Introduction

Oesophageal cancer (OC) ranks among the prevalent gastrointestinal tumours globally, with approximately 90% identified as oesophageal squamous cell carcinoma (OSCC).1 According to the Global Cancer Statistics in 2020, there were 604 000 new cases of OC, of which 84.42% were diagnosed as OSCC (512 469 cases).2 China consistently bears a substantial burden of OC, reporting 324 422 new cases in 2020, with 90.4% being OSCC (293 280 cases).1 A study indicated that the age-standardised incidence of OSCC was about 9.55/100 000 in China.3 Beyond high morbidity, OC is also accompanied by high mortality. In 2020, global OC-related deaths reached 544 000, with China contributing 55.35% (301 000).1 Recent decades have witnessed a decline in OC incidence and mortality in China, attributed to advancements in precision medicine. Between 2000 and 2016, OC incidence and mortality dropped by 4.2% and 4.4% per year, respectively, with OSCC incidence declining by 4.3% annually.3

The escalating total medical and health expenditure in the country, comprising government, social and personal health expenditure, places a substantial financial burden, especially attributable to OC. A study forecasted that the direct medical expenditure due to OC will increase from US$33.4 billion to US$76.4 billion from 2013 to 2030, an increase of 128.7%.4 Data from the 2021 China Health Statistical Yearbook revealed a severe economic burden, with the average hospitalisation cost for OC in Chinese public hospitals being US$3891.87, including a notably high average drug cost of US$1002.73.

For decades, platinum-based chemotherapy has stood as the primary first-line treatment for OC.5 6 However, its efficacy and safety were unsatisfactory, with an overall survival (OS) of 7–13 months.7–9 Consequently, there is a pressing need for improved treatment options to enhance the prognosis of patients with OC. In recent years, advances have been made in OC treatment, particularly the development of immunotherapy such as programmed cell death-ligand 1 (PD-L1)/PD-1 inhibitors. Clinical trials such as KEYNOTE-181, ESCORT and KEYNOTE-590 confirmed the efficacy and safety of immune checkpoint inhibitors (ICIs) for advanced OC.10–12 Serplulimab, a novel PD-1 inhibitor, showcased promising results in the ASTRUM-007 study when combined with chemotherapy (cisplatin + 5-fluorouracil) for advanced OSCC treatment. This study showed that serplulimab plus chemotherapy (serplulimab+fluorouracil+cisplatin (SCF) group) prolonged progression-free survival (PFS) (5.8 vs 5.3 months) and OS (15.3 vs 11.8 months) in comparison with chemotherapy (fluorouracil+cisplatin (CF) group) alone.13 Based on these findings, serplulimab coupled with chemotherapy earned a place in the OC diagnosis and treatment guidelines outlined by the Chinese Society of Clinical Oncology (CSCO).

Despite the superior efficacy and safety of serplulimab plus chemotherapy for advanced OSCC compared with chemotherapy alone, the cost of novel PD-1 inhibitors imposes a substantial economic burden on both patients and the country. Conducting a pharmacoeconomic evaluation becomes crucial to help policy-makers judiciously allocate limited medical resources. Hence, this study aims to assess the cost-effectiveness of serplulimab combined with chemotherapy in the treatment of advanced OSCC from the perspective of the Chinese healthcare system.

Materials and methods

This study was a secondary analysis of data from the ASTRUM-007 trial (NCT03958890). This study was designed and performed in accordance with the Consolidated Health Economic Evaluation Report Standards statement checklist developed by the International Society for Pharmacoeconomics and Outcomes Research (online supplemental table 1).14

Model overview

A partitioned survival model was built using the TreeAge software to simulate the clinical and economic outcomes of the two therapies. According to the development process of OSCC, this study established three states of PFS, progressive disease (PD) and death for advanced OSCC patients after chemotherapy, and the model structure is shown in figure 1. The baseline characteristics of patients were consistent with the ASTRUM-007 trial (online supplemental table 2), and drugs were administered as follows: serplulimab (3 mg/kg) or placebo (day 1) plus cisplatin (50 mg/m2, day 1) and 5-fluorouracil (1200 mg/m2) (days 1, 2), once every 2 weeks.13 Cisplatin was administered for a maximum of 8 cycles and 5-fluorouracil for a maximum of 12 cycles. All patients were assumed to enter the model in the PFS status and received two treatment regimens until disease progression. Since the ASTRUM-007 trial only gave the proportion of patients who received subsequent systemic anticancer therapy and subsequent immunotherapy after disease progression but did not give specific treatment regimens, this study selected therapeutic agents according to the CSCO guideline and made the following assumptions: (1) SCF group: 17% of patients received camrelizumab (200 mg, day 1, every 2 weeks), 21% of patients received irinotecan (160 mg/m2, day 1, every 2 weeks) and teysuno (50 mg, two times per day, days 1–10, every 2 weeks) and 62% of patients received best supportive care.13 (2) CF group: 33% of patients received camrelizumab (200 mg, day 1, every 2 weeks), 19% of patients received irinotecan (160 mg/m2, day 1, every 2 weeks) and teysuno (50 mg, two times per day, days 1–10, every 2 weeks) and 48% received best supportive care.13 With the exception of teysuno, which was administered orally, all drugs were administered by intravenous infusion.

Figure 1

A three-state partitioned survival model simulating OSCC. (A) Decision tree model diagram. (B) OSCC three-state transition diagram. CF group, fluorouracil+cisplatin; OS, overall survival; OSCC, oesophageal squamous cell carcinoma; PD, progressive disease; PFS, progression-free survival; SCF group, serplulimab+fluorouracil+cisplatin.

The model period was set at 2 weeks, in line with the ASTRUM-007 trial, and the time horizon was set at 10 years.13 15 16 Incremental cost-effectiveness ratio (ICER) was used as the main outcome measure, and other indicators included total cost, incremental cost, life years, quality-adjusted life years (QALYs) and incremental QALYs. The willingness to pay (WTP) was set to three times of China’s gross domestic product (GDP) per capita in 2022 (WTP=US$37 155/QALY), and the cost and utility values were discounted at an annual discount rate of 5%, according to the recommendations of the WHO and China Pharmacoeconomic Evaluation Guidelines.17 18

Survival estimate

The survival data were extrapolated from the Kaplan-Meier curves in the ASTRUM-007 study.13 First, points were taken from the original survival curves in the clinical trial using the Engauge Digitizer software (https://github.com/markummitchell/engauge-digitizer/releases).13 Then, based on the method of Guyot et al, the survHE package in the R software (https://www.r-project.org) was used to reconstruct the individual level data of the patient and fit the parameter distribution of the survival curve, including exponential distribution, Weibull distribution, log-normal distribution, log-logistic distribution and Gompertz distribution.19 Goodness-of-fit test was performed according to the Akaike information criterion (AIC) and Bayesian information criterion (BIC), where smaller AIC and BIC indicate better goodness of fit. Finally, the log-normal distribution was chosen as the best fit for the PFS curve in the SCF group, the log-logistic distribution was chosen as the best fit for the OS curve in the SCF group and the PFS curve and OS curve in the CF group, respectively, and the results are shown in online supplemental table 3. The R software was used to calculate the distribution parameters of curves. According to the distribution parameters of each curve (online supplemental table 4), the transition probability over time under the two treatment regimens was calculated. The fitting curves are shown in online supplemental figure 1.

Cost estimate

Only direct medical costs were involved, including drug costs, treatment costs for severe adverse events (AEs), supportive treatment costs, routine follow-up costs and laboratory and radiological examination costs. The cost of medicine was obtained from the China Medical Information Network (https://www.menet.com.cn), and the average price of listed prices in provinces and cities across the country was used. AEs management costs were calculated only for AEs of grade 3 or higher with an incidence greater than 5%, considering that grade 1–2 AEs are usually not treated with medication and that AEs management costs were included only in the first cycle to simplify the model. The costs of supportive treatment, follow-up and laboratory and radiological examinations were derived from the literature.20 All fees were expressed in dollars and converted at 2022 exchange rates (US$1=¥6.92). See table 1 for the cost information.

Table 1

Cost and utility parameters

When calculating the drug dose, this study referred to the literature and assumed that the weight of the patient was 65 kg and the body surface area was 1.72 m2.21 22

Utility estimate

Health utility value is a quantitative indicator of quality of life, and it is specified in pharmacoeconomics that the utility value of death is 0, the utility value of perfect health is 1 and the utility value of the rest of the state is between 0 and 1. Unfortunately, the ASTRUM-007 trial did not collect information on patients’ quality of life, so the health utility values used in this study were derived from a global multicentre quality of life study in people with gastric adenocarcinoma or gastro-oesophageal junction and were 0.74 for PFS and 0.56 for PD.23 The reduction in utility value due to AEs was also all referred to the literature (table 1).

Sensitivity analysis

To verify the stability of the model, it is necessary to perform one-way sensitivity analysis and probabilistic sensitivity analysis (PSA). This study intended to set the variation range of important parameters in the model, such as cost, utility value, the incidence of AEs and discount rate, so as to conduct the one-way sensitivity analysis, and the results were presented in a tornado chart. PSA was performed by using 1000 Monte Carlo simulations to sample the distribution corresponding to each parameter, and the results were presented as a scatter plot and cost-effectiveness acceptability curve (CEAC). In the univariate sensitivity analysis, the upper and lower limits of drug cost parameters were acquired from the China Medical Information Network. Parameters for which the upper and lower bounds were not obtained were set to a range of 80%–120% of the base value. Additionally, the discount rate was set to fluctuate in the range of 0%–8% according to the recommendation of the Guide to Pharmacoeconomics Evaluation in China.17 In the PSA, gamma distribution was used for cost parameters, beta distribution was applied for health utility value and the incidence of AEs and normal distribution was chosen for weight and body surface area.

Subgroup analysis

A series of subgroup analyses in the ASTRUM-007 trial were performed according to age, sex, PD-L1 expression, disease status and smoking history, but only survival curves and survival information were reported for the PD-L1 expression subgroup.13 Therefore, this study conducted a subgroup analysis of the economic difference between serplulimab combined with chemotherapy versus chemotherapy in patients with different PD-L1 expression.

Scenario analysis

The following three scenario analyses were performed in this study:

Scenario analysis 1: Shanghai Life Oasis Public Service Center provides patient assistance programme (PAP) for low income and subsistence patients. Considering that the cost of drugs has a large impact on the outcome, this study explored the economics of the two treatment schemes in the case of charitable drug donation. The drug donation plan of serplulimab is ‘6+6, 6+N’. ‘6+6’ means that patients can obtain 6 cycles of drug assistance if they purchase 6 cycles of serplulimab and ‘6+N’ means that patients can purchase 6 cycles of serplulimab at their own expense to obtain drug assistance until PD. In this scenario, it is assumed that all patients meet the charitable donation conditions of serplulimab.

Scenario analysis 2: serplulimab is not currently covered by Medicare, but similar PD-1 inhibitors, such as camrelizumab, sintilimab and toripalimab, were included within a few years of their launch. Considering the possible price reduction of serplulimab in the future, we calculated the ICER for the case of 50% and 80% price reduction of serplulimab.

Scenario analysis 3: the ICER changes with the time horizon, and we set the time limit to 5 and 20 years, respectively, to explore the effect of the study time limit on the results.24

ResultsBase case analysis results

The results (table 2) showed that the total cost of the SCF group was US$69 356 and 1.38 QALYs were generated, while the total cost of the CF group was US$27 749 and 1 QALY was obtained. Compared with the CF group, the SCF group gained 0.38 QALYs more at an incremental cost of US$41 607, resulting in an ICER of US$110 744.36/QALY. Since the ICER was much higher than the WTP, serplulimab combined with chemotherapy in the treatment of OSCC was not cost-effective.

Table 2

Base case results

Sensitivity analysis results

The results of the univariate sensitivity analysis are shown in figure 2. From the tornado plot, it can be seen that the price of serplulimab, the weight of patients and the utility value of the PFS stage have the greatest impact on the model. The utility value of the PD stage, the proportion of the CF and SCF groups receiving subsequent treatment, the cost of laboratory and radiological tests and the price of irinotecan also had moderate effects on the model. Other parameters had little effect on the model, such as the proportion of patients receiving immunotherapy in the CF and SCF groups, the cost of best supportive care, the cost of follow-up, the body surface area of patients and the price of camrelizumab. However, the ICER value did not exceed the WTP with the change of the above parameters, indicating that the fundamental analysis results were robust.

Figure 2

Tornado diagram of one-way sensitivity analysis of total people. CF group, fluorouracil+cisplatin; EV, expected value; ICER, incremental cost-effectiveness ratio; PD, progressive disease; PFS, progression-free survival; SCF group, serplulimab+fluorouracil+cisplatin.

CEAC (online supplemental figure 2) demonstrated that SCF started to be cost-effective when the WTP was US$66 667/QALY. When the WTP was US$105 000/QALY and US$133 333/QALY, the probability of SCF being economical was 50% and 90%, respectively. See online supplemental figure 3 for the scatter plot. It can be seen from this figure that in 1000 simulations, all the points fall above the WTP line, which means that SCF group does not have a cost-benefit advantage in 1000 sampling times.

Subgroup analysis results

Subgroup analysis results (table 3) showed that for patients with PD-L1 combined positive score (CPS) between 1 and 10, SCF provided 0.34 QALYs at a cost of US$36 938. The ICER was US$108 395.64/QALY. For PD-L1 CPS≥10, SCF produced 0.4 more QALYs at an incremental cost of US$47 108, resulting in an ICER of US$119 160.85/QALY. Therefore, the addition of serplulimab to chemotherapy was not an economical option regardless of PD-L1 expression.

Table 3

Subgroup and scenario analysis results

Univariate sensitivity analysis showed (online supplemental figure 4) that for patients with CPS between 1 and 10, the price of serplulimab, weight, utility value of PD and PFS stages had the greatest impact on the ICER. However, for patients with CPS≥10, the factors that most influenced the model included the utility value of the PFS stage, the price of serplulimab and weight (online supplemental figure 5).

The results of PSA were shown in online supplemental figures 2 and 3. Online supplemental figure 2 illustrated that the SCF group has the lowest probability of being cost-effective among the subgroup of CPS≥10 under the same WTP. As can be seen from online supplemental figure 3, neither of the 2 subgroups was economical in 1000 simulations.

Scenario analysis results

The results of the scenario analysis are shown in table 3.

Scenario analysis 1: even in the presence of the PAP, the cost of the SCF group (US$46 382) would be US$18 633 higher than that of the CF group (US$27 749). The outputs of the two groups were 1.38 QALYs and 1 QALY, respectively. The ICER was US$49 594.12/QALY.

Scenario analysis 2: the price reduction of serplulimab would not change the cost and output of the CF group, regardless of the price reduction of serplulimab, the total cost and output of the CF group was US$27 749 and 1 QALY, and likewise, the output of the SCF group would not change (1.38 QALYs). When serplulimab reduced prices by 50% and 80%, the total cost of the SCF group was US$52 269, US$42 106, and ICER was US$65 262.67/QALY and US$37 973.66/QALY, respectively.

Scenario analysis 3: when the time horizon was 5 years, CF generated 0.95 QALYs at a cost of US$25 757, while SCF produced 1.24 QALYs at an expense of US$63 940, with an ICER of US$130 348.71/QALY. When the time horizon was 20 years, the cost of CF and SCF was US$28 808 and US$72 427, respectively, and the output was 1.04 QALYs and 1.47 QALYs, respectively, resulting in an ICER of US$101 163.41/QALY.

Discussion

The first-line systemic treatment options for advanced or metastatic OSCC are limited. Palliative chemotherapy not only has limited survival benefits, but also has serious AEs. Fortunately, with the progress of medical research, immunotherapy has achieved more survival benefits. Existing clinical trials have shown that PD-1 inhibitor combined with chemotherapy prolongs OS in patients with OC by around 2–6 months compared with chemotherapy.25 26 As a humanised immunoglobulin G4 monoclonal antibody, serplulimab can block the binding of PD-1 to PD-L1 by binding to the PD-1 receptor expressed in T cells to enhance T-cell-mediated antitumour effects.13 In the treatment of small-cell lung cancer and advanced OSCC, serplulimab has shown effective antitumour activity and safety controllability.13 27 In recent years, new antitumour drugs represented by ICIs have brought clinical benefits to OSCC patients, but also brought heavy economic burden to patients and the country. Pharmacoeconomic evaluation has become important evidence to support national medical insurance negotiation. Therefore, this study analysed the economy of serplulimab in combination with chemotherapy in the treatment of advanced or metastatic OSCC from the perspective of China’s healthcare system. Similar to most of the new antitumour drugs such as PD-1 and PD-L1, our results showed that the combination of serplulimab with chemotherapy was not economical compared with traditional chemotherapy.

One-way sensitivity analysis showed that body weight had a large impact on the ICER. Because serplulimab is administered by weight, and lower weight patients receive fewer doses, it may be more economical to use serplulimab in the lower weight patient population, both nationally and individually. From the national perspective, there are both lower weight patients and overweight patients. Therefore, this study used the average body weight as the base value, and sets the upper and lower limit as ±20% of the base value, considering the impact of average body weight on the ICER. From the standpoint of individual patients, the selection of drugs should consider not only the economy, but also the effectiveness, safety and individual basic conditions.

Currently, the first-line immunotherapy drugs approved for OSCC in China include pembrolizumab, camrelizumab and sintilimab.28 Two studies evaluated the cost-effectiveness of pembrolizumab plus chemotherapy in the first-line treatment of advanced OSCC from the perspective of the Chinese healthcare system and both results showed that it was unlikely to be economical to add pembrolizumab to chemotherapy.15 29 Zhang et al assessed the economics of adding camrelizumab to the first-line chemotherapy from the viewpoint of the Chinese healthcare system and showed that camrelizumab plus chemotherapy was still unlikely to have a cost-effectiveness advantage.20 Four studies evaluated the cost-effectiveness of sintilimab in combination with chemotherapy in the treatment of OSCC in China, and the results of these studies all showed that sintilimab+chemotherapy may be an economical option.30–33 Pembrolizumab is an imported drug, and the price has never decreased since it was launched in China (US$2589.49 per 200 mg). Camrelizumab is a domestic PD-1 inhibitor developed by China. The initial price of camrelizumab was as high as US$2861.48/200 mg. After medical insurance negotiations, the price was reduced to US$423.15/200 mg. But unfortunately, even with an 85% reduction, camrelizumab+chemotherapy is still not economical for advanced OSCC. Sintilimab is also a domestic anti-PD-1 monoclonal antibody. The price at the beginning of the market is US$1132.74 per 100 mg, and the price after medical insurance negotiation is only US$156.08/100 mg. With a reduction of 86%, sintilimab combined with chemotherapy in the treatment of OSCC is economical. It is not surprising to conclude that serplulimab, a newly marketed PD-1 inhibitor, combined with chemotherapy in the treatment of OSCC is not economical when analysing the cost-effectiveness using its initial market price. However, according to the price reduction trend of domestic PD-1 inhibitors, there is a high possibility that the price reduction rate of serplulimab will exceed 80%. Scenario analysis 3 showed that if the price of serplulimab is reduced by 80%, resulting in an ICER of US$37 973.66/QALY, which is very close to the WTP (US$37 155/QALY). This means that serplulimab plus chemotherapy in the treatment of OSCC is likely to be cost-effective in the future.

From the economic evaluation of new antitumour drugs, it can be seen that most drugs were not cost-effective at the beginning of marketing, and some drugs were not economical even when the price reduction was more than 80%. The reasons for this phenomenon should be considered.

First, the large cost gap and weak survival benefit gap between immunotherapy and chemotherapy lead to higher ICER. In terms of the first-line PD-(L)1 inhibitors in OSCC, most of the published economic studies showed that the incremental QALYs generated by immunotherapy combined with chemotherapy were less than 1 QALY, and only 1 study achieved 1.23 QALYs, while the incremental cost was at least US$7110 and at most US$53 320, resulting in a high ICER.15 20 29–33

Second, the WTP threshold serves as a crucial parameter for assessing cost-effectiveness. If the cost-effectiveness ratio falls below WTP, the drug is considered cost-effective. While the WHO recommends (1–3)×GDP per capita as WTP, the present findings make relevant recommendation about WTP more consistent with real-world clinical scenarios.34 Certain studies posit that WTP thresholds should vary for chronic and life-threatening diseases, with a prevailing belief that the threshold for cancer treatment should surpass that of general diseases due to heightened sensitivity to increased life expectancy.35 36 In the Netherlands, the WTP threshold is contingent on disease severity, warranting a higher threshold for life-threatening conditions.37 Despite the defined WTP of US$50 000/QALY in the USA, in fact, the WTP range is usually set between US$100 000/QALY and US$150 000/QALY when the cost-effectiveness evaluation of anticancer drugs is carried out from the perspective of the American medical system.38 At present, some scholars in our country also propose a higher WTP for anticancer drugs compared with other medications. The future portends an enhancement in the economic landscape of tumour drugs, driven by economic development and policy improvements.

There are still some limitations in this study. First, the survival data for this study were derived from the ASTRUM-007 trial, but the survival data in the ASTRUM-007 trial are not mature. Second, there is no study on the utility value of serplulimab in the treatment of OSCC patients worldwide, and the health utility value of other treatment strategies used in this study may bias the model results. Third, to simplify the model, only the cost of drug therapy was taken into account when calculating the cost of AEs management, and only AEs of grade 3 and above with an incidence greater than 5% were calculated, which may lead to some reduction in costs. However, one-way sensitivity analysis showed that the cost of AEs management had almost no impact on the model.

Conclusion

This study evaluated the economy of serplulimab in combination with chemotherapy in the first-line treatment of patients with advanced or metastatic OSCC based on the perspective of the healthcare system in China, and the results showed that SCF is unlikely to be cost-effective.