Evaluation of margins and plan robustness for proton therapy of unilateral tonsil cancer.

Abstract: During proton therapy both target volumes and healthy tissue, including organs at risk (OARs), receives radiation dose. Thus, radiotherapy is a trade-off between good target coverage and OAR sparing. For protons, most of the dose is deposited right before it is stopped, a phenomenon termed the Bragg peak. Beyond this point no dose will be deposited, which is an advantageous feature since it enables more OAR sparing. However, this feature also makes proton therapy sensitive to variations in patient position, uncertainties in dose calculation and geometry/anatomy changes. Geometrical margins are therefore added around the target volume to ensure proper coverage.  An evaluation of the margins and plan robustness for proton therapy of unilateral tonsil cancer was conducted in this study, where the endpoint was to further optimize the margins to account for the trade-off between target coverage and doses to OAR. Verification CTs were compared with the plan CT of seven patients using the software Elastix and MICE toolkit. Dose volume histograms (DVHs) were evaluated together with Hausdorff distances (HdD), target coverage and dose differences along the craniocaudal direction, and related to the patient vertebrae.  Tendencies for the need of larger margins caudally of the cervical vertebra C3 was concluded from the HdD and two patients were selected for replanning. Four new treatment plans were created for each patient in the treatment planning system Eclipse and the resulting proton dose distributions were evaluated in MICE toolkit. Two plans utilized a uniform uncertainty of 4 and 5 mm respectively, and two plans utilized a varying uncertainty around the CTV.  Since Eclipse only allows the user to evaluate and optimize a plan with uniform setup uncertainty, new structures for the CTV had to be created for the varying uncertainty before the optimization. Caudally of C3 added PTV margins of 2 and 3 mm were created. Thereafter these new structures were evaluated and optimized with a setup uncertainty of 3 mm.  The limited available data suggests that the treatment plans with varying margins shows favorable characteristics and may improve the treatment quality. Tendencies for improved balance between target coverage and OAR sparing could be seen for the plan with a PTV margin of 3 mm caudally of C3 and a setup uncertainty of 3 mm. However, more patients need to be included in the study before certain conclusions can be made.