Breast cancer (BC) is the most frequently diagnosed cancer type in women worldwide, accounting for more than one quarter of all newly diagnosed female cancers (1, 2). Early detection of breast cancer is enhanced by nationwide accessibility of screening programs for women at risk (3). As a result, breast cancer is diagnosed more frequently at an early stage, leading to a beneficial prognosis. Concurrently, advances in radiotherapy have led to a significant reduction in local recurrence and BC mortality (4). Radiation is administrated to over half of all BC patients (5), and therefore, it is likely that radiotherapy contributes to high survival rates. In developed countries, such as Belgium, the 5-year relative survival rate is around 91% (6).

However, radiation from radiotherapy is not limited to tumor tissue. Despite the use of state-of-the-art radiation techniques, healthy organs -such as the heart- are exposed to minor radiation, leading to inflammatory responses in this tissue (7). These radiation-induced cardiotoxic effects expose BC survivors to an increased long-term risk for cardiovascular (CV) diseases (7–9). These events may have an additional impact on the health-related quality of life (HRQoL), in particular as HRQoL might already have been influenced by BC treatment prior to a potential CV event (10). Additionally, CV events are associated with an increased CV-related mortality risk (11). A meta-analysis demonstrated that the rate ratio for CV mortality was 1.3 (SE 0.09; 2p=0.0007) in BC patients receiving RT compared to controls undergoing the same treatment without RT. Although this meta-analysis provided evidence on benefits from reduction in breast cancer mortality by RT during treatment, the researchers demonstrated the moderate increase in death from CV events in particular in the first and second decade after treatment (4). Other studies indicated similar results (12). However, these studies mainly addressed CV mortality. Although research indicated that the economic burden of cardiovascular diseases soars compared to the health expenditure for other diseases, estimated around 9% of total health care expenditure across European countries (13), the long-term health outcome and costs of cardiovascular morbidity in breast cancer survivors are less considered. The application of health economic evaluation techniques poses an opportunity in order to model these long-term CV effects as it “provides a framework to make best use of clinical evidence through an organized consideration of the effects [ … ], health care costs, and other effects that are regarded as valuable” (17). Formerly, health economics research in the field of oncology have largely focused on costs and effects related to primary BC treatment rather than exclusively on long-term CV mortality and morbidity (14–16).

Most of these health economic studies did not include long-term cardiotoxic costs and effects in their analysis. There is still a lack in the more precise quantification of the societal burden of radiation-induced CV diseases. Our study aims to quantify the additional burden of radiation-induced cardiovascular diseases in BC survivors.

Our study focused on the incremental costs and effects attributed to radiation-induced cardiotoxicity. The analysis showed that radiotherapy is associated with four additional non-fatal and two additional fatal CV events in a cohort of 1,000 women over a time horizon of 20 years. In effect, the impact of taking radiation-induced cardiotoxic effects in account remains limited to an incremental cost of €275.10 per woman and incremental utility decrement of 0.017 per woman over 20 years.

Lundkvist et al. performed a CUA of proton beam therapy compared to conventional radiotherapy. As in our study, they choose non-fatal and fatal CV events as outcome measure in their Markov model. They expected 11.8 non-fatal CV events, and 3.3 fatal CV events per 100 55-year old women over a time horizon of 23 years (47). In our study, we observed, respectively, 9.6 events and 1.6 events in this age group over 20 years. Hence, the analyses produced a close match, supporting the assumptions in our model.

In our analysis, incremental costs increased with age, resulting in highest costs for older age groups. In effect, adjusting for age-specific incidence rates from different countries might reflect in higher costs for younger women. Direct costs were the main cost contributor in total costs. This is in line with results reported by Wilkins et al. who stated that direct health care costs contributed for over half of total CV-related costs in Europe (48). In our study the explanation lies in the increasing risk for non-fatal CV events at older age, accumulating in high inpatient care costs. The comparison between Lundkvist et al. and our results demonstrated earlier that the use of different data sources and extrapolating results over longer time horizons might reflect in -although small- differences between studies. As in our study, it remains a challenge to correctly extrapolate events over a long time horizon.

Our study is subject to several limitations. Firstly, it is important to notify that cardiotoxicity is not solely generated by radiation during BC treatment. Many studies in the field of cardio-oncology use the term ‘treatment-induced cardiotoxicity’ (49–51). Therefore, it could be considered a narrow approach to isolate radiotherapy from other cardiotoxic treatment factors (e.g. anthracyclines and HER-2 antagonists) (50, 52). On the other hand, in recent years the use of statins was widely introduced in order to reduce cardiovascular risk (53). Secondly, we assumed that a time horizon of 20 years would suffice in order to quantify radiation-induced cardiotoxicity since a reference paper in this field suggested an increased risk starting five years up until three decades after treatment (23). Hence, one might discuss a lifetime horizon in this matter is more appropriate. Especially for younger age groups in our model, the probability of an event after the analytic horizon of 20 years is plausible. In analogy to the EUROACTION study (54), we foresaw difficulties in modelling events in younger women because various covariables may intervene in longer time horizons. Especially in models concerning CV diseases, the probability of an event depends on several dependent parameters. Briggs et al. have suggested to introduce covariance in probabilistic analyses via Cholesky decomposition. Unfortunately, in most cases information is lacking regarding the underlying covariance structure to incorporate interdependency in the analyses. Therefore, it is common to treat all parameters as independent (55). We used numerous studies and data sources as vehicles for our economic evaluation. The possibility to combine data from different sources poses major opportunities for health economics, but this also increases the degree of uncertainty in the model (17). Thirdly, it is noteworthy that -though we strived to incorporate as much relevant evidence- some parameters were not included, possibly leading to an oversimplification of some aspects in the model. For example, the SCORE equations were updated in June 2021 allowing to add even more parameters to the prediction such as HDL-cholesterol and pre-existing diabetes mellitus (56). Also, the analyses of Darby et al. suggested that the risk for radiation-induced cardiotoxicity increased over time (23) whereas in our analyses the risk remains 4.1% per Gy. However, we employed the method which was proposed by Mulliez et al. to make estimations on the cardiovascular risk in breast cancer survivors (26). As argued by Briggs et al. highly complex model structures are not always more appropriate and simplified assumptions allow researchers to keep the model manageable (55). Finally, the authors would like to emphasize that results from health economic modelling need to be considered in the light of a specific context, meaning the cohort models contain various sources of evidence (e.g. transition probabilities, cost resources and health outcome measures) (17). Generalizability is only recommended to populations with a resembling context. Moreover, we intended to adopt a societal perspective, though we limited our cost resources to direct medical costs and indirect costs due to production loss. As with most CUA, it could be argued that we omitted some relevant cost resources (such as transportation costs, domestic help,…) or even costs in other sectors outside health care.

To our knowledge, our study is the first study which uses health economic modelling techniques to estimate the long-term costs and effects of radiation-induced cardiotoxicity. The use of these techniques can serve as a framework for the evaluation of the impact of a health risk on a population (57). Therefore, our results are applicable to future health economic research in the field of cardio-oncology. A noteworthy element in our analyses is the extent of MHD as influential parameter. Several studies revealed that heart doses were extensively higher in the past compared to current MHD (58, 59). Due to the awareness of long-term comorbidities in BC patients, substantial improvements on radiation therapy technology and protocols led to a reduction of the radiation dose to the heart (25). To date, research emphasizes the importance of continuing to invest in technological advances in heart-sparing techniques in order to further reduce the MHD (60). Our analyses showed that current radiation doses already evoke negligible incremental costs and disutility at population level. These findings emphasize the importance of previous technological innovations. Incremental costs and disutility must have been significantly higher in the past since MHD decreased substantially over the past decades. Hence, the cost of past investments in order to achieve current radiation dosages appears to be justified when considering the gains from cost and disutility reduction resulting from cardiotoxicity in BC survivors.

In conclusion, our analysis quantified more in detail the impact of radiation-induced cardiotoxicity in BC survivors. Our findings demonstrated that with current radiation techniques, minor overall mean incremental costs and disutility are to be expected over a time horizon of 20 years after primary BC treatment. Major research investments have been made in order to substantially decrease the MHD. Our findings highlight the importance of these investments when considered against the reduced costs and disutility from cardiovascular events as a result from these innovations. On the other hand, our analyses might also open a debate on potentially high opportunity costs of future investments since the expected marginal gains from further reducing the MHD may no longer outweigh the research budget.

All data referring to parameters is available in the article or supplementary material. Further inquiries should be directed to the corresponding author.

EK: primary investigator and author (first draft preparation). AW: health economist and author (review and editing). MR: radiation oncologist (medical expertise) and senior author (review and editing). KP: health economist, principal investigator and senior author (review and editing). All the authors contributed to and agreed with the submitted manuscript.

This research was established within the Strategic Research Programme (zwaartepunt, SRP 53, 2019-2024) ‘Societal Benefit of Markerless Stereotactic Body Radiotherapy: a Statistical Support based on Quantitative Imaging’ (SMARTQI of the Vrije Universiteit Brussel).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.