In breast cancer radiotherapy, substantial radiation exposure of organs other than the treated breast cannot be avoided, potentially inducing second primary cancer or heart disease. While distant organs and large parts of nearby ones receive doses in the mGy–Gy range, small parts of the heart, lung and bone marrow often receive doses as high as 50 Gy. Contemporary treatment planning allows for considerable flexibility in the distribution of this exposure. To optimise treatment with regards to long-term health risks, evidence-based risk estimates are required for the entire broad range of exposures. Here, we thus propose an approach that combines data from medical and epidemiological studies with different exposure conditions. Approximating cancer induction as a local process, we estimate organ cancer risks by integrating organ-specific dose–response relationships over the organ dose distributions. For highly exposed organ parts, specific high-dose risk models based on studies with medical exposure are applied. For organs or their parts receiving relatively low doses, established dose–response models based on radiation-epidemiological data are used. Joining the models in the intermediate dose range leads to a combined, in general non-linear, dose response supported by data over the whole relevant dose range. For heart diseases, a linear model consistent with high- and low-dose studies is presented. The resulting estimates of long-term health risks are largely compatible with rate ratios observed in randomised breast cancer radiotherapy trials. The risk models have been implemented in a software tool PASSOS that estimates long-term risks for individual breast cancer patients.

在乳腺癌放射治疗中，除了接受治疗的乳房以外的器官的大量辐射暴露是不可避免的，可能会诱发第二原发癌或心脏病。虽然远处器官和附近器官的大部分接受的剂量在 mGy-Gy 范围内，但心脏、肺和骨髓的小部分通常接受高达 50 Gy 的剂量。现代治疗计划允许在这种暴露的分布方面具有相当大的灵活性。为了针对长期健康风险优化治疗，需要对整个广泛的暴露范围进行基于证据的风险评估。在这里，我们因此提出了一种方法，将医学和流行病学研究的数据与不同的暴露条件相结合。将癌症诱导近似为局部过程，我们通过在器官剂量分布上整合器官特异性剂量-反应关系来估计器官癌风险。对于高度暴露的器官部位，应用基于医学暴露研究的特定高剂量风险模型。对于接受相对低剂量的器官或其部分，使用基于辐射流行病学数据建立的剂量反应模型。在中间剂量范围内加入模型导致在整个相关剂量范围内的数据支持的组合的、通常非线性的剂量反应。对于心脏病，提出了与高剂量和低剂量研究一致的线性模型。由此产生的长期健康风险估计与随机乳腺癌放射治疗试验中观察到的比率基本一致。