ESTRO 2024 - Abstract Book

S4640

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

ESTR0 2024

Results:

Considering a variable instead of a constant RBE resulted in a dose increase for all OARs (Figure 1.a), consequently leading to an increase in their respective NTCP values (Figure 1.b). In the most extreme scenario, this increase reached 32 percentage points. Notably, this effect was most prominent in small serial OARs (e.g. chiasm, optic nerves, and pituitary gland), located beyond to the SOBP and adjacent to the clinical target volume (CTV). When investigating the correlation between treatment-related parameters and ΔNTCP values, we observed a negative correlation between ΔNTCP and minimum opening angle between the fields (Spearman correlation down to -0.40), as well as a positive correlation between ΔNTCP and CTV volume (Spearman correlation up to 0.85) for most side effects. The ΔNTCP values for the 2-field plans were generally higher than those for the 3-field plans.

Both the LETopt and TEopt methods facilitated the reduction of D varRBE within the respective OARs. A reduction in D2 of up to 20.4 Gy(RBE) was evident, along with Dmean being decreased by an average of 2.1 Gy(RBE) (LETopt) and 4.2 Gy(RBE) (TEopt). This resulted in a median reduction of NTCP values for optimized OARs by 44%. The effectiveness of the optimization strategies depended on the selection of LET d and dose limits, as well as on the allowable maximum TE fraction within the OARs. In contrast to the reduction of D varRBE in OARs, for which TE or LET objectives were added, the dose in OARs not addressed by the novel optimization strategies or in normal brain tissue increased for some cases. When investigating how the number of fields impacts the potential of optimization strategies, 3-field plans showed a more effective NTCP reduction than 2-field plans (Figure 2). While all plans met clinical acceptance criteria under the assumption of a constant RBE, the application of the optimization strategies resulted in reduced CTV coverage when considering a variable RBE (α/β = 10 Gy). Furthermore, it was observed that none of the optimization strategies consistently outperformed the other, highlighting the need for individualized decision-making.