The accuracy of a delivered photon treatment is more of an isocentricity question in that if the isocenter is correctly aligned during setup, the delivered dose will still be comparable to the planned dose even if the patient’s body is not in exact alignment. In protons, the isocenter alignment is important as well, but just as important is the patient’s body alignment as minor depth changes in the beam path can under/overshoot the SOBP. Both the position of the physical body and the internal anatomy must be thoroughly accounted for. For example, the vac-lock is great for reproducing the isocenter, but not so much for reproducing the body as the patient’s skin may deform day-to-day especially if the patient is larger and/or elderly. Additionally, physiologic changes that are present on a planning CT, but not during treatment can affect the SOBP range (e.g., filled sinus for simulation vs empty sinus for treatment). It is important to understand the qualities that makes a photon plan robust, i.e., resistant to uncertainties, may be insufficient for a proton plan.
Multi-field optimization (MFO) allows for the optimization of multiple beams simultaneously. The optimizer understands the interplay between fields and decides the optimal spot weight based on which field is best suited to deposit dose to a particular region of the target. Prior to the development of robust optimization, MFO was not as frequently used because treatment uncertainty, specifically proton range uncertainty, was not factored into the calculation. Thus, TPSs at the time would produce plans that were only representative of planning CT conditions (nominal dose). The development of robust optimization allowed for users to plan while maintaining adequate coverage for the worst case scenarios, i.e., scenarios where uncertainties are accounted for. The TPS will initially calculate for the nominal dose and then put the dose through iterative perturbations which are instances involving varying levels of range and setup uncertainty.
To account for a potential mismatch between simulation and treatment setup, a setup uncertainty is assigned during optimization for each field individually (X, Y, or Z direction) or for the plan as a whole. Accounting for setup uncertainty can be pictured as the optimizer shaking the CT as dose distributes, and so the setup error will be simulated for in six directions to introduce HU variations. This setup uncertainty value can vary depending on the site and/or institutional preference. Patients treated for CSI will benefit from field-specific setup uncertainty to ensure that offsets in setup will still result in a smooth dose gradient between the upper and lower fields. The “matching” of fields is performed by the shallow gradient technique where shallow lateral gradients of adjacent fields are matched to achieve a uniform dose across the target. This technique is also used for large volumes that require that separate fields to encompass an entire target volume such as sarcoma, H&N, and pelvis cases.
There is also uncertainty associated with the proton beam range that arises from factors such as the accuracy of CT-to-HU number conversion-to-proton stopping power conversion and dose calculation. Normally, a stopping power value is assigned to a voxel based on its HU number that is obtained from a single calibration curve (HU_x = stopping power value_x). However, determining the HU number from the CT image is not a perfect process, and so separate voxels may be labeled as the same HU despite having different stopping power values in actuality. This issue is exacerbated in proton therapy due to its sensitivities over photons, and may result in a dose calculation that is not reflective of reality.
The range uncertainty value is determined by taking a percentage of the physical range which is accepted amongst the proton community to be 3.5%. Given these uncertainties ( range and setup), a rule of thumb is to use a beam arrangement with as short of a beam path to the target as possible while avoiding a beam that is directed through heterogeneities and/or an unstable surface. Furthermore, as the proton nears the end of its range, there exists the possibility that the BED may be deposited two to three millimeters deeper than what the TPS displays by LET enhancement. Thus, avoiding beams that stop just before serial critical organs is recommended and/or using orthogonal fields to spread the distal uncertainty as the uncertainty of each field will now be associated with half or less of the prescription dose.
|Khan Chapter 27 Summary|