A Dose-response Relationship for the Incidence of Radiation-related Heart Disease

Reviewer: Samuel Swisher-McClure, MD
The Abramson Cancer Center of the University of Pennsylvania
Last Modified: November 2, 2010

Authors: S. C. Darby, D. Brønnum, C. Correa, M. Ewertz, G. Gagliardi, B. Gigante, P. McGale, A. Nisbet, C. Taylor, P. Hall.
Affiliations: University of Oxford, Oxford, United Kingdom. Aalborg Hospital, Aalborg, Denmark. University of Odense, Odense, Denmark. Karolinska University Hospital, Stockholm, Sweden. Karolinska Institutet, Stockholm, Sweden. Royal Surrey County Hospital, Guildford, United Kingdom.


  • Post-operative radiation therapy performed either as part of breast conservation therapy (BCT) or after mastectomy in the setting of locally advanced disease, has been shown to significantly improve both local control and overall survival for women with breast cancer (Clarke et al. Lancet 2005).
  • However, long term results of randomized trials have also shown excess cardiac mortality in patients receiving radiation therapy particularly in older trials (Clarke et al. Lancet 2005).
  • Little is known about the effect of radiotherapy on incident heart disease or about the extent to which modern radiotherapy regimens, which tend to deliver lower cardiac doses, may still increase the risk.
  • In the current study the authors used population-based disease registries that have been available in Denmark and Sweden since the mid-1970s to relate mean cardiac dose to risk of developing heart disease.
  • That data was then used to develop a dose-response curve related to risk of cardiac disease (fatal or non-fatal) with mean heart dose delivered.


  • This was a retrospective case-control study using population based disease registries.
    • Cases were defined as women receiving radiation therapy for breast cancer who subsequently developed either fatal or non-fatal heart disease.
    • Control subjects were women who had also received radiation therapy for breast cancer and had no reported heart disease
    • Cases and Controls were matched for age, and country in which they received treatment.
  • Heart disease was defined as a hospital discharge code of myocardial infarction, a surgery code of coronary revascularization, or a death certificate with ischemic heart disease as underlying cause.
  • For both cases and controls, radiotherapy treatment charts were obtained and categorized according to regimen and laterality.
    • Treatment regimen varied substantially and included treatments delivered using Cobalt-60, Electrons, Orthovoltage equipment, or Megavoltage photon treatment.
  • Each regimen was reconstructed using a technique based on virtual simulation and computed tomography based 3-dimensional treatment planning with contouring of the heart and three main coronary arteries.
    • Of note, this virtual simulation and treatment planning was performed using a CT scan from a “typical” patient rather than a CT scan from the individual patient.
  • Mean cardiac doses were calculated using dose-volume histograms.


  • There were 1,018 cases of heart disease identified in irradiated breast cancer patients.
  • The authors identified 1,436 irradiated breast cancer patients who had not yet developed heart disease as controls.
  • Mean Cardiac dose was successfully estimated in 98% of the study subjects using the technique described.
  • The mean heart dose in left-sided breast cancer was 6.6 Gy (range, 0.0-20.4 Gy), while in right-sided breast cancer, it was 2.9 Gy (range, 0.0-27.7 Gy).
  • Mean heart dose was higher in subjects defined as cases compared to controls (5.3 Gy vs. 4.5 Gy, p < 0.0001) which was an expected finding.
  • The risk of heart disease was found to be 27% higher in left-sided than in right-sided breast cancer (95% confidence interval 7-50%, 2 p = 0.006).
  • When the estimated mean heart dose for each individual woman was considered, relative risk of heart disease was found to increase with dose in a linear fashion.
    • Compared with women whose estimated mean heart dose was < 5 Gy, women with estimated doses 5-14 and 15+ Gy had heart disease risks that were increased by 15% and 108%, respectively.
    • On average, there was a 4% increase in heart disease risk per 1 Gy increase in mean heart dose (95% confidence interval 2-6%; 2 p = 0.0002).

Author's Conclusions

  • This study has established the first dose-response curve relating mean cardiac dose to estimated risk of heart disease in patients undergoing radiation therapy for breast cancer.
  • The relative risk of heart disease was found to increase with dose in a linear fashion and there was no identifiable threshold dose associated with increased risk of heart disease
  • Dose-response relationships for radiation-induced heart disease that are free from bias can be derived from breast cancer patients who were irradiated in the past.
  • Such dose-response relationships may, in the future, be helpful in determining the likely risk of heart disease for patients with a variety of cancers where radiotherapy may involve some cardiac exposure.

Clinical Implications

  • This study was a retrospective case control study using population based disease registries to estimate a dose response relationship between mean cardiac dose and risk of heart disease among women receiving radiation therapy for breast cancer.
  • The results of the study provide further data that may be helpful in estimating a woman's subsequent risk of heart disease after radiation therapy.
  • The authors found that there was no threshold cardiac dose at which point risk of heart disease was increased, which suggests that any measurable cardiac dose will increase risk of heart disease to some degree.
  • This information could be helpful to consider when performing radiation treatment planning for such patients, and presents an opportunity for primary or secondary prevention of heart disease in women who receive significant cardiac dose despite optimal radiation treatment planning.
  • Limitations of the study include:
    • The study was retrospective in nature and used population based disease registries in order to identify patients with heart disease. Disease registry data provides somewhat limited information about any potential other events or conditions that may have been associated with the development of heart disease in these patients.
    • Case and control subjects were not matched for cardiac risk factors such obesity, tobacco use, hyperlipidemia, hypertension, or diabetes mellitus. There is potential that the two arms may have been imbalanced with respect to these factors.
    • Cardiac dose was estimated by re-creating a treatment plan through a virtual simulation technique using a CT scan from what the authors defined as a "typical" patient rather than from the individual subject. Individual patient anatomy can vary quite substantially and therefore calculations of mean cardiac dose using this method are likely to be inaccurate in many patients.
  • In summary, the study does establish a dose-response curve relating mean cardiac dose to subsequent risk of heart disease. However, the limitations of the study mentioned above may limit its wide applicability for all patients.