Nasopharyngeal carcinoma (NPC) is a prevalent malignant tumor of the head and neck with high-incidence foci, and unique geographical distribution mainly distributed in southern China and Southeast Asia. According to the World Health Organization’s International Agency for Research on Cancer, 40% of global NPC cases occur in China (Torre et al., 2015). Approximately 10% of new cases are metastatic patients, while another 15–30% of locally advanced NPC patients will develop locally recurrent or disseminated disease following treatment for locally advanced NPC (Lee et al., 2015; Yang et al., 2015; Tan et al., 2016). Cases experiencing recurrent or metastatic nasopharyngeal carcinoma (R/M NPC) have a poor prognosis, with a median overall survival (OS) of only 20 months (Zhang et al., 2016).

Platinum-based regimens have been considered standard first-line chemotherapies for R/M NPC. In 2016, a phase 3 randomized controlled study (GEM20110714) confirmed that gemcitabine plus cisplatin (GP) was more efficient compared with fluorouracil plus cisplatin (PF) for the first-line treatment of R/M NPC [progression-free survival (PFS), hazard ratio (HR), 0.55; 95% confidence interval (CI), 0.44 to 0.68; p < 0.0001]. Although the adverse events (AEs) in the GP and PF groups were different, the overall safety was controllable. This trial was a milestone in determining the first-line treatment preference for R/M NPC. lthough its clinical benefit is limited, as the median PFS is only 7 months (Zhang et al., 2016). Thus, novel therapeutic strategies are essential for this group of patients.

Since the 21st century, immune checkpoint inhibitor (ICI) has gradually attracted the attention of tumor community (Glisch et al., 2020; Singh et al., 2020; Watson et al., 2020; Luo et al., 2021). And studies have increasingly confirmed the effectiveness of these immunotherapies for malignant tumors (Ferris et al., 2016; Burtness et al., 2019; Paz-Ares et al., 2019; Schmid et al., 2020; Colombo et al., 2021; Hua et al., 2021; Jiang et al., 2021; Paz-Ares et al., 2021; Zhou et al., 2021). Toripalimab and carrellizumab are humanized high-affinity PD-1 monoclonal antibodies showing good clinical effectiveness and safety as first-line therapies for R/M NPC. In September 2015, the Incyte Corporation reached an agreement with Hengrui Pharmaceutical Co. to purchase the overseas rights to carrellizumab for $795 million USD. Breakthrough therapy designation for Triplel was granted by the US Food and Drug Administration (FDA) for NPC treatment in September 2020. In JUPITER-02 (NCT03581786), a phase 3 study, toripalimab plus GP (TGP) significantly improved PFS compared with GP chemotherapy (median, 11.7 vs. 8.0 months; HR, 0.52; 95% CI, 0.36 to 0.74; p = 0.0003) (Mai et al., 2021). In CAPTAIN-first (NCT03707509), a phase 3 study, camrelizumab plus GP (CGP) extended PFS significantly compared with GP chemotherapy (median, 9.7 vs. 6.9 months; HR, 0.54; 95% CI, 0.39 to 0.76; p = 0.0002). Overall survival (OS) in both groups was immature, and preliminary data suggested that patients receiving carrilizumab combined with chemotherapy tended to have improved survival (median OS, NR vs. 22.6 months; HR, 0.67; 95% CI, 0.41–1.11) (Yang et al., 2021a). Based on these data, CGP and TGP were approved as first-line treatment options for cases experiencing R/M NPC by the National Medical Products Administration of China and the Chinese Society of Clinical Oncology and included in the protocols for NPC interventions, version2021 ^{1 , 2} .

Although these treatment options have brought clinical benefits, the high cost and variety of ICIs means an analysis of their economics and efficacy is urgently essential to appraise which recently confirmed regimen presents the most clinical benefits at reasonable expenses and is more suitable for promotion. Therefore, the target of our study was to appraise the effectiveness and potential economic implications of TGP and CGP as first-line treatments for Chinese patients with R/M NPC from the Chinese citizen’s perspective.

We conducted a network meta-analysis and cost-effectiveness analysis based on two phase 3 clinical trials. Details of the network meta-analysis methods are given in the Supplementary Material. Cost-effectiveness analysis is guided by the Economic Assessment Report Standard Statement (CHEERS) checklist (Supplementary Table S1) and the details of its methods are presented below.

Over the past decade, immunotherapy has become one of the most important breakthroughs in cancer treatment (Sharon et al., 2014). PD-1 blocking antibody has good therapeutic effects on recurrent or metastatic head and neck cancer (R/M HNC), but there have been few corresponding economic evaluations of R/M HNC. The cost-effectiveness of pembrolizumab for R/M HNC patients demonstrated its cost-effectiveness compared to standard treatments in both China, the United States, and Argentina (Liu et al., 2019; Wurcel et al., 2021). A cost-effectiveness study by Robert et al. demonstrated that nivolumab was more cost-effective compared to chemotherapy in cases experiencing R/M HNSCC in the US (Haddad et al., 2020). In addition, network meta-analysis and cost-effectiveness analysis of R/M HNSCC patients in the US found that using nivolumab or pembrolizumab was more cost-effective based on WTP thresholds and patient weight (Pei et al., 2021). However, NPC is a type of HNC, and so far, there have been no cost-effectiveness analyses of immunotherapy for these patients; thus, a cost-effectiveness analysis of R/M NPC is crucial.

Individuals undergoing cancer treatment face a wide range of treatment options, but they and the societies they belong to face increasing economic burdens due to limited medical resources and high treatment prices. Economic assessment is a direct and theory-based approach that measures both the costs and outcomes in view of individual and social choices. Using the Markov model, we executed the first cost-effectiveness assessment of R/M NPC treatments from the Chinese citizens perspective and assessed the cost and efficacy of TGP, CGP, and GP chemotherapies at a WTP of $35,673 per QALY. Through our study, we found that TGP and CGP treatments resulted in an additional cost of $21,846 and $19,613, resulting in 1.108 and 0.960 QALYs compared to GP, respectively, which are clearly below the threshold for WTP. And ICER was significantly lower than the threshold of WTP. In addition, after comparing the two schemes, the ICER for TGP was deemed to be $15,103/QALY. Therefore, this cost-effectiveness study conducted for cases with R/M NPC in China demonstrated that TGP is the more efficient first-line treatment strategy and achieves the highest cost-effectiveness compared to CGP and GP.

The robustness of the model was verified through a sensitivity analysis of the model parameters. The most influential parameters in this model were the risk of thrombocytopenia from TGP and CGP, followed by the cost of toripalimab and camrelizumab. We found that the incidence of thrombocytopenia from TGP decreased by more than 15%, and the price of toripalimab decreased by more than 30% while that of camrelizumab rose by more than 30%, allowing TGP to dominate CGP economically. Because changing other parameters had no substantial effect on our results, mitigating drug-induced AEs and reducing the price of ICIs were considered the most practical measures for first-line TGP and CGP treatment to become absolutely cost-effective. It is worth noting that real-world evidence and many clinical studies have shown that grade 3/4 thrombocytopenia in chemo-immunotherapy is most likely to be caused by the chemotherapy. The incidences of grade 3/4 thrombocytopenia in NPC patients treated with GP chemotherapy alone and immunotherapy alone were over 10% and almost 0%, respectively (Zhang et al., 2016; Zhang et al., 2019; Yang et al., 2021a; Yang et al., 2021b; Mai et al., 2021). In clinical practice, if severe thrombocytopenia occurs during the treatment of NPC patients given chemo-immunotherapy, we first adjust the dose of chemotherapy to reduce its incidence, then reduce the ICER of the treatment regimen.

For a new drug to be approved by government and used correctly and widely in clinical situations, we need to rely on using clinical survival benefits, regional economic factors, and predictive markers to make judgements. For example, plasma Epstein-Barr virus deoxyribonucleic acid (EBV-DNA) levels have a substantial prognostic influence in cases with NPC. A retrospective analysis of 210 patients with NPC revealed a worse relapse-free survival rate (79.3%; p < 0.0001) and poorer OS (86%; p = 0.0003) in cases who were given high pretreatment EBV-DNA levels (Wang et al., 2013). A meta-analysis involving 22 investigations and 8,128 NPC cases showed that patients with high levels of EBV-DNA had a five to six times higher risk of death and metastasis than patients with low levels (Qu et al., 2020). However, programmed cell death ligand-1 (PD-L1) expression confirmed to be a proper biomarker for predicting the clinical effectiveness and prognosis of ICIs in HNC (Li et al., 2017; Cao et al., 2019; Huang et al., 2021). Li and others’ retrospective analysis proved that the 5-years OS and PFS of 120 nasopharyngeal carcinoma patients were 87.5 and 70.1%, respectively (Li et al., 2017). Another retrospective analysis showed a PD-L1 expression of 0%, 1–5%, 5–49%, and ≥50% in 154 NPC patients and 5-years OS and PFS of 75.5 and 85.7%, 72.7 and 72.7%, 55.9 and 68.3%, and 24.8 and 35%, respectively (Cao et al., 2019). Another meta-analysis suggested that high or positive expression of PD-L1 in head and neck squamous cell carcinoma (HNSCC) has a good predictive effect for OS at 6 and 12 months [relative risk (RR), 1.30; 95% CI, 1.02 to 1.65; p = 0.03; RR, 1.31; 95% CI, 1.05 to 1.62; p = 0.01] (Huang et al., 2021). In addition to the above two main biomarkers of clinical efficacy and prognosis, many studies have also indicated that the expression of epidermal growth factor receptor, Ki-67, vascular endothelial growth factor, and BRAF may also be good predicting biomarkers for NPC patients (Cheng et al., 2018; Cao et al., 2019; Chen et al., 2020; Li et al., 2021). Unfortunately, the JUPITER-02 and CAPTAIN-first studies lacked OS data for these two major prognostic markers, and PD-L1 expression was not grouped in CAPTAIN-first, hence, these could not be analyzed. The biomarkers that might lead specific patients to benefit from immuno-chemotherapy need to be confirmed through further research, which may make personalized treatment possible.

Some limitations were also evident in this study. First, an indirect comparison between first-line TGP and CGP was performed using network meta-analysis. Because we assumed no difference in patient characteristics between the two studies, there is potential uncertainty regarding the accuracy. Second, due to the short follow-up period of the two clinical trials, it was necessary to extrapolate the survival curve to obtain complete survival outcomes. The survival data will change over time, and the model will become more stable as more mature data becomes available. However, for now, this is an unavoidable limitation in our model. Third, to simplify the calculation, we assumed that follow-up treatment in the three groups only involved capecitabine chemotherapy, with the highest probability in the two studies, and ignored the other treatment options. On this basis, the analysis may have underestimated the cost of PD. However, sensitivity assessment demonstrated that changing the cost of capecitabine had little effect on the modelled results. Finally, considering that immunotherapy-related AEs are rare (the incidence was less than 10% in the two studies) and the cost of their treatment is quite high, we overlooked their administrative costs, which may overestimate the benefit of chemo-immunotherapy. However, including the cost of AE cases associated with immunotherapy will help to more accurately assess the overall cost of treating AEs using chemo-immunotherapy.

Conclusively, our achievements explain that TGP regimens could be more cost-efficient than GP and CGP regimens in China at a WTP threshold of $35,673 per QALY. It is necessary to provide patients the most efficacious treatment at the lowest cost, and the findings may help clinicians select the most appropriate drugs for patients and develop policies for medical reimbursement.