This economic analysis was based on Markov model and did not require approval from an institutional review board.
Clinical Parameters
Clinical patient characteristics and outcomes are from the NALA trial[11]. NALA is a randomized, active-controlled, phase III trial comparing neratinib, an irreversible pan-HER TKI, plus capecitabine (N + C) against lapatinib, a reversible dual TKI, plus capecitabine (L + C) in patients with centrally confirmed HER2-positive MBC with ≥ 2 previous HER2-directed MBC regimens.
The criteria for enrolling patients include the following: Inclusion criteria:(1)Aged ≥ 18 years at the signing of informed consent. (2)Histologically confirmed MBC, current stage IV.(3)Documented HER2 overexpression or gene-amplified tumor immunohistochemistry 3 + or 2+, with confirmatory fluorescence in situ hybridization (FISH) +.(4)Prior treatment with at least two HER2-directed regimens for metastatic breast cancer. Exclusion criteria: Received previous therapy with capecitabine, neratinib, lapatinib, or any other HER2-directed tyrosine kinase inhibitor.
Patients were randomly assigned 1:1 to neratinib (240 mg orally, once daily with food, continuously in 21-day cycles) plus capecitabine (1500 mg/m2 daily in 2 evenly divided doses, orally with water within 30 minutes after a meal, taken on days 1 to 14 of each 21-day cycle), or to lapatinib (1250 mg orally, once daily, continuously in 21-day cycles) plus capecitabine (2000 mg/m2 daily in 2 evenly divided doses, orally with water within 30 minutes after a meal, taken on days 1 to 14 of each 21-day cycle). Coprimary endpoints were centrally confirmed progression-free survival (PFS) and overall survival (OS). NALA was considered positive if either primary endpoint was met (a split between endpoints). Secondary endpoints were time to CNS disease intervention, investigator-assessed PFS, objective response rate (ORR), duration of response (DoR), clinical benefit rate, safety, and health-related quality of life (HRQoL).
A total of 621 patients from 28 countries were randomly assigned (N + C, n = 307; L + C, n = 314)[11].Centrally reviewed PFS was improved with N + C (hazard ratio [HR], 0.76; 95% CI, 0.63 to 0.93; stratified log-rank P = 0.0059). The median PFS is 5.6 months(4.9 to 6.9 months) in N + C and 5.5 months(4.3 to 5.6 months) in L + C. The OS HR was 0.88 (95% CI, 0.72 to 1.07; P = 0.2098). The ORR was 32.8% in N + C (95% CI, 27.1 to 38.9) and 26.7% in L + C (95% CI, 21.5 to 32.4; P = 0.1201); median DoR was 8.5 versus 5.6 months, respectively (HR, 0.50; 95% CI, 0.33 to 0.74; P = 0.0004). The most common all-grade adverse events were diarrhea (N + C, 83%; L + C,66%) and nausea (N + C,53%; L + C,42%). Discontinuation rates and health-related quality of life (HRQoL) were similar between groups.
Among all of the patients, 104 and 98 Asian patients were randomly assigned to receive N + C or L + C, respectively. The Median PFS of N + C and L + C was 7.0 and 5.4 months (P = 0.0011), respectively. Both median OS (23.8 versus 18.7 months; P = 0.185) and DoR (11.1 versus 4.2 months; P < 0.0001) were extended with N + C, compared to L + C. The incidences of grade 3/4 treatment emergent adverse events (TEAEs) and TEAEs leading to treatment discontinuation were mostly comparable between the two arms. Diarrhea and palmar-plantar erythrodysesthesia were the most frequent TEAEs in both arms, similar to the overall population in incidence and severity.
Markov Model
We established a Markov model with treeage Pro 2011 (treeage Software Inc., Williamstown, MA)to evaluate the cost-effectiveness of different treatment strategies. The Markov model of each therapeutic involved three mutually transferable health states: PFS, progressive disease (PD), and death. All patients were defined in the PFS state in the beginning and subsequently survived or died; patients who survived either remained in the PFS state or transferred to the PD state. Patients who transferred to the PD state either remained or died. The cycle length of transition probabilities was 21 days based on the period of therapy. The cost and utility values were calculated at a 3% annual discount rate[13]. The primary endpoints of the cost-effectiveness analysis were QALY and incremental cost-effectiveness ratio (ICER). Secondary endpoints were the average cost-effectiveness ratio (average CE) and net benefit (willing-to-pay [WTP] benefit-costs).
Costs Inputs
This study only considers direct medical costs, including drug costs, test and examination costs, doctor's diagnosis and treatment costs, material costs, bed costs, nursing costs, and treatment costs for common grade 3 or 4 adverse events. Tests cost included computed tomography, magnetic resonance imaging, single-photon emission computed tomography, and other blood tests.
Currently no listing of DS8201 (AstraZeneca&Daiichi Sankyo) in mainland of China, so we used its pricing at 15900 CNY per 100 mg in Hong Kong, China. Other drug costs were extracted from the drug price query website in China, and test costs, which were indirectly obtained from published literature[14], are shown in Table 1. All data in USD are converted into RMB at the exchange rate of 6.7394 (August 11, 2022).
Table 1
Parameters | Base-case value | Minimum | Maximum | Standard error | Distribution |
Drug cost(CNY/per cycle) | | | | | |
Neratinib | 11270.00 | 9016.00 | 13524.00 | 2254.00 | gamma |
Capecitabine | 1806.00 | 1444.80 | 2167.20 | 361.20 | gamma |
Lapatinib | 6839.91 | 5471.93 | 8207.89 | 1367.98 | gamma |
Capecitabine | 2531.62 | 2025.30 | 3037.94 | 506.32 | gamma |
DS8201(Enhertu) | 47700.00 | 38160.00 | 57240.00 | 9540.00 | gamma |
Routine follow-up of patients (CNY/per cycle) | 418.71 | 334.97 | 502.45 | 83.74 | gamma |
Expenditures on main AEs (CNY/per cycle) | | | | | |
N + C | | | | | |
Diarrhea | 9.13 | 7.31 | 10.96 | 1.83 | gamma |
Nausea | 4.06 | 3.25 | 4.87 | 0.81 | gamma |
PPE syndrome | 10.93 | 8.75 | 13.12 | 2.19 | gamma |
Vomiting | 3.77 | 3.02 | 4.53 | 0.75 | gamma |
Fatigue | 26.26 | 21.01 | 31.51 | 5.25 | gamma |
Total | 54.15 | 43.32 | 64.99 | 10.83 | gamma |
L + C | | | | | |
Diarrhea | 4.72 | 3.77 | 5.66 | 0.94 | gamma |
Nausea | 2.74 | 2.19 | 3.28 | 0.55 | gamma |
PPE syndrome | 12.87 | 10.29 | 15.44 | 2.57 | gamma |
Vomiting | 1.79 | 1.43 | 2.15 | 0.36 | gamma |
Fatigue | 28.01 | 22.41 | 33.61 | 5.60 | gamma |
Total | 50.12 | 40.10 | 60.15 | 10.02 | gamma |
Health State Utilities
QALY was used to quantify the health condition of patients. Health state utilities represented the life quality of patients based on disease state and adverse effects, ranged from 0 to 1. Utility values were all obtained from published pieces of literature [15–19]. We not only included the utility value of drugs but also considered the difference in adverse events with ≥ grade 3. The main adverse events are shown in Table 2.
Table 2
The main AEs and the costs.
Main AEs | Cost(CNY) | range | distribution | reference |
Diarrhea | 37.73 | 30.18–45.28 | gamma | [18] |
Nausea | 94.33 | 75.46–90.55 | gamma | [17] |
PPE syndrome | 113.99 | 91.19-109.43 | gamma | [19] |
Vomiting | 94.33 | 75.46–90.55 | gamma | [17] |
Fatigue | 875.29 | 700.23-1,050.35 | gamma | [16] |
Sensitivity Analysis
We performed a one-way sensitivity analysis to assess the impact of each parameter on model outputs. The ranges were set as ± 20% for utilities and ± 30% for costs [Table 1]. For the probabilistic sensitivity analysis, a Monte Carlo simulation was performed with 1000 iterations and each parameter was fitted to a specific distribution: a beta distribution for utilities and lognormal distribution for costs[20]. The WTP threshold was set to three times the per capita GDP of China in 2021: 243,000 CNY/QALY. The results are described as cost-effectiveness acceptability curves.