Elsevier

European Urology

Volume 60, Issue 2, August 2011, Pages 195-201
European Urology

Platinum Priority – Prostate Cancer
Editorial by R. Jeffrey Karnes on pp. 202–203 of this issue
Extent of Pelvic Lymph Node Dissection and the Impact of Standard Template Dissection on Nomogram Prediction of Lymph Node Involvement

https://doi.org/10.1016/j.eururo.2011.01.016Get rights and content

Abstract

Background

Our current lymph node involvement (LNI) nomogram was created using patients receiving both limited and standard lymph node dissection (LND). Over time, refinements in technique could affect the diagnostic yield from LND.

Objective

Our aim was to validate our existing LNI nomogram or develop a new nomogram with updated prediction coefficients that reflect the current standard LND template during radical prostatectomy (RP). We hypothesized that the existing nomogram would demonstrate good discrimination but poor calibration in a contemporary series of standard LND.

Design, setting, and participants

A retrospective analysis of 4176 consecutive primary RP patients was performed, including open procedures (3097 patients from 2000 to 2008) and laparoscopic procedures (1079 patients from 2005 to 2008). After excluding 127 patients (3%) with limited LND, 10 (0.2%) with pretreatment prostate-specific antigen (PSA) >50 ng/ml, and 318 (8%) with incomplete data, the final cohort totaled 3721 patients. The nomograms were evaluated using receiver operating characteristic analysis, calibration plots, and decision-curve analysis.

Interventions

Patients received open or laparoscopic (conventional and robot-assisted) RP and standard LND in our center.

Measurements

Assessments were obtained using preoperative PSA, biopsy Gleason score, and clinical stage.

Results and limitations

The median number of nodes removed was 11, with ∼60% of patients having at least 10 nodes removed (n = 2224). Overall, 5.2% of patients (n = 194) had positive lymph nodes. The new nomogram had very high discriminative accuracy (area under the curve: 0.862). The decision-curve analysis showed that the new nomogram had the highest clinical net benefit for all reasonable threshold probabilities.

Conclusions

The new nomogram shows improved calibration when predicting lymph node invasion in a contemporary cohort of patients with prostate cancer exclusively treated with RP and standard LND. This nomogram will be used as the preferred predictive model for counseling patients and developing studies at our institution.

Introduction

Radical prostatectomy (RP) is an effective method of treating men with clinically localized prostate cancer. Lymph node dissection (LND) is an important oncologic component of the procedure, and its role in staging and as a therapeutic intervention has been the focus of renewed interest. The presence of lymph node metastasis is associated with an elevated risk of systemic dissemination of disease and death [1], [2]. Additional diagnostic and therapeutic benefits have been recognized when performing LND adjunctively during RP [3], [4], [5], [6].

Stage migration has resulted in a greater proportion of cases detected at early stages with a lower incidence of positive lymph nodes. These factors, allied with the potential risk of morbidity from LND, may be contributing to a decrease in the number or extent of LND procedures performed in the United States [7], [8], [9], [10]. To better estimate the likelihood of occult nodal disease and guide clinical decision making, we previously created a nomogram to predict the risk of lymph node involvement (LNI) at the time of the RP [11]. Of critical importance is the accuracy of such a clinical predictive tool, which depends on the data used to impute its functions. In the case of pelvic LNI, the extent of LND and tissue processing are of importance. As reported in several studies [3], [4], [5], [12], we have established the boundaries of a standard LND template in our center that includes the bilateral dissection of the external iliac, obturator, and hypogastric nodes.

We hypothesized that the current nomogram would demonstrate good discrimination but poor calibration when applied to a contemporary series constituted exclusively of standard LND cases. Specifically, we expected it to underestimate the real incidence of LNI during RP. Discrimination refers to the ability of a nomogram to rank patients by their risk. It is typically described in terms of the probability that in a given pair of patients, one with and one without LNI, the nomogram will give the patient with affected nodes a higher probability. Calibration summarizes how well the predicted incidence of LNI matches the observed incidence, that is, whether in 100 men given x% probability of LNI, close to x do indeed have positive nodes. We undertook this study with the objective of rebuilding the nomogram coefficients, including only patients who have undergone the standard LND in our institution, either to validate the accuracy of the current nomogram or to update the nomogram prediction formula to reflect contemporary LND procedures and results as performed during RP.

Section snippets

Patients

We identified 3097 consecutive patients who underwent primary open RP between October 2000 and October 2008, and 1079 consecutive patients who underwent primary laparoscopic RP between February 2005 and October 2008. Robot-assisted laparoscopic prostatectomy cases were included in this series, with standard LND confirmed by chart review in all cases included in the analysis. We excluded 127 patients (3.0%) with limited LND (57 robotic patients and 70 others). We excluded patients with

Results

Table 1 lists the patient characteristics. The median number of nodes removed was 11, with ∼60% of patients having at least 10 nodes removed (n = 2224). Overall, 5.2% of patients (n = 194) had positive lymph nodes. The mean predicted probability of LNI was 3.2% from the previously published three-variable nomogram and 5.7% from the previously published four-variable nomogram. Both previously published nomograms had very high discriminative accuracy (area under the curve [AUC] of 0.861 for the

Discussion

Since 2002 we have adopted several nomograms for the prediction of clinical and pathologic outcomes after RP. One of these nomograms was developed to predict the preoperative risk of LNI during RP [11]. However, it was developed based on a cohort of patients that underwent both standard and limited LND templates across different centers. It was previously shown that more extended LND templates yield significantly higher number of total and positive lymph nodes, resulting in a better staging

Conclusions

Our updated nomogram demonstrates high discriminative accuracy and improved calibration relative to our prior nomograms when predicting LNI in a contemporary cohort of patients with prostate cancer managed with standard template pelvic LND during RP. This nomogram will be used as the preferred predictive model for counseling patients and developing studies at our institution.

References (22)

Cited by (91)

  • Temporal Trends and Predictors in Diagnosing Pathologic Node-Positive Prostate Cancer in Clinically Node-Negative Patients

    2021, Clinical Genitourinary Cancer
    Citation Excerpt :

    Our analysis suggests that PLND with fewer than 7 LN sampled may be inadequate for identifying pN+ disease. The clinical importance of extent of LN dissection has been studied previously and seems to improve survival outcomes in patients with a high risk of nodal involvement.8,18,19 Furthermore, our survival analysis reveals that patients at the highest risk for nodal involvement (GG4 and GG5) who are managed with PLND have significantly improved survival over those who do not undergo PLND.

View all citing articles on Scopus

Please visit www.eu-acme.org/europeanurology to read and answer questions on-line. The EU-ACME credits will then be attributed automatically.

View full text