SU‐E‐T‐46: Evaluation of Different Cancer Risk Models on Second Cancer Risk Calculation for Conventional Breast Cancer Radiation Therapy

Abstract

Purpose: To determine the influence of different cancer risk models on second cancer risk when these models are applied to full‐body dose distribution of two conventional breast treatments computed by Monte Carlo. Methods: Two conventional breast treatments (2‐fields and 4‐fields) were simulated in a full‐body voxelised anthropomorphic phantom with a Monte Carlo model of the Siemens Primus 6MV linac. Dose distribution for each organ at risk was extracted. Linear risk models from the ICRP 103 and BEIR VII were applied to the average dose of specific organs. Both cancer incidence risk and cancer mortality risk were determined for the BEIR calculations. Risks were determined when considering all organs together and out‐of‐field organs exclusively. In addition to these linear models, cancer risks were determined with non‐linear dose risk models applied to the full body dose distribution yielding “Organ Equivalent Doses” (OED). for all models used, both the absolute risks were determined for the two treatments and the relative risk between the two treatment plans. Results: The radiation‐induced second cancer risk ranged from 0.12% to 145% for the 2‐field treatment and from 0.84% to 458% for the 4‐field treatment depending on the choice of models. The calculated risks can be much higher than the values reported by epidemiological studies (2–4%). The ratio of risks between the two treatments varied from 1.2 to 9.1. Conclusions: Risk models should be applied with great caution. Great uncertainties underlie absolute risk calculations yielding unrealistic risks; therefore many investigators use the ratio of risk to compare different treatments. However, the ratio of risk between two treatments may also greatly vary depending on the risk models used. This may explain inconsistent findings on the potential increase in risk of IMRT techniques compared to conventional treatments.This work was supported by the Swiss National Science Foundation FNRS 320000‐114101