Patient stratification for dose scaling in cervical cancer: a model-based analysis using image-guided adaptive...

The purpose of this study was to identify the characteristics of cervical cancer patients who benefit most from dose scaling (DS) using a model-based approach in image-guided adaptive brachytherapy (IGABT). We retrospectively analyzed 57 cervical cancer patients who underwent IGABT and external beam radiotherapy. Model-based DS was simulated to increase the brachytherapy dose until predefined normal tissue complication probability thresholds were reached. Clinically significant therapeutic gain was predefined as a tumor control probability (TCP) increase of greater than 3%, consistent with prior radiobiological modeling studies. Patient characteristics were compared between groups with and without significant TCP gain. Of the 53 patients in whom dose escalation was feasible, 16 (30.2%) achieved a significant TCP gain. A significant gain was strongly associated with histopathological type and magnetic resonance imaging (MRI) findings. All four patients (100%) with adeno/adenosquamous carcinoma (AdSq) achieved a significant gain, compared with only 12 of 49 (24.5%) with squamous cell carcinoma (P = 0.006). Uterine corpus invasion (UCI) on MRI was also a significant predictor of benefit (P < 0.001). Conversely, larger high-risk clinical target volume limited the potential for dose increase. AdSq cases demonstrated substantial improvement, with TCP increases after DS often exceeding 10%. Patients with poor prognostic features on MRI, specifically UCI, and those with histological AdSq may benefit most from DS in IGABT for cervical cancer. This exploratory study suggests that model-based DS shows promise for personalized treatment planning, supporting a shift from fixed-dose prescriptions toward strategies tailored to individual patient and tumor characteristics.