Accepted: 29 June 2011; published online: 21 July 2011
Citation: J Med Econ 2011; 14:617–27
Abstract
Background:
With the addition of new agents for the treatment of multiple sclerosis (MS) (e.g., fingolimod), there is a need
to evaluate the relative value of newer therapies in terms of cost and effectiveness, given healthcare
resource constraints in the United States.
Objective:
To assess the cost-effectiveness of natalizumab vs fingolimod in patients with relapsing MS.
Methods:
A decision analytic model was developed to estimate the incremental cost per relapse avoided of
natalizumab and fingolimod from a US managed care payer perspective. Two-year costs of treating
patients with MS included drug acquisition costs, administration and monitoring costs, and costs of
treating MS relapses. Effectiveness was measured in terms of MS relapses avoided (data from AFFIRM
and FREEDOMS trials). One-way and probabilistic sensitivity analyses were conducted to assess uncertainty.
Results:
Mean 2-year estimated treatment costs were $86,461 (natalizumab) and $98,748 (fingolimod). Patients
receiving natalizumab had a mean of 0.74 relapses avoided per 2 years vs 0.59 for fingolimod. Natalizumab
dominated fingolimod in the incremental cost-effectiveness analysis, as it was less costly and more effective
in reducing relapses. One-way sensitivity analysis showed the results of the model were robust to changes in
drug acquisition costs, administration costs, and costs of treating MS relapses. Probabilistic sensitivity
analysis showed natalizumab was cost-effective 95.1% of the time, at a willingness-to-pay (WTP) threshold
of $0 per relapse avoided, increasing to 96.3% of the time at a WTP threshold of $50,000 per relapse
avoided.
Limitations:
Absence of data from direct head-to-head studies comparing natalizumab and fingolimod, use of relapse
rate reduction rather than sustained disability progression as primary model outcome, assumption of 100%
adherence to MS treatment, and not capturing adverse event costs in the model.
Conclusions:
Natalizumab dominates fingolimod in terms of incremental cost per relapse avoided, as it is less costly and
more effective.
Introduction
Multiple sclerosis (MS) is a chronic, disabling disease of the central nervoussystem, affecting 400,000 patients in the United States1. The progression,
! 2011 Informa UK Ltd www.informahealthcare.com/JME Cost-effectiveness of natalizumab vs fingolimod O’Day et al. 617
Jour
nal o
f M
edic
al E
cono
mic
s D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y U
nive
rsity
of
Lim
eric
k on
05/
11/1
3Fo
r pe
rson
al u
se o
nly.
severity, and symptoms of MS vary from patient to patient,but the disease can be grouped into four categories: relaps-ing–remitting MS (RRMS); primary–progressive MS; sec-ondary–progressive MS; and progressive–relapsing MS.RRMS is the most commonly diagnosed category, affectingas many as 85% of MS patients. Patients suffering fromRRMS incur episodes of neurological deterioration withsubsequent periods of stability without disease progressionin between exacerbations1. Relapses occur throughout thecourse of the disease, and the result is an accumulation ofphysical disability and cognitive decline. Despite thedebilitating effects of the disease, life expectancy is gener-ally not affected for patients with MS2. As such, given thechronic nature of the disease and the propensity for MSto develop in young adults between the ages of 20 and40 years, the clinical impact of the disease burden is sub-stantial over many years of a patient’s life3.
The economic impact of MS is also significant forpatients as well as for payers. Mean total lifetime costs(including direct and indirect costs) for patients withMS have been estimated to be as high as $2.2 millionper patient, based on 1994 dollars4. More than half ofthe costs associated with MS (57%) can be attributed toindirect costs such as lost income, equipment, and patientcare (including paid formal care and unpaid informalcare)5. When considering direct medical costs associatedwith MS, drug costs account for as much as 65% of medicalspending for MS patients5. From a managed care perspec-tive, patients with MS are a contributing factor to bothhigher per-member costs and costs that are increasing overtime. A retrospective analysis of pharmacy and medicalclaims data found that 12-month direct healthcare costsincreased for patients with MS compared to patients with-out MS6. A study conducted by Pope et al. found meandirect medical costs for members with MS to be two tothree times higher than for members without MS7.Kunze et al. analyzed cost trends and utilization of agentsfor the treatment of MS and found that per member permonth costs increased by nearly 60% during the timeperiod from 2004 to 20078. Schafer et al. looked at medicaland pharmacy cost trends from 2006 to 2009 and foundthat they increased by nearly 30% per member per year9.In addition, almost 60% of the costs during all four yearswere attributable to pharmacy costs9.
Though a significant portion of costs is attributable todrug therapy9, disease-modifying therapies (DMTs) havebeen shown to significantly reduce the number of MSrelapses in clinical trials and have subsequently becomean integral component of therapy for MS patients.Reducing the number of relapses is not only beneficialfor the patient, but the cost of treating MS relapses hasbeen shown to substantially increase the cost of medicalcare10. Costs associated with treating relapses vary basedon the severity of the relapse, but can range anywhere from$243 (2002 dollars, $329 inflated to 2010 dollars)
to $12,870 (2002 dollars, $17,420 inflated to 2010 dollars)per episode10,11. Management strategies that can be incor-porated to reduce the frequency and/or severity of relapseshave the potential to reduce the overall cost burden of thedisease10.
Healthcare decision makers in the United States mustcontinue to make coverage and reimbursement decisionsunder ever-increasing budget constraints. In light of thesignificant economic impact of MS and the evolving MStreatment landscape, there exists a need to critically eval-uate DMTs in terms of cost, safety, and clinical effective-ness. Interferons and glatiramer acetate are wellestablished in terms of efficacy and safety for the treatmentof RRMS and have been available for decades as first-linetherapy12. The two newest agents approved for the treat-ment of RRMS are natalizumab and fingolimod.Natalizumab is generally recommended for patients whohave had an inadequate response to, or are unable to tol-erate, an alternate MS therapy. Fingolimod is the first oralagent to reduce relapses and delay disability progression inpatients with relapsing forms of MS13. Fingolimod wasapproved in the European Union for patients with highlyactive RRMS despite treatment with beta interferon, or inpatients with rapidly evolving severe RRMS. In theUnited States, fingolimod was approved in 2010 for thetreatment of relapsing forms of MS. While the safety pro-files of beta interferons and glatiramer acetate are wellknown, the higher risks of potentially severe adverseevents associated with fingolimod, including bradyar-rhythmia and atrioventricular block at treatment initia-tion, infections, macular edema, respiratory effects, andhepatic effects14, as well as an observed higher rate ofskin cancers in clinical trials, may lead healthcare profes-sionals in the United States to use fingolimod as a second-line therapy until longer-term, real-world safety data areavailable14–17. Accordingly, there is a need to compare therelative value of newer MS therapies, such as fingolimodand natalizumab, in terms of cost, safety, and effectiveness,given the healthcare resource constraints.
Previous models comparing injectable DMTs, includingnatalizumab, have been developed18–21. Gani et al. evalu-ated natalizumab compared to other injectable DMTs andfound natalizumab to be the most cost-effective therapy interms of cost per quality-adjusted life year (QALY) forpatients with MS in the United Kingdom20. Chiao andMeyer evaluated natalizumab compared to other inject-able DMTs and found natalizumab to be the most cost-effective therapy as measured by cost per relapse avoidedfrom the perspective of a United States managed carepayer21. Given the cost-effectiveness demonstrated fornatalizumab in the models by Gani et al. and Chiaoet al., the fact that natalizumab has the highest efficacyof the injectable DMTs22–27, and the recent introductionof fingolimod to the MS treatment landscape, the objec-tive of this study was to develop an evidence-based
Journal of Medical Economics Volume 14, Number 5 October 2011
618 Cost-effectiveness of natalizumab vs fingolimod O’Day et al. www.informahealthcare.com/JME ! 2011 Informa UK Ltd
Jour
nal o
f M
edic
al E
cono
mic
s D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y U
nive
rsity
of
Lim
eric
k on
05/
11/1
3Fo
r pe
rson
al u
se o
nly.
economic model to assess the incremental cost-effectiveness of natalizumab and fingolimod, based onclinical trial data for patients with relapsing MS.
Methods
Model design
A decision analytic model was developed to estimate thecost-effectiveness of selected agents used in the treatmentof relapsing MS from the perspective of a managed careorganization in the United States. The time horizon for themodel was a static 2 years, representing the time frominitiation of therapy with the model comparator agentsforward. This time period is representative of the clinicaltrial data used to populate the model. It is recognized thatMS is a chronic disease that progresses after diagnosisthroughout the remainder of a patient’s lifetime; however,extrapolations beyond the available trial data were notexplored. Furthermore, this time period is representativeof the interests of a United States managed care organiza-tion, in which the average duration of continuous enroll-ment for individuals is approximately 2 years28. The twoagents included in this model for comparison were natali-zumab 300 mg administered intravenously (IV) once every4 weeks, and fingolimod 0.5 mg administered orally oncedaily. Dosing for each agent was based upon the approvedprescribing information13,22.
The model consisted of an incremental analysis usingrelapse rates from published clinical trials as the measure ofeffectiveness29,30. The AFFIRM and FREEDOMS trialswere pivotal phase 3 trials comparing natalizumab andfingolimod, respectively, to placebo29,30. Although natali-zumab is indicated as second-line therapy, AFFIRM stud-ied the use of natalizumab as first-line therapy, similar tothe way fingolimod was studied in FREEDOMS. Relapserates were selected as the primary clinical outcome ofinterest for inclusion in the model because relapses posea substantial burden to the patient and are often associatedwith considerable use of healthcare resources. Total 2-yearcosts of therapy were estimated based on acquisition costs,administration costs, and monitoring costs. Adverseevents were not included in the model, as serious adverseevents associated with therapy are rare. Though severeadverse events such as hepatotoxicity and progressive mul-tifocal leukoencephalopathy (PML) have occurred inpatients receiving natalizumab, the incidence of PMLhas been estimated to be 1.06 per 1000 patients (95% con-fidence interval [CI]: 0.85, 1.31 in 1000 patients). As ofJanuary 2011, there were 85 cases of PML identified from75,500 patients receiving natalizumab worldwide31. Todate, there is no consensus on effective management ofPML, and a literature search of PubMed did not identifyany studies assessing the costs associated with PML.
If estimates from a study of mean expenditures in the lastyear of life for patients receiving hospice care ($27,426)are used as a proxy measure for costs associated with sup-portive care for patients who develop PML, expected costsassociated with PML would be less than $30 per patient32.While the clinical and humanistic impact of PML is sig-nificant, the costs associated with PML managementare unlikely to have a significant economic impact on ahealth plan.
The relevant outcome of the model was the incremen-tal cost per relapse avoided for natalizumab as compared tofingolimod.
Model inputs
Drug acquisition costsThe costs per package, monthly drug costs, and 2-year drugcosts were determined based on the wholesale acquisitioncost as of September 2010. Published prescribing informa-tion was used to determine the number of doses per pack-age and number of packages per year13,22. The modelassumed there were no product rebates, discounts, orpatient copays or coinsurances. Drug acquisition costs fornatalizumab were calculated based on 13 packages peryear, one dose per package, and a cost of $3076 per pack-age, for a 2-year total of $79,977. Acquisition costs forfingolimod were calculated based on 13 packages peryear, 28 doses per package, and a cost of $3688 per package,for a 2-year total of $95,902. Real-world adherence pat-terns were not available for fingolimod at the time themodel was developed. To err on the side of a conservativeestimate, patients were assumed to be 100% adherent tonatalizumab and fingolimod in the model, thus maximizingthe drug costs for each agent.
Drug administration and monitoring costsTo account for differences in the routes of administrationfor natalizumab (1-hour IV administration requiring super-vision by a healthcare provider) and fingolimod (oral), themodel only included the 2-year costs of drug administra-tion for monthly infusions of natalizumab. Twenty-sixadministrations of natalizumab over the 2-year time hori-zon at a cost of $143.07 per administration33 yielded a totaladministration cost of $3720 (Table 1).
As treatment patterns and monitoring parameters maydiffer among clinical practices, the laboratory tests, imag-ing studies, and/or physician office visits required or rec-ommended (based on the prescribing information or theRisk Evaluation and Mitigation Strategy), for each agentwere included in the model when possible (Table 1).No explicit recommendations regarding the frequency oflaboratory monitoring are provided in the natalizumabprescribing information. However, given the possibilitythat administration of the agent can cause
Journal of Medical Economics Volume 14, Number 5 October 2011
! 2011 Informa UK Ltd www.informahealthcare.com/JME Cost-effectiveness of natalizumab vs fingolimod O’Day et al. 619
Jour
nal o
f M
edic
al E
cono
mic
s D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y U
nive
rsity
of
Lim
eric
k on
05/
11/1
3Fo
r pe
rson
al u
se o
nly.
immunosuppression/infection, hepatotoxicity, or labora-tory abnormalities22, the model assumed that patientswould receive a complete blood count and a liver functiontest at the time of therapy initiation. The model alsoassumed natalizumab patients would receive one magneticresonance imaging (MRI) scan over the 2-year timeperiod, as an MRI is recommended prior to initiating ther-apy with natalizumab22. Natalizumab is provided through arestricted distribution program, and it is recommendedthat prescribers evaluate patients after 3 months and6 months of therapy and every 6 months thereafter.Therefore, the model included a total of five neurologistvisits over 2 years for patients receiving natalizumab.These assumptions were confirmed by expert opinion.
In the absence of specific monitoring criteria, the modelemployed conservative estimates, with assumptions vali-dated and approved by expert opinion. For example, noexplicit recommendations around the frequency of labora-tory monitoring are provided in the fingolimod prescribinginformation. However, it was assumed that patients wouldreceive a complete blood count at the time of therapyinitiation. To calculate the number of liver functiontests associated with fingolimod for the model, one testwas assumed to be performed at baseline, and it wasassumed that 11% of patients would receive an additionaltest annually for experiencing fatigue, and 8.9% of patientswould receive an annual test for experiencing nausea,resulting in a total of 1.398 liver function tests per patient(1þ [2 * 11%]þ [2 * 8.9%]¼ 1.398)13,30. The modelassumed no MRIs were performed during the 2-yearperiod for fingolimod, which has no specific guidelinesfor MRI scans in its prescribing information. In addition,although patients with MS typically have regular follow-upvisits with their healthcare provider as a standard of care,there are no published guidelines or recommendations inthe fingolimod prescribing information regarding the fre-quency of neurologist visits. Therefore, the model conser-vatively assumes one neurologist visit after 6 months oftherapy and one every 6 months thereafter, for a total ofthree neurologist visits over 2 years for patients receivingfingolimod. These assumptions were confirmed by expertopinion.
The model assumes that all patients will receive anelectrocardiogram prior to initiating therapy with fingoli-mod based on a recommendation in the fingolimod pre-scribing information to identify underlying risk factors forbradycardia and atrioventricular block13. In addition, thefingolimod prescribing information states that patientsshould be observed for 6 hours after the first dose to mon-itor for signs and symptoms of bradycardia13, and thesecosts ($96)33 were incorporated into the model.Varicella zoster antibody testing and subsequent vaccina-tion for the varicella zoster virus, when appropriate, arerecommended prior to commencing therapy with fingoli-mod13. The model assumes that approximately 10% ofpatients are seronegative for varicella zoster, and, thus,0.1 patients would require varicella zoster virus antibodytesting and vaccination34. Fingolimod may be associatedwith a decrease in pulmonary function13. Therefore, themodel assumes 0.108 pulmonologist visits, spirometrytests, and diffusion lung capacity for carbon monoxidetests per patient based on the statistic that 5.4% ofpatients receiving fingolimod experienced dyspnea20
(5.4% * 2¼ 0.108). The fingolimod prescribing informa-tion suggests that an adequate ophthalmologic evaluationshould be performed at baseline and 3 to 4 months aftertreatment initiation, and that if patients report visual dis-turbances while on fingolimod, additional ophthalmologicevaluation should be undertaken13. In addition, MS
Table 1. Two-year drug administration and monitoring inputs fornatalizumab and fingolimod.
Parameter FrequencyOver 2 Years
UnitCost33
Total Costper Patient
Over 2 Years*
Drug administrationnatalizumab 26 $143 $3720fingolimod 0 0 0
CBC – complete blood count; LFT – liver function test; MRI – magneticresonance imaging; DLCO – diffusion lung capacity for carbon monoxide.*Some values may be off due to rounding.
Journal of Medical Economics Volume 14, Number 5 October 2011
620 Cost-effectiveness of natalizumab vs fingolimod O’Day et al. www.informahealthcare.com/JME ! 2011 Informa UK Ltd
Jour
nal o
f M
edic
al E
cono
mic
s D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y U
nive
rsity
of
Lim
eric
k on
05/
11/1
3Fo
r pe
rson
al u
se o
nly.
patients with diabetes mellitus or a history of uveitis shouldundergo an ophthalmologic evaluation prior to initiatingfingolimod and have regular follow-up ophthalmologicevaluations while receiving therapy13. Therefore, themodel assumes one ophthalmologist visit at baseline andone visit at 3 to 4 months, plus annual visits for the 7.8% ofthe population with diabetes35 and the 1.6% of patientswith macular edema, as reported in Kappos et al., resultingin a total of 2.188 visits per patient [1þ 1þ (2 * 7.8%)þ(2 * 1.6%)¼ 2.188]13,30,35. These assumptions were alsoconfirmed by expert opinion.
Efficacy inputsThe efficacy outcome included in the model was the antic-ipated reduction in relapse rates during the 2-year timehorizon. The definitions of clinical relapse differed slightlyin AFFIRM and FREEDOMS. In AFFIRM, relapses weredefined as new or recurrent neurologic symptoms, not asso-ciated with fever or infection, which lasted for at least24 hours and were accompanied by new neurologic signsfound by the examining neurologist29. In FREEDOMS,relapses were defined as symptoms accompanied by anincrease of at least half a point in the ExpandedDisability Status Scale (EDSS) score, of 1 point in eachof two EDSS functional-system scores, or of 2 points in oneEDSS functional-system score (excluding scores for thebowel–bladder or cerebral functional systems)30.Therefore, for comparison purposes, the model utilizedthe number of relapses over 2 years that occurred in theplacebo treatment arms during the phase 3 clinical trialsfor natalizumab and fingolimod29,30 and the number ofpatients in each placebo group29,30 to calculate theweighted average number of relapses per patient prior totreatment with either agent. As shown in Table 2, theweighted average number of relapses per patient prior totreatment was calculated to be 0.54.
The relative reduction in relapse rate was obtained fornatalizumab and fingolimod from their respective phase 3clinical studies29,30. The number of relapses avoided withtreatment was calculated as the weighted average numberof relapses per patient prior to treatment, multiplied by therelapse rate reduction for natalizumab (68.1%) and fingo-limod (54.0%)29,30. Because there are no head-to-headstudies comparing the efficacy of natalizumab and fingoli-mod, we conducted this indirect comparison in which weadjusted for the placebo relapse rate. The number ofrelapses per patient that could be anticipated to occurduring 2 years of therapy was then calculated as theweighted average number of relapses per patient prior totreatment, minus the number of relapses avoided withtreatment (Table 2).
The cost of treating an MS relapse was also included inthe model. As the severity of the relapse affects the leveland intensity of management required, the cost of a relapsein the model was divided into three categories as describedby O’Brien et al.: a relapse requiring low-intensity medicalmanagement (physician office visit or symptom manage-ment with medications); a relapse requiring medium-intensity medical management (emergency room [ER]visit, observational unit, or acute treatments such as IVmedications [e.g. steroids] in either an outpatient orhome setting); and a relapse requiring high-intensity med-ical management (hospitalization and subsequent care)10.Costs for each of the categories were based on publishedliterature10 and were adjusted using the Bureau of LaborStatistics’ medical care component of the Consumer PriceIndex from 2002 dollars to 2010 dollars11. Specificadjusted costs for each category of medical managementwere as follows: low-intensity, $329; medium-intensity,$2500; and high-intensity, $17,420. The proportion ofpatients requiring each category of care was obtainedfrom other published sources20 and was set at 40% forlow-intensity, 40% for medium-intensity, and 20% for
Table 2. Total cost per patient and relapses avoided per patient.
Natalizumab Fingolimod
Cost MeasureDrug costs $79,977 $95,902Administration costs $3720 NAMonitoring costs $1172 $550Cost of managing MS relapses per patient $1592 $2296Total cost per patient $86,461 $98,748
Effectiveness MeasureAnnualized relapse rate (placebo group) 0.73 0.40Number of patients receiving placebo 315 418Weighted average annualized relapse rate per patient prior to treatment 0.54Relative reduction in relapse rate 68.1% 54.0%
Two-year number of relapses per patient 0.35 0.50Two-year number of relapses avoided per patient 0.74 0.59Cost per relapse avoided $117,164 $168,754
MS – multiple sclerosis; NA – not applicable.
Journal of Medical Economics Volume 14, Number 5 October 2011
! 2011 Informa UK Ltd www.informahealthcare.com/JME Cost-effectiveness of natalizumab vs fingolimod O’Day et al. 621
Jour
nal o
f M
edic
al E
cono
mic
s D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y U
nive
rsity
of
Lim
eric
k on
05/
11/1
3Fo
r pe
rson
al u
se o
nly.
high-intensity medical management. The weighted aver-age cost of managing an MS relapse was calculated as$4616. The cost of treating relapses per patient was calcu-lated as the number of relapses per patient that could beanticipated to occur during 2 years of therapy, multipliedby the weighted average cost of managing an MS relapse.
The cost per relapse avoided was then calculated as thetotal 2-year cost of therapy (pharmacy costs, administra-tion costs, monitoring costs, and cost of treating MSrelapses) divided by the number of relapses avoided over2 years. The incremental cost per relapse avoided wascalculated as the difference between natalizumab and fin-golimod in the total 2-year cost of therapy divided by thedifference between natalizumab and fingolimod in thenumber of relapses avoided over 2 years.
Sensitivity analysis
A one-way sensitivity analysis was performed to determinethe impact of varying model inputs on the cost-effective-ness of natalizumab. Each model input (relapse rate reduc-tion, drug acquisition costs, drug administration costs, andcost of treating MS relapses) was varied around the base-case value, while keeping other inputs constant, to deter-mine the effect on the incremental cost per relapseavoided. The relapse rate reduction for natalizumab andfingolimod were varied around the base-case according tothe 95% CIs reported in the clinical trials. The values forthe cost of treating MS relapses were based on calculated95% CIs10. The drug acquisition costs and drug adminis-tration costs were varied around the base-case� 10%. Theinputs utilized for the one-way sensitivity analysis areshown in Table 3.
In addition, a probabilistic sensitivity analysis was con-ducted to assess the joint uncertainty of all model param-eters simultaneously. Relative relapse rate reductions weremodeled using a log-normal distribution, costs were mod-eled using normal distributions, the proportion of patientsreceiving low-, medium-, and high-intensity medical man-agement for MS relapses was modeled using a Dirichlet
distribution, and the monitoring resources utilized weremodeled using gamma distributions. The parameterranges for the probabilistic distributions were determinedusing the 95% CIs, where data were available (relativerelapse rate reductions and cost of treating MS relapses).For drug acquisition and administration costs, it wasassumed that the 95% CI was equivalent to �10% of thebase-case value. For resource utilization inputs, it wasassumed that the 95% CI was equivalent to �50% ofthe base-case value, to reflect the greater uncertaintyaround these parameters. The probabilistic sensitivityanalysis was run with 1000 Monte Carlo simulations,and a cost-effectiveness acceptability curve was gener-ated over a range of willingness-to-pay (WTP) thresh-olds for an MS relapse avoided from $0 to $100,000per relapse.
Results
The 2-year costs per patient, number of relapses avoidedper patient, and cost per relapse avoided are shown inTable 2. While natalizumab is associated with administra-tion costs and higher monitoring costs, it had the lowestaverage cost-effectiveness ratio in terms of reducingrelapses over 2 years. The 2-year drug costs and the costassociated with managing MS relapses per patient werehigher for fingolimod than for natalizumab, making thetotal 2-year cost per patient higher for fingolimod($98,748) vs natalizumab ($86,461), with the discrepancylargely attributable to drug costs. The incremental cost-effectiveness ratio (ICER) analysis showed that natalizu-mab dominates fingolimod, as it is less costly (a differenceof $12,287) and more effective (0.15 more relapsesavoided) over 2 years (Figure 1).
The results of the one-way sensitivity analysis areshown in Figure 2. The model inputs with the greatestinfluence on the ICER for natalizumab and fingolimodwere fingolimod relative reduction in relapse rate, fingoli-mod 2-year drug acquisition costs, natalizumab relative
Table 3. Inputs for one-way sensitivity analysis.
Input
Parameter Base Low High
Natalizumab annualized relapse rate (placebo) 0.73 0.62 0.87Fingolimod annualized relapse rate (placebo) 0.40 0.34 0.47Natalizumab relative reduction in relapse rate 68.1% 60.5% 75.7%Fingolimod relative reduction in relapse rate 54.0% 43.4% 64.6%Natalizumab 2-year drug acquisition costs $79,997 $71,979 $87,975Fingolimod 2-year drug acquisition costs $95,902 $86,312 $105,492Natalizumab 2-year drug administration costs $3720 $3348 $4092Cost of relapse requiring low-intensity management $329 $305 $354Cost of relapse requiring medium-intensity management $2500 $2086 $2914Cost of relapse requiring high-intensity management $17,420 $17,357 $17,483
Journal of Medical Economics Volume 14, Number 5 October 2011
622 Cost-effectiveness of natalizumab vs fingolimod O’Day et al. www.informahealthcare.com/JME ! 2011 Informa UK Ltd
Jour
nal o
f M
edic
al E
cono
mic
s D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y U
nive
rsity
of
Lim
eric
k on
05/
11/1
3Fo
r pe
rson
al u
se o
nly.
reduction in relapse rate, and natalizumab 2-year drugacquisition costs. The cost of relapses requiring low-,medium-, and high-intensity management had a minorimpact on the results, changing the ICER by less than$200. Resource utilization for patient monitoring had anegligible impact on the results, and, therefore, theseresults are not shown. None of the parameters over therange of tested values changed the results of the analysis,which found that natalizumab was more effective and lesscostly than fingolimod.
The probabilistic sensitivity analysis resulted in a meancost for natalizumab of $86,595 (95% CI: $77,790,$94,725) and a mean cost for fingolimod of $98,696(95% CI: $89,456, $108,521). Effectiveness was alsorobust, with the probabilistic analysis yielding 0.74relapses avoided per patient for natalizumab (95% CI:0.62, 0.87) and 0.59 relapses avoided per patient for fingo-limod (95% CI: 0.46, 0.74). Additional results from theprobabilistic sensitivity analysis found that in 97.5% ofsimulations natalizumab was less costly, in 97.7% of simu-lations it was more effective, and in 95.4% of simulations itwas dominant compared to fingolimod. Furthermore, nata-lizumab was cost-effective 95.1% of the time, at a WTPthreshold of $0 per relapse avoided, increasing to 96.3% ofthe time at a WTP threshold of $50,000 per relapseavoided.
Discussion
This model estimated the incremental cost per relapseavoided of natalizumab and fingolimod from a UnitedStates managed care payer perspective. Results of themodel demonstrated that natalizumab dominated fingoli-mod, as it was less costly and more cost-effective over a2-year time horizon. The difference in the total 2-year costof therapy per patient was primarily attributable to drugcosts; $79,977 for natalizumab vs $95,902 for fingolimod.The difference in drug costs outweighed the higher admin-istration and monitoring costs associated with natalizu-mab. Furthermore, natalizumab was more effective with0.74 relapses avoided per patient compared to 0.59 relapsesavoided per patient for fingolimod. Sensitivity analysesdemonstrated the model was robust, as none of the param-eters across the range of values tested in the one-way sen-sitivity analysis changed the results of the model. Theprobabilistic sensitivity analysis further demonstrated therobustness of the model based on the mean cost and effec-tiveness values calculated. In the case of a more costly andmore effective therapy, cost-effectiveness analyses must beevaluated with reference to a decision maker’s WTPthreshold for the outcome in question. However, the factthat natalizumab was found to dominate fingolimod in thisanalysis obviates the need to identify a specific WTPthreshold per relapse avoided.
Figure 1. Cost-effectiveness plane.
Journal of Medical Economics Volume 14, Number 5 October 2011
! 2011 Informa UK Ltd www.informahealthcare.com/JME Cost-effectiveness of natalizumab vs fingolimod O’Day et al. 623
Jour
nal o
f M
edic
al E
cono
mic
s D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y U
nive
rsity
of
Lim
eric
k on
05/
11/1
3Fo
r pe
rson
al u
se o
nly.
Although other models examining the cost-effective-ness of MS treatments have been developed, to the best ofour knowledge, our model is the first to compare the onlyapproved oral agent and an injectable agent for the treat-ment of relapsing MS. Given differences in comparators,methodological design, and model assumptions, it may bedifficult to compare our model to other published modelsexamining the cost-effectiveness of agents used for thetreatment of MS. Goldberg et al. evaluated first-line inject-able DMTs used for the treatment of relapsing MS in termsof cost per relapse avoided over a 2-year time horizon,using methodology similar to our model18. Other studieshave evaluated natalizumab against other injectableDMTs. Gani et al. examined the cost-effectiveness of nata-lizumab as compared to the other injectable DMTs withregard to estimating transitions between disability statesand the probability of relapse within each state20. UsingQALYs as the outcome measurement, they determinedthat natalizumab may be a cost-effective therapy forpatients with MS in the United Kingdom20. Natalizumabis the only MS treatment approved by the UnitedKingdom National Institutes of Clinical Excellence forthe management of highly active RRMS36. Chiao andMeyer found natalizumab to be the most cost-effectivetherapy with regard to relapses avoided when modeled
against other available injectable DMTs, includingIFN�-1a 30 mg intramuscular once weekly, IFN�-1b0.25 mg subcutaneous (SC) every other day, glatirameracetate 20 mg SC once daily, and IFN�-1a 44 mg SCthree times weekly, from the perspective of a UnitedStates managed care payer21. In their study, the cost perrelapse avoided was $56,594 for natalizumab and rangedfrom $87,791 to $103,665 for the other DMTs21.
MS is a chronic, disabling, costly disease that is oftendiagnosed in the mid-years of life. As the life expectancy ofpatients with MS is not typically reduced, the impact ofMS is often felt over several years of the patient’s life. As aresult, healthcare decision makers in the United Statesmay incur costs associated with these patients over severaldecades. Given that healthcare resources are constrained,these healthcare decision makers are often looking foropportunities to manage certain therapeutic categories,including MS, in an attempt to control costs.
In addition, as more agents become available for thetreatment of MS, demonstrating clinical efficacy whilecontrolling costs will be important to United Stateshealthcare decision makers. With the new oral agent fin-golimod entering the treatment landscape for MS, ques-tions will arise as to the potential benefits an oraltherapy may bring as compared to the physician- and
Figure 2. One-way sensitivity analysis results.
Journal of Medical Economics Volume 14, Number 5 October 2011
624 Cost-effectiveness of natalizumab vs fingolimod O’Day et al. www.informahealthcare.com/JME ! 2011 Informa UK Ltd
Jour
nal o
f M
edic
al E
cono
mic
s D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y U
nive
rsity
of
Lim
eric
k on
05/
11/1
3Fo
r pe
rson
al u
se o
nly.
self-administered injectable therapies that are alreadyapproved to treat MS. Adherence to therapy for MS hastraditionally been considered poor, and there is discussionamong healthcare professionals as to how the introductionof an oral agent may change this dynamic37,38. Adherencerates for injectable agents used for the treatment of MShave been reported to range from 50% to 80%37. Reasonscited for nonadherence to injectable MS therapies includea lack of treatment efficacy (perceived and real), adverseevents from therapy, costs, and injection anxiety38. Debatesurrounds the claim that injection anxiety contributes tononadherence. A study by Treadway et al. found thatinjection anxiety was not a mitigating factor for nonad-herence. Reasons cited in the study for nonadherenceincluded ‘forgot to take,’ ‘didn’t feel like it,’ and ‘tired ofshots’39. ‘Forgot to take’ is of particular concern forpatients with MS, as cognitive impairment is often a man-ifestation of the disease40. Natalizumab has the advantageof being administered as an IV infusion every 4 weeksunder the supervision of a healthcare professional, thuseliminating the need for patients to remember to take anoral agent on a daily basis. Patients with MS receivingnatalizumab may have improved adherence, because ifthey do not show up to receive their natalizumab infusion,the healthcare professional is aware and may potentiallyfollow up with patients to remind them that they are duefor their next natalizumab infusion. The ability of health-care professionals to monitor whether patients have takentheir daily oral medication, and to intervene if they havenot, is limited. Until studies pertaining to adherence withthe oral MS agent fingolimod can be conducted, this pointremains unclear. However, under the assumption thatnatalizumab has a higher adherence rate than fingolimod,our model may be considered to have used a conservativeapproach by assuming 100% adherence for both natalizu-mab and fingolimod.
In the absence of head-to-head comparative databetween natalizumab and fingolimod, a decision analysissuch as this provides insight into a comparison of theefficacy and costs associated with each of the therapies.This insight may be used to guide healthcare decisionmakers with regard to evaluating the benefits of onetherapy vs the other. The results of this model demon-strate that natalizumab is a less costly, more effectivetherapy than fingolimod for patients with relapsing MSand is still a viable therapeutic option for patients withrelapsing MS.
Limitations
The first limitation of the model was the absence of directhead-to-head studies comparing natalizumab and fingoli-mod. Relapse rates used in the model were derived fromthe respective clinical trials for natalizumab and
fingolimod. Therefore, differences in study designs, studyprocedures, or baseline characteristics may have had animpact on these rates, and results should be interpretedwith this in mind. However, the analysis attempted tocontrol for these differences through the use of placebo-adjusted rates. In addition, the definition of relapse inFREEDOMS includes an increase in EDSS score.However, relapses do not significantly impact the progres-sion of sustained disability41,42 and, thus, the relapse rate inFREEDOMS may be underreported. Our model thereforeused a conservative estimate by assuming that the relapserates in AFFIRM and FREEDOMS were commensurable.Due to the lack of head-to-head data for natalizumab andfingolimod, use of the relative reduction in relapse rate wasthought to be most appropriate and has been used in othermodels, such as the model by Goldberg et al.18.Furthermore, extensive sensitivity analyses were con-ducted to demonstrate the robustness of the model tochanges in inputs, such as the relative reduction inrelapse rate.
The second limitation was the use of relapse rate reduc-tion as the primary outcome of the model, rather thansustained disability progression. Disability progressionvaries among patients, and operational definitions for dis-ability progression were not consistent between natalizu-mab and fingolimod clinical trials29,30. The economicimpact of an MS relapse does not fully represent theburden of this chronic, debilitating disease that results indisability progression over time. Prevention of long-termdisability progression has been established as the mostimportant therapeutic goal by the American Academy ofNeurology43. Although long-term measures of cost-effectiveness may be appropriate for chronic conditionssuch as MS, the authors felt that the cost per relapseavoided may be most relevant to a United States health-care system because it is a short-term measure of cost-effectiveness and, although this may not be as commonin members with chronic conditions, the mean length ofcontinuous health plan enrollment is approximately2 years28.
The third limitation of the model is that it assumespatients are 100% adherent to treatment, and relapserates are calculated as such. Adherence data for natalizu-mab and fingolimod were not available at the time thismodel was developed. In theory, natalizumab may have ahigher adherence rate than other therapies as a result ofthe administration requirements that mandate the once-monthly IV infusion under the supervision of a healthcareprofessional. The administration of natalizumab by ahealthcare professional may increase the likelihood thatpatients receive treatment compared to a self-administeredoral agent that must be taken once daily. However, giventhe lack of available adherence data, the model assumed100% adherence for both agents.
Journal of Medical Economics Volume 14, Number 5 October 2011
! 2011 Informa UK Ltd www.informahealthcare.com/JME Cost-effectiveness of natalizumab vs fingolimod O’Day et al. 625
Jour
nal o
f M
edic
al E
cono
mic
s D
ownl
oade
d fr
om in
form
ahea
lthca
re.c
om b
y U
nive
rsity
of
Lim
eric
k on
05/
11/1
3Fo
r pe
rson
al u
se o
nly.
The fourth limitation, as described previously in theModel Design section, was that the costs of adverseevents were not captured in the model. PML has occurredin patients treated with natalizumab; however, the costsassociated with managing PML would be expected to haveminimal impact on costs to a health plan due to the rareoccurrence of this serious adverse event. In addition,although fingolimod is associated with adverse eventssuch as reduction in heart rate, atrioventricular block,infection, macular edema, and a decrease in pulmonaryfunction tests, until it has been on the market for alonger period of time and phase 4 data are collected, it isunclear how these potential adverse events may affecthealth plan costs or if any other adverse effects will becharacterized.
Conclusions
The results of this model suggest that natalizumab is lesscostly and more effective compared to the oral agent fin-golimod for the treatment of patients with relapsing MS.In the absence of head-to-head comparisons betweenagents, economic models may be useful tools for UnitedStates payers facing the challenges of offering the best carefor patients under limited financial resources.
TransparencyDeclaration of fundingFunding source: Biogen Idec Inc., Weston, MA, USA.
Declaration of financial/other relationshipsS.A. is employed by Biogen Idec Inc. R.M.M. was a paid consul-tant to Biogen Idec for the purposes of this project. K.M., K.O’D.,and M.F. are employed by Xcenda, a consulting company whichprovides services to Biogen Idec and other pharmaceuticalmanufacturers.
AcknowledgementsKylie Matthews, an employee of Xcenda, provided copyeditingsupport and formatted the manuscript for submission to theJournal of Medical Economics.
Previously presented at the Academy of Managed CarePharmacy 23rd Annual Meeting & Showcase, April 27–29,2011, Minneapolis, MN, USA.
References1. National Multiple Sclerosis Society. Who gets MS? Available at: http://www.
nationalmssociety.org/index.aspx [Last accessed 10 December 2010]
