EUROPEAN UROLOGY 61 (2012)

Transcription

1 EUROPEAN UROLOGY 61 (2012) available at journal homepage: Platinum Priority Prostate Cancer Editorial by A. Heidenreich on pp of this issue Updated Nomogram Predicting Lymph Node Invasion in Patients with Prostate Cancer Undergoing Extended Pelvic Lymph Node Dissection: The Essential Importance of Percentage of Positive Cores Alberto Briganti a, *, Alessandro Larcher a, Firas Abdollah a, Umberto Capitanio a, Andrea Gallina a, Nazareno Suardi a, Marco Bianchi a, Maxine Sun c, Massimo Freschi b, Andrea Salonia a, Pierre I. Karakiewicz c, Patrizio Rigatti a, Francesco Montorsi a a Department of Urology, Vita-Salute University, San Raffaele Scientific Institute, Milan, Italy; b Department of Pathology, Vita-Salute University, San Raffaele Scientific Institute, Milan, Italy; c Cancer Prognostics and Health Outcomes Unit, University of Montreal Health Centre, Montreal, Quebec, Canada Article info Article history: Accepted October 27, 2011 Published online ahead of print on November 7, 2011 Keywords: Prostate cancer Lymph node invasion Percentage of positive cores Nomogram Abstract Background: Few predictive models aimed at predicting the presence of lymph node invasion (LNI) in patients with prostate cancer (PCa) treated with extended pelvic lymph node dissection (eplnd) are available to date. Objective: Update a nomogram predicting the presence of LNI in patients treated with eplnd at the time of radical prostatectomy (RP). Design, setting, and participants: The study included 588 patients with clinically localised PCa treated between September 2006 and October 2010 at a single tertiary referral centre. Intervention: All patients underwent RP and eplnd invariably including removal of obturator, external iliac, and hypogastric nodes. Measurements: Prostate-specific antigen, clinical stage, and primary and secondary biopsy Gleason grade as well as percentage of positive cores were included in univariable (UVA) and multivariable (MVA) logistic regression models predicting LNI and formed the basis for the regression coefficient-based nomogram. The area under the curve (AUC) method was used to quantify the predictive accuracy (PA) of the model. Results and limitations: The mean number of lymph nodes removed and examined was 20.8 (median: 19; range: 10 52). LNI was found in 49 of 588 patients (8.3%). All preoperative PCa characteristics differed significantly between LNI-positive and LNI-negative patients (all p < 0.001). In UVA predictive accuracy analyses, percentage of positive cores was the most accurate predictor of LNI (AUC: 79.5%). At MVA, clinical stage, primary biopsy Gleason grade, and percentage of positive cores were independent predictors of LNI (all p 0.006). The updated nomogram demonstrated a bootstrap-corrected PA of 87.6%. Using a 5% nomogram cut-off, 385 of 588 patients (65.5%) would be spared eplnd. and LNI would be missed in only 6 patients (1.5%). The sensitivity, specificity, and negative predictive value associated with the 5% cut-off were 87.8%, 70.3%, and 98.4%, respectively. The relatively low number of patients included as well as the lack of an external validation represent the main limitations of our study. Conclusions: We report the first update of a nomogram predicting the presence of LNI in patients treated with eplnd. The nomogram maintained high accuracy, even in more contemporary patients (87.6%). Because percentage of positive cores represents the foremost predictor of LNI, its inclusion should be mandatory in any LNI prediction model. Based on our model, those patients with a LNI risk < 5% might be safely spared eplnd. # 2011 European Association of Urology. Published by Elsevier B.V. All rights reserved. * Corresponding author. Department of Urology, Vita Salute University, Via Olgettina 60, Milan, Italy. Tel ; Fax: address: briganti.alberto@hsr.it (A. Briganti) /$ see back matter # 2011 European Association of Urology. Published by Elsevier B.V. All rights reserved. doi: /j.eururo

2 EUROPEAN UROLOGY 61 (2012) Introduction Pelvic lymphadenectomy (PLND) is an essential staging procedure for the detection of nodal metastases in prostate cancer (PCa) [1]. Prevalence of lymph node invasion (LNI) at PLND ranges from 1.1% to 26% in contemporary patients [2 10]. Such variability in the rate of LNI relates to PCa characteristics at diagnosis as well as to the extent of PLND. Several reports have clearly shown that extended PLND (eplnd) is associated with a higher lymph node detection rate as compared with limited PLND, regardless of PCa aggressiveness [7 10]. For this reason, general agreement has been reached on the indication for performing eplnd whenever PLND is indicated [1,11]. Conversely, a limited PLND should be abandoned due to the high rates of false-negative findings [1]. Based on these considerations, we can state that only predictive models developed on eplnd series can reliably assess the risk of LNI in PCa. Despite this evidence, only three nomograms aimed at predicting LNI have been developed in patients treated with eplnd [5,6,12]. All of these eplnd-derived nomograms have demonstrated high accuracy when internally validated, but ideally the performance characteristics of any prediction model should be externally validated prior to use in clinical practice [13]. Unfortunately, of the three nomograms, only one [6] has been externally validated to date [14]. This nomogram is strengthened by the inclusion of percentage of positive cores as a covariate. This is key because the percentage of positive cores represents the foremost predictor of nodal metastases in PCa [6,14]. Therefore, inclusion of this biopsy-derived information should be mandatory in any LNI prediction model. For all these reasons, we consider this model a robust and accurate tool for the assessment of LNI risk. Besides external validation, this nomogram or any predictive tool requires periodic reappraisals in contemporary cohorts to ensure temporal validity [13]. Tools developed in historical patient cohorts might not be applicable in more contemporary patients, who are often diagnosed with more favourable PCa stage and grade [15,16]. Significant changes in PCa characteristics at presentation have also been demonstrated within the past decade [17]. Therefore, we considered it important to update the mentioned nomogram and to evaluate its performance characteristics in more contemporary patients treated with eplnd at radical prostatectomy (RP). 2. Materials and methods 2.1. Patient population Clinical and pathologic data were prospectively gathered for 588 patients treated with RP and eplnd between September 2006 and October 2010 at our institution. All patients were subjected to detailed preoperative evaluation consisting of prostate-specific antigen (PSA) assay (Abbott AxSYM PSA assay; Abbott Laboratories, Abbott Park, IL, USA), clinical stage assessed by the attending urologist (according to the 2002 American Joint Committee on Cancer staging system), and transrectal ultrasound-guided prostate biopsy. In addition to Gleason sum, the total number ofcorestaken aswell as the number and the percentage ofpositive cores were recorded for each patient. All men had histologically proven PCa and underwent RP preceded by eplnd, regardless of PCa characteristics. All cases were performed by six different surgeons who applied the same anatomic template during eplnd. The eplnds consisted of excision of fibrofatty tissue along the external iliac vein, with the distal limit the deep circumflex vein and the femoral canal. Proximally, eplnd was performed up to and including the bifurcation of the common iliac artery. All fibrofatty tissue within the obturator fossa was removed to skeletonise the obturator nerve completely. The lateral limit consisted of the pelvic sidewall, and the medial dissection limit was defined by perivesical fat. Lymph nodes along the internal iliac artery were also removed. Pelvic lymph node specimens were submitted for pathologic evaluation in multiple packages to improve the quality of specimen assessment [18]. Fat tissue containing lymph nodes were fixed in 10% buffered formalin. For each anatomic group, the number of nodes, the size of the largest node, and any gross features were described. Macroscopic specimen assessment was based on tactile and visual criteria. Large nodes (>2 cm) were sampled in multiple blocks. If no lymph nodes were macroscopically detected, all fat tissue was processed. Clearing solution was generally not used for pelvic lymph nodes. All blocks were embedded in paraffin, cut at 3 mm, and stained with haematoxylin-eosin. In selected cases immunohistochemical stain for cytokeratin and multiple sections were analyzed. Only cases submitted to eplnd with 10 nodes removed and examined were included [5,6,8]. No patient received neoadjuvant androgen-deprivation therapy Statistical analyses Descriptive statistics focused on frequencies and proportions for categorical variables. Means, medians, and ranges were reported for continuously coded variables. Chi-square test and t test were used to compare proportions and means, respectively. Univariable (UVA) and multivariable (MVA) logistic regression models predicting the presence of LNI at eplnd were fitted. Covariates consisted of preoperative PSA (coded as continuous variable), clinical stage, and primary and secondary biopsy Gleason grade (categorised as 3 vs 4) with or without percentage of positive cores (defined as the number of positive cores over the total number of cores taken; full vs base model, respectively). Regression coefficients were used to develop the updated nomogram that predicts the probability of LNI at eplnd. The area under curve (AUC) was used to quantify the tool s predictive accuracy. The nomogram was subjected to 200 bootstrap resamples for reduction of overfit bias and for internal validation. Bootstrap-corrected AUC of the model including percentage of positive cores as covariate (full model) was compared with the base model, in which information from biopsy cores was not considered. The Mantel-Haenszel test was used to assess the statistical significance of the difference in predictive accuracy between the two models. Preliminary analyses with restricted cubic spline models were conducted to check for the potential nonlinear relationship between the outcome and the continuously coded predictors (namely, PSA and percentage of positive cores). Because this approach did not improve the predictive accuracy of the model, only linear models were reported. For each nomogram cut-off, specificity, sensitivity, and negative predictive values (NPVs) were calculated. In addition, the extent of over- or underestimation of the observed LNI rate was explored graphically in logistic calibration plots for both models. Finally, a decision curve analysis was performed to evaluate and compare the net benefit associated with the use of the two models. All statistical analyses were performed using the R statistical package (R Foundation for Statistical Computing, Vienna, Austria). All tests were two sided, with a significance level set at 0.05.

3 482 EUROPEAN UROLOGY 61 (2012) Results Table 1 summarises the descriptive statistics of the nomogram development cohort (patients treated between November 2002 and August 2006) [7] and the update cohort (patients treated between September 2006 and October 2010). For the update population, data are tabulated according to the presence or absence of lymph node metastases. Mean number of lymph nodes removed and examined was 20.8 (median: 19). LNI was found in 49 of 588 patients (8.3%). Clinical stage, mean preoperative PSA, primary and secondary biopsy Gleason grade, and mean percentage of positive cores differed significantly between LNI-positive and LNI-negative patients (all p < 0.001). Conversely, the two populations of patients did not differ significantly in terms of mean age at surgery, extent of PLND, or biopsy sampling (all p 0.06; Table 1). Table 2 shows the UVA and MVA testing the association between predictors and presence of LNI. At UVA, all variables were significantly associated with the presence of LNI at eplnd (all p < 0001). In UVA predictive accuracy analyses, percentage of positive cores was the most accurate predictor of LNI (77.7%), followed by clinical stage (74.8%), primary biopsy Gleason grade (72.5%), preoperative PSA (67.4%), and secondary biopsy Gleason grade (64.9%). At MVA, clinical stage, primary biopsy Gleason grade, and percentage of positive cores achieved an independentpredictor status (all p 0.006), whereas PSA and secondary biopsy Gleason grade did not (all p 0.15). After 200 bootstrap resamples, the predictive accuracy of the full Table 1 Descriptive characteristics of patients treated between November 2002 and August 2006 (nomogram development cohort) [6] and patients treated between September 2006 and October 2010 (nomogram update cohort) * Overall nomogram development cohort ( ) [6] n = 278 Overall n = 588 (100%) Update nomogram cohort ( ) pn0 n = 539 (91.6%) pn1 n = 49 (8.3%) p value Age, yr 0.9 Mean (median) 66.2 (66.6) 65.2 (65.9) 65.2 (66.0) 65.3 (65.3) IQR PSA level, ng/ml <0.001 Mean (median) 9.9 (7.5) 8.0 (6.3) 7.5 (6.1) 12.8 (9.0) IQR Clinical stage (%) <0.001 T1 165 (59.4) 373 (63.4) 360 (66.8) 13 (26.5) T2 105 (37.8) 184 (31.3) 164 (30.4) 20 (40.8) T3 8 (2.9) 31 (5.3) 15 (2.8) 16 (32.7) Primary Gleason grade (%) < (91.7) 488 (83.0) 466 (86.5) 22 (44.9) 4 23 (8.3) 100 (17.0) 73 (13.5) 27 (55.1) Secondary Gleason grade (%) < (67.6) 406 (69.0) 386 (71.6) 20 (40.8) 4 90 (32.4) 182 (31.0) 153 (28.4) 29 (59.2) No. of cores taken 0.11 Mean (median) 12 (12) 17 (16.5) 17.3 (17) 16.1 (15.5) IQR No. of positive cores <0.001 Mean (median) 5.1 (4.0) 7.4 (6) 7.1 (6) 10.2 (10.5) IQR Percentage of positive cores, % <0.001 Mean (median) 44.7 (37.5) 43.6 (35.5) 41.0 (33.3) 71.7 (78.6) IQR Pathologic stage <0.001 pt2 199 (71.5) 431 (73.3) 427 (79.2) 4 (8.2) pt3a 48 (17.3) 97 (16.5) 84 (15.6) 13 (26.5) pt3b 31 (11.1) 58 (9.9) 28 (5.2) 30 (61.2) pt4 2 (0.3) 0 2 (4.1) Pathologic Gleason score < (42.1) 223 (37.9) 222 (41.2) 1 (2) (46.0) 282 (48) 260 (48.2) 22 (44.9) (11.9) 63 (10.7) 43 (8) 20 (40.8) Missing 20 (3.4) 14 (2.6) 6 (12.3) No. of removed and examined lymph nodes 0.06 Mean (median) 17.5 (16.5) 20.8 (19.0) 20.5 (19.0) 23.3 (21.0) IQR No. of positive lymph nodes in patients with LNI NA Mean (median) 2.6 (1.0) 2.9 (2) NA 2.9 (2) IQR IQR = interquartile range; PSA = prostate-specific antigen; LNI = lymph node invasion; NA = not applicable. * All patients underwent radical prostatectomy and extended pelvic lymph node dissection. The update cohort is stratified according to lymph node status.

4 EUROPEAN UROLOGY 61 (2012) Table 2 Univariable and multivariable logistic regression analyses predicting the presence of lymph node invasion * Covariates Univariable analysis Multivariable analysis Base model Full model Odds ratio (95% CI) p value AUC (%) Odds ratio (95% CI) p value Odds ratio (95% CI) p value Prostate-specific antigen level, ng/ml 1.10 ( ) < (1 1.1) ( ) 0.15 Clinical stage < T2 vs T ( ) ( ) ( ) 0.14 T3 vs T ( ) < ( ) ( ) Primary Gleason grade 4 vs ( ) < ( ) < ( ) <0.001 Secondary Gleason grade 4 vs ( ) < ( ) ( ) 0.31 Percentage of positive cores, % 1.04 ( ) < ( ) <0.001 AUC of multivariable models, % (gain: 5.1; p < 0.001) CI = confidence interval; AUC = area under the curve. * In 588 patients treated with radical prostatectomy and extended lymph node dissection at a single institution between 2006 and model was 87.6% versus 82.5% for the base model (5.1% gain; p < 0.001). Figure 1 graphically displays the MVA effect of predictor variables on the risk of LNI in the form of a nomogram. The calibration plot of predicted probabilities against observed LNI rates indicated better concordance for the full versus the base model (Fig. 2). Similarly, decision curve analysis [(Fig._1)TD$FIG] showed that the full model (Model 2) had the highest clinical net benefit across the entire range of threshold probabilities (Fig. 3). Table 3 further illustrates the predictive accuracy and error associated with the nomogram when it predicts that a patient has a low risk of LNI. Nomogram-derived predicted probabilities of positive lymph nodes are Fig. 1 Nomogram predicting the probability of lymph nodes invasion (LNI) in patients undergoing extended pelvic lymphadenectomy based on pretreatment prostate-specific antigen (PSA), clinical stage, primary and secondary biopsy Gleason score, and percentage of positive cores. Instructions: Locate the patient s pretreatment PSA on the PSA axis. Draw a line straight upward to the point axis to determine how many points toward the probability of positive lymph nodes the patient receives for his PSA value. Repeat the process for each additional variable. Sum the points for each of the predictors. Locate the final sum on the total point axis. Draw a line straight down to find the patient s probability of having LNI.

5 484 [(Fig._2)TD$FIG] EUROPEAN UROLOGY 61 (2012) [(Fig._3)TD$FIG] Fig. 2 Nomogram calibration plot. Dotted line indicates location of the ideal nomogram, in which predicted and actual probabilities are identical. Broken line indicates actual nomogram performance. Expected performance on future data is represented through the solid line. Calibration was tested for both models, with or without percentage of positive cores (PCC) (full and base model, respectively). Fig. 3 Decision curve analyses demonstrating the net benefit associated with the use of the nomogram with (model 2: full model) or without (model 1: base model) the inclusion of the percentage of positive cores as covariate. Net benefit can be interpreted as the proportion of all patients who have lymph node metastases and are recommended for extended pelvic lymph node dissection if no patients with negative lymph nodes were treated. categorised into strata. The number of patients actually having negative lymph nodes and those with LNI as well as sensitivity, specificity, and NPVs are depicted for each stratum. With this premise, using a nomogram cut-off of 5%, 385 of 588 patients (65.5%) would be spared eplnd, and LNI would be missed in only 6 patients (1.5%). All these six patients had two or fewer positive lymph nodes. Therefore, use of such a cut-off was associated with incorrect staging of a low number of patients with a low volume of nodal invasion. The sensitivity, specificity, and NPV associated with the 5% cut-off were 87.8%, 70.3%, and 98.4%, respectively. Table 3 Systematic analyses of the updated nomogram-derived cut-offs used to discriminate between patients with or without histologically confirmed lymph node invasion * Nomogram calculated probability of LNI, cut-off, % Patients in whom PLND is not recommended according to the cut-off (below cut-off) Patients below cut-off without histologic LNI ** Patients below cut-off with histologic LNI y Patients in whom PLND is recommended according to the cut-off (above cut-off) Patients above cut-off without histologic LNI z Patients above cut-off with histologic LNI NPV (17.3) 102 (18.9) 0 (0) 486 (82.7) 437 (81.1) 49 (100.0) (41.3) 241 (44.7) 2 (4.1) 345 (58.7) 298 (55.3) 47 (95.9) (53.1) 307 (57.0) 5 (10.2) 276 (46.9) 232 (43.0) 44 (89.8) (60.9) 353 (65.5) 5 (10.2) 230 (39.1) 186 (34.5) 44 (89.8) (65.5) 379 (70.3) 6 (12.2) 203 (34.5) 160 (29.7) 43 (87.8) (68.5) 396 (73.5) 7 (14.3) 185 (31.5) 143 (26.5) 42 (85.7) (71.6) 414 (76.8) 7 (14.3) 167 (28.4) 125 (23.2) 42 (85.7) (74.8) 431 (80.0) 9 (18.4) 148 (25.2) 108 (20.0) 40 (81.6) (77.6) 446 (82.7) 10 (20.4) 132 (22.4) 93 (17.3) 39 (79.6) (79.8) 458 (85.0) 11 (22.4) 119 (20.2) 81 (15.0) 38 (77.6) (84.5) 479 (88.9) 18 (36.7) 91 (15.5) 60 (11.1) 31 (63.3) (88.8) 497 (92.2) 25 (51.0) 66 (11.2) 42 (7.8) 24 (49.0) 95.2 LNI = lymph node invasion; PLND = pelvic lymph node dissection; NPV = negative predictive value. * In 588 patients treated with radical prostatectomy and extended pelvic lymph node dissection between 2006 and ** Percentage is indicative of specificity. y Percentage is indicative of 1, sensitivity. z Percentage is indicative of 1, specificity. Percentage is indicative of sensitivity.

6 EUROPEAN UROLOGY 61 (2012) Discussion Although the ideal candidate for PLND still needs to be defined, general agreement has been reached that if PLND is contemplated, it should be extended to reach an adequate nodal staging accuracy [1,11]. Consequently,if an LNI prediction model is developed, it must be based on eplnd series. Conversely, applying predictive tools developed on patients treated with limited PLND might significantly underestimate the true risk of LNI [2 4,19]. The need for such a tool might extend beyond the need of accurate staging because some investigators advocate a curative effect of eplnd in highly selected patients with LNI [20,21]. Despite these data, only three LNI nomograms developed on eplnd series are available to date [5,6,12]. Two of the three were developed and internally validated by Briganti et al. in European patients [5,6]. The first nomogram was 76% accurate and relied on PSA, clinical stage, and biopsy Gleason sum [5]. The second nomogram included the percentage of positive cores as a covariate [6]. Interestingly, inclusion of this biopsy information significantly increased the accuracy of LNI prediction. This nomogram was recently validated externally and demonstrated good performance characteristics in a large low-risk PCa patient population treated with eplnd [12]. However, nomograms as well as any other predictive model need periodic reassessment to address temporal validity [13]. Therefore, we decided to update our prediction model and test its current applicability in a more recent cohort of patients treated with eplnd and RP at a single tertiary referral centre between September 2006 and October We demonstrated that the nomogram is still highly accurate in detecting LNI (AUC: 87.6%) in these patients. Its calibration showed good concordance between predicted and observed LNI probabilities (Fig. 2). The application of this nomogram in clinical practice might contribute significantly to the selection of patient candidates for this extensive surgical approach and might spare the remaining patients eplnd. Using a cut-off of 5%, roughly two-thirds of patients (385 of 588; 65.5%) would be spared eplnd and LNI would be missed in only 6 men (1.5%), with a sensitivity, specificity, and NPV of 87.8%, 70.3%, and 98.4%, respectively. Although to date there is no consensus regarding what can be considered an optimal cut-off to perform eplnd, our results are in line to what was recommended by the 2010 National Comprehensive Cancer Network guidelines [22]. According to these guidelines, a cut-off that allows the avoidance of roughly 50% of PLNDs prior to RP at the cost of missing 12% of patients with LNI is considered acceptable [22]. In our analyses, a cut-off of 5% would warrant such results (Table 3). However, even in the absence of prospective randomised trials, previous studies have shown that more extensive PLND reduces the long-term risk of PCa recurrence and death [23,24]. In fact, traces of disseminated PCa cells may be found at the molecular level, albeit conventional histopathology turns negative. Therefore, one might speculate that eplnd should be recommended to everyone who is a true candidate for RP. Prospective randomised trials are needed to evaluate the effect of PLND in patients with PCa. Several results of our study deserve attention. First, our nomogram represents a highly accurate tool for the prediction of LNI in patients treated with eplnd. Second, we confirmed our previous finding supporting the role of percentage of positive cores as the foremost predictor of LNI [6,14]. Percentage of positive cores represented the most informative predictor of LNI. Its inclusion was able to significantly improve the AUC of a base model including only PSA, clinical stage, and biopsy Gleason grades as covariates (87.6 vs 82.5% for the full vs the base model; 5.1% gain; p < 0.001). The full model including percentage of positive cores had better concordance and higher net benefit as compared with the base model (Figs. 2 and 3). As a consequence, the lack of inclusion of the percentage of positive cores in LNI prediction models might significantly decrease the accuracy in assessing the risk of nodal invasion, even when models based on eplnd series are applied. Despite this evidence, no other tool included percentage of positive cores in LNI risk assessment. This represents a strength of our tool. Third, our predictive model is strengthened by the extensive nodal dissection performed (mean number of nodes removed: 20.8; median: 19) as well by the extent of biopsy sampling (mean number of total cores taken: 17; median: 16.5). Despite several strengths, our study is not devoid of limitations. Our sample size could ideally be larger and might not be representative of other cohorts. The performance characteristics of the nomogram might have been also influenced by the relatively low number of patients as well of events in the study population [25,26]. Because variations across the data sets can affect the accuracy of any prediction model, we cannot exclude that the performances of the updated nomogram might have increased by the presence of more aggressive cancer features at diagnosis (Table 1). The AUC of any prediction model indeed depends on the variation of the predictors as well as of the outcome rates [26]. Even in the presence of previous external validation, any updated model should also undergo new external validation [13,26]. Although internal validation represents an excellent substitute for an external cohort, the latter still represents the gold standard for confirming the validity of novel findings. Therefore, prior to suggesting its use in the everyday clinical practice, a formal external validation of the model is required. Other study limitations are related to differences between the number of lymph nodes removed and the number of lymph nodes actually examined by the pathologist. Variability of surgical technique and expertise [27] may contribute to differences in the number of lymph nodes removed, despite lymph node dissections within the same anatomic locations. Pathologic evaluation of node specimens may also account for different numbers of nodes identified and examined. Pathologic specimen dissection based on tactile findings may miss nonpalpable lymph nodes, especially when a large amount of fatty tissue surrounds them. Thus, at pathologic evaluation, an LNI diagnostic bias may be introduced if size and consistency are used as criteria for the identification of suspicious

7 486 EUROPEAN UROLOGY 61 (2012) nodes. This can be avoided by submitting pelvic lymph node specimens in multiple packages [18]. Despite these limitations, our study represents the most accurate tool available to date for LNI prediction in patients undergoing eplnd. 5. Conclusions We updated and internally validated the most accurate nomogram (AUC: 87.6%) predicting the probability of LNI in patients undergoing eplnd at RP. It is based on readily and routinely available clinical parameters, such as PSA, clinical stage, biopsy Gleason sum, and percentage of positive cores. Among these, the percentage of positive cores represents the foremost predictor of LNI and should be included in any LNI prediction model. Using a 5% nomogram cut-off, roughly two-thirds of patients would be spared eplnd, and LNI would be missed in only 1.5%. Therefore, in the absence of prospective data supporting the role of eplnd in PCa outcome, eplnd might be safely omitted in all patients with a nomogram-derived LNI risk <5%. Author contributions: Alberto Briganti had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Briganti, Karakiewicz, Montorsi, Rigatti, Suardi. Acquisition of data: Larcher, Abdollah, Capitanio, Gallina, Freschi. Analysis and interpretation of data: Briganti, Larcher, Abdollah, Capitanio, Gallina, Salonia. Drafting of the manuscript: Briganti, Larcher, Abdollah. Critical revision of the manuscript for important intellectual content: Briganti, Karakiewicz, Salonia, Rigatti, Montorsi. Statistical analysis: Briganti, Abdollah, Capitanio, Gallina. Obtaining funding: None. Administrative, technical, or material support: None. Supervision: Montorsi, Rigatti. Other (specify): None. Financial disclosures: I certify that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/ affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None. Funding/Support and role of the sponsor: None. References [1] Heidenreich A, Bellmunt J, Bolla M, et al. EAU guidelines on prostate cancer. Part 1: screening, diagnosis, and treatment of clinically localised disease. Eur Urol 2011;59: [2] Cagiannos I, Karakiewicz P, Eastham JA, et al. A preoperative nomogram identifying decreased risk of positive pelvic lymph nodes in patients with prostate cancer. J Urol 2003;170: [3] Kattan MW, Stapleton AM, Wheeler TM, et al. Evaluation of a nomogram used to predict the pathologic stage of clinically localized prostate carcinoma. Cancer 1997;79: [4] Makarov DV, Trock BJ, Humphreys EB, et al. Updated nomogram to predict pathologic stage of prostate cancer given prostate-specific antigen level, clinical stage, and biopsy Gleason score (Partin tables) based on cases from 2000 to Urology 2007;69: [5] Briganti A, Chun FK-H, Salonia A, et al. Validation of a nomogram predicting the probability of lymph node invasion among patients undergoing radical prostatectomy and an extended pelvic lymphadenectomy. Eur Urol 2006;49: [6] Briganti A, Karakiewicz PI, Chun FK-H, et al. Percentage of positive biopsy cores can improve the ability to predict lymph node invasion in patients undergoing radical prostatectomy and extended pelvic lymph node dissection. Eur Urol 2007;51: [7] Heidenreich A, Varga Z, Von Knobloch R. Extended pelvic lymphadenectomy in patients undergoing radical prostatectomy: high incidence of lymph node metastasis. J Urol 2002;167: [8] Schumacher MC, Burkhard FC, Thalmann GN, et al. Is pelvic lymph node dissection necessary in patients with a serum PSA < 10ng/ml undergoing radical prostatectomy for prostate cancer?. Eur Urol 2006;50: [9] Touijer K, Rabbani F, Otero JR, et al. Standard versus limited pelvic lymph node dissection for prostate cancer in patients with a predicted probability of nodal metastasis greater than 1%. J Urol 2007; 178: [10] Briganti A, Chun FK, Salonia A, et al. Critical assessment of ideal nodal yield at pelvic lymphadenectomy to accurately diagnose prostate cancer nodal metastasis in patients undergoing radical retropubic prostatectomy. Urology 2007;69: [11] Briganti A, Blute ML, Eastham JH, et al. Pelvic lymph node dissection in prostate cancer. Eur Urol 2009;55: [12] Godoy G, Chong KT, Cronin A, et al. Extent of pelvic lymph node dissection and the impact of standard template dissection on nomogram prediction of lymph node involvement. Euro Urol 2011;60: [13] Shariat SF, Karakiewicz PI, Suardi N, et al. Comparison of nomograms with other methods for predicting outcomes in prostate cancer: a critical analysis of the literature. Clin Cancer Res 2008;14: [14] Heidenreich A, Pfister D, Thuer D, et al. Percentage of positive biopsies predicts lymph node involvement in men with low-risk prostate cancer undergoing radical prostatectomy and extended pelvic lymphadenectomy. BJU Int 2011;107: [15] Gallina A, Chun FK, Suardi N, et al. Comparison of stage migration patterns between Europe and the USA: an analysis of men treated with radical prostatectomy for prostate cancer. BJU Int 2008; 101: [16] Shao YH, Demissie K, Shih W, et al. Contemporary risk profile of prostate cancer in the United States. J Natl Cancer Inst 2009;101: [17] Budäus L, Spethmann J, Isbarn H, et al. Inverse stage migration in patients undergoing radical prostatectomy: results of 8916 European patients treated within the last decade. BJU Int. In press. DOI: /j X x. [18] Bochner BH, Choi D, Herr HW, et al. Prospectively packaged lymph node dissections with radical cystectomy. Evaluation of node count variability and node mapping. J Urol 2004;172: [19] Abdollah F, Sun M, Thuret R, et al. Decreasing rate and extent of lymph node staging in patients undergoing radical prostatectomy may undermine the rate of diagnosis of lymph node metastases in prostate cancer. Eur Urol 2010;58: [20] Schumacher MC, Burkhard FC, Thalmann GN, et al. Good outcome for patients with few lymph node metastases after radical retropubic prostatectomy. Eur Urol 2008;54: [21] Cheng L, Zincke H, Blute ML, et al. Risk of prostate carcinoma death in patients with lymph node metastasis. Cancer 2001;91: [22] Mohler J, Bahnson RR, Boston B, et al. NCCN clinical practice guidelines in oncology: prostate cancer. J Natl Compr Canc Netw 2010;8:

8 EUROPEAN UROLOGY 61 (2012) [23] Joselyn Joslyn SA, Konety BR. Impact of extent of lymphadenectomy on survival after radical prostatectomy for prostate cancer. Urology 2006;68: [24] Masterson TA, Bianco Jr FJ, Vickers AJ, et al. The association between total and positive lymph node counts, and disease progression in clinically localized prostate cancer. J Urol 2006;175: [25] Vickers AJ, Cronin AM. Everything you always wanted to know about evaluating prediction models (but were too afraid to ask). Urology 2010;76: [26] Kattan MW. Factors affecting the accuracy of prediction models limit the comparison of rival prediction models when applied to separate data sets. Eur Urol 2011;59: [27] Briganti A, Capitanio U, Chun FK-H, et al. Impact of surgical volume on the rate of lymph node metastases in patients undergoing radical prostatectomy and extended pelvic lymph node dissection for clinically localized prostate cancer. Eur Urol 2008; 54: