Session Item

Diaphragm motion can be a significant factor influencing target coverage in irradiation of esophageal tumors. Proton therapy is potentially more sensitive to anatomical variations than photon radiotherapy. In this study we evaluated the dosimetric quality of proton treatments, taking into account the patient anatomy changes during the treatment course and the correlation with diaphragm movement changes.

In this study, 35 patients treated with proton therapy for esophageal cancer using a 23x1.8Gy fractionation scheme were evaluated. All patients received a 4D CT planning CT (pCT), as well as weekly repeat 4D CTs (reCT). The clinical target volume (CTV) was delineated on each phase of the pCT, and the CTVs were combined to an internal target volume (ITV). Three posterior beams with a separation of 20 degrees were used, with in some patients an additional fourth beam from anterior direction. Dose calculations were performed on the average CT. The proton plans were robustly optimized with 5mm setup uncertainty and 3% range uncertainty, combined with a fixed margin of 3mm around the ITV. During treatment, all plans were re-evaluated on the reCTs using a robust evaluation with 2mm setup uncertainty and 3% range uncertainty, accounting for residual errors. If necessary, plans were adapted based on evaluation on the reCT. After deformable registration to the pCT, for each robustness scenario a dose summation was created, deforming dose from reCT to pCT. Robust dose summation was performed by summing each scenario over all reCTs. Subsequently, voxel-wise minimum and maximum summed dose distributions were derived. Diaphragm positions in inhale and exhale phases of the 4D CT were extracted by measuring the diaphragm-lung intersection at the axial center of each lung. Therefrom a baseline diaphragm position (i.e. the average between inhale and exhale) and a diaphragm amplitude (i.e. the difference between inhale and exhale diaphragm position) were derived. 

Diaphragm average amplitude [standard deviation (SD)] on the pCT was 1.4 [0.8] cm. The amplitude change during treatment evaluated on each reCT was on average 0.2 [0.5] cm. The average baseline shift observed on the repeat CTs was -0.2 [0.5] cm (Figure 1).  
The average difference in voxel-wise minimum ITV D98% [SD] on repeat CTs with respect to the planning CT was -2.7 [9.4] %. We found a weak negative correlation between changes in ITV D98% and baseline shift (Pearson’s R = 0.54) and no correlation with changes in amplitude. After robust dose summation, the ITV D98% of the summed dose distributions is >94% for 32 of 35 evaluated patients, in accordance with the 90% aimed at in the margin recipe (Figure 2).

Despite changes in diaphragm amplitude up to 2.5cm and in baseline position up to 2 cm during patient treatment, robust dose summation showed that ITV coverage in proton therapy for esophageal cancer remained sufficient using a posterior beam setup.