ESTRO 37 Abstract book

S123

ESTRO 37

to the bulk of the tumor against an increased risk to miss a small part of the tumor. We will explore the balance between these competing effects for a spherical target volume and idealized 3D dose distributions, concentrating on two aspects inherent to the PTV: its geometrical rigidity (each volume element in the PTV is considered equally important, even though small geometrical errors are more likely that larger ones) and its dose-effect simplicity (increased/residual effect of dose above/below prescription level is not acknowledged). Also, the practice of margin expansion does not easily mix with the emerging dose painting by numbers, or biological dose painting initiatives. Cases in which the dose distribution notably changes as a result of geometric errors (e.g. in particle therapies) go beyond the PTV’s capabilities. Risks of PTV deprecation As the CTV and PTV margins simply add up, an overly tight CTV may be compensated for by a correct, yet overly safe PTV around it, and therefore go unnoticed. Substitution of this PTV by a method that more aptly addresses the geometrical errors, leading to a more concentrated dose, may reveal the underlying CTV error, and lead to reduced tumor control. Besides a tool to shape radiation dose distributions, the PTV is also used to evaluate a treatment plan. As long as the various implementations of probability based techniques differ in their principles and details, the plans generated by different initiatives are difficult to compare. Typically, an optimized plan will always outperform all other plans when evaluated for the quality it was optimized for (PTV coverage of a PTV-based plan, TCP of a TCP optimized plan). An objective method of plan evaluation and comparison in terms of tumor control (regardless optimization strategy) needs to be established, otherwise deprecation of the PTV might bring chaos. Practicalities A PTV-less technique has been developed to act as a drop-in replacement, providing the same minimum CTV dose confidence, but able to shape the dose to the individual patient anatomy. This leads to a clinical benefit, but increases the complexity of a customized radiotherapy chain. Commercial support would lower this threshold, however commercial partners need a guaranteed market before investments can be made. Conclusion While there is ample reason to replace the PTV, it will take a joint effort. SP-0241 Coverage probability based dose optimization - first clinical applications M. Alber 1 1 Heidelberg University Clinic, Department of Radiooncology, Stuttgart, Germany Abstract text The planning target volume (PTV) concept is a device to enforce an acceptable dose coverage of the clinical target volume (CTV) in treatment planning. It consists of two elements: an envelope volume and a dose prescription to every point of this volume. The prescription is unconditional: every point is treated the same. The coverage probability concept (CovP) of dose optimization makes the dose prescription conditional on the probability that the volume of interest occupies this point in a number of geometric scenarios. In other words, the concept does not alter the prescription dose for each point, but the strength with which this prescription is enforced during dose optimization. Points with low CovP are given up more readily when conflicting dose prescriptions demand. This is an attempt at quantifying the frequent practice to accept more readily cold spots in the outer shell of a PTV, especially in the vicinity of organs at risk. CovP brings the fuzzyness to the PTV concept that it always had in practice.

Strictly speaking, CovP ought only be applied for non- superficial tumours in a homogeneous density environment and for the compensation of systematic organ displacement- and setup errors. It fails where the PTV concept also fails, with tumour volumes extending to the patient surface, or with sizeable dose changes as a consequence of geometrical uncertainties. Since CovP can mediate PTV underdosages based on uncertainty probabilities, it is important to quantify these. Estimates about target volume shifts and setup errors can be derived for populations and individuals by repeated imaging. These error distributions are essential for shaping the dose gradients around targets correctly. One particularly well suited situation for the clinical application of CovP is that of simultaneous integrated boost (SIB) in pelvis or head-and-neck, especially to lymph nodes. Here, classic PTV-margin formulae result in a gross over-estimation of required margin around the boost volume, since they were derived for a 100-0% dose drop around the CTV, while in the SIB situation it is more likely a 100-85% dose drop. Further, the PTV-to-CTV ratio is particularly bad for small tagets, resulting in a huge amount of dose dumped in the patient for no other reason than to follow the dogma.First clinical and planning results with CovP in SIB for cervix and prostate patients indicate that substantial sparing of normal tissues, predominantly the bowel, can be achieved. Validation studies based on repeated CBCT images suggest that the loss in dose coverage of SIB volumes is marginal and occurs rarely. Treatment plans were obtained from research software or using gradually increasing dose prescriptions on a set of nested volumes to shape the dose gradients in a controlled, CovP-like fashion with conventional software. Thus, in practice CovP treatment plans boil down to a set of dose prescriptions for the PTV, that are more relaxed than usual, but are precisely adjusted to the geometrical uncertainties that are likely to be encountered. This makes CovP-based dose optimization arguably the smallest step away from strict PTV planning. SP-0242 Analysis and reporting of plan robustness F. Albertini 1 1 Paul Scherrer Institute PSI, Department of Radiotherapy, Villigen PSI, Switzerland Abstract text In conventional radiotherapy uncertainties due to motion and patient positioning are dealt with by defining adequate margins around the treated volume. For proton therapy, although for single field uniform dose (SFUD) plan, the concept pf planning target volume (PTV) can still be used, this is not valid anymore in case of highly modulated IMPT plans. This is mainly because the static dose cloud approximation (i.e. the assumption that the dose distribution in space does not change due to uncertainties) is violated (example shown in the picture). The effect of uncertainties on the (deterioration of) dose distribution has to be evaluated performing a robustness analysis. Examples on how to perform the robustness analysis will be presented (e.g. error bar distributions, dose volume histograms (DVHs) bands), also taking into account the impact of fractionation. For IMPT plans the robustness of a plan to uncertainties has to be guaranteed differently: the uncertainties should be explicitly accounted for during the optimization process. The robust optimization approach will guarantee that a plan is clinically acceptable even under the effect of specified uncertainties. However plan robustness comes often at the price of plan quality. Practically the challenge when planning is to find the optimal tradeoff between plan quality and robustness. The concept of a site specific robustness protocol (summarized by comparing the robustness of a given plan with that of a previously determined database) will be shown to be a