Patient population and treatment options
Patient data in this Study were obtained from the HER-2 negative subgroup analysis in STUDY-301 (NCT00337103). The inclusion criteria of STUDY-301 included: female sex; age≥18 years; histologically or cytologically confirmed breast cancer; up to three prior chemotherapy regimens and up to two prior chemotherapy regimens for advance dand/or metastatic disease; prior therapy with an anthracycline and a taxane; resolution of all chemotherapy-or radiation-related toxicities to≤grade 1 (except for stable sensory neuropathy≤grade 2 and alopecia); Eastern Cooperative Oncology Group performance status of 0 to 2; and adequate renal, bone marrow, and liver function. Measurable or nonmeasurable disease was allowed. Patients were randomly assigned (1:1) using a central interactive voice-response system to receivee ribulin mesylate 1.4 mg/m2(equivalent to eribulin 1.23 mg/m2[expressed as free base]) intravenously over 2 to 5 minutes on days 1 and 8, or capecitabine 1.25 g/m2 orally twice per day on days 1 to 14, both in 21-day cycles. Patients received study treatment until disease progression, unacceptable toxicity, or patient/investigator request to discontinue.
Model structure
EXCEL 2016 was used to construct the Partition Survival model, which included three states: progression-free survival (PFS), disease progression (PD) and Death. Patients in each states will convert to other states or stay in the original state within one cycle, and death is an absorbing state, that is, patients entering the death state shall not convert to other states. See Fig 1 for the conversion modes among states. In this Study, the initial state of all patients was assumed to be PFS, and the model cycle was set to 21 days, which was consistent with the administration period in STUDY-301. The study period was set at 10 years because more than 97 percent of the patients had died after 10 years of running the model.
Reconstruction of survival data
When the simulation time was within the follow-up period, the proportion of people in each state could be directly obtained from the OS curve and PFS curve. When the follow-up period is beyond, the parametric method should be used to reconstruct the survival curve, and then obtain the proportion of people in each state outside the follow-up period. GetData(version 2.20) software was used to obtain the original data from the OS and PFS curves of the HER-2 negative subgroup in STUDY-301, and R(version 4.1.1) software was used to reconstruct the OS and PFS curves of the ERI and CAP groups, respectively. The reconstructed curve was checked by visual inspection, and the median OS and median PFS of the two groups of reconstructed curves were compared with the data of the original literature, so as to judge whether the reconstructed curve was reliable. Finally, the optimal fitted distribution was judged according to Akaike information criterion (AIC) and Bayesian information criterion (BIC), and the distribution parameters of the fitted curve were obtained to calculate the survival rate. In this study, the log-logistic distribution was finally selected to fit the OS curve and PFS curve of the two groups, respectively. The data related to the reconstructed survival data are shown in Tab 1 and Tab 2.
Tab 1 The fitting results of different distribution of OS and PFS curve
Groups
|
Curves
|
Distribution
|
AIC
|
BIC
|
Median value(original/fitting)/months
|
ERI
|
PFS
|
Log-logistic
|
852.42
|
858.87
|
4.2/4.24
|
|
|
Weibull
|
896.52
|
902.96
|
|
|
|
Exponential
|
901.29
|
904.52
|
|
|
OS
|
Log-logistic
|
1198.37
|
1204.48
|
16.1/16.2
|
|
|
Weibull
|
1204.52
|
1211.13
|
|
|
|
Exponential
|
1219.07
|
1222.29
|
|
CAP
|
PFS
|
Log-logistic
|
986.13
|
992.79
|
4.0/3.99
|
|
|
Weibull
|
1018.73
|
1025.39
|
|
|
|
Exponential
|
1020.05
|
1023.37
|
|
|
OS
|
Log-logistic
|
1383.60
|
1389.94
|
13.5/13.6
|
|
|
Weibull
|
1386.21
|
1392.97
|
|
|
|
Exponential
|
6323.34
|
6323.36
|
|
AIC Akaike information criterion; BIC Bayesian information criterion; ERI eribulin; CAP capecitabine; PFS Progression-free survival; OS Overall survival
Tab 2 Distribution parameters of the survival curve
Paramaters
|
ERI group
|
CAP group
|
|
PFS curve
|
OS curve
|
PFS curve
|
OS curve
|
Shape
|
2.050
|
1.717
|
1.844
|
1.644
|
Scale
|
4.391
|
17.040
|
3.863
|
12.949
|
ERI eribulin; CAP capecitabine; PFS Progression-free survival; OS overall survival
Adverse events
Only grade 3 and 4 serious adverse events(AEs) with an incidence of ≥5% were included in this study, including neutropenia, leukopenia and peripheral neuropathy in ERI group and neutropenia, diarrhea, palmar-plantar erythrodysesthesia syndrome(PPESS) in CAP group. The formulae r=-ln(1-P1)/t1 and P2=1-exp(-rt2)were used to convert the incidence of adverse reactions during the follow-up period into the incidence of single cycle [14], where r is the instantaneous incidence and P1 is the probability of an event within a period of observation (namely, the incidence of adverse reactions during the follow-up period). P2 is the probability of metastasis in one cycle (that is, the incidence of single cycle), t1 is the length of observation (that is, the length of follow-up), t2 is the time of a single cycle (that is, 21 days), and the calculation results are shown in Table 3.
Tab 3 Single cycle incidence, disutility value and treatment cost of adverse events
Adverse events
|
Incidence per cycle/%(ERI )
|
Incidence per cycle/%(CAP)
|
disutility
|
Cost/$
|
Neutropenia
|
1.18
|
0.09
|
-0.15
|
2380.3
|
Leukopenia
|
0.18
|
-
|
-0.2
|
3214.93
|
Diarrhea
|
-
|
0.10
|
-0.103
|
200
|
Peripheral neuropathy
|
0.11
|
-
|
-0.18
|
694.4
|
PPESS
|
-
|
0.23
|
-0.016
|
0.23
|
ERI eribulin; CAP capecitabine; PPESS palmar-plantar erythrodysesthesia syndrome
Cost and utility
From the perspective of health system, this study only included direct medical costs, including drug costs, follow-up costs, disease administration costs, AE costs, and end-stage palliative care costs. The drug price is taken from the latest bidding price of yaozhi.com. According to the Report on Nutrition and Chronic Diseases in China (2020), the average height of Chinese women is 1.58 m, and the average body weight is 59 kg[21]. According to the body surface area formula, the body surface area is 1.62 m2. According to the relevant guidelines for the diagnosis and treatment of breast cancer, relevant diagnostic tests should be performed during the diagnosis and treatment of breast cancer, including physical examination, hematological examination (blood routine, liver and kidney function, blood electrolytes, tumor markers) and imaging examination (chest/abdomen ultrasound, bone scan, mammography). Therefore, the follow-up cost is mainly the above inspection cost, and the unit prices of inspection items are all from Price list of medical services in Anhui Province released by Anhui Provincial Medical Security Bureau(AMSB) in 2022. The cost of disease administration mainly includes hospitalization fee, nursing fee and so on. The costs of AEs were all derived from previous studies [15,16,17], and this study assumed that each patient would have only one AE during the treatment period. In this paper, the disutility values of AEs were included to evaluate the impact of AEs on quality of life, and the disutility values were all from previous studies [18,19,20]. The utility values of each state in this study are derived from a study based on Chinese population [21], and the utility values of PFS state and PD state are 0.87 and 0.71, respectively. However, the STUDY-301 also reported the results related to quality of life, so the utility values reported by STUDY-301 were included in the scenario analysis. The utility value of PFS status in ERI group was 0.717. The utility value of PFS status in CAP group was 0.715. The utility value of PD status in both groups was 0.695. The costs and disutility values of AEs are shown in Table 3, and other cost and utility parameters are shown in Table 4. According to the recommendations of China Guidelines for Pharmacoeconomic Evaluations (2020) [22], the cost and output of this study were discounted at a discount rate of 5%, and sensitivity analysis was conducted on discount rates of 0-8%.
Tab 4 Parameters of the partition survival model
parameters
|
Base value
|
Lower value
|
Upper value
|
Distribution
|
source
|
The price of medicine($)
|
|
|
|
|
|
Eribulin(2ml:1mg)
|
101.26
|
81.01
|
121.51
|
gamma
|
yaozh.com
|
Capacitabine(0.5g*12)
|
5.66
|
4.53
|
6.79
|
gamma
|
yaozh.com
|
Cost of disease administration($/day)
|
|
|
|
|
AMSB
|
The nursing fee of PFS
|
2.79
|
2.23
|
3.35
|
gamma
|
-
|
The nursing fee of PD
|
3.91
|
3.13
|
4.69
|
gamma
|
-
|
The hospitalization fee
|
5.86
|
4.69
|
7.03
|
gamma
|
-
|
Best supportive care
|
121.47
|
97.18
|
145.76
|
gamma
|
[21]
|
Palliative care
|
1874.64
|
1499.71
|
2249.57
|
gamma
|
[21]
|
The cost of follow-up($)
|
|
|
|
|
AMSB
|
Intravenous
|
0.84
|
0.67
|
1.01
|
gamma
|
-
|
Physical examination
|
4.18
|
3.34
|
5.02
|
gamma
|
-
|
Blood routine
|
0.42
|
0.34
|
0.50
|
gamma
|
-
|
Tumor markers
|
5.58
|
4.46
|
6.70
|
gamma
|
-
|
Chest/abdomen Ultrasound
|
12.55
|
10.04
|
15.06
|
gamma
|
-
|
Bone scan
|
6.28
|
5.02
|
7.54
|
gamma
|
-
|
Mammography
|
3.07
|
2.46
|
3.68
|
gamma
|
-
|
Blood electrolyte
|
4.88
|
3.90
|
5.86
|
gamma
|
-
|
Utility
|
|
|
|
|
|
PFS
|
0.87
|
0.696
|
1.00
|
beta
|
[22]
|
PD
|
0.71
|
0.568
|
0.852
|
beta
|
[22]
|
Discount rate
|
0.05
|
0
|
0.08
|
beta
|
[23]
|
AMSB Anhui Provincial Medical Security Bureau; PFS Progression-free survial; PD Disease progression
Outcomes
The primary outcome of this study was Quality Adjusted Life Years(QALYs). QALYs, which is obtained by multiplying the duration of a state by the utility value in that state, is a comprehensive index that takes into account both the length of life and the quality of life. The evaluation method adopts Cost-utility analysis (CEA), and takes Incremantal cost-utility ratio (ICUR) as the evaluation index. Comparing ICUR with Willing to pay (WTP), if ICUR is less than WTP, it indicates that eribulin is cost-utility; otherwise, capecitabine is cost-utility. According to the China Guidelines for Pharmacoeconomic Evaluations (2020), 3 times of China's per capita GDP was taken as the WTP. In 2021, China's per capita GDP was about $11,293.72, so the WTP of this study was $33881.17/QALY.
Sensitivity analysis
One-way sensitivity analyses (OWSA), Probability sensitivity analyses (PSA) and scenario analysis were used to evaluate the robustness of the study results. OWSA evaluates the impact of a single parameter on the result when it changes within a given range. Generally, the range of parameter change is 95% confidence interval (if the confidence interval is unknown, the baseline value will fluctuate by 20%). The range of parameter change is shown in Table 4. The results of OWSA are generally presented in the tornado diagram, in which the parameters are arranged from top to bottom according to their influence on the results.
PSA is to evaluate the impact of all parameters in the model on the results when they change according to a certain distribution. PSA is generally realized through Monte Carlo simulation. In this study, it is assumed that the cost parameters follow Gamma distribution, and the discount rate, utility value and other parameters follow Beta distribution. Based on the distribution of parameters, the sampling was repeated 1000 times, and the results of each sampling were recorded to evaluate the robustness of the study results. The results of PSA are shown in scatterplots and Cost-utility acceptability curve (CUAC).