Investigating the Efficacy of Optimization Structures Used In Adaptive Radiotherapy for SBRT Liver Cases
Abstract
Purpose
The Varian Ethos intelligent optimization engine (IOE) automates the generation of non-overlapping optimization structures and resolves planning conflicts via a prioritized goal hierarchy. While designed to minimize manual intervention, the influence of user-defined structures on adaptive SBRT liver plan quality remains unclear. This study compares IOE-only adaptive plans against clinical and modified plans incorporating user-defined structures to evaluate effects on target coverage and organ-at-risk (OAR) sparing.
Methods
Twenty retrospective SBRT liver cases treated with Ethos were analyzed. Three adaptive plans were generated per case: (1) IOE-only (no manual structures), (2) the clinical adaptive plan, and (3) a modified plan using alternative user-defined structures (e.g., planning risk volume (PRV) expansions, or OARs cropped from the planning target volume (PTV) with 0.3 cm margins). All plans utilized identical prescriptions and VMAT/IMRT techniques consistent with the clinical plan. Goal priorities were adjusted in modified plans to test alternative strategies. Evaluation metrics included PTV coverage, OAR doses, total monitor units, homogeneity and conformity indices.
Results
IOE-only plans generally achieved adequate target coverage while sparing OARs, provided the highest-priority OAR topped the hierarchy. However, under competing objectives, target coverage was typically maintained by sacrificing lower-priority OAR constraints. Plans using user-defined structures showed superior compliance across multiple OAR constraints without compromising target coverage. Early data suggest that PTV-derived structures with 0.3 cm cropping margins offer a more favorable balance between coverage and sparing than PRV-based approaches.
Conclusion
While Ethos IOE can generate adaptive SBRT liver plans without user-defined structures, supplemental user-defined structures improve consistency when PTV and OAR objectives compete. PTV-derived structures appear promising and warrant further evaluation.