To evaluate the lifetime economic and clinical outcomes of MR-HIFU adoption for patients with bone metastases, a patient simulation model was developed using TreeAge Pro 2024 (TreeAge Software, Williamstown, MA) and reported according to the Consolidated Health Economic Evaluation Reporting Standards 2022 checklist (10). The analysis was performed from a societal perspective, following the guidelines of the Netherlands National Institute for Medical Research (Zorginstituut Nederland -ZIN), with differential discounting for costs (4%) and QALYs (1.5%) (11). A lifetime period was chosen to reflect the impact of pain relief over the patient’s lifetime and the need for retreatment. The selected cycle length was 1 month to reflect the time needed to observe the full therapeutic effect of EBRT or MR-HIFU.
The patient population consisted of breast, prostate, or lung cancer patients with uncomplicated nonvertebral metastases who were preselected to evaluate eligibility for both EBRT and MR-HIFU (12). According to the ESTRO-ACROP guidelines, bone metastases are (i) painful; (ii) no pathological fracture is imminent or present; (iii) There is no compression of the spinal cord or cauda equina regardless of size or location (13). An impending pathological fracture is characterized as not requiring immediate surgery or stabilization, as assessed by a score of ≥9 on the Mirels scoring system recommended in the ESTRO-ACROP guidelines (13).
In particular, MR-HIFU is contraindicated in patients with spinal metastases and (impending) pathological fractures, and these patients are excluded from our analysis (14). The patient population was consistent with participants from the main study on MR-HIFU effectiveness, with a mean age of 64 years ± 13 (SD) (6).
Key Comparison Strategies
Three strategies were designed for the main comparison, all based on one first-line treatment and, if necessary, one salvage pain relief treatment. Salvage therapy is indicated when treatment fails or when the patient experiences recurrence of pain after initially successful pain relief (Figure 1). The first strategy, hereafter referred to as ‘early MR-HIFU’, is based on MR-HIFU as first-line treatment, followed by EBRT if necessary. The second treatment, later called ‘delayed MR-HIFU’, was based on EBRT as first-line treatment and followed by MR-HIFU when necessary.
The third strategy reflected the current standard of care: EBRT followed by reirradiation with EBRT. EBRT dose was defined as a single fraction (8 Gy per fraction). Multiple fractions were not included in our model because a previous study-based cost-utility analysis in the Netherlands concluded that multiple fractions were not cost-effective compared to single fractions (15). Additionally, more advanced radiotherapy techniques, such as stereotactic radiotherapy (SBRT), have not been considered standard treatments for bone metastases.
Key comparison strategies. The dotted line indicates that not all patients receive second-line treatment. This is because patients may achieve pain relief after the first treatment or may die before receiving the second treatment. EBRT = external beam radiotherapy, MR-HIFU = high-intensity focused ultrasound with magnetic resonance imaging.
model structure
The model structure was consistent with that of a previous economic modeling study developed for Germany (16) (Figure 2). After treatment with MR-HIFU or EBRT, patients may experience: (i) complete pain relief, defined as a pain score of 0 on the Numeric Ratio Scale (NRS), (ii) partial pain relief, defined as a pain score reduction of at least 2 points without an increase in analgesic intake, or (iii) persistent pain, and finally (iv) death. Although the risk of pathologic fractures is low in this patient population, fractures are costly and impact patients’ quality of life (17). Therefore, pathologic fracture was considered an event that could occur at any time, and patients suffering a pathologic fracture (iii) progressed to persistent pain.
input parameters
Several literature searches were performed in Medline (via PubMed) to identify appropriate input parameters. Details of the search strategy and key search results are provided in Online Resource 1. The main input parameters applied to the model are listed in Table 1.
event probability
Event probabilities for complete and partial pain relief were taken from a prospective nonrandomized study comparing MR-HIFU and EBRT that included 198 patients over a 12-month follow-up period (6). For EBRT, systematic reviews with meta-analyses have calculated the probability of recurrence and risk of fracture considering only single segment populations (2, 3). After EBRT, 30% of patients experienced pain recurrence at 1-year follow-up (3, 18). For MR-HIFU, the probability of recurrence was thought to be lower based on 12-month follow-up data from a prospective, open-label, nonrandomized phase 2 study (6).
status utility
Quality-adjusted life years (QALY) values were obtained from two main sources. Based on time trade-off (TTO) interviews, 187 patients from the United Kingdom (UK) and Canada were evaluated for the impact of skeletal-related events on their quality of life resulting from treatment/retreatment with MR-HIFU or EBRT (19).
Based on EQ-5D-5 L data for patients with chronic pain (20), a 50% and 15% increase in post-treatment utility was assumed for patients with complete and partial pain relief, respectively. This assumption has also been used in previous economic modeling studies (16, 21).
costs
MR-HIFU and EBRT treatment costs included resource use of pretreatment imaging. MR-HIFU procedure costs were calculated using a micro-costing approach (22) using costs from the ZIN reference list and the salary scales of the CAO Hospital Division (for salaried service staff) and Medical Staff (AMS) (for medical staff) (23, 24). The microcost calculation method is described in detail in Online Resource 2. EBRT costs were taken from a Dutch trial-based cost-utility study (15). Costs for nonvertebral fracture treatment were according to a previous Dutch cost-utility analysis for the acid zoledronic acid (25) . Costs were adjusted for inflation using the Consumer Price Index (26).
We included travel costs (e.g., transportation for outpatient treatment) and medication costs (e.g., painkillers, antiemetics, dexamethasone). According to the ZIN Cost Calculation Manual, the price per kilometer by car was assumed as a proxy because this is the most commonly used mode of transport when traveling to care facilities (24).
After MR-HIFU, all patients were assumed to receive dexamethasone for 3 days to prevent recurrence, according to the practice of one of the selected centers. After EBRT, 4% of patients received dexamethasone for 5 days only as needed (27). Oxycodone was used as a surrogate drug to reflect analgesic use, an assumption made previously in similar modeling studies (16, 21, 28). A dose of 20 mg every 4 hours was assumed for persistent pain and partial pain relief. Finally, we accounted for the cost of antiemetics in a proportion of patients (i.e., 50%) with nausea and vomiting due to radiation therapy. These patients were estimated to be taking 8 to 16 mg of ondansetron daily for 3 weeks (29). Drug prices were taken from the ZIN reference list (30).
Although a societal perspective was taken into account, productivity loss due to patient inability to work and early death was not included because the patient entered the model at age 64 and bone metastases were an advanced-stage disease with a high mortality rate. Because there is no evidence that pain relief reduces the need for mobility aids, formal and informal healthcare costs were assumed to be equal in both groups and excluded.
Model output and validation
To validate the model, experts were invited to check the suitability of the model concept and input data (i.e., face verification). Additionally, we performed cross-validation by comparing the assumptions applied in our model with those of similar models (16, 21) and by comparing model outputs with results from other modeling studies (16, 21) and trial-based cost-utility studies (15). Online Resource 3(31,32,33). The validation effort followed the ‘Assessing the validation status of health economic decision models’ (AdViSHE) checklist and is detailed in Online Resource 3 (34).
Model outcomes were lifetime costs, QALYs, and incremental cost-effectiveness ratios (ICERs). Net monetary benefit (NMB) was calculated taking into account a willingness to pay (WTP) threshold of €80,000 per QALY gained. WTP thresholds were determined by calculating the disease severity index for bone metastases (11). A detailed description of the condition and calculation of the disease severity index is provided in Online Resource 4. We also calculated the incremental NMB, which represents the difference in NMB between alternative interventions, with a positive incremental NMB indicating that the intervention is cost-effective compared to the next cheapest alternative (35).
Sensitivity analysis and subgroup analysis
In deterministic sensitivity analysis (DSA), we varied the parameters one by one to evaluate their impact on ICER. Parameters tested in DSA were response rate and fracture risk to MR-HIFU, probability of pain recurrence after MR-HIFU, any utility value, and cost of treatment of MR-HIFU, EBRT, and pathological fracture. The cost of MR-HIFU varied depending on the best and worst case scenarios as a result of microcost calculations.
Probabilistic sensitivity analysis (PSA) was performed to test the overall uncertainty of model results. In PSA, the model was run 10,000 times with all parameters varying simultaneously within an a priori defined distribution. Probabilities and utilities were specified as beta distributions, and costs were specified as gamma distributions. Increases and decreases in utility value were assigned a lognormal distribution ( 36 ).
Mortality rates are expected to vary between metastatic cancers depending on the primary disease. To examine the impact of patient heterogeneity on model outcomes, we performed subgroup analyzes for three patient subgroups: metastatic breast cancer, metastatic prostate cancer, and metastatic lung cancer.
