PLA: 28.8%

Sustainedprogression(all patients):

30 g: 21.9%

PLA: 34.9%

Sustainedprogressionby 36 months(increase o 1.0point on EDSSsustained or6 months):

30 g: 37%

60 g: 37%

Time (months)to progressiono disability(increase o 1.0point on EDSSsustained or3 months):

22 g: 18.5

44 g: 21.3

PLA: 11.9

N/A Confrmedprogressiono disability(increase o 1.0point on EDSSsustained or6 months):

8 MIU: 30%

30 g: 15%

Confrmedprogressiono disability(increase o 1.0point on EDSSsustained or6 months):

8 MIU: 6%

30 g: 8%

Study extensions(% o patientsinitiallyrandomizedto study drugcompletingextension)

35 months(77.6%) 20 6 years (61.6%) 24

8 years (57.6%) 25

10 years (51.2%) 28

8 years (N/A,LTFU)94

4 years (55.6%) 21 4 years (78.8%) 22

6 years (50.1%) 26


5 years (44.2%) 23


N/A N/A

C1MssG = C m -1 Mu sc s ud G u ; DB = d ub b d; eDss = ex a d d D ab s a u sca (Ku zk ); eiFnBDCs = eu a iFn b-1a D -C m a s ud ; eViDenCe = ev d c i D -r : eu a n Am ca C m a a v e fcac ; GA = g a am ac a ; iFn = ;iFnBMssG = i B a Mu sc s ud G u ; iM = amu cu a ; inCoMin = i d d C m a i ; itt = a ;ltFU = g- m -u v ; MiU = m a a u ; Mri = mag c a c mag g; Ms = mu c ; MsCrG = Mu sc C ab a vr a c G u ; n/A = ava ab ; n/r = d; plA = ac b ; prisMs = p v r a a d D ab b i b-1a subcu a u Musc ; rCt = a d m z d c ca a ; rr = a v k; sC = ubcu a u ; g = m c g am .

Table of Immunomodulatory Therapy Clinical Trials in Relapsing-Remitting Multiple Sclerosis (continued) TABLE 1

C1MSSG(1995)

MSCRG(1996)

EIFNBDCS(2002)

PRISMS(1998)

IFNBMSSG(1995)

INCOMIN(2002)

EVIDENCE(2002)

Be ore the immunomodulatory therapies or MS wereintroduced, treatment options or MS consisted o symptomaticmanagement (physical therapy and pharmacological treatmentor symptom management). Symptomatic management or MShas been supplemented in the past decade by 2 new classeso immunomodulatory therapies that have been approved asfrst-line treatments or RRMS: subcutaneous glatiramer acetate(SC GA) (Copaxone) and 3 b-inter erons: intramuscularinter eron b-1a (IM IFNb-1a [Avonex]), SC IFNb-1a (Rebi ),and SC IFNb-1b (Betaseron). Evidence rom randomized clinical

trials (1-2 years) 12-19; prospective extensions o the clinical trials(2-5 years)20-23; and long-term ollow-up studies o patientsinitially enrolled in clinical trials 24-27 (one o which extendsbeyond 10 years 28) have shown that there is good evidencedemonstrating the benefts o immunomodulatory therapies inreducing relapse rates, slowing the progression o disability, andreducing MS disease activity (Table 1).

However, the use o the immunomodulatory therapies ineveryday clinical practice has been a topic o substantial debate.The National Institute or Clinical Excellence (NICE) determined


4/17



that on the balance o their clinical and cost-e ectiveness,neither beta inter eron nor glatiramer acetate is recommendedand those therapies were provided to patients only a ter theDepartment o Health and the respective manu acturers agreedto a risk-sharing scheme. 29,30 In contrast to NICEs guidance,some clinical practice guidelines have recommended theimmunomodulatory therapies or the treatment o MS. 31 Theselection o specifc therapies is based on individual patientcharacteristics (e.g., disease severity, patient compliance) andtreatment characteristics (e.g., e fcacy, incidence o neutralizingantibodies [NAbs], side e ects).

The socioeconomic burden o MS is substantial given thedebilitating nature o this chronic, progressive, and li elongcondition that a ects individuals in the most productive years o li e. Drug acquisition cost or the immunomodulatory therapieswas estimated to exceed $16,000 per patient per year in 2006,a signifcant expenditure or health care payers. 32-34 The annualcost o illness o MS (in 1994 US$) is estimated to be between$6.8 and $13.6 billion, composed largely o indirect costs or

ormal/in ormal care and lost earnings. 35 Patwardhan et al. 36 assessed the link between disability levels and costs and oundthat costs rose at an exponential rate with increasing MS disabilitylevels, a fnding that was consistent with previously publishedstudies. 37-39 Given this evidence o increasing costs (direct andindirect) with increasing disease severity, we believe the abilityo the immunomodulatory therapies to reduce relapse rates andslow the progression o the MS may assist in reducing resourceuse and may, in turn, help to o set the cost o these therapies.

Cost-e ectiveness and cost-utility analyses (CEA/CUAs) areuse ul tools to assess the tradeo between the added costs andpotential benefts (e.g., improved patient outcomes) o newtherapies. In the current environment o cost-consciousness andlimited health care resources, CEA/CUA a ords decision makersan opportunity to evaluate new therapies rom an economicperspective and quanti y the budgetary implications o adopting

such therapies. A majority o the published CEA/CUA evaluationso immunomodulatory therapies or MS have been conductedrom perspectives outside the United States. 30,40-45 In a recentU.S.-based CUA evaluation, immunomodulatory therapy orthe treatment o nonprimary, progressive MS (e.g., RRMS andSPMS) was compared with no treatment over a 10-year timehorizon. 46 Cost-e ectiveness results rom this analysis, as wellas previously published CUA evaluations, were considerablyhigher than the arbitrary and commonly re erenced benchmarko $50,000 per quality-adjusted li e-year (QALY). 47

Recent published literature on the impact o theimmunomodulatory therapies in MS has provided key data

that have not been previously used in CEA/CUA evaluations.Specifcally, long-term ollow-up data o patients initiallyenrolled in clinical trials have been published, 24-28 one o which,or SC GA, extended beyond 10 years. 28 In previous CEA/CUAevaluations, long-term treatment outcomes (e.g., treatment e ectsand discontinuation rates) were extrapolated over time basedprimarily on data rom short-term clinical trials. In additionto long-term ollow-up data, there has been ocus on thedevelopment o NAbs among patients prescribed b-inter erons,which may inhibit or neutralize the e ectiveness o b-inter erontreatment. 48

Because o data limitations, previous models have madeassumptions regarding the impact o b-inter erons on treatmente ects (e.g., constant treatment e ects over time), which, in turn,made it di fcult to examine the impact that NAbs might have oncost-e ectiveness. With the availability o long-term data, cost-e ectiveness analysis o various therapies in the presence o theseclinical markers can be made more appropriately. In this regard,we examine the cost-e ectiveness o 5 treatment strategies inpatients diagnosed with RRMS (symptom management aloneand in combination with SC GA, IM IFN b-1a, SC IFNb-1a, orIM IFNb-1b). Cost-e ectiveness results (symptom managementvs. the 4 immunomodulatory therapies) were reported in termso cost per QALY gained as well as cost per outcome achieved

eDss = ex a d d D ab s a u sca (Ku zk ).

Schematic Representationof the Markov Model

FIGURE 1


5/17



(e.g., cost per year spent relapse ree or cost per year spent inless severe disease health states), thus providing decision makersrelevant data with which to evaluate the cost-e ectiveness o the4 immunomodulatory therapies versus symptom managementin treating RRMS.

nn MethodsModel Description

We developed a Markov model to assess the cost-e ectivenesso 5 treatment strategies to manage a hypothetical cohort o patients diagnosed with RRMS in the United States. The strategieswere symptom management alone (e.g., physical therapy/ exercise and pharmacological treatment [e.g., corticosteroidsor relapse, tizanidine or spasticity, and modafnil or atigue])and symptom management in combination with 1 o theollowing immunomodulatory therapies: SC GA, IM IFN b-1a,

SC IFNb-1a, or SC IFN b-1b. The clinical course o RRMS (e.g.,disease progression and relapse) was modeled in terms o theKurtzke Expanded Disability Status Scale (EDSS). 49 Specifcally,7 EDSS health states were modeled (Figure 1):1. EDSS 0.0-2.5: no or ew limitations in mobility2. EDSS 3.0-5.5: moderate limitations in mobility3. EDSS 6.0-7.5: walking aid or wheelchair required4. EDSS 8.0-9.5: restricted to bed5. Death (natural causes or EDSS 10)6. Relapse EDSS 0.0-2.5: relapse with a change in disability

within EDSS 0.0-2.57. Relapse EDSS 3.0-5.5: relapse with a change in disability

within EDSS 3.0-5.5Transitions among the health states occurred in 1-monthcycles. The baseline time horizon o the model was assumed to beli etime in order to capture the ull benefts o immunomodulatorytherapy. Costs and outcomes were estimated rom the societalperspective and were discounted at 3% per annum. 50 All costswere reported in 2005 U.S. dollars. The natural history o MSdisease progression, clinical e fcacy o MS therapies, mortality,resource use, costs, and utilities were obtained rom the publishedliterature. The model calculated the ollowing outcomes: averagenumber o years spent in EDSS 0.0-5.5; average number o yearsspent relapse ree; li e-years; QALYs; total costs and costs bycomponent (i.e., immunomodulatory therapy cost, MS-relatedmedical costs [e.g., drugs or symptom management], and lostworker productivity costs); and incremental cost-e ectivenessratios comparing symptom management alone with symptommanagement combined with each o the 4 immunomodulatorytherapies. Model parameters were varied in sensitivity analyses.

A number o underlying assumptions were adopted or thebase-case model:

1. A Web survey o patients (aged 18 years) treated withimmunomodulatory therapy in the United States and enrolled ina patient support program was used to determine the percentageo patients entering the model among the 4 nonrelapse EDSS

health states 51 (Table 2). The survey, based on previous surveysconducted in the United States and Europe, 52,53 was completed by711 MS patients and collected data such as disease in ormation(e.g., type o MS, therapy used), quality o li e, resource use, andassociated costs (direct and indirect). The majority o surveyedpatients were in the lower EDSS health states (e.g., EDSS


6/17



(i.e., only patients in the EDSS 8.0-9.5 health state couldtransition to EDSS 10, which is death). This assumption wasconsistent with a previous model 46 that was based on dataindicating that nearly 90% o MS patients were signifcantlydisabled (e.g., unable to walk) be ore death. 62 These data arepresented in Table 2.

Treatment Effects of SC GA and the b-InterferonsTreatment e ects associated with the immunomodulatorytherapies were estimated by adjusting (via a percentagereduction) the probabilities o relapse and disease progressionused in the symptom management arm o the model. Relapseand disease progression rates over time were obtained romrandomized clinical trials (SC GA, 12 IM IFNb-1a,13,14,63 SC IFNb-1a,15 SC IFNb-1b16,17); prospective extensions o the clinicaltrials (SC GA,20 IM IFNb-1a,21 SC IFNb-1a,22 SC IFNb-1b23);and long-term ollow-up studies (SC GA, 28 SC IFNb-1a26).

Relapse and disease progression rates were then mapped andftted to prediction curves over time to estimate the long-termtreatment e ects o SC GA and the b-inter erons (relapse: Figure 2;disease progression: Figures 3 and 4).

To account or the inherent di erences between the randomizedclinical trials (e.g., patient population, primary endpoint), a fxedpatient population was assumed or the base- case model. Inparticular, all patients entered the model on the basis o a fxedEDSS distribution (Table 2). Furthermore, short-term outcomeswere modeled by assuming, or all immunomodulatory therapies,a single percentage reduction or relapse and disease progressionin the frst 2 years o therapy (Table 2). This assumption wasbased on data rom several published review papers 64-66 thatindicated annual relapse and disease progression rates in therandomized clinical trials were similar among therapies. Insubsequent years o the model, treatment e ects were estimatedbased on the long-term, immunomodulatory-specifc prediction

Parameter DescriptionEstimate for

Base-Case Mode

Range for Sensitivity Ana ysis(25% Un ess Indicated Otherwise) Reference

Initial patient distribution among EDSS health statesEDSS 0.0-2.5

EDSS 3.0-5.5

EDSS 6.0-7.5

EDSS 8.0-9.5

26.4%

58.7%

13.8%

1.1%

Scenario 1100.0%

0.0%

0.0%

0.0%

Scenario 225.0%

25.0%

25.0%

25.0%

Oleen-Burkey [abstract] 51

Monthly probability o disease progression(symptom management)

EDSS 0.0-2.5 to 3.0-5.5

EDSS 3.0-5.5 to 6.0-7.5

EDSS 6.0-7.5 to 8.0-9.5

EDSS 8.0-9.5 to 10 (death)

0.004438

0.009189

0.003583

0.000952

N/C Weinshenker 11; Runmarker 60;Prosser46

Monthly probability o relapse (symptommanagement) 0.075500 N/C Goodkin58

; Weinshenker4,59

;Prosser46

Treatment e ects o SC GA and b-inter erons,

% reduction in:

Probability o disease progression

Probability o relapse

30%

27%

12.5%-47.5%

8.8%-45.2%

Filippini 64; Simpson90; Khan66

Incidence o NAbs

IM IFNb-1a

SC IFNb-1a

SC IFNb-1b

2.2%

17.4%

36.4%

1.7%-2.8%

13.1%-21.8%

27.3%-45.5%

Rossman48; IFNB MSStudy Group 16; EuropeanStudy Group 96; PRISMS15;SPECTRIMS97; Panitch 19;

Jacobs13,98; Clanet 14; Cohen 99

Monthly drug acquisition costs (WAC), 2005 U.S.$

SC GAIM IFNb-1a

SC IFNb-1a

SC IFNb-1b

$1,258.20$1,275.00

$1,469.64

$1,264.85

$943.65-$1,572.75$956.25-$1,593.75

$1,102.23-$1,837.05

$948.64-$1,581.06

r d B k 33

Summary of Parameters and Values Used in the Base-Case Markov Model TABLE 2 (continued)

(continued on next page)


7/17



curves. Sensitivity analyses were per ormed to assess the impacto varying the percentage reduction adjustments.

To account or the e ect o NAbs, the probabilities o relapse or patients in the b-inter eron arms o the model weresubsequently adjusted. The model assumed that NAbs woulda ect the probability o relapse and would occur only a terthe second year o continuous inter eron treatment. 67-69 Theincidence o NAbs was obtained rom the published literature 48 and was used to calculate an adjusted probability o relapse thatre ected a weighted average o NAb-positive (NAb+) and NAb-negative (NAb-) patients. A conservative approach to modelingthe impact o NAbs was adopted, where NAb+ patients (titer20) experienced a probability o relapse equal to that o theyear-2 probability in each subsequent year o the model, andNAb- patients experienced a probability o relapse equal to thatpredicted by the ftted curve described previously. The modeldid not consider dose escalation o the b-inter erons in responseto NAbs.

The probabilities o relapse (adjusted or NAbs) and diseaseprogression associated with the immunomodulatory therapieswere also adjusted to account or patients who discontinuedtherapy. Rates o discontinuation were obtained rom thepublished clinical trials and long-term ollow-up studies; arelative 3% annual discontinuation rate was assumed in the eventdata were not available. 30 Patients who discontinued therapywere assigned the probabilities o relapse and disease progressionused in the symptom management arm o the model. Similarto the calculation per ormed or NAbs, the adjusted probabilityo relapse and disease progression re ected a weighted average o patients continuing and discontinuing therapy. The ftted curvesor the long-term treatment e ects o the immunomodulatorytherapies adjusted or NAbs and discontinuation are re ectedin Figures 2-4. The fgures indicate an immediate short-term reduction in relapse and disease progression rates and astabilization in rates (relapse and disease progression) over thelong term.

Parameter DescriptionEstimate for

Base-Case Mode

Range for Sensitivity Ana ysis(25% Un ess Indicated Otherwise) Reference

Health state-specifc MS-related costs (monthly)EDSS 0.0-2.5

EDSS 3.0-5.5

EDSS 6.0-7.5

EDSS 8.0-9.5

Relapse EDSS 0.0-2.5

Relapse EDSS 3.0-5.5

$433.26

$838.8.3

$1,990.02

$3,499.59

$427.98

$1,094.80

$324.95-$541.58

$629.12-$1,048.54

$1,492.52-$2,487.53

$2,624.69-$4,374.49

$320.99-$534.97

$821.10-$1,368.50

Oleen-Burkey [abstract] 51

Monthly cost o lost worker productivity(no. o days missed x hourly wage x % o patients employed)Symptom management

SC GA

IM IFNb-1a

SC IFNb-1aSC IFNb-1b

$154.74

$109.39

$167.40

$167.40

$162.63

Scenario 1

N/C

$84.94-$133.82

$158.04-$176.77

$158.04-$176.77$159.47-$165.79

Scenario 2

$116.06-$193.43

$82.05-$136.74

$125.55-$209.26

$125.55-$209.26$121.97-$203.29

Lichtenberg74; Lage75; Busche76

Utility weights

EDSS 0.0-2.5

EDSS 3.0-5.5

EDSS 6.0-7.5

EDSS 8.0-9.5Utility decrement associated with relapse

0.824

0.679

0.533

0.4910.094

Scenario 1

0.618-1.000

0.509-0.849

0.400-0.666

0.368-0.6140.071-0.118

Scenario 2

0.710

0.590

0.420

0.125N/C

Scenario 3

N/C

N/C

N/C

N/C0.071-0.118

Prosser46; Kobelt77; Kobelt52

eDss = ex a d d D ab s a u sca (Ku zk ); eDss 0.0-2.5 = Ms m m m ma d ab 2 u c a a a ; eDss 3.0-5.5 = m d a d ab 1 u c a a a m d d ab u 4 a a d ab v ug c ud u da ac v ; eDss 6.0-7.5 = m u a a c a a a cqu d a k 100 m c d c a ; eDss 8.0-9.5 = c d b d ma -ca u c a d qu g a a c a ac v

da v g; GA = g a am ac a ; iFn = ; iM = amu cu a ; Ms = mu c ; n/A = a cab ; nAb = u a z g a b d ; n/C = c a g ;sC = ubcu a u ; m m ma ag m = m ma c ma ag m a ; wAC = a acqu c .

Summary of Parameters and Values Used in the Base-Case Markov Model TABLE 2 (continued)


8/17



CostsThe model accounted or immunomodulatory therapy acqui-sition costs, health state-specifc MS-related medical costs, andthe cost o lost worker productivity. Costs were obtained romthe published literature and were in ated to 2005 U.S. dollars,where necessary, by the medical component o the ConsumerPrice Index. 70 The annual acquisition cost o SC GA and eachb-inter eron was calculated using wholesale acquisition costs(WACs),32 days supply per prescription, the recommendeddosing schedule, patient compliance (assumed 70% in base-case model71-73), patient copayment ($25 per prescription inbase-case model), and the proportion o patients discontinuingtherapy as reported in clinical trials (see Table 1).

Health state-specifc MS-related medical costs were estimatedrom the previously described Web survey o 711 RRMS patientsin the United States. 51 The resource use and associated costs

component o the survey was used in the model and comprisedthe cost o inpatient care, outpatient care, community services,alterations and equipment, in ormal care, and medications usedto manage the symptoms o MS. Costs were estimated on thebasis o Current Procedural Terminology, 4th Edition (CPT-4)codes and corresponding physician ees or each CPT-4 code,diagnosis-related groups, and the Drug Topics r d B k 33 (WACprices). Health state and drug costs are presented in Table 2.

Worker ProductivityEstimates o lost worker productivity were based on the publishedliterature. Specifcally, Lichtenberg 74 estimated the average number

o days missed rom work or 47 major chronic conditions,including MS. SC GA and b-inter eron-specifc estimates o thenumber o days missed rom work (all reasons) were derivedvia regression analyses o retrospective administrative claimsdata75 and were subtracted rom the Lichtenberg baseline valueto determine the average number o days missed due to MS oreach therapy. The cost o lost worker productivity was estimatedas the number o work days missed (in hours) multiplied byan hourly wage and adjusted to account or the proportion o patients employed (Table 2). 75,76

The model assumed that productivity losses were limitedto patients in EDSS 0.0-5.5 and that patients in the later EDSShealth states (EDSS >5.5) were not employed. Since the resultso the Lage et al. analysis 75 identifed a modest nonsignifcantdi erence in work days missed or the b-inter erons relative tono treatment, the base-case model took these modest di erencesinto account even though they were not statistically signifcantby using the conservative assumption that the absolutereduction in days lost rom work was incurred in comparisonwith symptom management. Furthermore, while providing anestimate o absenteeism (within the context o a small samplesize and including paid time o [PTO]), the Lage et al. analysis 75 did not provide an estimate o overall lost worker productivity,which includes both absenteeism (lost productivity associated

0

0.1

0.2

0.3

0.4

0.5

1 6 11 16 21 26 31 36 41 46

Number of Years on Therapy

SC GA SC IFNB-1a IM IFNB-1a SC IFNB-1b

Prediction Curve of the Long-termProbability of Relapse While onImmunomodulatory Therapy

FIGURE 2

12-17,20-23,26,28,63

0

0.01

0.02

0.03

0.04

0.05

0.06

1 6 11 16 21 26 31 36 41 46


- -

Prediction Curve of the Long-termProbability of Disease Progression(EDSS 0.0-2.5: No or Few Limitationsin Mobility to EDSS 3.0-5.5: ModerateLimitations in Mobility) While onImmunomodulatory Therapy

FIGURE 3

12-17,20-23,26,28,63

P r o

b a

b i l i t y o

f R e

l a p s e


GA = g a am ac a ; iFn = ; iM = amu cu a ; sC = ubcu a u .

eDss = ex a d d D ab s a u sca (Ku zk ); GA = g a am ac a ;iFn = ; iM = amu cu a ; sC = ubcu a u .


SC GA SC IFNB-1a IM IFNB-1a SC IFNB-1bSC GA SC IFN b-1a IM IFN b-1a SC IFN b-1b



9/17



with missing a work day, excluding PTO) and presenteeism(lost productivity associated with being present at work butper orming below expectations). The results reported by Lageet al. provided an estimate o absenteeism that, given the lack o better data, might be used in the model to estimate the impacto MS and its treatments on worker productivity.

UtilitiesUtility weights allow an objective measurement o the desirabilityo a health state. Utility weights range rom 0.0 to 1.0, where autility value o 1.0 represents per ect health and a value o 0.0represents death. These utility values are used to estimate QALYsby multiplying the number o li e-years within a particular healthstate by that health states utility weight. Utilities were estimatedby health state and were obtained rom the published literature(Table 2). 46,52,77 The relapse health states were associated with a

0.094 decrement in utility (e.g., utility or relapse EDSS 0.0-2.5= 0.824 0.094 = 0.730).

Model CalculationsThe model estimated the ollowing outcomes: average numbero years spent in EDSS 0.0-5.5; average number o years spentrelapse ree; li e-years; QALYs; and total costs and costs bycomponent (i.e., immunomodulatory therapy cost, MS-relatedmedical costs, and lost worker productivity costs). Incrementalcost-e ectiveness ratios (ICERs) were assessed in the model bycomparing each o the individual immunomodulatory therapiesplus symptomatic management with symptom management.

The ICERs were calculated as the di erence in costs between2 treatments divided by the di erence in e ectiveness: (CostDrug A Cost Drug B) / (E ectiveness Drug A E ectivenessDrug B). The resulting ICERs described the relative cost o purchasing 1 additional unit o relative e ectiveness (e.g., cost o 1 additional year spent in the lower EDSS health states). ICERswere calculated or 4 e ectiveness measures (cost per QALY; costper li e-year; cost per year spent in the lower EDSS health states[EDSS 0.0-5.5]; and cost per year spent relapse ree).

Sensitivity AnalysesTo test the robustness o the model assumptions and specifcparameters, univariate sensitivity analyses were per ormedby increasing and decreasing values or key parameters inthe model. Plausible ranges o values were obtained rom thepublished literature or by varying the estimates by up to 25%in each direction. Parameters analyzed included health stateand relapse state utilities; symptomatic treatment costs, healthstate costs, and drug costs; percentage employed and work dayssaved; percentage reductions in relapse and disease progressionrates in the frst 2 years o therapy; changes in relapse and diseaseprogression over time; change in treatment discontinuation overtime; EDSS distribution; incidence o NAbs; change in NAbs overtime; and discount rates. Sensitivity results or each model input

assessed were ranked rom most sensitive to least sensitive and

plotted on a tornado diagram. Results (Figure 5) indicated thatchanges in health state utilities were the most sensitive modelparameter. To urther test the sensitivity o the model to changesin health state utilities, 3 analyses were per ormed: (1) changingall utility values by a relative 25%; (2) replacing the utilityvalues in the base-case model with European-derived utilitiesused in previous economic evaluations 78; and (3) changing onlythe disutility values associated with the relapse health states bya relative 25%. Additional scenario-based sensitivity analyseswere per ormed to evaluate the impact on model results o changes to multiple parameters (e.g., simulating assumptionsused in previously published MS models). Ranges used insensitivity analyses (univariate and scenario) are presented inTable 2.

nn Results

Total costs per patient over the time horizon o a patients li etimewere estimated at $295,586, $352,760, $364,267, $377,996,and $358,509 or symptom management, SC GA, IM IFN b-1a,SC IFNb-1a, and SC IFN b-1b, respectively (Table 3). MS-relatedmedical costs were the largest cost component (approximately95% o total costs in the symptom management arm and70%-75% o total costs in the immunomodulatory arms), ollowedby the cost o immunomodulatory therapy (approximately

0.00

0.02

0.04

0.06

0.08

0.10

0.12

1 6 11 16 21 26 31 36 41 46


- -

Prediction Curve of the Long-termProbability of Disease Progression(EDSS 3.0-5.5: Moderate Limitations in

Mobility to EDSS 6.0-7.5: Walking Aidor Wheelchair Required) While onImmunomodulatory Therapy

FIGURE 4

12-17,20-23,26,28,63


eDss = ex a d d D ab s a u sca (Ku zk ); GA = g a am ac a ;iFn = ; iM = amu cu a ; sC = ubcu a u .



10/17



Sensitivity Analyses ResultsFIGURE 5

n : s v a a ba d m d c a g g va u : (1) b +25% a d -25%; a d (2) b c f d va u . t g a ba u m m d #1a d ba u m m d #2. t a k a c fc ba :

* t m z : ba ca = m ; v a a = 5 a .** C m a c : ba ca = 70%; v a a = 100%.*** U va u : ba ca = U.s.-ba d va u ; v a a = eu a -ba d va u .**** eDss d bu : ba ca = ac a eDss a a ; v a a = a a a f eDss a a (eDss 0.0-2.5).

eDss = ex a d d D ab s a u sca (Ku zk ); GA = g a am ac a ; iFn = ; iM = amu cu a ; Ms = mu c ; nAb = u a z g a b d ;sC = ubcu a u .

25% Increase in Input Value

25% Decrease in Input Value


11/17



20%-25% o total costs). Comparing direct medical costs(i.e., MS-related medical costs and immunomodulatory therapycosts) o the symptom management arm with the SC GA, IMIFNb-1a , SC IFNb-1a, and SC IFN b-1b arms o the model( rom Table 3), we ound the added cost o immunomodulatorytherapy was partially o set by savings in MS-related medicalcosts, and the greatest cost o sets occurred in the SC GA arm(24% o the cost o SC GA was o set by savings in MS-relatedmedical costs versus 17%-22% o the cost o the b-inter erons).

Outcomes over the li etime time horizon assessed in themodel were similar across the 4 immunomodulatory therapiesand were generally improved compared with outcomes withsymptom management (Table 3). The exception was or theestimated li e-years gained, as immunomodulatory therapy hadlittle impact on survival other than the slight delay in diseaseprogression to EDSS 10 (death). Comparisons among the

4 immunomodulatory therapies indicated that patients onSC GA had slightly better outcomes relative to patients on theb-inter erons.

Overall, the model indicated that patients on SC GAexperienced greater cost benefts compared with patients onany o the 3 b-inter erons. The incremental cost per QALY was$258,465, $337,968, $416,301, and $310,691 or SC GA,IM IFNb-1a, SC IFNb-1a, and SC IFN b-1b, respectively,compared with symptom management. ICERs comparingsymptom management with the 4 immunomodulatory therapiesor the other outcomes o interest are presented in Table 3.

Sensitivity AnalysesOverall, univariate sensitivity analyses showed results to besensitive to changes in health state utilities, the percentagereduction in disease progression rates in the frst 2 years o therapy used to estimate immunomodulatory therapy treatmente ects, model time horizon, and immunomodulatory therapyacquisition costs (Figure 5). Further assessment o changes tohealth state utilities revealed that changes to the disutility valuesassociated with the relapse health states had minimal e ect oncost-e ectiveness results. However, changes to utility valuesover all health states had a substantial impact. We observed thegreatest impact when we changed all utility values by a relative25%.

In a scenario in which the model assumed no change inSC GA e ects on disease progression in the frst 2 years and a25% improvement or the b-inter erons, we observed an overallimprovement in the cost-e ectiveness o the b-inter eronscompared with symptom management, and the ICERs or theb-inter erons (vs. symptom management) were more avorablethan those o SC GA (vs. symptom management). However,ICERs or the b-inter erons (vs. symptom management) did notimprove compared with SC GA (vs. symptom management)when we assumed no change in SC GA e ects on relapse in thefrst 2 years and a 25% improvement or the b-inter erons.

As expected, shorter time horizons (e.g., 10-20 years) resultedin larger ICERs when we compared symptom management witheach o the 4 immunomodulatory therapy arms o the model, asthe shorter time horizons did not ully account or all beneftsassociated with immunomodulatory therapy. Assuming nochange in the acquisition cost o SC GA and a 25% decrease in theacquisition cost o the b-inter erons resulted in more avorableICERs or the b-inter erons (vs. symptom management) comparedwith SC GA (vs. symptom management). Finally, changes inthe incidence o NAbs had minimal impact on the ICERs o theb-inter erons compared with symptom management.

nn DiscussionThe present analysis is the frst economic model in MS to(1) incorporate long-term data on treatment e ects, (2) includethe e ect o NAbs on the b-inter erons, (3) account or the

inherent di erences among clinical trial study designs o theimmunomodulatory therapies (e.g., via fxed EDSS distributionand single percentage reduction in relapse and diseaseprogression rates in the frst 2 years o therapy), and (4) presentresults in terms o cost-utility (cost per QALY gained) and cost-e ectiveness (e.g., cost per year spent relapse ree or cost per yearspent in less severe disease health states). Model results highlightthe potential long-term health and economic benefts o treatingMS patients with immunomodulatory therapy and indicatethat all 4 immunomodulatory therapies are associated withincreased benefts compared with symptom management, albeitat higher costs and without consideration o the cost o adverse

events. The model indicated that, o the 4 immunomodulatorytherapies used to manage MS and in comparison with symptommanagement, SC GA was the best strategy in terms o outcomesand costs. Sensitivity analysis indicated that the model wassensitive to changes in a number o key parameters, and thuschanges in these key parameters would likely in uence theestimated cost-e ectiveness results.

In a previous U.S.-based cost-e ectiveness model conductedby Prosser et al. 46 the authors concluded that IM IFN b-1acompared with no treatment (i.e., symptomatic treatment)yielded the largest gain in QALYs with an ICER between $1.8 and$2.2 million per QALY gained. These results were signifcantlydi erent rom that reported in the current analysis and can beattributed to the underlying methodology used to model MS, themost notable aspects o which include assumptions regardingthe treatment e ects associated with the immunomodulatorytherapies, model time horizon, and utility values. In terms o the treatment e ects o the immunomodulatory therapies overtime, the Prosser model is based on relapse rates and diseaseprogression rates reported in the pivotal clinical trials or therespective drugs. The current analysis supplements the pivotalclinical trials with data rom patients initially enrolled in thepivotal trials and ollowed over time. While both approacheshave limitations (pivotal trials in MS are based on a 2-year


12/17



snapshot o a chronic, li elong condition, and the ollow-

up studies are nonrandomized), they represent alternativemethods or modeling MS outcomes. It should be noted thatthe published estimate o e ectiveness or IM IFN b-1a is basedon those subjects who completed 2 years in the trial at the timeit was stopped as opposed to being based on the intent-to-treatpopulation. This approach likely overestimates the true e ects o IM IFNb-1a in this particular trial, which is apparent in resultsrom head-to-head clinical trials o IM IFN b-1a.18,19

Sensitivity analyses conducted in the Prosser model, thecurrent analysis, and other MS models have clearly indicated thatresults are in uenced by time horizon, with shorter time horizonsassociated with less avorable ICERs (e.g., Prosser 46 and othermodels30,44,78 ) and longer time horizons associated with moreavorable ICERs (e.g., current analysis, Nuijten and Hutton 41).Guidelines or the conduct o cost-e ectiveness analyses (e.g.,

Academy o Managed Pharmacy, 79 Canadian Coordinating O fceor Health Technology Assessment, 80 NICE81) recommend thatpharmacoeconomic models should be su fciently long to re ectall relevant costs and outcomes, which would suggest that anMS model should adopt a li etime perspective, given the chronicnature o MS. While guidelines may suggest that a longer timehorizon is more appropriate, the managed care perspective o tenre ects much shorter time horizons (e.g., 1-3 years), whichwould translate into increasingly less avorable ICERs as the

time horizon is shortened. However, as the enrollment base o

managed care plans shi ts over time, plans are likely to gain andlose MS patients across the spectrum o disability levels. Thus,therapies that reduce relapse and disease progression rates, suchas the immunomodulatory therapies or MS, may reduce theburden o MS.

A comparison o the utility values used in the current analysiswith that o the Prosser model indicated substantial di erences,which are re ected in the ICER results o the individual models.Utilities used in the Prosser model were derived rom a U.S.sample o MS patients and members o the general community 82 and were signifcantly higher than those reported in a subsequentU.S. study, 52,77 which were used in the current analysis. Thehigher utilities contributed to the higher ICERs reported in theProsser model. The Prosser utilities were also signifcantly higherthan those reported in the United Kingdom and Europe 30,78 (e.g., utility assigned to the initial EDSS health state 0.0-2.5 was0.95 in the Prosser model compared with 0.71 derived roma European population or the same health state 78), which isre ected in the more avorable ICERs calculated in analyses romother countries. 30,40,41 It is also o interest that the percentagereduction in utilities rom one health state to the next wassmaller in the Prosser U.S.-based analyses (e.g., 9% reduction inutilities rom EDSS 0.0-2.5 to EDSS 3.0-5.5 and 12% reductionin utilities rom EDSS 3.0-5.5 to EDSS 6.0-7.5 46,52,77 ) compared

Cost ComponentSymptom

Management SC GA IM IFN b-1a SC IFN b-1a SC IFN b-1b

Li etime drug acquisition costs(average no. o years on therapy)

$0

(N/A)

$77,340

(13.54)

$82,635

(13.05)

$95,208

(12.93)

$76,957

(13.21)

MS-related medical costs $282,950 $264,351 $265,366 $266,839 $265,940

Productivity costs $12,636 $11,069 $16,266 $15,948 $15,611

Total costs $295,586 $352,760 $364,267 $377,996 $358,509

Average no. o years spent in EDSS 0.0-5.5 12.28 14.92 14.71 14.29 14.54

Average no. o years spent relapse- ree* 11.42 14.67 14.24 13.98 14.15

Li e years 14.791 14.819 14.818 14.815 14.817

QALYs 9.081 9.303 9.285 9.279 9.284

Incremental cost per year spent in EDSS 0.0-

5.5 $21,667 $28,293 $41,008 $27,860

Incremental cost per year spent relapse- ree $17,599 $24,327 $32,207 $23,065

Incremental cost per li e-year gained $2,076,622 $2,588,087 $3,378,626 $2,452,616

Incremental cost per QALY gained $258,465 $337,968 $416,301 $310,691

* Cumu a v umb m a - c v d a .GA = g a am ac a ; iFn = ; iM = amu cu a ; Ms = mu c ; n/A = a cab ; QAly = qua -adju d - a ;sC = ubcu a u ; m m ma ag m = m ma c ma ag m a .

Base-Case Discounted Costs per Patient (Lifetime Perspective) TABLE 3


13/17



with analyses per ormed in other countries that estimated a 30%reduction in utilities rom EDSS 0.0 to EDSS 3.0 and a similarreduction rom EDSS 3.0 to EDSS 7.0. 30,78

Overall, the majority o economic evaluations in MS,including the current analysis, resulted in ICERs well abovethe arbitrary and commonly re erenced benchmark o $50,000per QALY, even in the best-case scenarios used in sensitivityanalyses.30,40-46,78,83 This was, in part, a re ection o (1) thechronic nature o the disease, (2) survival not being signifcantlya ected by the disease, (3) the modest QALY benefts associatedwith immunomodulatory therapy in MS versus symptommanagement, and (4) the high drug acquisition costs o theimmunomodulatory therapies. A review o the published cost-e ectiveness literature revealed a number o analyses o healthcare interventions that resulted in ICERs above the $50,000 perQALY benchmark, including $1.8 to $2.2 million per QALY

as reported in the Prosser MS model46

; $91,000 per QALY orosteoarthritis or rheumatoid arthritis patients using diclo enacversus ibupro en 84; $110,000 per QALY or patients usingmet ormin in a diabetes prevention program 85; $200,000 perQALY or osteoarthritis or rheumatoid arthritis patients usingdiclo enac and a proton pump inhibitor versus celecoxib 84;$370,000 per QALY or women with irritable bowel syndromeusing alosetron versus no treatment 86; and $56,000 to $840,000per QALY or the use o high-dose erythropoietin versus normaldosages to maintain increased hemoglobin levels (e.g., 12-14 g/ dL).87

A review o the cost-e ectiveness literature or blood sa etyinterventions (e.g., blood screening and pathogen inactivation)identifed a median ICER o $355,000 per QALY, 88 with someresults exceeding several million dollars per QALY. 89 While directcomparisons between these studies and the current analysis arelimited, the results indicate that not all economic evaluationsare bounded by the $50,000 per QALY benchmark and thatnumerous interventions with ICERs well above this thresholdhave been deemed valuable by patients, health care decisionmakers, and society. Thus, evaluating the cost-e ectiveness o the immunomodulatory therapies in terms o the $50,000 perQALY benchmark may not be appropriate.

Limitations

First and oremost among the limitations o this study is itsreliance on clinical trial data. While clinical trial data areconsidered the pre erred source or the basis o parameterinputs used in cost-e ectiveness analyses, the MS clinical trialshave been criticized or a number o methodological issues. Forexample, in the Cochrane review o the b-inter eron randomized,placebo-controlled clinical trials, 65 Rice et al. commented that thequality o the trials was variable with substantial methodologicalinadequacies. In its review o all immunomodulatory therapyrandomized clinical trials or MS, which included SC GA, 29 NICE noted similar methodological issues concerning

randomization, 30,40 allocation concealment, 30,36,40 intent-to-treatanalysis,30,37,40 and last observation carried orward analysis. 36 Finally, incomplete description o treatment dropouts in theshort-term clinical trials likely a ected trial results, where i dropouts were deemed to progress (worst-case scenario), thee ect o these drugs on relapse and disease progression rateswould likely be lost. 65

Second, our economic analyses did not include the impact o adverse events (e.g., cost and disutility) except to the extent thatthese might be captured indirectly in the proportion o patientswho discontinue therapy (Table 1). However, the cost o treatingadverse events would not likely a ect the overall results o thisanalysis since the most common adverse events reported or all4 immunomodulatory therapies were injection site reactions andin uenza-like symptoms, which were generally sel -limiting andsignifcantly decreased over time. 90,91 Nevertheless, these adverse

events would likely have an impact on patient utilities, whichwould in uence the cost-e ectiveness results. In the Prossermodel, treatment-specifc disutilities were assigned to accountor patient discom ort associated with each treatment, wherevalues ranged rom 0.066 or SC GA to 0.204 or SC IFN b-1b.These disutilities avored SC GA over the b-inter erons and thuswould not likely have changed the model results.

A third limitation arises rom the assumption in the model thatpatients who discontinued immunomodulatory therapy werenot allowed to switch to another therapy nor reinitiate therapyat a later time. It is expected that a proportion o patients willdiscontinue therapy due to worsening o MS symptoms (e.g., asspecifed by the Association o British Neurologists guidelinesor stopping therapy: 2 disabling relapses within 12 months,secondary progression with increased disability over 6 months,and loss o ability to walk that persists or 6 months 92); however,some patients may discontinue therapy and experience nochange in their disability status. On the basis o general clinicalpractice patterns, these patients would be likely candidates orswitching or reinitiating therapy. Had patients been allowed toswitch or reinitiate therapy in the model, the calculated ICERswould likely be less avorable or all the immunomodulatorytherapies. This assumption (no switching and no reinitiation o therapy) was also adopted in other MS models. 30,40,41,46 Evidencerom real-world observational studies in MS has indicatedthat rom 5% 73,93 to 13% o patients 72 switched therapy upondiscontinuation.

A ourth limitation arises rom the assumption in the modelregarding the estimation o relapse and disease progression ratesassociated with the immunomodulatory therapies. To accountor di erences in the immunomodulatory clinical trials, a fxedpatient population was assumed to enter the model (e.g., basedon fxed EDSS distribution), and in the frst 2 years o therapy,the model assumed that the probabilities o relapse and diseaseprogression used in the symptom management arm wereadjusted by a common percentage reduction (relapse: 27%;


14/17



disease progression: 30%). A ftted prediction curve o treatmente ects was used to estimate long-term treatment e ects o theimmunomodulatory therapies. This is a conservative approachto modeling treatment e ects over time that uses both the clinicaltrial and long-term data to predict uture outcomes. PreviousCEA/CUA studies or the immunomodulatory therapies reliedon data rom short-term clinical trials and made assumptions toextrapolate treatment e ects over time. At the time o the presenteconomic analyses, SC GA was the only immunomodulatorytherapy with long-term data collected in a systematic mannerover more than 10 years (although it is important to note thatby 10 years, 51.2% o patients eventually withdrew rom studyollow-up). 28 Thus, the ftted prediction curve or SC GA maybe a better representation o patient outcomes compared withthe curves estimated or the b-inter erons, which were based onsignifcantly shorter time horizons (e.g., the longest ollow-up

was 6 years or SC IFN- b1a).Fi th, the data used in the present economic analysis are

not rom comparative, head-to-head clinical trials o theimmunomodulatory therapies. Comparative trials o this type,particularly one in which all 4 therapies were evaluated head tohead, would be the ideal source o in ormation or the model,but the similarity in outcomes across the 4 therapies suggeststhat a very large and expensive randomized clinical trial wouldbe necessary in order to observe signifcant di erences among thetherapies. Nevertheless, reviews o the individual randomizedclinical trials or the 4 immunomodulatory therapies suggestthat the therapies have similar short-term e ects on relapse and

disease progression rates.66

Sixth, clinical trials are designed to specifcally test thee fcacy and tolerability o a particular therapy in a selectpatient population. The specifed inclusion and exclusioncriteria o clinical trials may not result in a population that isrepresentative o all patients (e.g., real-world situations). Thisis especially relevant in MS clinical trials in which patients areselected on the basis o a number o criteria, including havingexperienced an average o 2 relapses in the previous 2 years,a rate that is relatively higher than that reported or the generalMS population (range o 0.14 to 1.1 relapses per year). Fromthe modeling perspective, the clinical trial population is thebest available data; however, cost-e ectiveness results based ondata derived rom this population may or may not comport withpopulation-based care and the projection o average economiccosts.

nn Conclusions

All 4 immunomodulatory therapies used in the treatmento RRMS patients are associated with increased beneftscompared with symptom management alone, albeit at highercosts. The pharmacoeconomic model indicated that, o the4 immunomodulatory therapies used to manage MS and incomparison with symptom management, SC GA was the best

strategy. Sensitivity analysis indicated that the model was sensitiveto changes in a number o key parameters, and thus changesin these key parameters would likely in uence the estimatedcost-e ectiveness results. While the results o this analysisprovide decision makers with health economic in ormationsupporting the use o the immunomodulatory therapies, MSis a heterogeneous disease and physicians must select the mostappropriate treatment based on the disease characteristics o individual patients. Comparative head-to-head, randomizedclinical trials o the immunomodulatory therapies or thetreatment o MS are needed to confrm the results predicted byeconomic models.

What is already known about this subject A number o CEA/CUA evaluations o the immunomodulatory therapies

or MS have been published. However, the majority o these evaluationshave been conducted rom perspectives outside the United States. 30,40-45 The most recent U.S.-based analysis was the recent CUA evaluation o the immunomodulatory therapies or the treatment o nonprimary,progressive MS (e.g., RRMS and SPMS) over a 10-year time horizon. 46 Cost-e ectiveness results rom this analysis, as well as previous evaluations,were considerably higher than the arbitrary and commonly re erencedbenchmark o $50,000 per QALY.

What this study adds The present analysis is the frst economic model in MS to (1) incorporate

long-term data on treatment e ects, (2) include the e ect o NAbs on theb-inter erons, (3) account or the inherent di erences among clinical trialstudy designs o the immunomodulatory therapies, and (4) present resultsin terms o cost-utility (cost per QALY gained) and cost-e ectiveness(e.g., cost per year spent relapse ree.

Compared with the U.S.-based model published by Prosser et al. 46 in 2004,the results rom the present analysis are signifcantly di erent rom thatreported by Prosser and can be attributed to the underlying methodologyused to model MS, the most notable aspects o which include assumptionsregarding the treatment e ects associated with the immunomodulatorytherapies, model time horizon, and utility values.

DISClOSURESFunding or this research was provided to RTI-Health Solutions by TevaNeuroscience, Inc., and was obtained by authors Christopher Bell, JonathanGraham, and Stephanie Earnshaw Bell, Graham, and Earnshaw are employedby RTI-Health Solutions, an independent contract research organization thathas received research unding rom Teva Neuroscience, Inc., or this and other

studies and rom other pharmaceutical companies that market drugs or thetreatment o patients with multiple sclerosis and other medical conditions. Authors MerriKay Oleen-Burkey and Jane Castelli-Haley are employed by TevaNeuroscience, Inc., which manu actures glatiramer acetate. Kenneth Johnsonserves as a consultant to Teva Neuroscience, Inc. Oleen-Burkey discloses thatshe owns stock options in Teva Neuroscience, Inc., as well as stock in Pfzerand other health care companies.

Bell served as principal author o the study. Study concept and design werecontributed primarily by Bell and Earnshaw, with input rom the coauthors.Data collection was the work o Graham, Castelli-Haley, and Bell, with inputrom Oleen-Burkey and Earnshaw; data interpretation was the work o Bell,

Johnson, and Earnshaw with input rom the coauthors. Writing o the manu-script was primarily the work o Bell, Oleen-Burkey, and Castelli-Haley, withinput rom Earnshaw and Graham; its revision was the work o Bell, Graham,and Johnson, with input rom the coauthors.


15/17



REFERENCES1. National Multiple Sclerosis Society. National MS Society In ormationSourcebook. 2005. Available at: http://nationalmssociety.org/sourcebook.asp.

Accessed November 1, 2005.

2. Mayr WT, Pittock SJ, McClelland RL, Jorgensen NW, Noseworthy JH,Rodriguez M. Incidence and prevalence o multiple sclerosis in OlmstedCounty, Minnesota, 1985-2000. n u g . 2003;61(10):1373-77.

3. Noonan CW, Kathman SJ, White MC. Prevalence estimates or MS in theUnited States and evidence o an increasing trend or women. n u g . 2002;58(1):136-38.

4. Weinshenker BG, Bass B, Rice GP, et al. The natural history o multiplesclerosis: a geographically based study. 2. Predictive value o the early clinicalcourse. B a . 1989;112(pt 6):1419-28.

5. Con avreux C, Aimard G, Devic M. Course and prognosis o multiplesclerosis assessed by the computerized data processing o 349 patients. B a . 1980;103(2):281-300.

6. OConnor P. Key issues in the diagnosis and treatment o multiple sclerosis. An overview. n u g . 2002;59(6 suppl 3):S1-S33.

7. Lublin FD, Reingold SC. Defning the clinical course o multiple sclerosis:results o an international survey. National Multiple Sclerosis Society (USA) Advisory Committee on Clinical Trials o New Agents in Multiple Sclerosis.n u g . 1996;46(4):907-11.

8. Kremenchutzky M, Cottrell D, Rice G, et al. The natural history o multiplesclerosis: a geographically based study. 7. Progressive-relapsing and relapsing-progressive multiple sclerosis: a re-evaluation. B a . 1999;122(pt 10):1941-50.

9. Richards RG, Sampson FC, Beard SM, Tappenden P. A review o the naturalhistory and epidemiology o multiple sclerosis: implications or resource alloca-tion and health economic models. h a t c A . 2002;6(10):1-73.

10. Rice GPA, Kremenchutzky M, Cottrell DA, Baskerville J, Ebers GC.Observations rom the natural history cohort o London, Ontario. In: FilippiM, Comi G, eds. t c n u c c : p ma p g v Mu sc .Milano: Springer Verlag Italia; 2002.

11. Weinshenker BG, Bass B, Rice GP, et al. The natural history o multiplesclerosis: a geographically based study. I. Clinical course and disability. B a . 1989;112( pt 1):133-46.

12. Johnson KP, Brooks BR, Cohen JA, et al. Copolymer 1 reduces relapse rateand improves disability in relapsing-remitting multiple sclerosis: results o aphase III multicenter, double-blind placebo-controlled trial. The Copolymer 1Multiple Sclerosis Study Group. n u g . 1995;45(7):1268-76.

13. Jacobs LD, Cook air DL, Rudick RA, et al. Intramuscular inter eron beta-1aor disease progression in relapsing multiple sclerosis. The Multiple SclerosisCollaborative Research Group (MSCRG). A n u . 1996;39(3):285-94.

14. Clanet M, Radue EW, Kappos L, et al. A randomized, double-blind, dose-comparison study o weekly inter eron beta-1a in relapsing MS. n u g .2002;59(10):1507-17.

15. Randomised double-blind placebo-controlled study o inter eron beta-1ain relapsing/remitting multiple sclerosis. PRISMS (Prevention o Relapses andDisability by Inter eron beta-1a Subcutaneously in Multiple Sclerosis) StudyGroup. la c . 1998;352(9139):1498-1504.

16. Inter eron beta-1b is e ective in relapsing-remitting multiple sclerosis. I.Clinical results o a multicenter, randomized, double-blind, placebo-controlledtrial. The IFNB Multiple Sclerosis Study Group. n u g . 1993;43(4):655-61.

17. Paty DW, Li DK. Inter eron beta-1b is e ective in relapsing-remitting mul-tiple sclerosis. II. MRI analysis results o a multicenter, randomized, double-blind, placebo-controlled trial. UBC MS/MRI Study Group and the IFNBMultiple Sclerosis Study Group. n u g . 1993;43(4):662-67.

18. Durelli L, Verdun E, Barbero P, et al. Every-other-day inter eron beta-1bversus once-weekly inter eron beta-1a or multiple sclerosis: results o a2-year prospective randomised multicentre study (INCOMIN). la c . 2002;359(9316):1453-60.

19. Panitch H, Goodin DS, Francis G, et al. Randomized, comparative study o inter eron beta-1a treatment regimens in MS: The EVIDENCE Trial. n u g .2002;59(10):1496-1506.

20. Johnson KP, Brooks BR, Cohen JA, et al. Extended use o g latiramer acetate(Copaxone) is well tolerated and maintains its clinical e ect on multiple sclerosisrelapse rate and degree o disability. Copolymer 1 Multiple Sclerosis StudyGroup. n u g . 1998;50(3):701-08.

21. Clanet M, Kappos L, Hartung HP, Hohl eld R. Inter eron beta-1a in relapsingmultiple sclerosis: our-year extension o the European IFNbeta-1a Dose-Comparison Study. Mu sc . 2004;10(2):139-44.

22. PRISMS-4: Long-term e fcacy o inter eron-beta-1a in relapsing MS.n u g . 2001;56(12):1628-36.

23. Inter eron beta-1b in the treatment o multiple sclerosis: fnal outcomeo the randomized controlled trial. The IFNB Multiple Sclerosis Study Groupand The University o British Columbia MS/MRI Analysis Group. n u g .1995;45(7):1277-85.

24. Johnson KP, Brooks BR, Ford CC, et al. Sustained clinical benefts o glatiramer acetate in relapsing multiple sclerosis patients observed or 6 years.Copolymer 1 Multiple Sclerosis Study Group. Mu sc . 2000;6(4):255-66.

25. Johnson KP, Ford CC, Lisak RP, Wolinsky JS. Neurologic consequence o delaying glatiramer acetate therapy or multiple sclerosis: 8-year data. Ac an u sca d. 2005;111(1):42-47.

26. Kappos L, Stam Moraga M, Alsop JC. Long-term tolerability o inter eronbeta-1a in relapsing-remitting multiple sclerosis: 6-year sa ety ollow-up o the PRISMS study [abstract P334]. Presented at: Joint Meeting o the Americasand European Committees or Treatment and Research in Multiple Sclerosis;September 18-22, 2002; Baltimore, MD.

27. Kappos L, Traboulsee A, Constantinescu C, et al. Long-term subcutaneousLong-term subcutaneousinter eron beta-1a therapy in patients with relapsing-remitting MS. n u g .2006;67(6):944-53.

28. Ford CC, Johnson KP, Lisak RP, et al. A prospective open-label study o glatiramer acetate: over a decade o continuous use in multiple sclerosispatients. Mu sc . 2006;12:309-20.

29. National Institute or Clinical Excellence. Beta Inter eron and Glatiramer Acetate or the Treatment o Multiple Sclerosis. Technology Appraisal GuidanceNo. 32. London: National Institute or Clinical Excellence; 2002.

30. Chilcott J, McCabe C, Tappenden P, et al. Modelling the cost e ectivenesso inter eron beta and glatiramer acetate in the management o multiple sclero-sis. Commentary: evaluating disease modi ying treatments in multiple sclerosis.BMJ.2003;326(7388):522.

31. Goodin DS, Frohman EM, Garmany GP Jr, et al. Disease modi yingtherapies in multiple sclerosis: report o the Therapeutics and Technology

Assessment Subcommittee o the American Academy o Neurology and the MSCouncil or Clinical Practice Guidelines. n u g . 2002;58(2):169-78.

32. First Data Bank. Data on fle. t va n u c c ; 2006.

33. Drug Topics r d B k or Windows. Version 61127. Vol. 36. Montvale, NJ:Thomson PDR 2006; April 2005 release date.

34. Lundy J, Craig BM. The use o disease-modi ying agents among multiplesclerosis patients enrolled in Medicare rom 1995 to 2002 and the impact o Medicare Part D: analysis o claims data rom the Medicare current benefciarysurvey. C t . 2006;28(1):140-45.

35. Whetten-Goldstein K, Sloan FA, Goldstein LB, Kulas ED. A comprehensiveassessment o the cost o multiple sclerosis in the United States. Mu sc .1998;4(5):419-25.

36. Patwardhan MB, Matchar DB, Samsa GP, McCrory DC, Williams RG, Li TT.Cost o multiple sclerosis by level o disability: a review o literature. Mu sc . 2005;11(2):232-39.

37. Grima DT, Torrance GW, Francis G, Rice G, Rosner AJ, La ortune L. Costand health related quality o li e consequences o multiple sclerosis. Mu sc . 2000;6(2):91-98.


16/17



38. OBrien JA, Ward AJ, Patrick AR, Caro J. Cost o managing an episodeo relapse in multiple sclerosis in the United States. BMC h a s v r . 2003;3(1):17.

39. Orlewska E, Mierzejewski P, Zaborski J, et al. A prospective study o thefnancial costs o multiple sclerosis at di erent stages o the disease. eu J n u . 2005;12(1):31-39.

40. Kobelt G, Jonsson L, Fredrikson S. Cost-utility o inter eron beta1b in thetreatment o patients with active relapsing-remitting or secondary progressivemultiple sclerosis. eu J h a ec . 2003;4(1):50-59.

41. Nuijten MJ, Hutton J. Cost-e ectiveness analysis o inter eron beta inmultiple sclerosis: a Markov process analysis. Va u h a . 2002;5(1):44-54.

42. Iskedjian M, Walker JH, Gray T, Vicente C, Einarson TR, Gehshan A.Economic evaluation o Avonex (inter eron beta-Ia) in patients ollowing asingle demyelinating event. Mu sc . 2005;11(5):542-51.

43. Bose U, Ladkani D, Burrell A, et al. Cost-e ectiveness analysis o glatirameracetate in the treatment o relapsing-remitting multiple sclerosis: frst estimates. J M d ec . 2001;4:207-19.

44. Parkin D, McNamee P, Jacoby A, Miller P, Thomas S, Bates D. A cost-utilityanalysis o inter eron beta or multiple sclerosis. h a t c A . 1998;2(4):iii-54.45. Phillips CJ, Gilmour L, Gale R, et al. A cost utility model o beta-inter-eron in the treatment o relapsing-remitting multiple sclerosis. J M d ec . 2001;4:35-50.

46. Prosser LA, Kuntz KM, Bar-Or A, Weinstein MC. Cost-e ectiveness o inter-eron beta-1a, inter eron beta-1b, and glatiramer acetate in newly diagnosednon-primary progressive multiple sclerosis. Va u h a . 2004;7(5):554-68.

47. Hirth RA, Chernew ME, Miller E, Fendrick AM, Weissert WG. Willingnessto pay or a quality-adjusted li e year: in search o a standard. M d D c Mak g. 2000;20(3):332-42.

48. Rossman HS. Neutralizing antibodies to multiple sclerosis treatments. J Ma ag Ca p a m. 2004;10(3)(suppl):S12-S19. Available at: http://www.amcp.org/data/jmcp/June04Supplement1.pd . Accessed March 5, 2007.

49. Kurtzke JF. Rating neurologic impairment in multiple sclerosis: anexpanded disability status scale (EDSS). n u g . 1983;33(11):1444-52.50. Gold M, Siegel JE, Russell LB, et al. C -e c v h a a d M d c . New York: Ox ord University Press; 1996.

51. Oleen-Burkey M, Kobelt G, Borgstrom F. Costs and quality o li e o patients with relapsing-remitting multiple sclerosis currently on immuno-modulatory therapy in the United States [abstract]. International Committeeon Databases in Multiple Sclerosis; 2003.

52. Kobelt G. Economic evidence in multiple sclerosis: a review. eu J h aec . 2004;(5 suppl 1):S54-S62.

53. Kobelt G, Lindgren P, Parkin D, et al. Costs and quality o li e in multiplesclerosis: a cross-sectional observational study in the United Kingdom. Stock-holm: Stockholm School o Economics; 2000. EFI Research Report No. 398.

54. Kobelt G, Berg J, Atherly D, Hadjimichael O. Costs and quality o li ein multiple sclerosis: a cross-sectional study in the United States. n u g .

2006;66(11):1696-1702.55. Rodriguez M, Siva A, Ward J, Stolp-Smith K, OBrien P, Kurland L. Impair-ment, disability, and handicap in multiple sclerosis: a population-based studyin Olmsted County, Minnesota. n u g . 1994;44(1):28-33.

56. Pittock SJ, Mayr WT, McClelland RL, et al. Change in MS-related disabilityin a population-based cohort: a 10-year o llow-up study. n u g . 2004;62(1):51-59.

57. MacLurg K, Reilly P, Hawkins S, Gray O, Evason E, Whittington D. A primary care-based needs assessment o people with multiple sclerosis.B J G p ac . 2005;55(514):378-83.

58. Goodkin DE, Hertsgaard D, Rudick RA. Exacerbation rates and adherenceto disease type in a prospectively ollowed-up population with multiple sclerosis.Implications or clinical trials. A c n u . 1989;46(10):1107-12.

59. Weinshenker BG, Ebers GC. The natural history o multiple sclerosis.Ca J n u sc . 1987;14(3):255-61.60. Runmarker B, Andersen O. Prognostic actors in a multiple sclerosisincidence cohort with twenty-fve years o ollow-up. B a . 1993;116 (pt 1):117-34.

61. Centers or Disease Control and Prevention. U.S. vital statistics reports.Table 3. Number o deaths and death rate, by age, race, and sex. 2003;52(3).

Available at: http://www.cdc.gov/nchs/data/nvsr/nvsr52/nvsr52_03.pd . Accessed June 17, 2005.

62. Phadke JG. Survival pattern and cause o death in patients with multiplesclerosis: results rom an epidemiological survey in northeast Scotland. J n u n u u g p c a . 1987;50(5):523-31.63. Simon JH, Jacobs LD, Campion M, et al. Magnetic resonance studies o Magnetic resonance studies o intramuscular inter eron beta-1a or relapsing multiple sclerosis. The MultipleSclerosis Collaborative Research Group. A n u . 1998;43(1):79-87.64. Filippini G, Munari L, Incorvaia B, et al. Inter erons in relapsing remittingmultiple sclerosis: a systematic review. la c . 2003;361(9357):545-52.65. Rice GP, Incorvaia B, Munari L, et al. Inter eron in relapsing-remitting mul-tiple sclerosis. C c a Da aba s r v. 2001;(4):CD002002.

66. Khan O, Zabad R, Caon C, Zvartau-Hind M, Tselis A, Lisak R. Comparativeassessment o immunomodulating therapies or relapsing-remitting multiplesclerosis. Cns D ug . 2002;16(8):563-78.67. Kappos L, Clanet M, Sandberg-Wollheim M, et al. Neutralizing antibodiesNeutralizing antibodiesand e fcacy o inter eron beta-1a: a 4-year controlled study. n u g . 2005;65(1):40-47.

68. Francis GS, Rice GP, Alsop JC. Inter eron beta-1a in MS: results ollowingdevelopment o neutralizing antibodies in PRISMS. n u g . 2005;65(1):48-55.

69. Giovannoni G, Munschauer FE III, Deisenhammer F. Neutralising anti-bodies to inter eron beta during the treatment o multiple sclerosis. J n un u u g p c a . 2002;73(5):465-69.

70. Medical Consumer Price Index. Gabriel, Roeder, Smith & Company. Avail-able at: www.grsnet.con/grs_library/index.html. Accessed June 17, 2005.

71. Frohman EM, Stuve O, Havrdova E, et al. Therapeutic considerations ordisease progression in multiple sclerosis: evidence, experience, and utureexpectations. A c n u . 2005;62(10):1519-30.72. Tremlett HL, Oger J. Interrupted therapy: stopping and switching o thebeta-inter erons prescribed or MS. n u g . 2003;61(4):551-54.73. Rio J, Porcel J, Tellez N, et al. Factors related with treatment adherence tointer eron beta and glatiramer acetate therapy in multiple sclerosis. Mu sc . 2005;11(3):306-09.

74. Lichtenberg FR. Availability o new drugs and Americans ability to work. J occu e v M d. 2005;47(4):373-80.75. Lage MJ, Castelli-Haley J, Oleen-Burkey MA. E ect o immunomodulatorytherapy and other actors on employment loss time in multiple sclerosis. w k. 2006;27(2):143-51.

76. Busche KD, Fisk JD, Murray TJ, Metz LM. Short term predictors o unem-ployment in multiple sclerosis patients. Ca J n u sc . 2003;30(2):137-42.77. Kobelt G. Costs and quality o li e in multiple sclerosis: a cross-sectionalstudy in the USA. Stockholm: Stockholm School o Economics; 2004. EFI

Working Paper Series in Economics and Finance No. 594.

78. Parkin D, McNamee P, Jacoby A, Miller P, Thomas S, Bates D. A cost-utilityanalysis o inter eron beta or multiple sclerosis. h a t c A . 1998;2(4):iii-54.79. Academy o Managed Care Pharmacy. AMCP F ma F mu a subm -

, version 2.1. 2005. Available at: http://www. mcpnet.org/data/resource/ Format~Version_2_1~Final_Final.pd . Accessed January 2, 2007.80. Canadian Coordinating O fce or Health Technology Assessment. Guide-lines or economic evaluation o pharmaceuticals: Canada. 2nd ed. Ottawa:Canadian Coordinating O fce or Health Technology Assessment (CCHOTA);1997.


17/17


81. National Institute or Clinical Excellence. Gu d M d t c g A a a . London: National Institute or Clinical Excellence (NICE); 2004.

82. Prosser LA, Kuntz KM, Bar-Or A, Weinstein MC. Patient and communitypre erences or treatments and health states in multiple sclerosis. Mu sc .2003;9(3):311-19.

83. Phillips CJ. The cost o multiple sclerosis and the cost e ectiveness o disease-modi ying agents in its treatment. Cns D ug . 2004;18(9):561-74.

84. Maetzel A, Krahn M, Naglie G. The cost e ectiveness o ro ecoxib andcelecoxib in patients with osteoarthritis or rheumatoid arthritis. Ar um. 2003;49(3):283-92.

85. Within-trial cost-e ectiveness o li estyle intervention or met ormin or theprimary prevention o type 2 diabetes. D ab Ca . 2003;26(9):2518-23.

86. Ladabaum U. Sa ety, e fcacy and costs o pharmacotherapy or unctionalgastrointestinal disorders: the case o alosetron and its implications. A mp a mac t . 2003;17(8):1021-30.

87. Tonelli M, Winkelmayer WC, Jindal KK, Owen WF, Manns BJ. The cost-e ectiveness o maintaining higher hemoglobin targets with erythropoietin inhemodialysis patients. K d i . 2003;64(1):295-304.

88. Stone PW, Teutsch S, Chapman RH, Bell C, Goldie SJ, Neumann PJ. Cost-utility analyses o clinical preventive services: published ratios, 1976-1997. Am J p v M d. 2000;19(1):15-23.

89. Bell CE, Botteman MF, Gao X, et al. Cost-e ectiveness o trans usion o Cost-e ectiveness o trans usion o platelet components prepared with pathogen inactivation treatment in theUnited States. C t . 2003; 25(9):2464-86.

90. Simpson D, Noble S, Perry C. Glatiramer acetate: a review o its use inrelapsing-remitting multiple sclerosis. Cns D ug . 2002;16(12):825-50.

91. McCormack PL, Scott LJ. Inter eron-beta-1b: a review o its use in re-lapsing-remitting and secondary progressive multiple sclerosis. Cns D ug . 2004;18(8):521-46.

92. Association o British Neurologists. Guidelines or the use o beta inter-erons and glatiramer acetate in multiple sclerosis; 2001. Accessed January 2,2007.

93. ORourke KE, Hutchinson M. Stopping beta-inter eron therapy in multiplesclerosis: an analysis o stopping patterns. Mu sc . 2005;11(1):46-50.

94. Rudick RA, Cutter GR, Baier M, et al. Estimating long-term e ects o Estimating long-term e ects o disease-modi ying drug therapy in multiple sclerosis patients. Mu sc .2005;11(6):626-34.

95. Ebers G, Rice G, Wol C, et al. 16-year long-term ollow-up o inter eronbeta-1b treatment in RRMS. n u g . 2005;64(suppl 1):A385.

96. Placebo-controlled multicentre randomised trial o inter eron beta-1b intreatment o secondary progressive multiple sclerosis. European Study Groupon inter eron beta-1b in secondary progressive MS. la c . 1998;352(9139):1491-97.

97. Randomized controlled trial o inter eron-beta-1a in secondary progressiveMS: Clinical results. n u g . 2001;56(11):1496-1504.

98. Jacobs LD, Beck RW, Simon JH, et al. Intramuscular inter eron beta-1atherapy initiated during a frst demyelinating event in multiple sclerosis.CHAMPS Study Group. n e g J M d. 2000;343(13):898-904.

99. Cohen JA, Cutter GR, Fischer JS, et al. Beneft o inter eron beta-1a onMSFC progression in secondary progressive MS. n u g . 2002;59(5):679-87.

Documents

J Manag Care Pharm. 2007 Apr;13(3) 245-61