Model-based cardiac dose estimation in radiation treatment of left breast cancer

To develop a mathematical model for cardiac dose estimation for patients who have been treated for left-sided breast cancer without CT data.

After obtaining institutional review board approvals, 147 patients with left-sided breast cancer were selected that were treated supine with opposed tangents. The heart blocks from the tangential fields were removed and dose calculations were performed with 6 MV beams using an advanced algorithm. This study was performed with CT data with DRR to represent a radiographic simulator image of yesteryear treatments. The beam's eye-view images showing delineated breast, lung and heart were created to represent views on radiograph. The maximum heart distance (MHD) was recorded and irradiated heart area (AREA) was computed by combination of triangles and rectangles. Based on accurate 3D dose calculation, mean dose (D_mean) and V₁₀ to V₅₀ of heart were analyzed with respect to MHD and AREA for dosimetric parameters using linear and quadratic fit.

The treatment parameters calculated by MHD and segments using 2D radiographs were within 2% of the actual dosimetric parameters computed from the 3D planning system. The MHD and AREA vs D_mean, V₁₀, V₂₀, V₃₀ and V₅₀ showed very good correlation with linear model (R² > 0.91); however the correlation was significantly better with quadratic model (R² > 0.92). The analysis of the dosimetric error with our linear and quadratic model is remarkable within <3% error for most cases.

The proposed mathematical model for the cardiac dose estimation is accurate within ±3% using a radiograph without CT data. This provides avenues for patient pooling in future studies related to radiation dose and cardiac toxicity. These results will help in estimating cardiac dose analysis accurately from previous studies as well as in centers still using 2D planning.

The evidence of cardiac risk following radiotherapy continues to be one of the important considerations for the management of left-sided breast cancer patients. One of the problem in the estimation of dose-risk effects is the reconstruction of heart dose for pre-CT treatments. In this study, a simple mathematical model is presented that could estimate cardiac dose within ±3% in left breast cancer treatment from 2D radiograph where CT data do not exist.