Spatial Dosimetric-Based Prediction of Long-Term Urinary Toxicity After Permanent Prostate Brachytherapy

Area of Science:

• Radiation oncology
• Medical physics
• Urology

Background:

• Urinary toxicity is a common side effect of low-dose-rate (LDR) prostate brachytherapy.
• Predicting long-term toxicity is crucial for improving patient outcomes and treatment planning.

Purpose of the Study:

• To explore the correlation between spatial dose distribution and post-implant urinary toxicity.
• To develop a predictive model for long-term urinary toxicity in prostate LDR patients.

Main Methods:

• Eighty-five prostate LDR patients with over 12-month follow-up were analyzed.
• Patient-reported urinary toxicity was assessed using the International Prostate Symptom Score (IPSS) questionnaire.
• A machine learning model integrated baseline IPSS and dosimetric features (DVHs) to predict toxicity.

Main Results:

• 48% of patients experienced long-term urinary toxicity.
• The ML model identified baseline IPSS and six dosimetric features (posterior prostate subzones) as key predictors.
• The model achieved a high predictive performance with an AUC of 0.81.

Conclusions:

• Machine learning model effectively predicts long-term urinary toxicity after prostate LDR.
• Integrating spatial dose distribution and baseline symptoms aids in patient risk stratification.
• This approach can guide preventative measures and optimize follow-up care.