ESTRO 2025 - Abstract Book

S1443

Clinical - Lung

ESTRO 2025

References: 1. Wallace ND, Hardcastle N, Bressel M, et al. A Prospective Study of Gallium Ventilation and Perfusion PET/CT Before, During, and After Radiotherapy in Patients with NSCLC. JTO 2023; 18(11):S578 2. McIntosh L, Jackson P, Hardcastle N, et al. Automated Assessment of Functional Lung Imaging with 68 Ga Ventilation/Perfusion PET/CT Using Iterative Histogram Analysis. EJNMMI Phys 2021 11;15(6):1726 3. Lucia F, Hamya M, Pinot F, et al. A Feasibility Study of Functional Lung Volume Preservation during Stereotactic Body Radiotherapy Guided by Gallium- 68 Perfusion PET/CT. Cancers (Basel) 2023 11;15(6):1726

4506

Poster Discussion Predicting tumour volume reduction in non-small cell lung cancer: Independent validation of a single parameter PSI model Sarah Barrett 1,2 , Mohammad U Zahid 3 , Heiko Enderling 3,4 , Conor McGarry 5,6 , Gerard M Walls 5,6 , Laure Marignol 1,2 1 Applied Radiation Therapy Trinity, Discipline of Radiation Therapy, Trinity College Dublin, Dublin, Ireland. 2 Trinity St. James’s Cancer Institute, Trinity College Dublin, Dublin, Ireland. 3 Department of Radiation Oncology, Division of Radiation Oncology, The University of Texas MD Anderson Cancer Center, Houston, USA. 4 Institute for Data Science in Oncology, The University of Texas MD Anderson Cancer Center, Houston, USA. 5 Patrick G. Johnston Centre for Cancer Research, Queen’s University Belfast, Belfast, United Kingdom. 6 Northern Ireland Cancer Centre, Belfast Health & Social Care Trust, Belfast, United Kingdom Purpose/Objective: The Proliferation Saturation Index (PSI) model, which models tumour dynamics in response to radiation as an instantaneous reduction in tumour volume, has been shown to predict non-small cell lung cancer (NSCLC) tumour volume regression in response to conventionally fractionated radiation therapy (RT) [1-3]. Prior work has solved for a radiation sensitivity parameter (α) and tumour growth parameter (λ) [1, 2]. This study seeks to validate the performance of the model, as a single parameter model, in an independent dataset. Material/Methods: Seventy-one patients with T1-3 N0 M0 disease treated with 55Gy/20# RT alone, from the NI-HEART cohort [4], were included. Model inputs were tumour volume measures from the cone beam CT (CBCT) imaging (days 1–3 or days 1 3 and day 10 RT). Tumour volumes were delineated using a semi-automated method [5]. Model prediction of tumour regression over the remainder of RT was simulated using days 1-3 CBCTs, and days 1-3 plus day 10 CBCTs as inputs. Absolute tumour volume measured on remaining weekly CBCTs (mean acquisition days were 10, 17 and 24) were compared to the model simulated volumes at the same timepoint using scatter plots. R 2 values and Pearson Correlation Coefficients (PCC) were calculated for all predicted timepoints combined. Model sensitivity to parameter variation was tested by varying the α and λ parameters +/- 20% and evaluating the impact on correlations. Governance approvals were provided, and ethical approval waived by the Belfast Health & Social Care Trust (IRAS 293181). Results: Data prediction using volume measures from day 1-3 CBCTs showed fair agreement between the measured and simulated volumes (R 2 = 0.81, PCC = 0.9) for the whole cohort (Figure 1). Inclusion of the day 10 CBCT, improved performance (R 2 = 0.91, PCC = 0.95), Model fit to the measured volumes for individual patients can be seen in Figure 2. The model was robust to parameter variation with the maximum change in either R 2 or PCC being -1.53% in R 2 when α and λ were increased by 20%.