Dosimetric Validation of Ethos Online Adaptive Radiotherapy for Prostate Cancer: An End-to-End Phantom Study

Background/Objectives: Online adaptive radiotherapy (oART) re-optimizes treatment on the anatomy of the day, but automatically generated plans require dosimetric verification and dose–volume assessment. We performed an end-to-end validation of the Varian Ethos workflow for prostate cancer, including cross-adaptation between two phantom anatomies. Methods: A 3 Gy × 20-fraction reference plan was generated for a RUBY phantom. Scheduled, adapted, and cross-adapted plans were evaluated in two prostate, bladder, and rectum configurations. Absolute dose was measured with a Semiflex 3D ionization chamber at the clinical target volume (CTV), rectum, and bladder and compared with Ethos and independent Mobius calculations. Plan quality was assessed from cumulative dose–volume histograms using ICRU 83-oriented descriptors. Results: For scheduled, adapted, and cross-adapted Ethos deliveries, local differences were within 1.0% at the CTV and 3.6% at the rectum. Bladder percentages were larger, but the underlying absolute differences did not exceed 0.048 Gy. When a cross-scheduled plan was deliberately evaluated on the changed anatomy, dose discrepancies of 0.345 Gy per fraction in the rectum and 0.245 Gy in the bladder exceeded the predefined acceptance limits; adaptation reduced the corresponding SF_IS discrepancies to within 0.015 Gy. Adaptation preserved target coverage and reduced rectal dose in same-anatomy comparisons, with small bladder trade-offs. The representative combined standard uncertainty was 2.71%, with an expanded uncertainty of 5.42% (k = 2); these values were used for uncertainty interpretation and not as location-specific acceptance limits. Conclusions: The Ethos workflow delivered accurate target dose and corrected clinically relevant plan–anatomy mismatch. Dose-volume benefits were anatomy- and objective-dependent rather than uniform.