Author: Robbie Beckert, Austen N. Curcuru, Farnoush Forghani, Yi Huang, Geoffrey D. Hugo, Hyun Kim, Eric Laugeman, Luke Christian Marut, Thomas R. Mazur, Allen Mo, Emily Sigmund 👨🔬
Affiliation: Washington University in St. Louis School of Medicine, WashU Medicine, Washington University School of Medicine in St. Louis, Wash U Medicine, Washington University in St. Louis, Department of Radiation Oncology, Washington University School of Medicine in St. Louis 🌍
Purpose: Adaptive SBRT is resource intensive, requiring additional personnel for online planning, and should be reserved for cases where it is most beneficial. The purpose of this research is to create a reliable decision support tool for physicians to predict the benefit of adaptive SBRT treatment for patients with left adrenal tumors.
Methods: The twenty patients included in this study were treated to a dose of 50 Gy in 5 fractions on either MR-guided (17 patients) or CT-guided (3 patients) adaptive radiotherapy systems. A LASSO model was created to predict the number of fractions adapted for a specific patient using dose and geometric parameters. The chosen parameters were PTV volume and distance from PTV and D0.5cc for small bowel, large bowel, and stomach, all at the time of simulation. A leave one out cross validation method was used to test the model.
Results: The most important influence on the model was D0.5cc for the stomach at simulation, with an individual R2 value of 0.5182. The least predictive element was PTV volume with an R2 value of 0.0971. The R2 value for the predictive equation was 0.49 and the correlation coefficient was 0.74. The leave one out cross validation used to test the model reported an R2 of 0.29 and a correlation coefficient of 0.54. The model displayed an average sensitivity of 100% and specificity of 60%.
Conclusion: The reported LASSO R2 and correlation coefficient indicate moderate fit and strong correlation between the prediction parameters and outcome. Our results suggest that patient specific metrics at the time of simulation can provide a reliable predictive tool for physicians to decide between conventional and adaptive SBRT treatment for left adrenal patients. It would be instrumental both in streamlining workflow and allocating resources for adaptive radiotherapy in a clinical setting.