Cost-effectiveness analysis of proton beam therapy for treatment decision making in paranasal sinus and nasal cavity cancers in China

Paranasal sinus and nasal cavity cancers are rare types of cancer accounting for about 10% of all head and neck cancers [1, 2]. Surgery is only suitable for early-stage diseases (T1-T2N0). For the vast majority of cases, which are already in advanced stage (T2-T4N+) at diagnosis, radiotherapy is the indispensable approach either as post-operative treatment to improve local control or as the radical treatment [3]. Developments of radiation techniques have been a key approach to improve the local control and survival of this disease. With the introduction of conformal targeted radiotherapy and intensity-modulated photon-radiation therapy (IMRT), the 5-year overall survival rate has been improved from 28% in the 1960s to ~ 50% till presently [4].

However, further improvement in the local control of this disease using IMRT seems difficult due to contradiction between its need for high required radical dose to the tumor and the dose limits of the surrounding normal tissues. As a result, the dominance of IMRT is being challenged by a new charged particle radiotherapy mode, termed as proton beam therapy (PBT), which possesses superior dose distribution afforded by protons’ “Bragg peaks” [5, 6]. According to an authoritative systematical review, compared to IMRT, PBT has been found to improve the 5-year overall survival rate of paranasal sinus and nasal cavity cancer from 45.1 to 69.7% [7]. For this reason, paranasal sinus and nasal cavity cancers were included as the “Group 1″ indication of PBT in the “Proton Beam Therapy Model Policy” by the American Society for Radiation Oncology [8].

The advanced form of PBT, intensity-modulated proton radiation therapy (IMPT), has the ability to apply pencil beam scanning technique to scan or “paint” the tumor volume voxel-by-voxel and layer-by-layer, and has undoubtedly become a more advantageous radiotherapy type for paranasal sinus and nasal cavity cancer. However, IMPT is much more expensive than IMRT due to technique complexity of particle therapy. Taking higher capital investment costs and operational costs into account, the cost ratio (IMPT/IMRT) ranges from 3.2 to 4.8 [9]. In china, the incidence of paranasal sinuses and nasal cavity cancer is similar to rest of the world, PBT has been introduced to treat this tumor type for few years. As a developing country with a gross domestic product (GDP) per capita of ~ 10,000 US dollars ($), PBT related costs are not yet covered by Chinese public medical insurance due to limited medical resources. Hence, cost-effectiveness analysis (CEA) is urgently-needed in clinical decision making for the appropriate radiotherapy mode [10].

The Markov model is a stochastic model that can track the natural process of chronic disease (i.e. tumor) by multiple cycles of operation and has become an ideal method to assess the cost-effectiveness of cancer treatment in a long-term period [11]. In this study, we designed a 3-state Markov model to evaluate the cost-effectiveness of PBT with Chinese settings and to facilitate the decision making for paranasal sinus and nasal cavity cancer treatment.

According to current data from the Particle Therapy Co-operative Group, only 1 proton center is in operation for the 1.4 billion Chinese population in China, and at least 10 new proton centers will start to treat patients in next 3 years due to huge demands [21]. Considering high treatment cost as a limiting factor hindering the use of PBT, cost-effectiveness evaluation is of urgent need for the treatment decision making when PBT become more available in the near future. To our knowledge, this is the first study on CEA modeling of PBT which was designed specifically for China.

Due to deficiency of valid data and lacking of a uniform model pattern, effective and reliable CEA studies of PBT are rare worldwide [22]. Some classical studies such as a 3-state Markov model designed for breast cancer, which focused on the advantages of IMPT in reducing the incidence of irradiation-induced coronary heart disease, concluded that IMPT was cost-effective for patients with 1 cardiac risk factors when photon are unable to achieve mean dose of heart < 5 Gy [23]. Another 6-state Markov model designed for Stage IVa oropharynx cancer, which was based on a hypothesis that IMPT could make a 25% reduction of xerostomia, dysgeusia and the need for gastrostomy tube, concluded that IMPT was cost-effective only in younger patients who could benefit from profound reductions of long-term morbidity [24]. These two Markov models focused on the late toxicities’ reduction of IMPT and could hardly be popularized to the tumor types for which IMPT could obviously improve tumor control compared with IMRT. Due to the anatomical location and a relatively low radiosensitivity of paranasal sinus and nasal cavity cancer, the advantage of IMPT compared with IMRT mainly lies in the improvement of tumor control, and no significant reduction of acute and late toxicities was found in report of Samir H Patel et al. [7]. Therefore, toxicities were assumed to be identical between the two strategies in our model building. This 3-state Markov model could be easily applied to the tumor types with similar behavior.

For validating model robustness, probabilistic sensitivity analyses were used to evaluate the uncertainties of parameters by their value variations. The Tornado diagram showed that only the probability of IMPT eradicating cancer, the probability of IMRT eradicating cancer and the cost of IMPT had significant impact on ICER. These results did correspond to our model design in which we focused on the advantage of IMPT in improving tumor control compared with IMRT so that the difference in probabilities for eradicating tumor between the two strategies would be the source of the difference in effectiveness and cost. Furthermore, the 2-, 5- and 10-year overall survival rates of the model were calculated and found to be within the ranges of previously published survival data of paranasal sinus and nasal cavity cancer, which demonstrated that our proposed model did abide with the natural process of this disease.

The specific scenarios that may benefit from IMPT were further analyzed, and results of CEA further supported that the application of IMPT for this type of tumor was reasonable from the perspective of the daily Chinese clinical practice. For the base case, IMPT should at least achieve a 0.859 probability of eradicating cancer. This probability is readily achieved as the actual 5-year locoregional control rate of paranasal sinus and nasal cavity cancer treated by PBT is 89.5% [7]. On the other hand, IMRT could hardly obtain the expected eradication probability of 0.771 in general. Hence, the specific scenarios in favor of IMPT in clinical practice would be when IMRT is considered as inability to well eradicate the tumor. The threshold of maximum cost of IMPT ($52,163.9) demonstrated the current price ($50,000) as reasonable considering the economic situation of China. Stratified analyses of different age levels showed that IMPT was considered as cost-effective only in patients ≤56-years old using the current WTP threshold of China. And this age threshold would increase with the growth of WTP threshold (the growth of GDP per capita) and the cost reduction of IMPT. With an assumption of a 20% cost reduction (with a cost of IMPT $40,000), IMPT could be cost-effective in patients ≤63-year-old at the WTP of China, in patients ≤69-year-old at a WTP of $50,000 / QALY, and in patients ≤72-year-old at a WTP of $100,000 / QALY. Therefore, we presume that IMPT would be more recommended in the future.

There are three main limitations to this model. First, the HSUVs applied in our model were derived from previously published studies for head and neck cancer because of the lack of specific HSUVs for paranasal sinus and nasal cavity cancer. As previously reported, the overall mean HSUV of head and neck cancer survivors after radiotherapy was consistently considered as 0.7 [25, 26]. In this study, the survivors after radiotherapy had two different states, namely as “alive with cancer” and “no cancer”. Therefore, we chose 0.47 as HSUV of “alive with cancer” for describing the utility of salvage treatment relative to chemotherapy, as reported by Ward MC et al. [13]; and HSUV of “no cancer” was assumed as 0.94 for describing the situation of no cancer and no need of treatment, as reported by Noel CW et al. [14]. Second, our CEA results indicated that the benefits obtained from IMPT (the improvement of tumor control in comparison with IMRT) was a principal consideration for the clinical decision of prescribing IMPT, but the current analyses with base case setting had not involved evaluating such benefits of an individual patient. So, we plan to use model-based approaches (such as tumor control probability model) to estimate personal benefits from IMPT for supporting individualized decision making in the future [27]. Third, due to the current deficiency of valid data about adverse events between IMPT and IMRT for paranasal sinus and nasal cavity cancers, such as acute/late toxicities and hospital admission rates during treatment, the effectiveness gain of IMPT in reducing these adverse events were not taken into the effectiveness calculation. We assumed that this limitation might be the cause of the negative results in trials of Monte Carlo simulations (only 13.5% of trials favored IMPT to IMRT), which were more susceptible to long-term differences [28]. With more data from the underway clinical trials evaluating toxicities and other clinical outcomes (such as NCT00797498 (photon/proton radiation therapy for cancers of the nasal cavity and/or paranasal sinuses)), the advantages of IMPT for this tumor type would be further evaluated, and IMPT might be more recommended with respect to patient’s life quality after radiotherapy.

Cost-effectiveness is a pivotal consideration for decision making of PBT in developing countries. To the best of our knowledge, this is the first CEA study of PBT designed for the specific situation of China. The 3-state Markov model designed for paranasal sinus and nasal cavity cancer was found to be practical, reliable and representative and could be easily applied to CEA studies of other similar tumor types. Our CEA results further evinced that IMPT was cost-effective for paranasal sinus and nasal cavity cancer treatment in the Chinese settings. The probability of IMPT eradicating cancer, the probability of IMRT eradicating cancer, the cost of IMPT and the patient’s age were found to be decisive components influencing ICER. Although our findings point that IMPT was cost-effective for paranasal sinus and nasal cavity cancer in patients ≤56-year-old, as the growth of WTP threshold and the cost reduction of IMPT, IMPT is expected to become more recommended in the future with respect to cost-effectiveness.