Main Session
Sep
29
PQA 06 - Genitourinary Cancer, Gynecological Cancer, and Health Care Access and Engagement
Presenter(s)
Wenbin Gao, PhD - Cancer Hospital of Shandong First Medical University, Shandong,
W. Gao1, J. Ma1, J. Chen1, X. LI2, C. Fan2, H. Xiao3, and Y. Yin4; 1Cancer Hospital of Shandong First Medical University, Shandong, China, 2Cancer Hospital of Shandong First Medical University, jinan, China, 3Shandong Provincial Key Medical and Health Laboratory of Pediatric Cancer Precision Radiotherapy (Shandong Cancer Hospital), Jinan, Shandong, China, 4Department of Radiation Oncology Physics and Technology, Shandong Cancer Hospital and Institute, Shandong First Medical University and Shandong Academy of Medical Sciences, Jinan, China
Purpose/Objective(s):
Concurrent chemoradiotherapy combined with image-guided brachytherapy is the standard for locally advanced cervical cancer, yet failures persist due to heterogeneous radioresistant sub-volumes. While Dose Painting by Numbers (DPbN) enables voxel-level modulation, current strategies relying on static baseline metrics often ignore temporal biological evolution. Since longitudinal 18F-FDG changes better reflect intrinsic radiosensitivity, we propose a voxel-level Biologically-guided Adaptive Radiotherapy (BgART) framework using sequential PET/CT to quantify early metabolic response and guide patient-specific dose prescriptions.Materials/Methods:
We retrospectively analyzed 18 cervical cancer patients treated with radical intensity-modulated radiation therapy (IMRT). Sequential 18F -FDG PET/CT scans were acquired at baseline (SUV0) and mid-treatment (after 12–15 fractions). Using in-house software and deformable image registration, we extracted voxel-level metabolic changes to construct a Dose Response Matrix (DRM). The DRM quantifies intrinsic radiosensitivity by mapping the rate of SUV change to the in vitro surviving fraction index (SF2). In the feature space of (SUV0, DRM), a boundary curve distinguishing controlled from uncontrolled voxels was fitted using clinical outcome data. Subsequently, voxels were stratified by SUV0 and DRM, and a dose-Tumor Voxel Control Probability (TVCP) model was established using maximum likelihood estimation. Finally, a voxel-wise prescription function was inversely derived to calculate the minimum dose required for each voxel to achieve a target control probability, thereby generating individualized dose distribution maps for adaptive planning.Results:
Model-predicted Tumor Control Probability (TCP) showed high concordance with 3-month clinical outcomes (AUC=0.86). Under a uniform reference dose of 56 Gy, predicted TCP varied substantially (range: 0.27–0.89). To attain 95% TCP, calculated voxel-specific minimum doses ranged from 86.3 to 100.4 Gy, aligning closely with typical cumulative clinical doses (external beam plus brachytherapy). Consequently, the tumor voxel fraction controlled by standard prescriptions varied widely (41%–94%). The resulting voxel-based dose maps accurately identified radioresistant "hotspots," offering essential guidance for personalized adaptive strategies.Conclusion:
This study validates a voxel-level BgART framework that effectively translates early metabolic response into actionable dose prescriptions. By targeting intrinsic radioresistance identified via longitudinal PET/CT, this approach offers a promising strategy to individualize treatment and potentially improve local control in locally advanced cervical cancer.