Indications and complications of pelvic lymph node dissection for prostate cancer: are currently available nomograms accurate to predict lymph node invasion?

Objectives: to externally validate currently available nomograms for predicting lymph node invasion (LNI) in prostate cancer (PCa) patients and to assess the potential risk of complications of extended pelvic lymph node dissection (ePLND) when using the recommended cut-off. Methods: 14.921 patients who underwent radical prostatectomy with ePLND at eight European tertiary referral centers were retrospectively identified. After exclusion of patients with incomplete biopsy or pathologic data, 12.009 were included. Of these, 609 had undergone mpMRI-targeted biopsies. Among ePLND-related complications we included lymphocele, lymphedema, hemorrhage, infection, and sepsis. The performances of MSKCC, Briganti 2012, Briganti 2017, Briganti 2019, Partin 2016 and Yale models were evaluated using the receiver operated characteristic curve (AUC), calibration plots, and decision curve analysis (DCA). Results: overall, 1158 (9.6%) patients had LNI with a mean of 17.7 and 3.2 resected and positive nodes, respectively. No significant differences in AUCs were observed between MSKCC (0.79), Briganti 2012 (0.79), Partin 2016 (0.78), Yale (0.80), Briganti 2017 (0.81) and Briganti 2019 (0.76). A direct comparison of older models showed a better discrimination for MSKCC and Briganti 2012 nomograms. A tendency for underestimation was seen for all the older models, whereas Briganti 2017 and 2019 nomograms tended to overestimate LNI risk. DCA analysis showed a net benefit for all models, with a lower net benefit for Partin 2016 and Briganti 2019 models. ePLND-related complications were experienced by 1027 patients (8.9%), and 12.6% of pN1 patients. Conclusions: currently available nomograms have similar performances and limitations for the prediction of LNI. Miscalibration was present for all nomograms that showed however a net benefit. In patients with only systematic biopsy, MSKCC and Briganti 2012 nomograms are better to predict LNI.