Purpose/Objective(s): Organ-at-risk (OAR) sparing is a critical challenge for many cancers, but is particularly important for head and neck cancer patients, who experience significant acute and late toxicity. Prescription to the CTV with robust optimization against setup and range uncertainty has replaced planning target volume (PTV) optimization for proton therapy at our institution. We hypothesized that optimizing head and neck IMRT plans to the CTV, but ensuring robustness of target coverage against setup uncertainty, would allow for clinically significant reductions in OAR doses compared to traditional PTV-based optimization.

Materials/Methods: We identified 10 head and neck cancer patients who completed radiotherapy: 4 with squamous cell carcinoma (SCC) of the right oropharynx, 3 with SCC of the supraglottic larynx, and 3 with SCC of the nasopharynx. For the control arm, we first optimized to the PTV (typically a 3 mm uniform expansion from CTV), with target goals of 95% PTV receiving 100% prescription (Rx) dose and 99% PTV receiving 93% Rx dose. For the experimental arm, we optimized to the CTV to achieve equivalent target goals (95%-100% and 99%-93%) for the second-to-worst case scenario under 3 mm isocenter shift uncertainty analysis. The worst-case scenario was rejected as an outlier; this is current practice for proton robust multi-field optimization (rMFO) planning. Differences in OAR doses were compared with 2-tailed paired Student’s t-tests.

Results: Across all patients, substantial dose reductions were seen. Parotid gland dose was reduced by a mean of 293 cGy (range -96-813, p<0.01), submandibular glands by 210 cGy (-192-654, p<0.01), constrictors by 410 cGy (102-624, p<0.01), esophagus by 223 cGy (-48-463, p<0.01), oral cavity-less-PTV by 296 cGy (-33-466, p<0.01), and mandible-less-PTV by 263 cGy (-109-680, p<0.01). In specific cases, major reductions in maximum OAR dose were achieved: brainstem by 1005 cGy in a nasopharynx cancer patient, optic nerve by 1257 cGy in a nasopharynx cancer patient, and cochlea by 1012-2076 cGy in 3 patients (2 oropharynx and 1 nasopharynx); mean cochlea dose was reduced by 821-1207 cGy in 3 patients.

Conclusion: IMRT plan optimization to the CTV that achieves robust target coverage against setup uncertainty may yield clinically significant improvements in OAR sparing for head and neck patients, as compared to standard PTV-based planning.