Presentation Authors: Timo Soeterik*, Utrecht, Netherlands, Harm van Melick, Lea Dijksman, Nieuwegein, Netherlands, Heidi Kusters, Nijmgen, Netherlands, Saskia Stomps, Almelo, Netherlands, Douwe Biesma, Nieuwegein, Netherlands, Fred Witjes, Jean-Paul van Basten, Nijmegen, Netherlands
Introduction: The main challenge in robot assisted radical prostatectomy (RARP) is to achieve an optimal oncological result without compromising the functional outcomes by using a nerve-sparing approach. Extracapsular tumor extension (ECE) makes a nerve sparing RARP oncological unsafe. Therefore, it is important to know the likelihood of side-specific ECE. Since prostate MRI underestimates the presence of ECE, we determined the potential additional value of a nomogram, including oncological as well as MRI findings, to predict side-specific ECE.
Methods: Patients with prostate cancer who underwent RARP in 2014-2017 in two Dutch teaching hospitals were included for analysis. Predictors included in the multivariable logistic regression model were: Epstein grading, PSA density, presence of side specific ECE on MRI, percentage of biopsies containing cancer per side, and side-specific suspect findings on digital rectal examination (cT2 or cT3 vs. cT1). Primary outcome was side specific ECE established by experienced uropathologists during PA evaluation of the prostate specimen. Diagnostic accuracy of our prediction model was compared with results of prostate MRI alone. In our model, a calculated probability of >20% for presence of side-specific ECE was used as threshold to define a positive test result (ECE present).
Results: A total of 932 patients were included for analysis. The developed model was superior to prostate MRI alone regarding sensitivity of ECE presence, negative predictive value (ECE absence) and area under the curve (Table). Compared with MRI alone, our prediction model identified side specific ECE in more than a quarter of the patients (sensitivity 79% vs. 53% for left-sided ECE and 81% vs. 46% for right-sided ECE). Moreover, the negative predictive value (NPV) was 92%, using the model with a cut-off of >20%. Thus, nerve-sparing can be done safely in patients with a probability of ECE < 20% (Table).
Conclusions: The nomogram is able to detect more cases of ECE preoperatively in comparison to MRI alone. Combined with its high NPV, it is a useful tool in the preoperative planning of a side specific safe nerve sparing radical prostatectomy.
Source of Funding: Independent grants from Amgen and Astellas