3409 - Prediction Model Construction and Clinical Utility of Positive Surgical Margins in Apical Prostate Tumors Using Multimodal MRI

Purpose/Objective(s):

Apical prostate tumors, given their anatomical proximity to critical structures such as the urethral sphincter, present a heightened risk of positive surgical margins (PSM) following radical prostatectomy. This often necessitates adjuvant or salvage radiotherapy, increasing the therapeutic burden on patients. Currently, clinical practice lacks robust preoperative tools to quantitatively predict PSM risk. This study aims to integrate multimodal MRI-derived parameters with clinical characteristics to develop a predictive model for PSM in apical prostate tumors, thereby facilitating individualized treatment planning.

Materials/Methods: We retrospectively analyzed 1,029 patients who underwent radical prostatectomy at our institution between October 2020 and December 2024 and had preoperative MRI within six months prior to surgery. After excluding 721 patients who received neoadjuvant therapy, had non-apical tumors, or presented with metastatic disease, 308 prostate cancer patients were enrolled. Baseline characteristics including preoperative PSA, postoperative clinicopathological types, BMI, age, and Gleason score were recorded. Key MRI-derived anatomical parameters related to urinary continence were measured, including prostate volume, tumor volume and percentage of prostate involvement, maximum tumor diameter, distance from the tumor to the levator ani muscle, distance from the tumor to the superior/inferior margins of the membranous urethra, and membranous urethral length. Multivariate logistic regression was used to identify independent predictors of PSM and construct a predictive model. Model performance was evaluated using receiver operating characteristic (ROC) curve analysis.

Results: Of the 308 patients, 90 (29.2%) had apical PSM postoperatively. Univariate analysis identified the following significant predictors: distance from the tumor to the inferior margin of the membranous urethra (OR = 0.294, 95% CI: 0.176–0.489, *p* < 0.001) , prostate volume (OR = 0.982, 95% CI: 0.965–0.998, *p* < 0.05), gleason score (OR = 2.549, 95% CI: 1.352–4.807, *p* < 0.02), BMI (OR = 0.916, 95% CI: 0.833–1.008, *p* < 0.1). Multivariate analysis confirmed the distance from the tumor to the inferior margin of the membranous urethra (ß = -1.116, *p* < 0.01) and pathological type (ß = -1.028, *p* < 0.001) as independent predictors. These were combined with BMI, Gleason score, and prostate volume to construct the final model, which demonstrated an AUC of 0.737 (95% CI: 0.674–0.800). Calibration curves indicated strong model fit.

Conclusion: The predictive model developed in this study, incorporating MRI-derived quantitative parameters and clinical variables, effectively stratifies the risk of PSM in apical prostate tumors following radical prostatectomy. This tool provides an objective foundation for preoperative decision-making, enabling tailored therapeutic strategies and advancing the principles of precision medicine in prostate cancer management.