ESTRO 37 Abstract book

ESTRO 37

S479

painting and how they can be accounted for using robust optimization. Material and Methods For this purpose, robust dose-painting was developed and implemented in our fully automated clinical TPS, such that the expected TCP is optimized for both uncertain dose-response relations and Gaussian systematic (∑=3 mm) and random (σ=5.5 mm) set-up uncertainties. The TCP of the tumor was modeled as the product of the TCPs over all voxels, which were described by sigmoid shaped dose-response relations with uncertain parameters that followed Gaussian distributions. The effect of uncertainty was modeled for different geometrical situations and for a range of TCP parameters. The geometrical situations consisted of 3 phantoms (Fig 1) and a patient case (Fig 2). The TCP parameters varied with ΔTD50 (difference in TD50 between the resistant and sensitive regions) ranging from 0 to 35 Gy. Optimizations were performed for different levels of uncertainty, expressed by the SD of the TD50 (TD50 σ ), that varied from 0 (no uncertainty) to 10 Gy. For a fair evaluation, dose-painting was compared to non-specific dose escalation with the same dose constraints.

Conclusion In the absence of uncertainty, the gains of dose-painting are highly variable (between 1- 29%) depending on the orientation of the resistant and sensitive tumor regions. Uncertainty in dose response relations strongly reduces the expected gains, and can only be partly compensated by robust dose-painting. To achieve with dose-painting a benefit of > 10% in TCP, the TD50 σ should be below 10 Gy in for the most favorable geometry and lower otherwise. Due to the choice of TCP parameters in this study, the gains presented here should be interpreted as best case gains. PO-0900 Spatiotemporal fractionation schemes for liver stereotactic body radiotherapy J. Unkelbach 1 , D. Papp 2 , M. Gaddy 2 , N. Andratschke 1 , T. Hong 3 , M. Guckenberger 1 1 Universitätsspital Zürich, Radiation Oncology, Zürich, Switzerland 2 North Carolina State University, Mathematics, Raleigh, USA 3 Massachusetts General Hospital, Radiation Oncology, Boston, USA Purpose or Objective Most radiotherapy treatments are fractionated because normal tissues can tolerate much higher radiation doses if the total dose is split into several fractions. However, more fractions also require that the total dose must be increased to maintain the same level of tumor control. Hence, fractionation decisions face the trade-off between increasing the number of fractions to protect normal tissues and increasing the total dose to maintain tumor control. In that sense, the ideal treatment would hypofractionate in the tumor and simultaneously split the dose evenly into many fractions in normal tissues. While this may appear impossible, it turns out that some degree of hypofractionation in the tumor can be achieved along with near-uniform fractionation in normal tissues. This is possible by delivering distinct dose distributions in different fractions. We refer to this concept as spatiotemporal fractionation. Material and Methods

Results For the most favorable phantom geometry (Phantom C), without uncertainty in dose-response, the maximum gain of dose-painting (ΔTCP) was 29% and occurred at a TD50 of 20 Gy. Uncertainty reduced ΔTCP to 22% (TD50 σ =5 Gy) and 7% (TD50 σ = 10 Gy) , to even a negative ΔTCP of -3% in the least favorable geometry (phantom B, TD50 σ = 10 Gy). Including uncertainty directly in the robust IMRT optimization could restore the gains but by only 2% to 4%. For the patient case, dose-painting led to a similar plan as non-specific dose escalation and therefore only a modest gain in ΔTCP < 1% was observed, independent on the level of uncertainty (Fig. 2).