Author: Daniel A. Alexander, Anish Butala, Anthony Austin Kassaee, Neil Malhotra, Ian Messing, Gabrielle W Peters, Ryan M. Scheuermann 👨🔬
Affiliation: Department of Radiation Oncology and Applied Sciences, University of Pennsylvania 🌍
Purpose: Stereotactic Body Radiation Therapy (SBRT) for spinal metastases uses CBCT-based localization and highly conformal treatment delivery to increase the biologically equivalent dose to the target while minimizing normal tissue toxicity. Due to steep dose gradients near critical dose-limiting OARs, intrafraction motion may carry significant clinical and dosimetric impact. Reference dose deviations from known shifts observed on mid- or post-treatment CBCT may support time-sensitive clinical decision-making and reduce the need for case-specific physics evaluations.
Methods: Spine SBRT cases at our institution are planned following the CCTG SC.24 protocol, resulting in highly standardized dose distributions constrained by dose falloff and critical OAR proximity to the target. We hypothesize that for small shift magnitudes, dose deviations arise primarily from relative shifts between the dose distribution and contours rather than changes in the dose distribution shape. We developed a Python script to shift the dose distribution relative to the structure set in 6 degrees of freedom, in steps of 0.5, 1, 1.5, and 2 mm/degrees. This was applied to 18 spine SBRT plans from our institution, and results were compared to forward-calculated dose deviations in the TPS to evaluate clinical viability.
Results: Dose deviations were reported in percentage of prescription (12 Gy x 2). D0.03cc[cGy] was evaluated for SpinalCord, SpinalCord_PRV, and ThecalSac. Vertical shifts generated the largest median dose deviation, with a 2mm shift corresponding to a 20.55% (493.2 cGy) increase for ThecalSac and a 6.50% (155 cGy) increase for SpinalCord. The mean difference between script-generated and TPS-generated dose deviations was -1.28% ± 3.46%, (-31.8668 ± 44.1928 cGy) attributed to interpolation and DVH-sampling variation.
Conclusion: This method provides an automatable alternative to resource-intensive forward calculations and was used to generate reference values to support clinical decision-making. Future work includes development of a real-time application for patient-specific dose deviations based on observed intrafraction shifts.