Abstract
BACKGROUND AND PURPOSE: Intensity-modulated proton therapy (IMPT) offers dose/volume advantages for nasopharyngeal cancer (NPC), however is complex, time-consuming, and the Bragg-peak's broad lateral penumbra can compromise target coverage near critical structures. Transmission proton beams (TBs) may mitigate this, while fully automated planning can ensure high-quality plans with reduced planning times. This study aimed to evaluate the added value of TBs in robust, fully automated IMPT (autoIMPT) planning for NPC, and benchmarking these against manually generated clinical IMPT plans. MATERIALS AND METHODS: Twenty-four NPC patients previously treated with IMPT with 70.00 Gy(RBE) to the primary tumor and 54.25 Gy(RBE) to the elective volumes were included. AutoIMPT- and autoIMPT+TB plans were generated by Erasmus-iCycle, using the same 6-beam configuration and clinical constraints. Dose/volume-based comparison to clinical plans was performed for target coverage, OAR doses, and normal tissue complication probabilities (NTCP). RESULTS: Voxel-wise (vw) minimum target D98% of automated plans was similar or improved compared to clinical plans, especially in cases with compromised target coverage. In serial OARs, significant dose reductions were observed, especially in brainstem and optical system. Dose reductions to serial OARs were more pronounced in the autoIMPT+TB plans, up to 22.5 Gy for vw-maximum optical nerve D0.03cm(3). Similarly, dose reductions in most parallel OARs were observed (-13.2 Gy in pituitary Dmean). NTCP for dysphagia remained equal; xerostomia NTCP was significantly lower in the automated plans. CONCLUSIONS: Fully automated IMPT and IMPT+TB plans achieved similar or improved robust target coverage and OAR sparing compared to clinical plans. Incorporation of TBs to IMPT showed similar target coverage and enhanced dose reductions in serial OARs.