Patient population
In order to better reflect the characteristics and disease progression of pediatric Ph + ALL patients in China, a real-world cohort survey was conducted in our hospital to collect baseline characteristics, diagnosis and treatment information, cost information and so on. A total of 32 pediatric Ph + ALL patients were included from the electronic medical record, 14 of whom were treated with imatinib, while the others were received dasatinib [8]. The median age was 7 years old. In the cohort study, the pediatric patients received approximately a 4-week induction therapy and a 44-week intensive chemotherapy.
Model Structure
It was assumed that the patients were separated into two groups receiving either imatinib (300mg/ day, per square meter of body surface area) or dasatinib (70mg/ day, per square meter of body surface area) during the treatment period. According to the diagnosis and treatment standard of childhood ALL, Ph + ALL patients should undergo induction therapy, intensive chemotherapy, and maintenance treatment. A combined decision tree and 10-year Markov model was developed to estimate the increase cost-effectiveness between imatinib and dasatinib from the health system perspective, seen Fig. 1. In the model, the disease progression was simplified to a few disease states, and all the patients would be in only one state in any cycle. The decision tree model was used to simulate the progression in the induction therapy and intensive chemotherapy. The total duration for these two periods was about one year. At the end of the two period of therapy, patients would be in one of three states: (1) CR, (2) non-CR, and (3) death, and then turned into the 10-years Markov simulation phase. In the Markov model, the state of each patient was decided by the end of intensive chemotherapy and the single cycle was one year. If non-CR patients or those who relapsed from CR state have passed doctor's assessment, they would receive hematopoietic stem cell transplantation (HSCT), otherwise they might stay at the relapse or non-CR state and receive the chemotherapy again to keep alive. If transplantation failed, they might choose other possible therapeutic schedules, like immunotherapy. In each state, the patients had the possibility to die.
Transition Probabilities
In this study, the transition probabilities were consisted of two parts. For the decision tree model, the parameters were collected from the real-world retrospective cohort study from our hospital. The rate of each transition was calculated. At the end of the 4-week induction chemotherapy, none of the patients treated with imatinib reached the CR state, while 3 of those treated with dasatinib (16.7%) became CR. At the end of the 44-week intensive chemotherapy, the CR rate for imatinib group and dasatinib group was 50% and 83.5% respectively [11]. For the Markov model, the parameters were from published articles and experts’ opinions. The rates from CR state to Relapse or Death state in maintenance therapy of the two group were from the published data of head-to-head clinical trial in China [8], and were estimated to the annual probabilities via transformation formula [12]. The parameters after transplantation were estimated from a relevant article [13]. And other parameters like the death rate of non-CR patients in maintenance therapy, the transplantation rate, the successful rate for transplantation were from experts’ opinions, seen Table 1.
Table 1
Base Case Parameter Values and Clinically Plausible Ranges for Model
Parameter | Base-case value (range) | Distribution | Reference |
Transition Probabilities |
Imatinib |
non-CR after induction therapy | 1 | Beta | Cao 2021[11] |
non-CR after intensive chemotherapy | 0.5 | Beta | Cao 2021[11] |
Relapse in maintenance therapy | 0.1000 (0.0415, 0.1729) | Beta | Shen 2020[8] |
Death of CR patients in maintenance therapy | 0.0879 (0.0364,0.1365) | Beta | Shen 2020[8] |
Dasatinib |
non-CR after induction therapy | 0.833 | Beta | Cao 2021[11] |
CR of CR patients in intensive chemotherapy | 1 | Beta | Cao 2021[11] |
CR of non-CR patients in intensive chemotherapy | 0.875 | Beta | Cao 2021[11] |
Relapse in maintenance therapy | 0.0537 (0.0107, 0.1035) | Beta | Shen 2020[8] |
Death of CR patients in maintenance therapy | 0.0304 (0.0099, 0.0504) | Beta | Shen 2020[8] |
Death of non-CR patients in maintenance therapy | 0.6 (0.54, 0.66) | Beta | Experts' opinions |
Transplantation |
Transplantation in non-CR patients | 0.1 (0.09, 0.11) | Beta | Experts' opinions |
Success in transplantation | 0.4 (0.3, 0.5) | Beta | Experts' opinions |
Relapse after transplantation | 0.0582 (0.0354, 0.0806) | Beta | Lin 2019[13] |
Death in CR patients after transplantation | 0.23 (0.21, 0.25) | Beta | Lin 2019[13] |
Death in non-CR/ relapse patients after transplantation | 0.57 (0.49, 0.64) | Beta | Lin 2019[13] |
Costs, CNY, per year |
Imatinib |
Drug | 104857.2 (83885.76, 125828.64) | Gamma | Expense list |
Other costs in induction therapy | 71019.18 (10168.61, 170406.88) | Gamma | Expense list |
Other costs in intensive chemotherapy | 266871.23 (120615.16, 398733.09) | Gamma | Expense list |
Dasatinib |
Drug | 51100 (40880, 61320) | Gamma | Expense list |
Other costs in induction therapy | 91338.95 (26639.48, 176128.92) | Gamma | Expense list |
Other costs in intensive chemotherapy | 208152.13 (141714.97, 274986.24) | Gamma | Expense list |
Other costs for maintenance in CR state | 57500 (46000, 69000) | Gamma | Expense list |
Maintenance in non-CR state | 375953.99 (244489.6, 531322.71) | Gamma | Expense list |
Transplantation | 250000 (200000, 300000) | Gamma | Experts' opinions |
Maintenance in CR state after transplantation | 399953.99 (268489.6, 555322.71) | Gamma | Expense list |
non-CR/ relapse after transplantation | 1000000 (800000, 1200000) | Gamma | Experts' opinions |
Time horizon, year |
Imatinib |
Induction therapy | 0.0795 | | Cao 2021[11] |
Intensive chemotherapy | 0.8164 | | Cao 2021[11] |
Dasatinib |
Induction therapy | 0.0849 | | Cao 2021[11] |
Intensive chemotherapy | 0.6822 | | Cao 2021[11] |
Utilities, QALYs |
CR | 0.88 (0.82, 0.93) | Beta | Lin 2019[13] |
non-CR/ relapse | 0.76 (0.7, 0.82) | Beta | Lin 2019[13] |
CR after transplantation (first 5 years) | 0.8 (0.74, 0.86) | Beta | Lin 2019[13] |
CR after transplantation (second 5 years) | 0.86 (0.8, 0.91) | Beta | Lin 2019[13] |
non-CR/ relapse after transplantation | 0.73 (0.67, 0.79) | Beta | Lin 2019[13] |
CR complete response, non-CR non-complete response, QALY quality-adjusted life year. |
Cost And Utility
From the health system perspective, direct health care costs were included, such as drug costs, other medical-related treatment costs like testing costs and hospitalization costs during induction therapy and intensive chemotherapy, the cost in CR or non-CR or relapse state during the maintenance period, and the cost of bone marrow transplantation and in the following states. The annual original drug costs were calculated according to the usage and dosage specified in the instructions. In the base case analysis, a 7-years-old child who weighed 20 kg with a 0.8m2 body surface area was set as a sample. Other cost parameters were collected from the expense list in the hospital, and the experts’ opinions would be chosen if there was no relevant list, as shown in the Table 1. The costs were expressed in 2019 China yuan (CNY).
The outcome used in this study was quality-adjusted life years (QALYs). QALYs were calculated by multiplying the health utility of a specific health state by the number of years lived in that state. Utility score of the states in different periods were collected from published related literature, as presented in Table 1 [13]. It was assumed that non-CR and relapse states have the same utility. The discount rate for cost and utility was 5% as recommended [14].
Base Case Analysis
In base case analysis, total cost and total QALYs of the two group over a decade time horizon were calculated. ICER was defined as the differences in costs divided by the differences in health outcomes of dasatinib and imatinib. The willingness to pay (WTP) threshold value for QALY was set at one capita of the gross domestic product (GDP) according to the 2020 China Guidelines for Pharmacoeconomic Evaluations [14]. As reported in China statistical Bulletin of National Economic and Social Development 2019, the 1 time GDP per capita is CNY 70,892 [15], which was the set threshold in the incremental analysis through TreeAge Pro 2011 software.
Sensitivity Analysis
One-way sensitivity analysis was conducted to alter variations such as costs, the recurrence rate, the mortality rate, and utilities in model input to assess the influence on our conclusions. If there was not reported the variance or the range of the parameters in reference, the cost and utility parameters were fluctuated by ± 20% of the base value, and the transfer probability parameters were ± 10% according to experts’ consultation. The annual discount rate fluctuated between 3% and 8%.
A probability sensitivity analysis was also conducted with Monte Carlo simulation to explored the association of input parameter uncertainty with model outcomes. It was assumed that the costs obeyed the Gamma distribution, the utility and transition probability value obeyed the Beta distribution, and the discount rate obeyed the Triangle distribution. The value of each model input was randomly drawn from the assigned parametric distributions in one thousand Monte-Carlo simulation.
Considering the using of cheaper generic drugs in China, which have already passed the consistency evaluation, a scenario analysis was conducted to calculate the ICER of generic drug usage with other conditions remain unchanged.