Purpose
Proton therapy (PT) is challenged by anatomical variations during the treatment course. In this study, the performance of a standard robust optimization technique is evaluated for patients with nasopharyngeal cancer.
Material and Methods
PT plans using simultaneous integrated boost (68/60/50Gy) were optimized for five nasopharyngeal cancer patients, treated originally with photons, using Eclipse TPS v13.7. Planning was performed using multiple-field optimization with three beam directions, and robust optimization for CTV1 (68Gy) with 5 mm perturbations in all directions. Using the Smart Adapt tool, virtual CTs (VCTs) were obtained based on deformable image registration to the planning CT from the daily CBCTs. Doses were recalculated with the original plans in the VCTs, and plan robustness was evaluated comparing CTV1 coverage (V95) and OAR doses.
Results
Throughout treatment, CTV1 coverage was acceptable (V95 > 99%). In only one fraction for one patient, target coverage was compromised (V95 = 98%). The maximum dose to the brainstem increased approximately 13% for 2/5 patients and 15% to the spinal cord for 3/5 patients. The mean dose to the ipsilateral parotid gland increased 15% for 2/5 patients. For two patients, a 30% volume reduction of the parotid gland was observed.
Conclusions
Target volume coverage was not significantly compromised for the patients studied. An increase in dose to the OARs during treatment was observed which might give rise to increased normal tissue toxicity. Standard robust optimization designed for rigid translation may not account sufficiently for anatomical variations during treatment course.

Evaluation of robustly optimized IMPT plans using daily CBCT scans for nasopharyngeal cancer