Cardiac Structure Doses in Women Irradiated for Breast Cancer in the Past and Their Use in Epidemiological Studies

Incidental cardiac exposure during radiation therapy may cause heart disease. Dose-response relationships for cardiac structures (segments) may show which ones are most sensitive to radiation. Radiation-related cardiac injury can take years to develop; thus, studies need to involve women treated using 2-dimensional planning, with segment doses estimated using a typical computed tomography (CT) scan. We assessed whether such segment doses are accurate enough to use in dose-response relationships using the radiation therapy charts of women with known segment injury. We estimated interregimen and interpatient segment dose variability and segment dose correlations.

The radiation therapy charts of 470 women with cardiac segment injury after breast cancer radiation therapy were examined, and 41 regimens were identified. Regimens were reconstructed on a typical CT scan. Doses were estimated for 5 left ventricle (LV) and 10 coronary artery segments. Correlations between cardiac segments were estimated. Interpatient dose variation was assessed in 10 randomly selected CT scans for left regimens and in 5 for right regimens.

For the typical CT scan, interregimen segment dose variation was substantial (range, LV segments <1-39 Gy; coronary artery segments <1-48 Gy). In 10 CT scans, interpatient segment dose variation was higher for segments near field borders (range, 3-47 Gy) than other segments (range, <2 Gy). Doses to different left-anterior descending coronary artery (LADCA) segments were highly correlated with each other, as were doses to different LV segments. Also, LADCA segment doses were highly correlated with doses to LV segments usually supplied by the LADCA. For individual regimens there was consistency in hotspot location and segment ranking of higher-versus-lower dose.

The scope for developing quantitative cardiac segment dose-response relationships in patients who had 2-dimensional planning is limited because different segment doses are often highly correlated, and segment-specific dose uncertainties are not independent of each other. However, segment-specific doses may be reliably used to rank segments according to higher-versus-lower doses.