This study adhered to the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) checklist (Husereau et al., 2022). It targeted patients with advanced pleural mesothelioma deemed unsuitable for surgery and who had not undergone previous systemic therapy for their advanced condition. However, [neo] adjuvant chemotherapy was allowed more than a year before treatment. Eligibility criteria included an Eastern Cooperative Oncology Group performance status score of 0 or 1, measurable disease as defined by the Response Evaluation Criteria in Solid Tumours (RECIST) version 1.1, modified for use in pleural mesothelioma (mRECIST) and the provision of a tumor sample for correlative analysis. Exclusion criteria included patients with untreated CNS metastases, pneumonitis, current glucocorticoid treatment equivalent to more than 10 mg daily of prednisone (within 7 days before the first dose of study treatment), or those suffering from concurrent serious illnesses or other cancers.

Eligible patients were treated with intravenous cisplatin (75 mg/m²) and pemetrexed (500 mg/m²) every 3 weeks for up to six cycles. Those randomly assigned to the pembrolizumab group also received intravenous pembrolizumab (200 mg) every 3 weeks for up to 2 years. Patient evaluations were conducted before each cycle, 4 weeks after discontinuation, every 12 weeks until progression, and biannually until death. Imaging for both groups was scheduled every 6 weeks for the first three assessments, followed by 12-week intervals.

Subsequent treatment regimens were based on the Chinese expert consensus on diagnosing and treating malignant pleural mesothelioma (2023 edition) (Wang et al., 2023), the IND227 trial, and clinical practice. In the pembrolizumab plus chemotherapy group, patients received bevacizumab plus gemcitabine plus carboplatin, while the chemotherapy group received nivolumab. The doses of chemotherapy agents were calculated using a reference patient weighing 65.00 kg with a height of 1.64 m (body surface area of 1.72 m²). The current pembrolizumab and nivolumab patient assistance programs were considered for their effect on price reductions for these drugs.

The cost implications of adverse events (AEs) were determined based on data from the IND227 trial. In our analysis, only grade 3/4 AEs, classified as serious adverse events (SAEs), were considered, and this was restricted to those with incidence rates exceeding 5%. These SAEs included fatigue, anemia, thrombocytopenia, neutropenia, increased lipase levels, and hyperglycemia.

The development and outcome analysis of the model was conducted using TreeAge Pro 2022 software (Williamstown, MA, USA) and R software (version 4.2.3, Vienna, Austria). A three-state Markov model was implemented (Freitag et al., 2024; Wei et al., 2024), which comprised the following health states: progression-free survival (PFS), progressive disease (PD), and death (Figure 1).

All subjects entered the model in the PFS state and were either treated with chemotherapy alone or a combination of pembrolizumab and chemotherapy until disease progression or the emergence of intolerable toxicity. Patients whose disease advanced were transitioned to the PD state, where they received additional therapies following the discontinuation of pembrolizumab or placebo. The fraction of patients in the PD state was determined based on the area under the OS curve, the proportion of surviving patients, the proportion of patients with PFS, and the difference between the OS and PFS curves.

The simulation was performed over a period of 10 years, thereby encompassing over 99% of the mortality events in both treatment cohorts. The analysis was conducted from the vantage point of the Chinese healthcare systems, integrating all costs associated with healthcare services.

The primary outcomes measured were quality-adjusted life years (QALYs) and costs. To account for the value of time and adjust future costs and benefits, a 5% annual discount rate was used, as recommended by the World Health Organization (WHO) guidelines for pharmacoeconomic evaluations (Chinese Pharmaceutical Association, 2024). All costs were adjusted for 2023 prices using the local consumer price index and converted to US dollars at an exchange rate of $1 = ¥7.0467. The analysis included a cost-effectiveness evaluation, expressed in terms of incremental cost-effectiveness ratios (ICERs). The ICER was calculated as follows: ICER = [Cost (pembrolizumab plus chemotherapy) - Cost (chemotherapy)]/[QALY (pembrolizumab plus chemotherapy) - QALY (chemotherapy)]. The willingness-to-pay (WTP) threshold was three times China’s per capita gross domestic product (GDP) in 2023, $38,042.49, according to WHO recommendations. Furthermore, this study incorporated assessments of the incremental net health benefit (INHB) and the incremental net monetary benefit (INMB). These values were computed using the following equations: INHB (λ) = (μE1 - μE0) - (μC1 - μC0)/λ = ΔE - ΔC/λ and INMB (λ) = (μE1 - μE0) × λ - (μC1 - μC0) = ΔE × λ - ΔC. In these equations, μCi and μEi represent the costs and utility associated with pembrolizumab plus chemotherapy (i = 1) or chemotherapy alone (i = 0), respectively, with λ denoting the WTP threshold.

Survival curves for OS and FS in the IND227 trial were constructed using the method proposed by Guyot et al. (2012). Time-to-event data points for the Kaplan-Meier survival curves for OS and PFS were extracted using GetData Graph Digitizer version 2.26 (www.getdata.graph.digitizer.com). These data points were then applied to fit a range of parametric survival models, including Exponential, Weibull, WeibullPH, Gamma, Log-normal, Gompertz, Generalized Gamma, and Log-logistic distributions.

To determine the most suitable survival curves for PFS and OS, the fits were evaluated using both the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC), along with a visual review of the results (as presented in Tables 3-5). The estimated shape parameters (g) and scale parameters (l) for these curves are outlined in Table 1. Additionally, extensive details on long-term survival data can be found in Table 1 and Figures 2–7.

This analysis was conducted from the perspective of the Chinese healthcare system, emphasizing direct costs. These include drug expenses, laboratory test fees, costs associated with enhanced chest and abdomen CT scans, preventive medications for each intravenous administration, end-of-life expenses, infusion fees per intravenous administration, subsequent treatment costs, and expenses related to managing grades 3 and 4 AEs. Drug costs were derived from national medical insurance negotiation prices and local charges, while other cost data were from previously published research studies and relevant literature.

The drug doses for the patients were determined using the IND227 study protocol. Treatment costs per cycle were determined based on dose schedules and the local price (Table 1) (Wu et al., 2014; Li et al., 2021; Lin et al., 2021; Liu L. et al., 2023; Zhu et al., 2023). The cost associated with each AE was calculated by multiplying the incidence rate of the AE by the per-occurrence cost of managing these events (Wu et al., 2020; Shu et al., 2022; Liu S. et al., 2023). We assumed that all AEs occurred during the first cycle of treatment. Table 2 presents the incidence rates for each AE in detail.

The health utility scores in our model were assigned on a scale ranging from death (0) to perfect health (1). Due to the unavailability of the European Quality of Life-5 Dimensions-5 Level (EQ-5D-5L) data from the IND227 trial, direct quality-of-life data could not be obtained. Consequently, the utility values were from the relevant published literature (Table 1). The utility values for PFS and PD were set at 0.706 and 0.565, respectively (Zhan et al., 2017; Scherpereel et al., 2022; Yang L. et al., 2022). AEs were considered to decrease health utility, a concept known as disutility (Morris et al., 2014; Yang J. et al., 2022; Zhu et al., 2022; Liu S. et al., 2023). This disutility associated with AEs was incorporated only in the first cycle of the model and was assumed to occur once every 3 weeks.

Two sensitivity analyses were conducted to address the inherent uncertainty: one-way sensitivity analysis (OWSA) and probabilistic sensitivity analysis (PSA). In the OWSA, the parameter ranges were from the existing published literature, and the parameter variations were set to ±20% of the baseline values (Morris et al., 2014; Wu et al., 2014; Zhan et al., 2017; Wu et al., 2020; Li et al., 2021; Lin et al., 2021; Scherpereel et al., 2022; Shu et al., 2022; Yang J. et al., 2022; Yang L. et al., 2022; Zhu et al., 2022; Liu L. et al., 2023; Liu S. et al., 2023; Zhu et al., 2023). For PSA, the model parameters were altered simultaneously across 10,000 Monte Carlo simulations, allowing an evaluation of the probability that each intervention would be cost-effective in varying the WTP thresholds for an additional QALY. In these simulations, beta distributions were assigned to utility parameters, and gamma distributions were used for cost variables. The outcomes of this analysis are shown in a scatter plot and a cost-effectiveness acceptability curve.

In subgroup analyses, the cost-effectiveness of pembrolizumab plus chemotherapy versus chemotherapy alone as first-line treatment for untreated advanced pleural mesothelioma was assessed separately for patients with epithelioid and non-epithelioid histologies. This assessment used base-case analysis, OWSA, and PSA methods. Due to the lack of distinct data in the IND227 trial in the follow-up treatment plan, drug selection, and the occurrence of adverse reactions specific to the epithelioid and non-epithelioid subgroups, it was assumed that these factors were consistent with those observed in the general study population.

During a 10-year analysis horizon, base-case findings revealed that the pembrolizumab plus chemotherapy group achieved an additional 0.99 QALYs at an incremental cost of $35,560.57. In contrast, the chemotherapy-only group gained 0.77 QALYs, with associated costs totaling $17,360.94. Comparative analysis between pembrolizumab plus chemotherapy and chemotherapy alone indicated a mean incremental effect of 0.23 QALYs and an additional cost of $18,199.63. This resulted in an ICER for pembrolizumab plus chemotherapy versus chemotherapy of $80,557.23 per QALY, as shown in Table 5. In particular, when evaluated against the China WTP cost-effectiveness threshold of $38,042.49 per QALY, pembrolizumab plus chemotherapy was not more cost-effective than chemotherapy. Additionally, at this WTP threshold, pembrolizumab plus chemotherapy, relative to chemotherapy, demonstrated an INHB of −0.25 QALYs and an INMB of $-9,605.00 (Table 5).

In the subgroup analyses, the ICER for pembrolizumab plus chemotherapy versus chemotherapy alone was $33,917.61 per QALY gained for patients with non-epithelioid histology and $99,536.73 per QALY for those with epithelioid histology. In particular, in the non-epithelioid subgroup, ICER was below the China WTP threshold of $38,042.49 per QALY, as shown in Table 6. Furthermore, the INHB associated with pembrolizumab plus chemotherapy was 0.05 QALYs for non-epithelioid patients and −0.29 QALYs for epithelioid patients. Regarding INMB, the values were $2,085.28 and $-11,127.42, respectively, at the WTP threshold of $38,042.49 per QALY compared to chemotherapy, as detailed in Table 6.

Figure 8 presents a tornado diagram from OWSA in the analysis that covers the entire population. This diagram shows the factors that exert the most significant influence on the base-case outcomes. The cost of pembrolizumab, the utility value of PD, and the cost of bevacizumab had the most pronounced impact on the base-case results. Figure 9 indicates that, for patients with non-epithelioid histology, ICER was mainly influenced by the utility of PFS, and the costs of pembrolizumab and nivolumab. In contrast, in patients with epithelioid histology, ICER was predominantly affected by three factors: the utility of PD, the cost of pembrolizumab, and the cost of bevacizumab, as shown in Figure 9. However, due to the marked differences in health outcomes between the two treatment strategies in all three groups, no variations in parameter values altered the study outcomes for these groups.

Figures 10–13 show acceptability curves and probabilistic scatter plots, which indicate the cost-effectiveness landscape. PSA outcomes revealed significant probabilities that pembrolizumab plus chemotherapy is deemed cost-effective. These probabilities were 0.55% for the entire population, 69.41% for those with non-epithelioid histology, and 0.14% for those with epithelioid histology, evaluated against a WTP threshold of three times China’s GDP per capita ($38,042.49).