Knowledge-Based Models for NRG HN001 Proton Plan Quality Analysis.

PURPOSE/OBJECTIVE(S) Radiotherapy treatment plan quality is essential to the success of clinical trials. Knowledge-based models have been used extensively for photon-based plan quality evaluation. With an increasing number of intensive modulated proton therapy (IMPT) plans submitted to clinical trials, we built knowledge-based proton planning (KBP) models for nasopharyngeal cancer cases. We report on the testing and the implementation of the models for the review of the quality of NRG-HN001 proton plans. MATERIALS/METHODS IMPT plans were manually created for 50 cases submitted to NRG-HN001. These plans were used to train a KBP model. Iterations of KBP-based re-optimization of model plans were performed to enhance the model performance. The final model was then used for generating 20 IMPT plans on the original proton cohort of patients to compare with submitted plans for quality evaluation. RESULTS Among the 20 IMPT plans to date submitted to HN001, 11 plans showed deviation unacceptable for target coverage and 7 plans for OARs sparing. Compared with submitted plans, all KBP plans meet per protocol or variation acceptable values for both target and OAR volumes. The 20-plan average target volume VRxGy[%] increased 8% (P = 0.00371), meanwhile, brain stem D0.03 cc [Gy] decreased 8 Gy (P < 0.00001), optic nerves D0.03 cc [Gy] decreased 11 Gy (P = 0.0002), chiasm D0.03 cc [Gy] decreased 12 Gy (P = 0.0006), parotids mean dose decreased 4Gy (P = 0.002). Doses to other OARs (spinal cord, mandible, TM joint, temporal lobes, eyes, lens, larynx, cochlea, and esophagus) showed no statistically significant changes. CONCLUSION This is the first successful application of KBP for clinical trial IMPT plan quality assurance. KBP-aided IMPT plans demonstrate better target coverage with reduced OAR doses compared with submitted plans from multi-institutions, showing improvement opportunities for IMPT planning. It is feasible to use these KBP models for current and future nasopharyngeal cancer clinical trials to improve proton radiotherapy plan quality.