3231 - Development of Salvage Brachytherapy-Specific Genitourinary Toxicity Nomograms after Prostate Reirradiation
Presenter(s)
M. Bilski1,2, G. Salfi3, A. Gomez De Iturriaga4, P. Saniewski5, M. Stankiewicz6, D. Buchser7, F. Mastroleo8, T. Soeterik9, M. Masiuk10, B. A. Jereczek-Fossa11,12, T. Zilli13, and M. Pedrani14; 1Affidea Nu-med Center of Oncology Diagnostics and Therapy, Zamosc, Poland, 2Department of Radiotherapy, Medical University of Lublin, Lublin, Poland, 3Oncology Institute of Southern Switzerland, Ente Ospedaliero Cantonale, Bellinzona, Switzerland; Institute of Oncology Research, Bellinzona, Switzerland, 4Cruces Unisersity Hospital / Biocruces Health Research Institute, Spain, Bilbao, Spain, 5Radiotherapy Department. Maria Sklodowska-Curie Institute-Oncology Centre, Warsaw, Poland, 6Brachytherapy Department, Maria Sklodowska-Curie National Research Institute of Oncology, Gliwice, Poland, 7Hospital Universitario de Cruces, Barakaldo, Spain, 8Division of Radiation Oncology, IEO, European Institute of Oncology, IRCCS, Milan, Italy, 9Department of Radiation Oncology, University Medical Center Utrecht, Utrecht, Netherlands, 10Radiotherapy Department, Affidea Nu-med Center of Oncological Diagnostics and Therapy, Zamosc, Poland, 11Radiation Oncology, IEO European Institute of Oncology IRCCS, Milan, Italy, 12Oncology and Hemato-Oncology, University of Milan, Milan, Italy, 13Department of Radiation Oncology, Oncology Institute of Southern Switzerland, Ente Ospedaliero Cantonale, Bellinzona, Switzerland, 14Oncology Institute of Southern Switzerland, Ente Ospedaliero Cantonale, Bellinzona, Switzerland
Purpose/Objective(s):
Salvage HDR brachytherapy is a potentially curative option for locally recurrent prostate cancer after prior radiotherapy, but its wider use is limited by GU and GI toxicity. Existing salvage brachytherapy series consistently implicate baseline urinary function, treated volume, and urethral/bladder dose as key drivers of toxicity, yet no widely validated, salvage brachytherapy–specific prediction models or nomograms for GU/GI toxicity exist. Clinicians, therefore, rely on broad risk factors rather than on individualized, model-based estimates.Materials/Methods: We retrospectively analyzed 236 patients treated with salvage prostate HDR brachytherapy at four international centers after heterogeneous primary radiotherapy modalities. Acute and late GU/GI toxicities were coded as binary endpoints (grade =2 and, for one GU endpoint, grade =3). A prespecified candidate set included primary RT type, volume category, clinical target volume (CTV), summed EQD2, log-transformed CTV and EQD2, and age (continuous and =70). Missing data were handled using multiple imputation; multivariable logistic models were compared by cross-validated Brier score, with internal bootstrap validation of the best model for each endpoint (AUC and calibration slope). Pooled odds ratios (ORs) and nomograms were derived.
Results: For acute GU grade =2 toxicity, the best model included volume category, logCTV, logEQD2, and age. Higher volume category was strongly associated with acute GU toxicity (OR 4.73, 95% CI 1.62–13.79; p=0.0046), and age had a modest independent effect (OR 1.06 per year, 95% CI 1.01–1.12; p=0.026). Optimism-corrected AUC was 0.66 with a calibration slope of 0.88, supporting moderate discrimination and acceptable calibration. For late GU grade =2 toxicity, the optimal model included volume category, primary RT fractionation/type, EQD2 tertiles, CTV tertiles, and age =70. Higher volume category (OR 4.37, 95% CI 1.76–10.82; p=0.0016), highest vs lowest EQD2 tertile (OR 3.17, 95% CI 1.30–7.73; p=0.012), and prior moderately hypofractionated vs conventional EBRT (OR 3.16, 95% CI 1.47–6.80; p=0.0034) were independently associated with increased late GU toxicity. Optimism-corrected AUC was 0.68 with a calibration slope of 0.70. In contrast, both acute and late GI grade =2 models showed no robust independent predictors, low optimism-corrected AUCs (˜0.5–0.6), and substantial overfitting, limiting their clinical utility.
Conclusion: In a multiinstitutional salvage HDR brachytherapy cohort, we developed the first GU-focused, salvage-specific, internally validated prediction models and nomograms, demonstrating that treated volume, cumulative EQD2, and prior moderately hypofractionated EBRT are major drivers of GU toxicity. These models begin to address an unmet need for individualized GU risk estimation in salvage brachytherapy, while GI toxicity prediction remains unreliable and requires larger, more granular datasets and external validation.