Validation of Automated Script-Based Proton Therapy Optimisation for Comparative Head and Neck Planning

Marte Kåstad Høiskar et al.

This retrospective study evaluated an automated, rule-based intensity-modulated proton therapy (IMPT) optimisation script developed within a commercial treatment planning system for comparative planning in head and neck cancer (HNC). Twenty patients previously treated with manually optimised IMPT plans were included, predominantly with oropharyngeal cancer. For each patient, two automated IMPT plans were generated: one using a standard five-field arrangement and another reproducing the patient-specific field configuration used clinically. The plans were compared with the clinically approved IMPT plans regarding target robustness, dose distribution, and normal tissue complication probabilities (NTCP) for dysphagia and xerostomia. Automated planning required approximately 64 minutes on average. Both automated approaches produced comparable or lower mean doses to several organs at risk, with median dose reductions of up to 3.3 Gy for the glottis. NTCP estimates were also similar or slightly improved, with median reductions of 0.1–0.3 percentage points for dysphagia and approximately 0.3 percentage points for xerostomia. Robust target coverage was generally maintained, although some differences in nominal CTV60 coverage were observed.

In conclusions, the automated IMPT optimisation script generated clinically acceptable plans with robustness, dose distributions, and NTCP values comparable to manually optimised clinical plans. Importantly, a standard five- or six-field configuration was sufficient for comparative planning in most patients, reducing planner-dependent variability. The approach substantially shortened planning time from days to hours and may facilitate faster, more consistent identification of HNC patients who are likely to benefit from proton therapy. Further optimisation and prospective clinical validation are warranted.

Published by Physics and Imaging in Radiation Oncology 2026.

https://doi.org/10.1016/j.phro.2026.101058