Optimization of the Dose-Volume Effect Parameter "a" in EUD-Based TCP Models for Breast Cancer Radiotherapy.

IntroductionRadiotherapy treatment plans traditionally rely on physical indices like Dose-volume histograms and spatial dose distributions. While these metrics assess dose delivery, they lack consideration for the biological effects on tumors and healthy tissues. To address this, radiobiological models like tumor control probability (TCP) and Normal tissue complications probability (NTCP) are increasingly incorporated to evaluate treatment efficacy and potential complications. This study aimed to assess the predictive power of radiobiological models for TCP in breast cancer radiotherapy and provide insights into the model selection and parameter optimization.MethodsIn this retrospective observational study, two commonly used models, the Linear-Poisson and Equivalent uniform dose (EUD)-based models, were employed to calculate TCP for 30 patients. Different radiobiological parameter sets were investigated, including established sets from literature (G and G) and set with an optimized "a" parameter derived from clinical trial data (a and a). Model predictions were compared with clinical outcomes from the START trials.ResultsThe Linear-Poisson model with es lished parameter sets from the literature demonstrated good agreement with clinical data. The standard EUD-based model (a = -7.2) significantly underestimated TCP. While both models exhibited some level of independence from the specific parameter sets (G vs. G), the EUD-based model was susceptible to the "a" parameter value. Optimization suggests a more accurate "a" value closer to -2.57 and -5.65.ConclusionThis study emphasizes the importance of clinically relevant radiobiological parameters for accurate TCP prediction and optimizing the "a" parameter in the EUD-based model based on clinical data (a1 and a2) improved its predictive accuracy significantly.