ESTRO 2024 - Abstract Book

S3519

Physics - Dose prediction, optimisation and applications of photon and electron planning

ESTRO 2024

Australia. 11 Liverpool Hospital, Radiation Oncology, Liverpool, Australia. 12 University of Wisconsin, School of Medicine and Populatoin Health, Madison, USA

Purpose/Objective:

Focal boost prostate radiotherapy encompasses a prescription of escalated doses to intra-prostatic lesions (IPLs). The prescription is conventionally applied homogeneously to the boost volume. However, the heterogeneity of the cancer cell distribution and the radiobiological impact of that heterogeneity within the IPLs is not considered for current focal boost radiotherapy. To enable a heterogeneous IPL boost that incorporates tumour spatial characteristics, this study aimed to develop a pipeline to generate optimal heterogeneous dose prescriptions that lead to maximum tumor control based on the voxel-wise distribution of tumor cells. The practical limits and robustness of treatment planning for both homogeneous and heterogeneous dose escalation were investigated for multiple radiotherapy treatment modalities. Within the developed pipeline, dose prescriptions were structure-based, encompassing sub-structures with varying prescribed doses. This allowed original voxel-level tumour characteristics and optimal dose distributions to be accepted by commercial treatment planning systems. For three selected patients with IPL volumes of 3.22 cc, 4.63 cc, and 10.37 cc, their voxel-wise optimal dose distributions, for a given integral dose, were computed using a previously introduced methodology [1] based on the cell density distribution and tumour annotation masks of IPLs derived from multiparametric MR and histology [2, 3]. A novel methodology was developed to generate optimal structure-based dose prescriptions that maximize total tumor control probability (TCP) by utilizing voxel-wise tumor cell distributions and a linear-quadratic (LQ) TCP model. The method used two optimizers to find the optimal volume and prescribed dose of each sub-structure in an equivalent structure-based dose prescription. A prescription of homogenously-distributed 35 Gy to the prostate and up to 50 Gy to IPLs in 5 fractions was generated. With the same integral doses to the IPLs, heterogeneous structure-based dose prescriptions were generated with 4 sub-structures for the three patients. Treatment plans with homogeneous and heterogeneous IPL boost doses were created for five treatment modalities which included Elekta Versa, Varian TrueBeam, CyberKnife, Helical Tomotherapy, and intensity modulated proton therapy (IMPT). Resulting plans, that met defined normal tissue dose constraints, were compared using dose metrics and estimated TCP. Pearson correlation coefficients were computed to assess the linear correlation between planned achievable dose and tumour cell numbers. Material/Methods:

Results:

Across the five treatment modalities and three patients, a total of 30 plans (including plans for homogeneous and heterogeneous IPL boosts) were generated without violating any dose constraints. An example is given in Figure 1, depicting the employed prescriptions, resulting planned dose distributions, and spatial TCP distributions for CyberKnife treatment plans of the patient with the largest IPL volume. All plans with homogeneous escalation achieved TCPs above 90%. Within those, the Elekta Versa plans demonstrated the highest average TCP. In comparison to the homogeneous dose escalation plans, 10 plans (67%) with a heterogeneous dose escalation achieved higher TCPs. Specifically, the CyberKnife and Elekta Versa modalities exhibited increased TCPs for heterogeneous dose escalation plans for all patients. The most substantial