Author: Rajeev K. Badkul, Ronald C Chen, Ying Hou, Harold Li, Chaoqiong Ma, Jufri Setianegara 👨🔬
Affiliation: Department of Radiation Oncology, University of Kansas Medical Center 🌍
Purpose:
Postimplant urinary toxicity is common in prostate low-dose-rate (LDR) brachytherapy. We developed a machine learning (ML) model to explore the correlation between spatial dose distribution and postimplant urinary toxicity, aiming to assist decision making in LDR treatment planning thereby improving patient outcomes
Methods:
One-hundred-fiften prostate LDR patients with >12-month follow-up were included. Patient-reported urinary toxicity was collected prospectively using the International Prostate Symptom Score (IPSS) questionnaires, from before implant (baseline) to post-implant follow-up. Patients were then grouped into those whose symptom scores returned to ≤2 points above baseline by 12 months (no long-term toxicity) vs those who did not (long-term toxicity). Eighty-five features were extracted for each patient, including principal components of dose-volume histograms (DVHs) from multiple prostate subzones, the whole prostate and urethra, as well as baseline IPSS and implantation characteristics. An ML model incorporating backward feature selection algorithm was developed to predict long-term toxicity status, using a shuffle-and-split validation strategy for model evaluation during feature selection. A univariate statistics analysis was conducted to the model selected features
Results:
Sixty-three of 115 patients (55%) had long-term urinary toxicity. Seven features were selected during model training, including baseline IPSS and six dosimetric features from several prostate subzones primilary located in the posterior prostate. The model achieved a high mean area under the receiver operating characteristic curve (AUC) of 0.83, with a balanced sensitivity and specificity of 0.78 by adjusting the probability threshold. In univariate analysis, only baseline IPSS and one selected dose feature were significantly correlated with long-term toxicity with AUC<0.71.
Conclusion:
The proposed ML model, integrating baseline IPSS and spatial dosimetric features, effectively predicts long-term urinary toxicity after prostate LDR. Future research will test whether reducing LDR planning dose to identified zones will reduce patient toxicity