Integrating Genomic Classifiers and Nonsuspicious Magnetic Resonance Imaging Findings in Predictive Modelling for...

Area of Science:

• Urology
• Oncology
• Medical Imaging

Background:

• Accurate prediction of lymph node involvement (LNI) is crucial for staging and treatment planning in prostate cancer.
• Current methods may not fully capture the risk of LNI, potentially leading to overtreatment or undertreatment.

Purpose of the Study:

• To develop and validate a predictive model for LNI in men undergoing radical prostatectomy.
• To integrate clinical, radiological (magnetic resonance imaging - MRI), and genomic classifier (GC) data for improved prediction accuracy.

Main Methods:

• Retrospective analysis of patients undergoing robot-assisted radical prostatectomy (RARP) with extended pelvic lymphadenectomy (ePLND).
• Patients stratified into three cohorts based on preoperative work-up: suspicious MRI, nonsuspicious MRI, and GC testing.
• Logistic regression and artificial neural network (ANN) models were used to predict LNI, with performance evaluated by receiver operating characteristic (ROC) and decision curve analysis (DCA).

Main Results:

• The developed nomogram demonstrated high predictive accuracy for LNI across all cohorts (ROC: 0.84-0.92).
• Key predictors included prostate-specific antigen (PSA), Gleason Grade Group (GGG), positive biopsy cores, MRI findings, and GC results.
• ANN models showed comparable performance to the nomogram, with ROC values ranging from 0.82-0.91.

Conclusions:

• A novel nomogram integrating diverse parameters effectively predicts lymph node metastases in prostatectomy patients.
• This model can help personalize treatment strategies, potentially avoiding unnecessary lymphadenectomies while minimizing the risk of missed metastases.