External Validation of Nomograms for Predicting Pelvic Lymph Node Metastases in Patients with Prostate Cancer and the Added Value of the Prostate-specific Membrane Antigen Positron Emission Tomography–based PRIMARY Score

Our study results suggest that the Amsterdam-Brisbane-Sydney nomogram performs better than other models in predicting pathological lymph-node involvement in patients with prostate cancer. The PRIMARY score showed a significant association with lymph node involvement and represents a potential imaging-based marker to support clinical decision-making. The aim of our study was to predict pathological lymph node involvement (stage pN1) in patients undergoing radical prostatectomy (RP) and extended pelvic lymph node dissection (ePLND) for prostate cancer (PC) and to compare the performance of nomograms used in ePLND decision-making. Data for 191 patients with PC who underwent and ePLND between 2018 and 2023 were analyzed retrospectively. Demographics, prostate-specific membrane antigen (PSMA) positron emission tomography (PET)/computed tomography (CT) results, and multiparametric magnetic resonance imaging (mpMRI) findings were assessed in relation to pN1 prediction and nomogram comparison. Statistical analyses included χ 2 and Mann-Whitney U tests. New models in which PSMA PET/CT parameters (PRIMARY score, mean and maximum intraprostatic standardized uptake values) were incorporated in the Amsterdam-Brisbane-Sydney nomogram were evaluated using the area under the receiver operating characteristic curve (AUC) in a second analysis for a subset of 139 patients. The performance of the new models was analyzed using likelihood ratio tests. In the primary cohort of 191 patients, 35 (18.3%) had stage pN1 at RP. AUC values were 0.751 (95% confidence interval CI 0.676–0.826) for the Briganti 2012 nomogram, 0.722 (95% CI 0.641–0.803) for the Briganti 2017 nomogram, 0.725 (95% CI 0.643–0.807) for the Memorial Sloan Kettering Cancer Center nomogram, and 0.862 (95% CI 0.794–0.929) for the Amsterdam-Brisbane-Sydney nomogram. In the second analysis for the subset of 139 patients, the PRIMARY score was significantly higher in the pN1 group ( p = 0.011). The AUC for a new model incorporating the PRIMARY score in the Amsterdam-Brisbane-Sydney nomogram was 0.870 (95% CI 0.791–0.949), which surpasses AUC results for the other models. The performance of the newly developed models was significantly better than the original nomogram according to likelihood ratio tests ( p < 0.001 for all models). The Amsterdam-Brisbane-Sydney nomogram outperformed other nomograms in predicting pN1 status. The PRIMARY score was significantly higher for a pN1 subgroup versus pN0, which indicates its potential clinical value. These findings suggest that the PRIMARY score and the Amsterdam-Brisbane-Sydney nomogram show promise for pN1 prediction. To the best of our knowledge, this is the first study to investigate the relationship between the PRIMARY score and pN1 status. Further research is needed to clarify the role of the PRIMARY score in clinical decision-making. We looked at how well different tools predict whether patients with prostate cancer are likely to have metastasis in their pelvic lymph nodes. We found that combining a tool called the Amsterdam-Brisbane-Sydney nomogram with a score for PET (positron emission tomography) scans called the PRIMARY score gave the best prediction performance. This combined tool could help in planning the extent of surgery for patients with intermediate-risk or high-risk prostate cancer if our results are confirmed in other studies.