Author: Kaelyn Becker, Xenia Ray 👨🔬
Affiliation: University of California, San Diego, University of California San Diego 🌍
Purpose: Ethos 2.0 with HyperSight (Varian Medical Systems) imaging enables nearly fully automated offline adaptive radiotherapy including automated deformation of targets and recalculation on the Hounsfield unit-accurate CBCT acquired on-treatment. However, to make use of this capability, disease-site specific planning templates must first be developed to guide the optimization. We sought to develop a robust Ethos planning template for HN cancers and guidelines for its clinical use.
Methods: Seventeen HN cancer patients, initially planned manually in Eclipse, were replanned using a fine-tuned Ethos 2.0 planning template with the Intelligent Optimization Engine (IOE). The template was optimized for three-dose level bilateral oropharynx cases to produce clinical-quality VMAT plans with minimal adjustments. Its versatility was then tested and refined for unilateral cases, varying prescriptions, and other HN disease sites. Template-generated plans were compared to clinical plans using standard dose metrics.
Results: The Ethos 2.0 templated plans were comparable to the manually created Eclipse plans. Notably, mean salivary gland dose was lower in the Ethos templated plans (18.16 Gy vs. 23.56 Gy for parotid glands, and 43.84 Gy vs. 45.90 Gy for salivary glands) while 105% prescription dose coverage was higher at intermediate and lower dose levels (75.2% vs. 61.8% and 51.8% vs. 39.6%, respectively). Optimization structures, created by subtracting the lowest level PTV from the nearby OARs with a 3 mm overlap, helped shape the dose around critical structures while preserving target coverage. When OARs and PTVs overlapped completely, removal of the structure’s goals from the template was necessary.
Conclusion: We developed and validated a robust Ethos planning template for HN, applicable to bilateral and unilateral oropharynx and nasopharynx cases. Future work will evaluate its ability to generate clinically acceptable plans from mid-treatment CBCTs and recontoured targets and OARs, potentially eliminating mid-treatment CTs and significantly reducing re-planning time for HN cancers.