Nomograms to identify prostate cancer (PCa) patients who should be considered for an extended pelvic lymph node dissection (ePLND) during radical prostatectomy (RP) do not account for information obtained at PSMA-PET. We hypothesized that available tools might be suboptimal and that they should be recalibrated in cN0 disease at PSMA-PET due to the high negative predictive value (NPV) for lymph node invasion (LNI) of PET.
Overall, 662 PCa patients who received a PSMA-PET and a subsequent RP and ePLND at 9 referral centres between 2016 and 2022 were identified. All patients received an anatomically defined ePLND regardless of PSMA-PET. All patients had available details on preoperative MRI, targeted and systematic biopsy and pathologic data. The calibration and discrimination of available models predicting LNI (Briganti 2012, Briganti 2017, and Briganti 2019 nomograms) was assessed at external validation. The same process was repeated in men with cN0MO disease at PSMA-PET. The Briganti 2019 model was recalibrated using the multivariable logistic regression coefficients of the variables included in the nomogram in men cN0MO disease at PSMA-PET. Calibration plots, the ROC-derived AUC, and decision-curve analyses (DCAs) were used to determine the calibration, discrimination, and net benefit associated with the recalibrated nomogram and to compare it with available tools.
Overall, 115 (17.4%) patients had positive spots in the pelvis at PSMA PET/CT. The rate of LNI was 19.5%. The discrimination of the Briganti 2012, 2017, and 2019 nomograms was 75, 75, and 71%. When focusing on patients without positive spots at PSMA-PET (n= 537, rate of LNI: 10%), the discrimination of the Briganti 2012, 2017, and 2019 nomograms was even lower (72, 72, and 70%). In these patients, the maximum diameter of the index lesion, seminal vesicle invasion at MRI, and a target biopsy ISUP grade group 4-5 were associated with a risk of LNI (all p<0.05). A nomogram based on the coefficients of this multivariable regression depicted a discrimination of 80%, better calibration characteristics, and a higher net-benefit compared to available models. The use of a 7% cut-off of the recalibrated nomogram in men with a cN0M0 disease at PSMA-PET would have allowed for sparing 55% ePLNDs (vs. 29% for the Briganti 2019 nomogram) at the cost of missing only 3.4% (vs. 2.9%) LNIs.
The use of the Briganti nomograms in men staged with PSMA-PET results in a high number of unnecessary ePLNDs due to the NPV of molecular imaging. A novel recalibrated version of the Briganti nomogram should be used to identify the candidates for ePLND to reduce the risk of unnecessary procedures without missing additional LNIs in men with cN0M0 PCa at PSMA-PET.