Purpose: To develop and implement a fully automatic, clinical-criteria-driven treatment planning solution for locally advanced non-small-cell lung cancer (LA-NSCLC).
Methods: 21 patients with LA-NSCLC of various tumor locations and sizes with prescription of 60Gy in 30 fractions were retrospectively included in this study. Clinical plans were created manually using static IMRT. Automated plans with the same beam arrangements were designed using in-house-developed automated planning system: ECHO. ECHO uses hierarchical constrained optimization technique and is integrated into Eclipse using API scripting. Department clinical criteria were separated into 2 categories: limits and guidelines. Limits were strictly enforced by ECHO as constraints, while guidelines were optimized as much as reasonably achievable in objective functions. Dosimetric metrics were evaluated for plans in terms of tumor coverage and normal tissue sparing.
Results: Planning Target Volume (PTV) sizes ranged from 179cc to 1371cc. Planning time for ECHO averaged 59 minutes (range: 17min-107min). Both clinical and ECHO plans met all clinical criteria. Compared with clinical plans, ECHO plans were more consistent in target coverage and normal tissue sparing. Percentage dose to 95% of PTV for ECHO plans ranged 96.7%-100% (median 100%), compared to clinical plans with 87.1%-101.3% (median 100%). Median volume of total lungs excluding GTV receiving >5Gy and >20Gy for ECHO was 59.7% (46.9%-78.9%) and 31.8% (19.5%-36.9%), and for the clinical plans was 60.3% (42.7%-73.4%) and 33.3% (16.9%-36.3%) respectively. Doses to other organs at risk (OARs), including cord, esophagus and heart did not vary significantly between ECHO and clinical plans.
Conclusion: We developed a flexible automated treatment planning system for LA-NSLC, driven by modifiable constraints and planning limits/guidelines, and integrated with Eclipse. The automated ECHO plans met established clinical criteria with better consistency for tumor coverage and OARs sparing. Use of ECHO saves time and effort while resulting in improved or non-inferior plans.
Funding Support, Disclosures, and Conflict of Interest: This work was partially supported by the MSK Cancer Center Support Grant/Core Grant (NIH P30 CA008748)