Cost per response analysis of strategies for chronic immune thrombocytopenia

Abstract

Objectives:
This analysis estimated the cost per response and the incremental cost per additional responder of romplostim, eltrombopag, and the "watch-and-rescue" (monitoring until rescue therapies are required) strategy in adults with chronic immune thrombocytopenia (ITP).

Study design:
The decision tree is designed to estimate the total cost per response for romiplostim, eltrombopag, and watch and rescue over a 24-week time horizon; cost-effectiveness was evaluated in terms of incremental cost per additional responder.

Methods:
Model inputs including response rates, bleeding-related episode (BRE) rates, and costs were estimated from registrational trial data, an independent Bayesian indirect comparison, database analyses, and peer-reviewed publications. Costs were applied to the proportions of patients with treatment response and nonresponse (based on platelet count). The total cost per response and the incremental cost per additional responder for each treatment were calculated. Sensitivity analyses and alternative analyses were performed.

Results:
With higher total costs and greater treatment efficacy, romiplostim and eltrombopag had a lower 24-week cost per response and a lower average number of BREs than watch and rescue. Eltrombopag was weakly dominated by romiplostim. The incremental cost-effectiveness ratio of romiplostim versus watch and rescue was $46,000 per additional responder. The model results are most sensitive to response rates of romiplostim and watch and rescue and the BRE rate for splenectomized nonresponders. Alternative analyses results were similar to the base case.

Conclusions:
In adults with chronic ITP, romiplostim represents an efficient way to achieve response, with lower costs per response than eltrombopag; both romiplostim and eltrombopag had lower costs per response than watch and rescue.

Full text

SP294 JULY 2018 www.ajmc.com
ORIGINAL
RESEARCH
C
hronic immune thrombocytopenia (ITP) is an autoimmune
disorder characterized by a low platelet count and increased
risk of bleeding. Two thrombopoietin-receptor agonists
(TPO-RAs), romiplostim (once-weekly subcutaneous injection)
and eltrombopag (once-daily oral agent), are indicated for the
treatment of adults with chronic ITP who have had an insucient
response to corticosteroids, immunoglobulins, or splenectomy.
-
In clinical practice, patients are sometimes monitored until rescue
therapies, like intravenous (IV) immunoglobulin, are required,
commonly referred to as the “watch-and-rescue” strategy.
Although some patients undergo splenectomy to treat their
ITP, nonsplenectomized patients account for the majority of adult
patients with ITP seen by clinical practices in the United States.
The primary goal of ITP therapy is to help achieve a platelet count
that prevents major bleeding. Both of the available TPO-RAs have
been shown to increase and maintain platelet counts
,
and reduce
the incidence of bleeding-related episodes (BREs). A BRE is dened
as the occurrence of a bleeding event and/or use of rescue therapy,
including intravenous immunoglobulin (IVIg), anti-D, corticoste-
roids, platelet transfusions, and dosage increases.
,
There is limited
evidence related to the economics of TPO-RA therapy currently
available in published literature.
,
This analysis was designed
to evaluate the cost-eectiveness (in terms of incremental cost
per additional responder) and cost per treatment response of the
 TPO-RAs and the watch-and-rescue strategy for treating adults
with chronic ITP in the United States.
METHODS
Overview and Model Structure
The target patient population consists of both splenectomized
() and nonsplenectomized () adults with chronic ITP. Model
comparators included romiplostim, eltrombopag, and watch and
rescue. The model was developed in Microsoft Excel  using Visual
Basic for Applications (Microsoft Corp; Redmond, Washington).
Cost Per Response Analysis of Strategies for
Chronic Immune Thrombocytopenia
Kelly Fust, MS; Anju Parthan, PhD; Xiaoyan Li, PhD; Anjali Sharma, MD; Xinke Zhang, MS; Marco Campioni, PhD;
Junji Lin, PhD, MS; Xuena Wang, PhD; Richard Zur, PhD; Karynsa Cetin, MPH; Melissa Eisen, MD; and David Chandler, PhD
ABSTRACT
OBJECTIVES: This analysis estimated the cost per response
and the incremental cost per additional responder of
romplostim, eltrombopag, and the “watch-and-rescue”
(monitoring until rescue therapies are required) strategy in
adults with chronic immune thrombocytopenia (ITP).
STUDY DESIGN: The decision tree is designed to estimate
the total cost per response for romiplostim, eltrombopag,
and watch and rescue over a 24-week time horizon;
cost-effectiveness was evaluated in terms of incremental
cost per additional responder.
METHODS: Model inputs including response rates,
bleeding-related episode (BRE) rates, and costs were
estimated from registrational trial data, an independent
Bayesian indirect comparison, database analyses, and
peer-reviewed publications. Costs were applied to the
proportions of patients with treatment response and
nonresponse (based on platelet count). The total cost per
response and the incremental cost per additional responder
for each treatment were calculated. Sensitivity analyses and
alternative analyses were performed.
RESULTS: With higher total costs and greater treatment
efficacy, romiplostim and eltrombopag had a lower 24-week
cost per response and a lower average number of BREs
than watch and rescue. Eltrombopag was weakly dominated
by romiplostim. The incremental cost-effectiveness ratio
of romiplostim versus watch and rescue was $46,000 per
additional responder. The model results are most sensitive
to response rates of romiplostim and watch and rescue and
the BRE rate for splenectomized nonresponders. Alternative
analyses results were similar to the base case.
CONCLUSIONS: In adults with chronic ITP, romiplostim
represents an efficient way to achieve response, with lower
costs per response than eltrombopag; both romiplostim
and eltrombopag had lower costs per response than watch
and rescue.
Am J Manag Care. 2018;24(Spec Issue No. 8):SP294-SP302

THE AMERICAN JOURNAL OF MANAGED CARE® VOL. 24, SPECIAL ISSUE NO. 8 SP295
Cost Per Response Analysis for Chronic Immune Thrombocytopenia
The model begins with the decision to treat
patients with ITP with either romiplostim or
eltrombopag or to adopt the watch and rescue
strategy. The analysis was based on a decision
tree that stratied patients into response or no
response, followed by the presence or absence
of a BRE (Figure 1). Costs were applied to each
group of patients in the decision tree. The
patients were followed over a -week time
horizon, consistent with the trial durations
for romiplostim and eltrombopag., For each
strategy, the average number of BREs, BRE costs, percentage of
patients who responded, and total costs, including drug, physician,
and lab test costs, were estimated. The total cost per response for
each treatment was calculated. Cost-eectiveness was evaluated
in terms of incremental cost per additional responder from the
US payer perspective.
Treatment Response Rates
Overall platelet response was dened in the romiplostim trials as the
percentage of patients with a platelet count   /L for at least
 weeks during the trial, excluding responses within  weeks after
use of rescue medications., Overall platelet response was dened
in the eltrombopag trial as the percentage of patients: ()with a
platelet count of - × /L for at least  consecutive weeks
during treatment, including all data up to time of withdrawal for
patients who prematurely withdrew, excluding responses during
rescue treatment and up to the time platelet counts fell below
 × 

/L after cessation of rescue treatment; or () with a platelet
count of - × /L for at least  of the last  weeks of treatment,
excluding premature withdrawals and patients using rescue therapy
at any time on treatment.
-
In the model, treatment response was
dened by overall platelet response based on the number of weeks
with a platelet count  × 

/L. The response rates for romiplostim
were estimated using trial data.
,
The International Society for
Pharmacoeconomics and Outcomes Research (ISPOR) Task Force
on Indirect Treatment Comparisons Good Research Practices
report suggests that data from head-to-head trials are preferred
in economic evaluations of active comparators; in the absence of
these data, evidence from an indirect treatment comparison may
be considered.

The results from an independent Bayesian indirect
comparison analysis suggested that the overall response rate with
romiplostim was signicantly higher than with eltrombopag
(odds ratio [OR], .).

Accordingly, the eltrombopag response
rates (. for nonsplenectomized and . for splenectomized
patients) were estimated using the romiplostim response rates
(. for nonsplenectomized and . for splenectomized
patients) and the OR of . estimated from Cooper et al.

The
watch-and-rescue response rates for nonsplenectomized and
splenectomized patients of . and ., respectively, were
estimated from pooled placebo response rates., Response rates
are presented in Table 1.-,,,-
BRE Rates
A BRE was dened as a discrete and identiable event of bleeding
and/or the use of rescue therapy occurring within close proximity
of one another ( days). A composite end point, such as a BRE,
tends to be more clinically relevant because the bleeding events in
phase  trials are likely to be confounded by increased use of rescue
medication in the placebo arms.

According to Weitz et al, applying
the BRE method to the romiplostim trial shows that treatment was
associated with a reduction in the rate of unique clinical episodes
related to bleeding compared with placebo. In the model, BREs were
estimated from a post hoc analysis of phase  placebo-controlled
studies of romiplostim in patients with chronic ITP and were
calculated by pooling the placebo and romiplostim data.

BREs
were assumed to depend on response and splenectomy status only.
TAKEAWAY POINTS
›
Limited evidence evaluating the economic efficiency of thrombopoietin-receptor agonist
(TPO-RA) therapy in the United States is currently available in published literature.
›
Results of this analysis provide information on the efficiency (cost per response) and
cost-effectiveness (incremental cost per additional responder) of the 2 available TPO-RAs
(romiplostim and eltrombopag) and of “watch and rescue” in adults with chronic immune
thrombocytopenia in the United States.
›
Romiplostim represents an efficient way to achieve response, with lower costs per response
than eltrombopag and watch and rescue.
FIGURE 1. Cost Consequence Analysis Decision Tree
Structure
BRE indicates bleeding-related episode.
1 column
BRE
Response
No BRE
Romiplostim
BRE
No Response
No BRE
BRE
Response
No BRE
Eltrombopag
BRE
No Response
No BRE
BRE
Response
No BRE
Watch and Rescue
BRE
No Response
No BRE
Patients With Immune
Thrombocytopenia

SP296 JULY 2018 www.ajmc.com
ORIGINAL RESEARCH
Because there were no published BRE data for eltrombopag, BRE
rates were assumed to be the same as those for romiplostim and
watch and rescue (Table ).
Costs
Wholesale acquisition costs of eltrombopag (tablet strengths, .mg,
mg, mg, and mg) and romiplostim were obtained from the
EncoderPro database.

Although the romiplostim prescribing infor-
mation indicates that the maximum weekly dose of romiplostim is
 mcg/kg per week,

in clinical trials of romiplostim, the maximum
allowed dose was  mcg/kg per week.

Accordingly, in the base-case
analysis, the maximum weekly dose of romiplostim was allowed to
exceed  mcg/kg and an alternative analysis limiting the maximum
weekly dose to  mcg/kg was also performed. It was assumed that
patients in the watch and rescue treatment arm do not incur drug
acquisition costs. Drug acquisition costs and dosing parameters for
romiplostim and eltrombopag are presented in Table .
In the real-world setting, patients are on dierent tablet strengths
of eltrombopag. Therefore, the proportions of patients utilizing the
various eltrombopag tablet strengths were estimated from published
TABLE 1. Model Parametersa
Base-Case Estimate DSA Range (95% CI) Reference
Eltrombopag vs romiplostim response OR 0.15 Did not vary Cooper et al (2014)11
Response rates
Nonsplenectomized
Romiplostim 87.8% 73.0%-95.4% Kuter et al (2008)3
Eltrombopag 51.9% 28.8%-75.8%bEstimated based on data from Kuter
et al 20083 and Cooper et al 201411
Watch and rescue 14.52% 7.3%-26.3% Kuter et al (2008)3; Cheng et al (2011)4
Splenectomized
Romiplostim 78.6% 62.8%-89.2% Kuter et al (2008)3
Eltrombopag 35.5% 20.2%-55.2%bEstimated based on data from Kuter
et al (2008)3 and Cooper et al (2014)11
Watch and rescue 4.76% 0.8%-17.4% Kuter et al (2008)3; Cheng et al (2011)4
Weekly BRE rates
Nonsplenectomized
Nonresponder 0.128 0.104-0.158 Weitz et al (2012)8
Responder 0.031 0.020-0.047 Weitz et al (2012)8
Splenectomized
Nonresponder 0.151 0.126-0.179 Weitz et al (2012)8
Responder 0.039 0.026-0.059 Weitz et al (2012)8
Wholesale acquisition costs
Romiplostim $5.826/mcg Not included EncoderPro Database15 (July 2015)
Eltrombopag (12.5 and 25 mg) $4.082/mg Not included EncoderPro Database15 (July 2015)
Eltrombopag (50 and 75 mg) $3.988/mg Not included EncoderPro Database15 (July 2015)
Eltrombopag (weighted average of all
strengths/unit costs) $4.008/mg Not included EncoderPro Database15 (July 2015)
Dosing parameters
Romiplostim (without top-coding) 317 mcg/week Not included Kuter et al (2008)3
Eltrombopag 54.875 mg/day Not included Cheng et al (2011)4
Physician and lab test costs
Administration of romiplostim visit
(injection) $25.51/visit Not included CMS17,c
Physician visit $73.30/visit Not included CMS17,c
Platelet count test $6.09/test Not included CMS (Laboratory)18,c
Hepatic function panel $11.11/test Not included CMS (Laboratory)18,c
(continued)

THE AMERICAN JOURNAL OF MANAGED CARE® VOL. 24, SPECIAL ISSUE NO. 8 SP297
Cost Per Response Analysis for Chronic Immune Thrombocytopenia
literature (. on  mg, . on  mg, and . on  mg).
The average cost of eltrombopag (. per mg) was estimated by
calculating a weighted average of the unit costs and proportion of
patients on each tablet strength. Patients on romiplostim incurred
a weekly drug acquisition cost, whereas patients on eltrombopag
incurred a daily drug acquisition cost. Both responders and
nonresponders were assumed to receive treatment for the entire
model horizon and accordingly incurred  weeks of drug costs.
The average treatment cost per BRE (Table ) was estimated
from a retrospective study of a large US administrative healthcare
claims database that was sponsored by Amgen. Adult patients
with newly diagnosed ITP were identied between the years 
and  by having at least  outpatient claims separated by at
least  days or  inpatient claim with International Classication
of Diseases, Ninth Revision, Clinical Modication (ICD-9-CM) diag-
nosis code . for primary ITP. A BRE was dened as  actual
bleeding event and/or use of rescue therapy (IV immunoglobulin
and anti-D; IV steroids; and/or platelet transfusion). In the study,
Lin et al did not consider an increase in dose or frequency of a
concurrent ITP medication as rescue therapy. Average BRE costs
for both nonsplenectomized and splenectomized patients were
estimated; however, due to the high variability of the estimates,
the dierence in BRE costs between the  patient groups was not
statistically signicant. Therefore, the average total cost among
both splenectomized and nonsplenectomized patients was used
in the base-case analysis.
The costs and frequency of physician oce visits for adminis-
tration of romiplostim and monitoring patients with chronic ITP,
platelet count tests for patient monitoring, and hepatic function
panels for patients on eltrombopag are also presented in Table.,,,
Thetotal costs of physician oce visits and lab tests were calculated
by multiplying the frequency of testing by the time horizon and
the cost of individual visits. All cost estimates are presented in
 US dollars.
Model Analyses
In the model, the total costs at  weeks, proportion of patients with
response, and average number of BREs were calculated for each
comparator. The -week cost per response for each comparator was
calculated by dividing the total cost at  weeks by the proportion of
patients with response. The cost-eectiveness of the  TPO-RAs and
the watch and rescue strategy for treating adults with chronic ITP
in the United States was evaluated in terms of incremental cost per
additional responder. An alternative analysis was also performed
using incremental cost per BRE avoided as the outcome of interest.
When conducting cost-eectiveness analyses (CEAs), if a strategy
is both more costly and less eective compared with an alternative
strategy, then it is said to be dominated by the alternative strategy and
no incremental cost-eectiveness ratio (ICER) is calculated., If a
more costly strategy provides additional benet, then the  strategies
are compared by dividing the additional cost (ie, incremental cost)
by the additional benet (ie, incremental eectiveness)., Weak
dominance (also called extended dominance) occurs when the
ICER for a strategy is greater (ie, the strategy is less cost-eective)
than that of a more costly alternative.- Strategies that are weakly
dominated are excluded, and then ICERs of the remaining strategies
are recalculated., Given the -week time horizon of the model,
costs and outcomes were not discounted.
Deterministic, or -way, sensitivity analyses (DSAs) were performed
to assess how changes in key model parameters, and parameter
TABLE 1. (Continued) Model Parametersa
Base-Case Estimate DSA Range (95% CI) Reference
Frequency of physician visits
Romiplostim administration visits 1 per week Not included Assumption
Physician visits (for romiplostim,
eltrombopag, and watch and rescue)
1 per week in weeks 1-4;
1 per 4 weeks in weeks 5-24 Not included Romiplostim and eltrombopag
prescribing information1,2
Frequency of clinical tests
Platelet count test (for romiplostim,
eltrombopag, and watch and rescue)
1 per week in weeks 1-4;
1 per 4 weeks in weeks 5-24 Not included Romiplostim and eltrombopag
prescribing information1,2
Hepatic function panel (for
eltrombopag)
1 per 2 weeks in weeks 1-4;
1 per 4 weeks in weeks 5-24 Not included Assumption
BRE cost (per event)
Average total cost $6022 $5421-$6623 Lin et al (2017)16
BRE indicates bleeding-related episode; CPT, Current Procedural Terminology; DSA, deterministic sensitivity analysis; OR, odds ratio.
aAll cost estimates are in 2015 US$.
bDSA ranges for eltrombopag response rates were estimated using the upper and lower bounds for the romiplostim response rates3 and applying the eltrombopag
response OR (0.15).11
cThe costs of administration/injection visits and physician office visits were estimated from the July release of the 2015 National Physician Fee Schedule Relative
Value File,17 using CPT codes of 96372 (therapeutic, prophylactic, or diagnostic injection [specify substance or drug]; subcutaneous or intramuscular) and 99213
(office or other outpatient visit for the evaluation and management of an established patient), respectively. The costs of platelet counts and hepatic function panels
were estimated from the 2015 Clinical Diagnostic Laboratory Fee Schedule18 using CPT codes of 85049 (blood count; platelet, automated) and 80076 (hepatic func-
tion panel), respectively.

SP298 JULY 2018 www.ajmc.com
ORIGINAL RESEARCH
uncertainty, impact cost-eectiveness results. In the sensitivity
analyses, parameters were varied using  CIs derived from the
clinical trial data or database analyses (Table ). The model was
analyzed with each parameter varied individually to its corresponding
upper or lower limit, and results were calculated. Results of the DSA
are presented visually in the form of tornado diagrams. The DSA
was performed with incremental cost per additional responder as
the outcome measure. Alternative analyses examining the impact
of diering assumptions related to romiplostim dosing, response
rates, nonsplenectomized patients, and frequency of platelet count
tests and physician visits were also performed.
RESULTS
Base Case
The total -week costs per patient ranged from , for watch and
rescue to , for romiplostim (Table 2). Compared with the watch
and rescue strategy, use of either of the  TPO-RAs was associated with
fewer BREs and thus a lower BRE treatment cost (, for watch
and rescue; , for eltrombopag; and  for romiplostim) and
was associated with a lower cost per response (, for watch
and rescue; , for eltrombopag; and , for romiplostim).
With better treatment ecacy, romiplostim was associated with a
TABLE 2. Base-Case and Alternative Analysis Results: Cost Per Response, by Treatment Strategy
Comparator
Proportion
With Response
Average
Number of BREs
24-Week Costs 24-Week Cost
Per ResponseDrug Visits and Tests BREs Total
Base-case analysis
Watch and rescue 9.5% 3.12 $0 $715 $18,788 $19,503 $204,403
Eltrombopag 43.5% 2.27 $36,949 $792 $13,684 $51,425 $118,113
Romiplostim 83.1% 1.27 $44,321 $1327 $7670 $53,318 $64,165
Alternative analysis 1: maximum dosage of 10 mcg/kg/week for romiplostim and corresponding response rate (80.8%)
Watch and rescue 9.5% 3.12 $0 $715 $18,788 $19,503 $204,403
Eltrombopag 39.7% 2.34 $36,949 $792 $14,112 $51,853 $130,639
Romiplostim 80.9% 1.31 $43,762 $1327 $7915 $53,003 $65,541
Alternative analysis 2: response rates for eltrombopag (60.0% for splenectomized and 71.8% for nonsplenectomized) estimated using
registrational trial data
Watch and rescue 9.5% 3.12 $0 $715 $18,788 $19,503 $204,403
Eltrombopag 65.8% 1.71 $36,949 $792 $10,302 $48,043 $73,053
Romiplostim 83.1% 1.27 $44,321 $1327 $7670 $53,318 $64,165
Alternative analysis 3: patient population limited to nonsplenectomized patients only
Watch and rescue 14.5% 2.74 $0 $715 $16,507 $17,222 $118,639
Eltrombopag 51.9% 1.87 $36,949 $792 $11,236 $48,978 $94,327
Romiplostim 87.8% 1.03 $44,321 $1327 $6180 $51,828 $59,027
Alternative analysis 4: platelet tests and physician visits are weekly during weeks 5 to 24 for watch and rescue, eltrombopag, and
romiplostim
Watch and rescue 9.5% 3.12 $0 $1905 $18,788 $20,694 $216,884
Eltrombopag 43.5% 2.27 $36,949 $1983 $13,684 $52,616 $120,848
Romiplostim 83.1% 1.27 $44,321 $2518 $7670 $54,509 $65,598
Alternative analysis 5: watch and rescue has zero physician visits and zero platelet count tests
Watch and rescue 9.5% 3.12 $0 $0 $18,788 $18,788 $196,914
Eltrombopag 43.5% 2.27 $36,949 $792 $13,684 $51,425 $118,113
Romiplostim 83.1% 1.27 $44,321 $1327 $7670 $53,318 $64,165
Alternative analysis 6: response rates for eltrombopag (62.1% for splenectomized and 56.8% for nonsplenectomized patients)
estimated from Bussel et al (2009)22
Watch and rescue 9.5% 3.12 $0 $715 $18,788 $19,503 $204,403
Eltrombopag 59.5% 1.85 $36,949 $792 $11,163 $48,904 $82,188
Romiplostim 83.1% 1.27 $44,321 $1327 $7670 $53,318 $64,165
BRE indicates bleeding-related episode.

THE AMERICAN JOURNAL OF MANAGED CARE® VOL. 24, SPECIAL ISSUE NO. 8 SP299
Cost Per Response Analysis for Chronic Immune Thrombocytopenia
lower cost per response than eltrombopag. The incremental cost
per additional responder is presented in Table3.

Eltrombopag was
weakly dominated by romiplostim, and the ICER for romiplostim
versus watch and rescue was , per additional responder.
Sensitivity Analyses
Given that romiplostim and watch and rescue are the  strate
-
gies on the cost-eectiveness frontier, the DSA was performed
comparing romiplostim with watch and rescue only. Results of
the DSA indicated that model results were most sensitive to the
response rate of patients on romiplostim, the response rate of
patients on watch and rescue, and the BRE rate for splenectomized
nonresponders (Figure 2). Varying the response rate for patients on
romiplostim to the lower and upper bounds of the  CI yielded
ICERs of , and ,, respectively. Varying the response
rate of patients on watch and rescue to the lower and upper bounds
of the  CI yielded ICERs of , and ,, respectively.
Varying the BRE rate for splenectomized nonresponders to the
lower and upper bounds of the  CI yielded ICERs of ,
and ,, respectively.
TABLE 3. Base-Case and Alternative Analysis Results: Incremental Cost Per Additional Respondera
Comparator
Total
Costs
Proportion With
Response
Incremental
Costs
Incremental Proportion
With Response ICER
Base-case analysis
Watch and rescue $19,503 9.5% Reference Reference Reference
Eltrombopag $51,425 43.5% $31,922 34.0% Weakly dominatedb
Romiplostim $53,318 83.1% $33,815 73.6% $45,973
Alternative analysis 1: maximum dosage of 10 mcg/kg/week for romiplostim and corresponding response rate (80.8%)
Watch and rescue $19,503 9.5% Reference Reference Reference
Eltrombopag $51,853 39.7% $32,350 30.2% Weakly dominatedb
Romiplostim $53,003 80.9% $33,500 71.3% $46,966
Alternative analysis 2: response rates for eltrombopag (60.0% for splenectomized and 71.8% for nonsplenectomized) estimated using
registrational trial data
Watch and rescue $19,503 9.5% Reference Reference Reference
Eltrombopag $48,043 65.8% $28,540 56.2% Weakly dominatedb
Romiplostim $53,318 83.1% $33,815 73.6% $45,973
Alternative analysis 3: patient population limited to nonsplenectomized patients only
Watch and rescue $17,222 14.5% Reference Reference Reference
Eltrombopag $48,978 51.9% $31,756 37.4% Weakly dominatedb
Romiplostim $51,828 87.8% $34,607 73.3% $47,219
Alternative analysis 4: platelet tests and physician visits are weekly during weeks 5 to 24 for watch and rescue, eltrombopag,
andromiplostim
Watch and rescue $20,694 9.5% Reference Reference Reference
Eltrombopag $52,616 43.5% $31,922 34.0% Weakly dominatedb
Romiplostim $54,509 83.1% $33,815 73.6% $45,973
Alternative analysis 5: watch and rescue has zero physician visits and zero platelet count tests
Watch and rescue $18,788 9.5% Reference Reference Reference
Eltrombopag $51,425 43.5% $32,637 34.0% Weakly dominatedb
Romiplostim $53,318 83.1% $34,530 73.6% $46,945
Alternative analysis 6: response rates for eltrombopag (62.1% for splenectomized and 56.8% for nonsplenectomized patients; 59.0%
for combined splenectomized and nonsplenectomized populations) estimated from Bussel et al (2009)22
Watch and rescue $19,503 9.5% Reference Reference Reference
Eltrombopag $48,904 59.5% $29,401 50.0% Weakly dominatedb
Romiplostim $53,318 83.1% $33,815 73.6% $45,973
ICER indicates incremental cost-effectiveness ratio.
aTime horizon: 24 weeks. All costs presented in 2015 US$.
bWeak dominance occurs when the ICER for a strategy is greater than that of a more costly alternative (also called extended dominance).

SP300 JULY 2018 www.ajmc.com
ORIGINAL RESEARCH
Alternative Analyses
Results of all alternative analyses are presented in Table  (cost
per response), Table  (incremental cost per additional responder),
and Table 4 (incremental cost per BRE avoided). When a maximum
dosage of  mcg/kg/week for romiplostim (ie,with top-coding)
was considered and the corresponding response rate was included
in the analyses, the cost per response for romiplostim increased
from , (base-case) to , and the cost per response
for eltrombopag increased from , (base-case) to ,.
The cost per response for watch and rescue remained unchanged.
The ICER for romiplostim versus watch and rescue was ,
per additional responder; eltrombopag was weakly dominated
by romiplostim.
When the eltrombopag response rates were
estimated using registrational trial data, the
cost per response for eltrombopag decreased
from , (base-case) to ,, and the
cost per response results for the other strate-
gies remained unchanged. As in the base-case
analysis, eltrombopag was weakly dominated
by romiplostim and the ICER for romiplostim
relative to watch and rescue did not change
from the base-case result.
When the patient population was limited
to nonsplenectomized patients only, results
were similar to those of the base-case analysis.
The incremental cost per additional responder
for romiplostim relative to watch and rescue
was ,; eltrombopag continued to be
weakly dominated by romiplostim. Cost per
response estimates were , for watch
and rescue, , for eltrombopag, and
, for romiplostim.
When the frequency with which patients
received platelet count tests and physician visits
was increased to weekly during weeks  to ,
the cost per response for romiplostim increased
from , (base-case) to ,; the cost per
response for eltrombopag increased from ,
(base-case) to ,; and the cost per response
for watch and rescue increased from ,
(base-case) to ,. The incremental cost per
additional responder for romiplostim relative
to watch and rescue remained unchanged from
the base case; eltrombopag continued to be
weakly dominated by romiplostim.
When the watch and rescue patients were not
assigned costs for physician visits and platelet
test counts throughout the -week period,
the cost per response for watch and rescue
decreased from , (base-case) to ,, and the cost per
response results for the other strategies remained unchanged. The
incremental cost per additional responder for romiplostim relative
to watch and rescue increased from , (base-case) to ,,
and eltrombopag continued to be weakly dominated by romiplostim.
When the eltrombopag response rates were estimated from Bussel
et al,

eltrombopag was weakly dominated by romiplostim and the
incremental cost per additional responder for romiplostim relative
to watch and rescue did not change from the base-case result. The
cost per response for eltrombopag decreased from , (base
case) to ,, and the cost-per-response results for the other
strategies remained unchanged. Additionally, when the eltrom-
bopag response rates were varied according to the lower and upper
TABLE 4. Alternative Analysis Results: Incremental Cost Per BRE Avoideda
Comparator
Total
Costs
Average
Number
of BREs
Incremental
Costs
Incremental
BREs
Avoided ICER
Watch and rescue $19,503 3.12 Reference Reference Reference
Eltrombopag $51,425 2.27 $31,922 0.85 Weakly
dominatedb
Romiplostim $53,318 1.27 $33,815 1.85 $18,278
BRE indicates bleeding-related episode; ICER, incremental cost-effectiveness ratio.
aTime horizon: 24 weeks. All costs presented in 2015 US$.
bWeak dominance occurs when the ICER for a strategy is greater than that of a more costly alternative
(also called extended dominance).
FIGURE 2. Results of Deterministic Sensitivity Analysis (incremental cost per
additional responder): Romiplostim Versus Watch and Rescuea
BC indicates base-case; BRE, bleeding-related episode; ICER, incremental cost-effectiveness ratio.
aThe vertical axis represents the BC ICER, the horizontal bars represent the difference between the BC
ICER and the ICER generated when the model is run using the high and low values of the plausible range,
and the entire length of each horizontal bar represents the magnitude of variation in cost-effectiveness
results. Bars that touch the vertical axis indicate parameters whose most favorable value results in domi-
nance. Bars that extend off the tornado diagram (to the right) represent parameters whose least favorable
value yields a dominated result.
$47,090
$47,445
$47,482
$47,708
$47,778
$58,152
$62,052
$45,214
$44,968
$44,465
$43,911
$43,898
$41,698
$39,208
$0 $10,000 $20,000 $30,000 $40,000 $50,000 $60,000
$70,000
BRE Rate for Nonsplenectomized
Responders (0.02-0.05); BC: 0.03
BRE Rate for Splenectomized
Responders (0.03-0.06); BC: 0.04
Cost of BREs ($6623-$5421); BC: $6022
BRE Rate for Nonsplenectomized
Nonresponders (0.16-0.1); BC: 0.13
BRE Rate for Splenectomized
Nonresponders (0.18-0.13); BC: 0.15
Combined Response Rate for Patients
on Watch and Rescue (0.04-0.22); BC: 0.1
Combined Response Rate for Patients
on Romiplostim (0.92-0.68); BC: 0.83
Incremental Cost Per Additional Responder

THE AMERICAN JOURNAL OF MANAGED CARE® VOL. 24, SPECIAL ISSUE NO. 8 SP301
Cost Per Response Analysis for Chronic Immune Thrombocytopenia
bounds of the CIs for the base case scenario (. to . for
splenectomized patients; . to . for nonsplenectomized
patients), eltrombopag remained weakly dominated by romiplostim.
Accordingly, the ICER for romiplostim relative to watch and rescue
remained unchanged from the base-case scenario.
Lastly, when cost-eectiveness was assessed in terms of incre-
mental cost per BRE avoided (Table ), the ICER for romiplostim
relative to watch and rescue was ,, and eltrombopag was
weakly dominated by romiplostim.
DISCUSSION
The cost per response and the incremental cost per additional
responder were evaluated for  TPO-RA treatments and a watch and
rescue strategy in both splenectomized and nonsplenectomized
adults with chronic ITP. The use of either TPO-RA resulted in lower
costs per treatment response and fewer BREs than the watch and
rescue strategy. In the base-case analysis, eltrombopag was weakly
dominated by romiplostim and the ICER of romiplostim relative to
watch and rescue was , per additional responder. DSA results
suggest that model results are most sensitive to the response rates
of romiplostim and the watch and rescue strategy, as well as the
BRE rate for splenectomized nonresponders.
Results of alternative analyses examining () a maximum
dosage of  mcg/kg/week for romiplostim (ie, with top-coding)
and corresponding response rate, () eltrombopag response rates
estimated using registrational trial data, () nonsplenectomized
patients only, () additional platelet count tests and physician
visits for patients on all treatments, () zero platelet count tests and
physician visits for patients on the watch and rescue strategy, and
() eltrombopag response rates estimated from Bussel et al yielded
similar results to the base case. An alternative analysis examining
the incremental cost per BRE avoided found that eltrombopag was
weakly dominated by romiplostim and the ICER of romiplostim
relative to watch and rescue was ,.
Limitations
Results of this analysis should be interpreted in light of the following
assumptions and limitations. The ecacy of eltrombopag was
estimated using an OR obtained from an independent Bayesian
indirect comparison performed by Cooper et al,

who noted that the
clinical trials included in the analysis may have diered in terms
of study population and design.

Despite these dierences, Cooper
et al concluded that the romiplostim and eltrombopag clinical
trials included in the indirect comparison were suciently similar.
Nonresponders were assumed to continue treatment for  weeks,
which may overestimate drug costs. BRE rates were assumed to
depend on platelet levels, independent of whether patients were on
active TPO-RA treatment or watch and rescue. Adverse events were
not included in the model due to limited evidence in the literature.
Rituximab was not included in the model due to inconsistent use
in treatment and the identication of literature to determine doses
per patient that prevent bleeding events or predict a response. In
the model, patients on watch and rescue were assumed to incur
zero medication costs; however, in the real-world setting, patients
might be receiving concurrent medication other than the TPO-RAs.
Therefore, the model is likely to underestimate the total costs for
patients on watch and rescue. There are currently no well-established
willingness-to-pay thresholds for the incremental cost per additional
responder in this clinical context; accordingly, it is ultimately up
to the payer to determine whether TPO-RAs are cost-eective in
the treatment of ITP. Finally, patients receiving TPO-RAs were
assumed to be  compliant according to product labels. The
eltrombopag prescribing information states that, due to drug–drug
and drug–food interactions, patients must not take or ingest any
antacids, dairy products, or mineral supplements within  hours
of administration.

Noncompliance with these recommendations
would cause a signicant reduction of eltrombopag bioavailability,
consequently impacting the ecacy of the drug. According to the
results for other drugs with similar drug–drug and drug–food
interactions,, in which noncompliance was about , it is
unlikely that patients will be  compliant; however, data on
eltrombopag compliance are not currently available.
CONCLUSIONS
In adults with chronic ITP, romiplostim represents an ecient way
to achieve response, with lower costs per response than eltrom-
bopag and watch and rescue. Eltrombopag was weakly dominated
by romiplostim, and the ICER for romiplostim versus watch and
rescue was , per additional responder. n
Acknowledgments
The authors would like to acknowledge Mark Bensink of Amgen, Inc, for
his contribution to this manuscript.
Author Aliations: Optum (KF, AP), Boston, MA; Amgen, Inc (XL, AS,
XZ, MC, XW, KC, ME, DC), Thousand Oaks, CA; Pharmerit International (JL),
Bethesda, MA; Creative-Ceutical (RZ), Chicago, IL.
Source of Funding: Funding for this study was provided by Amgen, Inc.
The role of Amgen, Inc, as study sponsor included the provision of data and
review of and comment on the draft and nal manuscript.
Author Disclosures: Ms Fust reports employment with Optum and a paid
consultancy to Amgen, Inc, which commercializes romiplostim,  of the 
TPO-RAs under discussion in this manuscript, at the time of the study. Dr
Parthan reports employment with Optum, which received funding from
Amgen to conduct the study. Dr Li reports employment with and stock
ownership in Amgen during study development. Dr Zhang reports employ-
ment with Amgen. Dr Campioni, Ms Cetin, Dr Eisen, and Dr Chandler report
employment with and stock ownership in Amgen. Dr Lin reports employment
with Amgen during study development. The remaining authors report no
relationship or nancial interest with any entity that would pose a conict
of interest with the subject matter of this article.
Prior Presentation: Results of this analysis for nonsplenectomized
patients only were presented at the th Annual Meeting of the American
Society of Hematology (Orlando, FL; December ). Results including
splenectomized patients have not been presented or published previously.

SP302 JULY 2018 www.ajmc.com
ORIGINAL RESEARCH
Authorship Information: Concept and design (KF, AP, XL, XZ, MC, JL, KC,
ME, DC); acquisition of data (DC); analysis and interpretation of data (KF,
AP, XL, AS, XZ, MC, JL, XW, RZ, KC, ME, DC); drafting of the manuscript (KF,
AP, KC, DC); critical revision of the manuscript for important intellectual
content (AP, XL, AS, XZ, MC, JL, RZ, KC, ME, DC); statistical analysis (XL, XZ,
RZ, ME); obtaining funding (JL); and supervision (AS).
Address Correspondence to: Kelly Fust, MD, Optum,  Boylston St,
Boston, MA . Email: kelly.fust@optum.com.
REFERENCES
1. NPLATE (romiplostim) [prescribing information]. Thousand Oaks, CA: Amgen, Inc; 2016. pi.amgen.com/~/
media/amgen/repositorysites/pi-amgen-com/nplate/nplate_pi_hcp_english.pdf. Accessed May 11, 2018.
2. PROMACTA (eltrombopag) [prescribing information]. East Hanover, NJ: Novartis Pharmaceuticals Corp; 2015.
pi.amgen.com/~/media/amgen/repositorysites/pi-amgen-com/nplate/nplate_pi_hcp_english.pdf. Accessed
May 11, 2018.
3. Kuter DJ, Bussel JB, Lyons RM, et al. Efficacy of romiplostim in patients with chronic immune thrombocyto-
penic purpura: a double-blind randomised controlled trial. Lancet. 2008;371(9610):395-403.
doi: 10.1016/S0140-6736(08)60203-2.
4. Cheng G, Saleh MN, Marcher C, et al. Eltrombopag for management of chronic immune thrombocyto-
penia (RAISE): a 6-month, randomised, phase 3 study [erratum in Lancet. 2011;377(9763):382]. Lancet.
2011;377(9763):393-402. doi: 10.1016/S0140-6736(10)60959-2.
5. Cetin K, Wasser J, Wettten S, Altomare I. Rate of bleeding-related episodes (BREs) in adult patients
with primary immune thrombocytopenic pupura (ITP): a population-based retrospective cohort study of
administrative medical claims data in the United States (US). Blood. 2014;124(21):202. bloodjournal.org/
content/124/21/202. Accessed May 11, 2018.
6. Rodeghiero F, Stasi R, Gernsheimer T, et al. Standardization of terminology, definitions and outcome criteria
in immune thrombocytopenic purpura of adults and children: report from an international working group. Blood.
2009;113(11):2386-2393. doi: 10.1182/blood-2008-07-162503.
7. Stasi R, Murali M, Michel M, et al. Evaluation of bleeding-related episodes in patients with immune
thrombocytopenia (ITP) receiving romiplostim or medical standard of care. Int J Hematol. 2012;96(1):26-33.
doi: 10.1007/s12185-012-1088-8.
8. Weitz I, Sanz MA, Henry D, et al. A novel approach to the evaluation of bleeding-related episodes in patients with
chronic immune thrombocytopenia. Curr Med Res Opin. 2012;28(5):789-796. doi: 10.1185/03007995.2012.684046.
9. Kikuchi K, Miyakawa Y, Ikeda S, Sato Y, Takebayashi T. Cost-effectiveness of adding rituximab to splenec-
tomy and romiplostim for treating steroid-resistant idiopathic thrombocytopenic purpura in adults. BMC Health
Serv Res. 2015;15:2. doi: 10.1186/s12913-015-0681-y.
10. Lee D, Thornton P, Hirst A, Kutikova L, Deuson R, Brereton N. Cost effectiveness of romiplostim for
the treatment of chronic immune thrombocytopenia in Ireland [erratum in Appl Health Econ Health Policy.
2013;11(6):687]. Appl Health Econ Health Policy. 2013;11(5):457-469. doi: 10.1007/ s40258-013-0044-y.
11. Cooper K, Matcham J, Helme K, Akehurst R. Update on romiplostim and eltrombopag indirect comparison.
Int J Technol Assess Health Care. 2014;30(1):129-130. doi: 10.1017/S0266462313000767.
12. Eltrombopag for treating chronic immune (idiopathic) thrombocytopenic purpura. National Institute for Health
and Care Excellence website. nice.org.uk/guidance/ta293. Published July 24, 2013. Accessed May 11, 2018.
13. Cummins E, Fielding S, Scott N, et al. Eltrombopag for the treatment of chronic immune thrombocytopenic
purpura (ITP): a single technology appraisal. National Institute for Health and Care Excellence website.
nice.org.uk/guidance/ta293/documents/thrombocytopenic-purpura-eltrombopag-rev-ta205-evidence-review-
group-report2. Published October 19, 2012. Accessed May 11, 2018.
14. Jansen JP, Fleurence R, Devine B, et al. Interpreting indirect treatment comparisons and network meta-
analysis for health-care decision making: report of the ISPOR Task Force on Indirect Treatment Comparisons
Good Research Practices: part 1. Value Health. 2011;14(4):417-428. doi: 10.1016/j.jval.2011.04.002.
15. EncoderPro.com website. encoderpro.com/epro. Accessed July 2015.
16. Lin J, Zhang X, Li X, et al. Cost of bleeding related episodes in adult patients with primary immune
thrombocytopenia: a population-based retrospective cohort study of administrative claims data for commercial
payers in the United States. Clin Ther. 2017;39(3):603-609.e1. doi: 10.1016/j.clinthera.2017.01.023.
17. 2015 National Physician Fee Schedule relative value file (July release). CMS website. cms.gov/Medicare/
Medicare-Fee-For-Service-Payment/PhysicianFeeSched. Accessed November 2015.
18. 2015 Clinical Diagnostic Laboratory Fee Schedule. CMS website. cms.gov/Medicare/Medicare-Fee-For-
Service-Payment/Clinicallabfeesched. Accessed November 2015.
19. Drummond MF, Sculpher MJ, Claxton K, Stoddart GL, Torrance GW. Methods for the Economic Evaluation of
Health Care Programmes. 4th ed. Oxford, UK: Oxford University Press; 2015.
20. Gold MR, Siegel JE, Russell LB, Weinstein MC, eds. Cost-Effectiveness in Health and Medicine. New York, NY:
Oxford University Press; 1996.
21. Bala MV, Zarkin GA. Application of cost-effectiveness analysis to multiple products: a practical guide. Am J
Manag Care. 2002;8(3):211-218.
22. Bussel JB, Provan D, Shamsi T, et al. Effect of eltrombopag on platelet counts and bleeding during treat-
ment of chronic idiopathic thrombocytopenic purpura: a randomised, double-blind, placebo-controlled trial.
Lancet. 2009;373(9664):641-648. doi: 10.1016/S0140-6736(09)60402-5.
23. Cooper KL, Fitzgerald P, Dillingham K, Helme K, Akehurst R. Romiplostim and eltrombopag for immune
thrombocytopenia: methods for indirect comparison. Int J Technol Assess Health Care. 2012;28(3):249-258.
doi: 10.1017/S0266462312000414.
24. Vytrisalova M, Blazkova S, Palicka V, et al. Self-reported compliance with osteoporosis medication—
qualitative aspects and correlates. Maturitas. 2008;60(3-4):223-229. doi: 10.1016/j.maturitas.2008.07.009.
25. Hamilton B, McCoy K, Taggart H. Tolerability and compliance with risedronate in clinical practice. Osteopo-
ros Int. 2003;14(3):259-262. doi: 10.1007/s00198-002-1370-3.
Full text and PDF at www.ajmc.com