Effects of setup errors and shape changes on breast radiotherapy.

PURPOSE

The purpose of the present study was to quantify the robustness of the dose distributions from three whole-breast radiotherapy (RT) techniques involving different levels of intensity modulation against whole patient setup inaccuracies and breast shape changes.

METHODS AND MATERIALS

For 19 patients (one computed tomography scan and five cone beam computed tomography scans each), three treatment plans were made (wedge, simple intensity-modulated RT [IMRT], and full IMRT). For each treatment plan, four dose distributions were calculated. The first dose distribution was the original plan. The other three included the effects of patient setup errors (rigid displacement of the bony anatomy) or breast errors (e.g., rotations and shape changes of the breast with respect to the bony anatomy), or both, and were obtained through deformable image registration and dose accumulation. Subsequently, the effects of the plan type and error sources on target volume coverage, mean lung dose, and excess dose were determined.

RESULTS

Systematic errors of 1-2 mm and random errors of 2-3 mm (standard deviation) were observed for both patient- and breast-related errors. Planning techniques involving glancing fields (wedge and simple IMRT) were primarily affected by patient errors (∼6% loss of coverage near the dorsal field edge and ∼2% near the skin). In contrast, plan deterioration due to breast errors was primarily observed in planning techniques without glancing fields (full IMRT, ∼2% loss of coverage near the dorsal field edge and ∼4% near the skin).

CONCLUSION

The influences of patient and breast errors on the dose distributions are comparable in magnitude for whole breast RT plans, including glancing open fields, rendering simple IMRT the preferred technique. Dose distributions from planning techniques without glancing open fields were more seriously affected by shape changes of the breast, demanding specific attention in partial breast planning techniques.