Room: Stars at Night Ballroom 1
Purpose: To develop and implement an intensity-modulated radiation therapy (IMRT) optimization plug-in for a commercial treatment planning system (TPS) that allows for inclusion of a customized radiation therapy (RT) plan optimization methodology incorporating biomarker and dose-based statistical models to estimate personalized toxicity and efficacy.
Methods: Recently developed statistical models can be leveraged to develop more personalized RT plans by incorporating patient-specific biomarkers for toxicity and efficacy into the dose selection process. In previous work, these statistical models have been incorporated into an optimization model formulation to calculate optimal dose for an individual patient by maximizing the utility of a plan. The â€˜utilityâ€™ metric of the plan is defined by the probability of efficacy minus the sum of weighted toxicity probabilities. In this work, we integrate this optimization methodology as a software plug-in to evaluate its clinical feasibility. Using a commercially available dose calculation engine, beamlet dose contributions to points within individual patient geometries are calculated using predefined beam arrangements. This information is combined with patient-specific statistical models to define the plan utility function. Individual beamlet fluences are then optimized to maximize the utility metric in a third-party commercial optimization engine. After optimization, ideal beam fluences are transferred back to the commercial TPS for clinical dose calculation and plan review.
Results: A new IMRT optimizer that incorporates variability in patient radiation sensitivity was developed, and implemented, and resulting plans retrospectively analyzed in comparison to previously treated lung RT plans. For lung cases, the patient-specific utility metric incorporated statistical models for local-regional progression-free survival, cardiac events, pneumonitis, and esophagitis.
Conclusion: This method facilitates IMRT optimization using a novel utility approach which can naturally accommodate baseline or mid-treatment biomarkers. Implementation of this method allows for direct exploration and inclusion of patient-specific trade-offs between efficacy and toxicity within RT planning in a quantitative manner.
Funding Support, Disclosures, and Conflict of Interest: This work is funded by NIH P01CA059827 and supported in part by Varian Medical Systems Inc.