Session Item

Proton therapy has made a large impact on the concept of planning target volume (PTV) margins. The concept to expand the clinical target volume (CTV) with a margin to achieve CTV coverage in case of patient setup errors has some clear limitations. The assumption that errors can be described as a shift of the CTV relative to a static dose cloud is not always accurate. An example is in photon treatments where patient setup errors result in dose distribution changes in build-up regions at low to high density interfaces. In proton therapy, the inaccuracy of the static dose cloud assumption results from the dependence of proton range on material in the beam path, e.g. caused by patient shifts. Furthermore, range errors arise from the inaccuracy of conversion of planning CT Hounsfield units to proton stopping powers. In intensity modulated proton treatments, steep inter-beam dose gradients can exist centrally in the CTV, and dose variations that originate from range errors cannot be handled with a CTV to PTV margin.

A solution is robust planning, i.e., explicitly taking into account uncertainties during robust optimization and robustness evaluation of treatment plans, in so called treatment scenarios. Robust planning avoids the static dose cloud assumption by performing a dose calculation in each treatment scenario and assessment of dose to the CTV in these scenarios. As most proton centers clinically introduced robust planning in recent years, it is clear that PTVs are not needed in proton therapy. Although conceptually straightforward, robust planning introduces difficulties, e.g. in the scenario selection, evaluation method of the scenario dose distributions and plan acceptance criteria. The purpose of this talk is to give background information and provide practical choices in the transition from PTV based planning to robust treatment planning. In line with van Herks margin recipe based on normal distributions of systematic and day-to-day patient setup errors, worst case scenarios can be determined at a 90% confidence level. So, even after abandoning PTV margins as such, the PTV margin size is still being used in treatment scenarios in robust planning.

A difficulty in robust planning is that the treatment scenarios result in numerous dose distributions per treatment plan that need to be assessed. A solution for the determination of the (near) minimum CTV dose under uncertainties is to review the voxel-wise minimum CTV dose. The voxel-wise minimum dose shows for each voxel the minimum dose over all scenarios. It can be shown that in radiotherapy treatments where the static dose cloud assumption is valid, a voxel-wise minimum dose evaluation is equivalent to a PTV minimum dose evaluation. This is an important observation for the consistency between proton treatments and (historical) photon treatments planned on a PTV. Plan acceptance criteria determined for the PTV minimum dose were found to require only minor adjustments (e.g. 1%) for application to voxel-wise minimum dose evaluations.  

So, even though we don’t need PTV margins in proton therapy, robust planning methods that replaced PTV planning include setup errors with magnitudes in agreement with PTV margins. Voxel-wise minimum dose evaluations are highly correlated to PTV dose evaluations, including corresponding plan acceptance criteria. Despite major changes introduced with robust planning, links to PTV planning remain.