Reduction of organ of interest dose in proton therapy for oesophageal cancer through optimized setup robustness settings and online adaptation

Background and purpose: Oesophageal radiotherapy may cause side effects due to high organs of interest (OOIs) doses. While proton therapy (PT) reduces OOI dose compared to photon therapy, setup robustness settings used may limit OOI sparing. This study assessed how optimizing robustness settings for adequate target coverage affected OOI dose, and evaluated the potential of online adaptive PT (OAPT). Materials and methods: Twenty oesophageal cancer patients treated with intensity-modulated PT were analysed. Plans with 8 (clinical standard), 7, 6 and 5 mm setup robustness settings were generated on the planning 4D-CT (pCT) and robustly evaluated on weekly repeat 4D-CTs (rCTs). OAPT plans with individualized robustness settings were optimized and robustly evaluated on rCTs. Clinical target volume (CTV) V95% and OOI doses were assessed, and two-year mortality estimated using a validated model. Results: A 5 mm setup robustness setting maintained CTV V95% ≥ 98% for 80% of patients, with 92.4–97.7% in the remaining 20%. With OAPT, V95% ≥ 98% was achieved in 75%, with 97.6–97.9% in the remaining 25%. Mean heart dose decreased from 11.8 Gy (RBE) (Inter Quartile Range (IQR) 10.9–13.9) to 9.2 Gy (RBE) (IQR 8.2–11.6) when reducing settings from 8 to 5 mm. OAPT provided a reduction of 3.7 Gy (RBE), reducing two-year mortality by 6.9% compared to the 8 mm setting. Conclusions: Reducing setup robustness settings to 5 mm was feasible for most patients and reduced OOI dose. Large anatomical changes required plan adaptation. OAPT allowed further OOI dose reductions while preserving acceptable target coverage.