AUA 2018: The First Nomogram to Identify Candidates for Extended Pelvic Lymph Node Dissection Among Men Staged with mpMRI for Clinically Localized Prostate Cancer
For this study, the authors identified 2,502 prostate cancer patients treated with radical prostatectomy and extended pelvic lymphadenectomy between 2005 and 2017 at five tertiary care referral centers. The population was randomly divided into a development (n=1,876) and a validation cohort (n=626) for their model. In the development cohort, two multivariable models to predict lymph node invasion were developed:
1. A base model using PSA, clinical stage assessed by DRE (T1 vs. T2 vs. T3), primary Gleason grade (3 vs. ≥4), secondary Gleason grade (3 vs. ≥4) and percentage of positive cores
2. An advanced model including MRI T-stage [no lesion visible/lesion without capsular penetration (T1-2) vs. lesion with suspicion of extra-capsular extension (3a) vs. lesion with suspicion of seminal vesicle involvement (3b)] to the base model instead of DRE.
Covariates were used to develop the nomogram for predicting the probability of lymph node invasion. The predictive accuracy of the two models was compared using the AUC. Decision curve analyses were performed to compare the net benefit. Finally, for each nomogram cut-off, specificity, sensitivity, and negative predictive values (NPVs) were calculated.
In the cohort, lymph node invasion rates were 4.7% and 5.6%, in the development and validation cohort. All variables included into both the base and advanced models were independent predictors of lymph node invasion (all p≤0.01). The predictive accuracy of the basic model was 0.845 and of the advanced models in the external validation cohort was 0.85. The advanced model including MRI resulted into higher net-benefit relative to the base model, although marginally. Using a 5% nomogram cut-off, 514 patients (82.1%) would be spared extended pelvic lymphadenectomy and lymph node invasion would be missed in 11 patients (2.1%). The sensitivity of this model was 68.6%, specificity was 85.1%, and NPV was 97.9%.
The strength of this study is the development and validation cohort design for developing the clinical model. Certainly, mpMRI assists in preoperative planning, particularly in high risk patients. This may include tumor location and assessing risk of extra-capsular extension (which may lend one to not necessarily perform nerve-sparing on that side). Although this is the first model predicting lymph node invasion using local stage information derived from mpMRI, the predictive accuracy of the model including MRI was only marginally better than the model using DRE. The authors suggest that mpMRI, when available, should be preferred in the assessment of lymph node invasion in order to better identify candidates to extended pelvic lymphadenectomy, however it is likely more appropriate to reserve mpMRI for the highest risk preoperative patients rather than using mpMRI resources for all patients undergoing radical prostatectomy.
Presented by: Armando Stabile, Vita-Salute San Raffaele University, Milan, Italy
Co-Authors: Paolo Dell'Oglio, Milan, Italy, Matteo Soligo, Rochester, MN, Pietro Grande, Paris, France, Giorgio Brembilla, Giulia Cristel, Nicola Fossati, Giorgio Gandaglia, Antonio Esposito, Francesco De Cobelli, Milan, Italy, Bernhard Grubmüller, Vienna, Austria, Raphaele Renard-Penna, Paris, France, Laurent Salomon, Créteil, France, Shahrokh F. Shariat, Vienna, Austria, Jeffrey R. Karnes, Rochester, MN, Francesco Montorsi, Milan, Italy, Alexandre De La Taille, Créteil, France, Morgan Roupret, Paris, France, Alberto Briganti, Milan, Italy
Written by: Zachary Klaassen, MD, Urologic Oncology Fellow, University of Toronto, Princess Margaret Cancer Centre, Twitter: @zklaassen_md, at the 2018 AUA Annual Meeting - May 18 - 21, 2018 – San Francisco, CA USA