S1708
Physics - Dose prediction/calculation, optimisation and applications for particle therapy planning
ESTRO 2026
1 Radiation oncology, Leiden University Medical Center, Leiden, Netherlands. 2 Holland, PTC, Delft, Netherlands. 3 Radioterapy, Erasmus MC Cancer Institute, Rotterdam, Netherlands Purpose/Objective: Manual intensity-modulated proton therapy (IMPT) planning for skull base chordoma is challenging due to the proximity of critical structures to the target volume and the presence of postoperative stabilization materials. In this study, we investigated whether automated IMPT planning can produce plans of comparable or superior quality to manually generated clinical plans. Material/Methods: Thirteen patients with skull base- or cervical chordoma treated with IMPT (74.0 Gy[RBE] to the primary tumor (CTV7400) and 59.2 Gy[RBE] to the elective volume (CTV5920)) were retrospectively included. Three patients were used to develop a site-specific wish-list for Erasmus-iCycle, using the same beam configuration and dose objectives and constraints as in manual planning. Automated IMPT plans were then generated for the remaining ten patients. Dose- volume parameters for targets and organs at risk (OARs) were compared between automated and clinical plans. When automated plans showed inferior target coverage, patient-specific adjustments to the wish-list were performed, allowing limited exceedance of serial OAR constraints similar to clinical plans. Statistical significance was assessed using Wilcoxon signed-rank tests. Results: Median target coverage was slightly higher for automated IMPT plans than for manual plans (CTV7400: 6834 vs 6717 cGy; CTV5920: 5847 vs 5784 cGy), with no statistically significant differences. For serial OARs, statistically significant dose reductions were observed for the brain, brainstem, and carotid arteries, with the largest reduction observed in the brainstem surface D0.03cc (681 cGy). Most parallel OARs showed non-significant reductions. Clinical target coverage goals were achieved in 4 of 10 patients using both methods (Figure 1).
Notably, while manual clinical plans overall required concessions leading to exceedance of serial OAR constraints, automated planning consistently maintained these limits (Figure 2). Among the six patients with compromised target coverage, automated plans improved coverage in three cases but showed deterioration in the other three. After patient-specific wish-list adjustments in three cases, automated plans achieved improved target coverage compared with manual plans, with equal or lower OAR doses.
Made with FlippingBook - Share PDF online