S1969
Physics - Dose prediction/calculation, optimisation and applications for photon and electron planning
ESTRO 2026
Conclusion: The scorecard-based evaluation proved useful for objective, reproducible assessment of plan quality and can support clinical approval decisions. RapidPlan achieved comparable or superior scores, confirming its ability to enhance plan consistency, quality, and efficiency. Iterative scorecard assessment can establish percentage-based reference baselines for clinical use in both manual and automated planning workflows. References: 1. Nelms BE, Robinson G, Markham J, Velasco K, Boyd S, Narayan S, et al. Variation in external beam treatment plan quality: an inter-institutional study of planners and planning systems. Pract Radiat Oncol. 2012;2(4):296–305. doi:10.1016/j.prro.2011.11.0122. Moore KL, Brame RS, Low DA, Mutic S. Experience- based quality control of clinical intensity-modulated radiotherapy planning. Int J Radiat Oncol Biol Phys. 2011;81(2):545–51. doi:10.1016/j.ijrobp.2010.11.0303.Rayn K, Freedman D, Buchberger D, Ajlouni M, Cook R, Lee J, et al. Scorecards: quantifying dosimetric plan quality in pancreatic ductal adenocarcinoma stereotactic body radiation therapy. Adv Radiat Oncol. 2023;8(4):101229. doi:10.1016/j.adro.2023.101229 Keywords: scorecard, RapidPlan, knowledge-based Collaborative multi-institutional approach for external validation and update of hierarchical Bayesian eGFR predictive models for kidney cancer SABR Vanessa Panettieri 1,2 , Antony Carver 3 , Nicholas Hardcastle 1,2 , Muhammad Ali 4,2 , Helen Howard 3 , Anjali Zarkar 5 , Shankar Siva 4,2 1 Department of Physical Sciences, Peter MacCallum Cancer Centre, Melbourne, Australia. 2 Sir Peter MacCallum Department of Oncology, The University of Melbourne, Melbourne, Australia. 3 Department of Medical Physics, University Hospitals Birmingham NHS Foundation Trust, Birmingham, United Kingdom. 4 Department of Radiation Oncology, Peter MacCallum Cancer Centre, Melbourne, Australia. 5 Department of Oncology, University Hospitals Birmingham NHS Foundation Trust, Birmingham, United Kingdom Digital Poster 3965 Purpose/Objective: Stereotactic ablative radiotherapy (SABR) is emerging as a promising, non-invasive treatment for primary renal cell carcinoma (RCC), demonstrating both acceptable side-effect profiles and preservation of renal function [1]. As a relatively new treatment option, there is the need for robust, generalizable tools to predict post-treatment adverse events.
Developing such models is challenging due to the limited number of patients treated worldwide. Collaborative data pooling across institutions offers a powerful strategy to increase statistical power for model development while maintaining patient privacy. In this work, we present iterative external updating and validation of a hierarchical Bayesian models for predicting eGFR following RCC SABR, using data from two international institutions. Material/Methods: Data of 30 patients with primary RCC treated with SABR at a single institution A (with 26 Gy in a single or 42 Gy in three fractions) were used to build the initial model with PyMC 5.26.1, assuming that eGFR (mL/min per 1·73m ² ) declines from baseline following treatment before reaching a plateau. Within this framework, eGFR at any given time point was modelled as a function of a rate of decline and a final plateau value. These two parameters were in turn a function of kidney equivalent uniform dose (EUD), patient age, tumour volume, and number of kidneys. Posterior distributions for the model parameters were then shared with an independent institution B. These models were externally validated using clinical data and dose-volume histograms from 68 RCC patients treated in the TROG 15.03 FASTRACK II trial [1]. Then the model was iteratively updated using the data of these 68 patients and model parameters were compared. Results: Model performance of the external validation was evaluated by means of a calibration plot (slope=0.95, R2=0.86 (Fig 1). Figure 2-b1 shows the posterior distributions for the mean and standard deviation of the regression coefficient linking healthy kidney EUD to the rate of decline and the final value. After the initial fit by institution A the 94% interval for the β EUD for the rate of decline was –1.2< β EUD<1.8. After the model update by B (Fig.2-b2) this interval decreased to –0.8< β EUD<1.9, narrowing the interval and increasing the tension with zero, strengthening the evidence that dose affects the rate of eGFR decline.
Made with FlippingBook - Share PDF online