ESTRO 2024 - Abstract Book

S4660

Physics - Optimisation, algorithms and applications for ion beam treatment planning

ESTR0 2024

Purpose/Objective:

The use of high linear energy transfer (LET) radiation such as carbon ion radiotherapy (CIRT) benefits from a higher relative biological effectiveness (RBE) and a lower oxygen enhancement ratio (OER), making it suitable to treat deep seated and hypoxic targets such as sacral chordomas (SC) [1]. This is especially true for dose-averaged LET (LET d ) values larger than 50 keV/μm, as suggested by Matsumoto et al.[2]. The standard approach to CIRT treatment plan optimization relies solely on dose-based objectives. In this case, LET d typically peaks at the distal edge of the spread-out Bragg peak, with most of the high dose in the target being delivered through low/medium-LET radiations (LET d < 50keV/μm ). As the suboptimal levels of LET d within the target are suggested to be related to inferior treatment outcome [2,3], multiple LET d optimization approaches have been proposed to address this issue[4]. In this regard, LET d based optimization functions have been integrated in RayStation 2023B and a technical investigation revealed a conflict between dose uniformity, plan robustness and LET d maximization [5]. In this context, this work aims to define the major benefits and limitations of employing LET based optimization in a clinical scenario, with high requirements on an acceptable dose coverage to the clinical target volume (CTV), sufficient organs at risk (OARs) sparing and a preservation of plan robustness against setup and range uncertainties. Seventeen SC cases treated with CIRT at a single institution were retrospectively selected for this study. A reference plan was robustly-optimized (5mm setup error and 3.5% range uncertainty) for each case in the research version of RayStation 12B with a prescription dose (Dp) of 73.6Gy(RBE) in 16 fractions (4.6 Gy(RBE)/fraction), with a 3-beams configuration (2 opposite lateral + 1 vertical) and clinical goals according to the institution's current planning guidelines. RBE was calculated with the local effect model version I (LEM-I). Once a clinically acceptable plan was available, a minimum LET d function (minLET d ) on the gross tumor volume (GTV) was added to the objective function, and multiple plans were optimized with succesively increasing minLET d in steps of 3keV/μm, covering the range 37 55 keV/μm. All plans were robustly evaluated assuming error scenarios of +/- 3mm/3%. The target dose at 1% (D1) and 95% (D95), as well as the adherence to clinical goals on the sacrum, rectum and bowel OARs (CG OARs ), were investigated. For each case, the highest minLETd value (optimal minLET d ) that guaranteed both the CG OARs on the nominal plan and a robust target coverage was selected and considered for the following analysis. In particular, a plan was considered acceptable with at least 80% of robust scenarios fulfilling D1 < 1.05Dp or 60% to 80% of scenarios fulfilling D1 < 1.05Dp with worst-case D1 < 1.1Dp (Figure 1). A possible relation between the optimal minLET d and the CTV volume was investigated. The LET-optimized plans where compared to the reference plans considering the volume that receives high LET radiation (> 50keV/μm, V50 LET ) and the LET d at 98% and 50% (L98, L50) of both CTV and GTV. Eventually, the reference and LET-optimized plans were recalculated with the modified microdosimetric kinetic model (mMKM) to investigate the discrepancies between the two models in such a setting. A homogeneity index was computed for all nominal plans and Mann-whitney statistical tests were applied to assess statistical relevance of the evaluations. Material/Methods: