Dose-response relationship for breast cancer induction at radiotherapy dose

Abstract

Purpose: Cancer induction after radiation therapy is known as a severe side effect. It is therefore of interest to predict the probability of second cancer appearance for the patient to be treated including breast cancer.

Materials and methods: In this work a dose-response relationship for breast cancer is derived based on(i) the analysis of breast cancer induction after Hodgkin's disease,(ii) a cancer risk model developed for high doses including fractionation based on the linear quadratic model, and(iii) the reconstruction of treatment plans for Hodgkin's patients treated with radiotherapy,(iv) the breast cancer induction of the A-bomb survivor data.

Results: The fitted model parameters for an α/β = 3 Gy were α = 0.067Gy-1 and R = 0.62. The risk for breast cancer is according to this model for small doses consistent with the finding of the A-bomb survivors, has a maximum at doses of around 20 Gy and drops off only slightly at larger doses. The predicted EAR for breast cancer after radiotherapy of Hodgkin's disease is 11.7/10000PY which can be compared to the findings of several epidemiological studies where EAR for breast cancer varies between 10.5 and 29.4/10000PY. The model was used to predict the impact of the reduction of radiation volume on breast cancer risk. It was estimated that mantle field irradiation is associated with a 3.2-fold increased risk compared with mediastinal irradiation alone, which is in agreement with a published value of 2.7. It was also shown that the modelled age dependency of breast cancer risk is in satisfying agreement with published data.

Conclusions: The dose-response relationship obtained in this report can be used for the prediction of radiation induced secondary breast cancer of radiotherapy patients.

Figure 1

Plot of the modelled excess absolute risk (solid line) to the epidemiological data of Travis et al [8]for α/β = 1 Gy. The dashed line represents the LNT-model for breast cancer with the corresponding error [10].

Figure 2

Plot of the modelled excess absolute risk (solid line) to the epidemiological data of Travis et al [8]for α/β = 3 Gy. The dashed line represents the LNT-model for breast cancer with the corresponding error [10].

Figure 3

Plot of the modelled excess absolute risk (solid line) to the epidemiological data of Travis et al [8]for α/β = 5 Gy. The dashed line represents the LNT-model for breast cancer with the corresponding error [10].

Figure 4

Plot of the modelled age dependence of the standardized incidence ratio (normalised to the De Bruin data) as the solid lines for the age at treatment groups <20, 21-30, 31-40 and 41-50, respectively. The corresponding epidemiological data from De Bruin are plotted as the symbols together with the corresponding 95% confidence interval.