Session Item

Swelling of the breast during radiotherapy sometimes occurs in breast cancer patients. With the use of more conformal techniques like Intensity Modulated Radiotherapy Therapy (IMRT) or Volumetric Modulated Arc Therapy (VMAT) this can lead to under-dosing of the target volume. Common workarounds, like virtual bolus or “skin flash” methods, often are not robust enough to prevent this. However, the dosimetric consequences of swelling of the breast might be mitigated by using robust optimization.
In this study we investigated if there is a potential gain for robust optimization for elective breast radiotherapy.

Retrospectively, eleven patients with swelling of the breast with more than 0.8cm, as measured on the CBCT, were selected. The prescribed dose for the breast CTV was 40.05Gy in 15 fractions. A PTV margin of 5 mm was used. The clinically used plan was used for this study, which was an IMRT plan with two tangential  fields (6MV and/or 10MV). For each patient a non-robust plan, according to the clinical workflow with a skin flash method was made.
For creation of a robust plan, a new feature called simulated organ motion, was used. Four synthetic CT-scans were generated by deforming the breast CTV in different directions towards the air over a distance of 1cm on average. These synthetic CTs were used as scenarios during optimization, besides the original planning CT.
Evaluation of the planned dose was done on a repeat CT-scan, when available, or else on a synthetic CT-scan generated by deforming the CBCT to the planning CT. Both CT-scans were registered to the planning CT-scan with only translations and no rotations according how the patient would be treated. The difference between planned dose and evaluated dose of the coverage of the breast CTV V_95% (of the prescribed dose) and the D_1cc(Gy) (dMax) was evaluated. The difference of the V_95% between the planned and the evaluation dose, of both non-robust and robust plans, was tested with a Wilcoxon signed-rank test with a p-value of 0.05.

For eight out of eleven patients V_95% coverage with the robust plan was better than with the non-robust plan (P=0.026). The average difference between planned and evaluated dose of the V_95%  was -0.88% and 0.02% for the non-robust and robust plan, respectively. The differences for each patient are shown in figure 1. 
For five out of eleven patients with a non-robust plan the V_95% coverage decreased with more than -1% and only one out of eleven patients with a robust plan. In figure 2 an example is shown for which the coverage of a non-robust plan isn’t enough but for a robust plan it was.
The D_1cc increased with an average of 0.6Gy and 0.8Gy, for the non-robust and robust plan, respectively.

Organ motion-based robust optimization results in more robustness against swelling of the breast than using a skin flash method.