External validation of the updated nomogram predicting lymph node invasion in patients with prostate cancer...

The aim of our study was to determine the validity of the updated nomogram [Briganti et al.: Eur Urol 2012;61:480-487] as a prediction tool for pelvic lymph node invasion (LNI) in the current era by using a large multicentric population of men who underwent extended pelvic lymph node dissection (ePLND) at the time of radical prostatectomy (RP) at tertiary referral centers.

Between 2000 and 2011, 896 consecutive patients underwent RP and ePLND at two tertiary referral centers for clinically localized prostate cancer (PCa). Uni- and multivariable logistic regression models predicting the presence of LNI at ePLND were built in. Covariates consisted of preoperative PSA, clinical stage, primary and secondary biopsy Gleason grade with or without percentage of positive cores. Patients' data were entered into a logistic model formula derived from the original publication of Briganti. The nomogram was assessed by comparing its predicted probability of LNI with the actual presence of LNI. The area under the curve was used to quantify its predictive accuracy.

Mean preoperative PSA, clinical and pathological stage, primary and secondary biopsy and pathological Gleason grade, such as mean number of total cores, positive cores and percentage of positive cores differed significantly between LNI-positive and LNI-negative patients (all p < 0.001 except for number of total cores, p = 0.019). The mean number of lymph nodes removed was 14.8, and LNI was found in 101 patients (11.8%). In the univariate analysis the percentage of positive cores was the most accurate predictor of LNI (72%), followed by PSA (69%), primary biopsy Gleason grade (64%), clinical stage (60%), and secondary biopsy Gleason grade (59%). The predictions of the nomogram were virtually perfect when the predicted probability was ≤20%. We tested the performance characteristics of various Briganti nomogram-derived cut-offs (1-14%) for discriminating between patients with and without LNI. In our population, 41.6% of patients were classified below the 5% cut-off proposed in the original Briganti et al. report. In the multivariate analysis these variables remained statistically significant predictors for the presence of lymph node metastases. The predictive accuracy of the full model reached 79%.

The updated nomogram predicting LNI in patients with PCa undergoing ePLND has been externally validated, demonstrating excellent accuracy and calibration characteristics and a general applicability for predicting the presence of LNI.