ESTRO 2024 - Abstract Book

S4599

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

ESTR0 2024

generate for each patient a single Pareto-optimal plan obeying pre-defined hard constraints and prioritised objectives as specified in a generic wish-list [2]. The same wish-list was used for both IMPT and IMPT&TPB.

The optimisation includes large-scale sparsity-induced BPs/TPBs selection (SISS) in the first phase and multicriteria optimisation with the 20% of candidate beamlets with the highest monitor units from phase 1 in Erasmus-iCycle [2,3] in the second phase. Pencil beams with monitor units (MU) below the clinically minimum deliverable MU are removed in the third phase. For IMPT plans, 20,000 candidate BPs were uniformly distributed in the CTVs with a 5 mm expansion, while in IMPT&TPB, 20.000 BPs and 18.000 TPBs were similarly distributed. TPBs were obtained by using 244 MeV shoot through beams. The best combination of BPs and TPBs was chosen by the optimiser. IMPT&TPB and IMPT were compared for eight nasopharynx and eight oropharynx patients, clinically treated with conventional IMPT. All 16 patients were planned using the same wish-list, minimum MU and 21 robustness scenarios. For plan comparisons, the voxelwise minimum D98% for the high-dose CTV (CTV70) was scaled to 95% of the prescription dose in the voxel-wise minimum dose distribution. NTCPs were computed according to the Dutch National Protocol for Model-Based Selection for Proton Therapy in Head and Neck Cancer [4]. Both for the oropharynx and nasopharynx cases, overall differences between IMPT and IMPT&TPB in target doses were clinically insignificant. For oropharynx, observed small differences in target coverage and target near-maximum dose were not statistically significant. In nasopharynx cases, a statistically significant average increase of 0.5 Gy was observed for IMPT&TPB in the voxelwise maximum (VWmax) D2% of CTV70, while a statistically significant decrease of 0.6 Gy was noted in the Vwmax D2% of the low-dose CTV (CTV54.25). In all nasopharynx IMPT&TPB plans, CTV70 VWmax D2% remained below the 105% constraint level. IMPT&TPB was superior for most OARs (Figure 1). Both nasopharynx and oropharynx showed statistically significant reductions in mean parotid and submandibular gland doses, which is probably related to the high priority in the wish list. For nasopharynx/oropharynx cases, NTCPs for xerostomia grade II and III were reduced by 3.6%-point/3.1%-point and 1.3%-point/1.1%-point, respectively. For dysphagia, differences were not significant. For nasopharynx, also cochlea doses showed substantial reductions for IMPT&TPB (Figure 1). Results: