Related Experiment Video
Updated: May 20, 2026
Comparison of Predictive Performance of Three Lymph Node Staging Systems in Colorectal Signet Ring Cell Carcinoma Based on Machine Learning Model
Published on: April 18, 2025
External validation of the updated Briganti nomogram to predict lymph node invasion in prostate cancer patients
Jens Hansen1, Michael Rink, Marco Bianchi
1Martini Clinic, Prostate Cancer Centre at University Medical Centre Hamburg-Eppendorf, Hamburg, Germany. jenshansen81@gmail.com
Purpose:
We aimed to test accuracy and generalizability of a recently updated nomogram to assess the probability of lymph node invasion (LNI), when applied to a different European cohort of men undergoing radical prostatectomy (RP) with extended pelvic lymph node dissection (ePLND).
Materials And Methods:
The study cohort consisted of 1,282 men with clinically localized PCa who underwent RP and ePLND, including removal of obturator, external iliac, and hypogastric lymph nodes, between 01/2007 and 08/2011. Descriptive measurements included preoperative clinical and biopsy variables, such as prostate-specific antigen (PSA), clinical stage (CS), primary and secondary biopsy Gleason pattern, and percentage of positive cores. We used the area under curve (AUC) of the receiver operator characteristic analysis to quantify accuracy of the model to predict LNI. The extent of over- or under-estimation was explored graphically within loess calibration plots.
Results:
The median number of removed lymph nodes was 15 with an interquartile range of 12-20. Twelve percent (n = 155) of men had LNI. Preoperative clinical and biopsy characteristics differed significantly (all P ≤ 0.002) between men with LNI and those without. External validation of the previously reported updated LNI nomogram showed very good accuracy (AUC: 0.829). A nomogram-derived cut-off of 4% could lead to a reduction of 48% of lymph node dissection, while missing 10% of patients with LNI.
Conclusions:
We report the external validation of an updated LNI nomogram, demonstrating accuracy and applicability in a different European cohort. A nomogram-derived cut-off of 4% confirmed good performance characteristics within a different external validation cohort.
