Publicatiedatum
Naam tijdschrift
Urologic Oncology
Background: Prediction of lymph node invasion (LNI) after radical prostatectomy has been rarely assessed in robotically assisted laparoscopic radical prostatectomy (RALP) series. We aimed to develop and externally validate a pretreatment nomogram for the prediction of LNI following RALP in patients with high- and intermediate-risk prostate cancer.
Methods: 1654 RALP patients were prospectively collected between 2009 and 2016 from academic and community hospitals. We included patients with intermediate- and high-risk prostate cancer who underwent pelvic lymph node dissection (e-PLND). Logistic regression analysis was applied to construct a nomogram to predict LNI. Centers were randomly assigned to the training cohort (80%) and validation cohort (20%). The discriminative accuracies were evaluated by the areas under the curve and by the calibration plot. The net benefit of the nomogram to predict LNI was assessed by decision curve analysis and a cut-off was proposed.
Results: In total, 14% of the patients in our cohort had pN1 disease. Applying logistic regression analysis, the following covariates were chosen to develop the nomogram: initial PSA, clinical T stage, biopsy Gleason sum, and proportion of positive biopsy cores. The nomogram showed a median discriminative accuracy of 73% and excellent calibration. The net benefit of the model ranged between 7% and 51% predicted risk of LNI. A cut-off to perform e-PLND was set at 7%. This would permit a 29% of avoidable e-PLND, missing 9.4% of patients with LNI.
Conclusions: We developed and externally validated a nomogram to predict LNI in patients treated with RALP from a prospective, multi-institutional, nationwide series. A risk of LNI > 7% is proposed as cut-off above which e-PLND is recommended.
Methods: 1654 RALP patients were prospectively collected between 2009 and 2016 from academic and community hospitals. We included patients with intermediate- and high-risk prostate cancer who underwent pelvic lymph node dissection (e-PLND). Logistic regression analysis was applied to construct a nomogram to predict LNI. Centers were randomly assigned to the training cohort (80%) and validation cohort (20%). The discriminative accuracies were evaluated by the areas under the curve and by the calibration plot. The net benefit of the nomogram to predict LNI was assessed by decision curve analysis and a cut-off was proposed.
Results: In total, 14% of the patients in our cohort had pN1 disease. Applying logistic regression analysis, the following covariates were chosen to develop the nomogram: initial PSA, clinical T stage, biopsy Gleason sum, and proportion of positive biopsy cores. The nomogram showed a median discriminative accuracy of 73% and excellent calibration. The net benefit of the model ranged between 7% and 51% predicted risk of LNI. A cut-off to perform e-PLND was set at 7%. This would permit a 29% of avoidable e-PLND, missing 9.4% of patients with LNI.
Conclusions: We developed and externally validated a nomogram to predict LNI in patients treated with RALP from a prospective, multi-institutional, nationwide series. A risk of LNI > 7% is proposed as cut-off above which e-PLND is recommended.