Modeling heart and lung complication data in radiation therapy of the breast

Abstract

In breast cancer, radiotherapy represents the most effective adjunct to surgery to prevent loco‐regional recurrences. Serious side‐effects following breast cancer radiotherapy are both cardiac and pulmonary. In this thesis, complication and dose distribution data were modeled to quantitatively describe the complication probability after a given treatment. The relative seriality model was used for the parameterization. Excess long‐term cardiac mortality data for breast cancer patients treated with various radiotherapy techniques were derived from the randomized trials in Stockholm (1971–1976) and Oslo (1964–1972). Dose distribution data were reconstructed with a three‐dimensional treatment planning system. A dose‐response relationship describing the data was determined. This served as a model to predict the probability of excess cardiac mortality in a group of 100 stage I breast cancer patients planned for radiotherapy using a conventional technique. A subgroup of the patients had a significant risk of excess long‐term cardiac mortality, due to the irradiation of a relatively large volume of the heart; this could be decreased either by an extended blocking of the beams or by applying a treatment method based on intensity modulation. Radiation pneumonitis data were obtained from patients treated at the Radiumhemmet (1993–1994). Individual three‐dimensional dose distributions were available. Model parameters plus uncertainties were determined using the maximum likelihood method. The uncertainty in the dose‐response curve due to the parameter uncertainties was also considered. The resulting parameters were then used to calculate the pneumonitis probability in five groups of breast cancer patients treated with radiotherapy. The predicted values of pneumonitis were not significantly different from the clinical incidence in three of the groups.