High Incidence of Prostate Cancer Detected by Saturation Biopsy after Previous Negative Biopsy Series

Similar documents
Zonal Origin of Localized Prostate Cancer Does not Affect the Rate of Biochemical Recurrence after Radical Prostatectomy

Prostate Cancer Gene 3 (PCA3): Development and Internal Validation of a Novel Biopsy Nomogram

Insignificant Prostate Cancer in Radical Prostatectomy Specimen: TimeTrends and Preoperative Prediction

european urology 52 (2007)

Cancer. Description. Section: Surgery Effective Date: October 15, 2016 Subsection: Original Policy Date: September 9, 2011 Subject:

Introduction. Key Words: high-grade prostatic intraepithelial neoplasia, HGPIN, radical prostatectomy, prostate biopsy, insignificant prostate cancer

BJUI. Follow-up of men with an elevated PCA3 score and a negative biopsy: does an elevated PCA3 score indeed predict the presence of prostate cancer?

External Validation of Urinary PCA3-Based Nomograms to Individually Predict Prostate Biopsy Outcome

A Nomogram Predicting Long-term Biochemical Recurrence After Radical Prostatectomy

Anatomic distribution and pathologic characterization of small-volume prostate cancer (o0.5 ml) in whole-mount prostatectomy specimens

EUROPEAN UROLOGY 58 (2010)

Owing to the widespread use of prostate specific antigen (PSA)

Preoperative Gleason score, percent of positive prostate biopsies and PSA in predicting biochemical recurrence after radical prostatectomy

THE DILEMMA OF PROSTATE CANcer

Are extended biopsies really necessary to improve prostate cancer detection?

Clinical Utility of the PCA3 Urine Assay in European Men Scheduled for Repeat Biopsy

1. Introduction. Department of Urology, Graduate School of Medicine, Chiba University, Inohana, Chuo-ku, Chiba , Japan 2

THE SIGNIFICANCE OF HYPOECHOIC LESION DIRECTED AND TRANSITION ZONE BIOPSIES IN IMPROVING THE DIAGNOSTIC ABILITY IN PROSTATE CANCER

Since the beginning of the prostate-specific antigen (PSA) era in the. Characteristics of Insignificant Clinical T1c Prostate Tumors

Fisher s exact test for contingency tables. A two-tailed p-value <0.05 was accepted as statistically significant.

BJUI. Study Type Prognosis (individual cohort study) Level of Evidence 2b OBJECTIVES CONCLUSIONS

MR-US Fusion Guided Biopsy: Is it fulfilling expectations?

Prostate Cancer. Axiom. Overdetection Is A Small Issue. Reducing Morbidity and Mortality

Understanding the risk of recurrence after primary treatment for prostate cancer. Aditya Bagrodia, MD

EUROPEAN UROLOGY 60 (2011)

Correlation of Gleason Scores Between Needle-Core Biopsy and Radical Prostatectomy Specimens in Patients with Prostate Cancer

concordance indices were calculated for the entire model and subsequently for each risk group.

Approximately 680,000 men are diagnosed with prostate

Predictive Models. Michael W. Kattan, Ph.D. Department of Quantitative Health Sciences and Glickman Urologic and Kidney Institute

Information Content of Five Nomograms for Outcomes in Prostate Cancer

The Chances of Subsequent Cancer Detection in Patients with a PSA > 20 ng/ml and an Initial Negative Biopsy

Prostate Overview Quiz

Contribution of prostate-specific antigen density in the prediction of prostate cancer: Does prostate volume matter?

Can nomograms be superior to other prediction tools?

INTRADUCTAL LESIONS OF THE PROSTATE. Jonathan I. Epstein

Outcomes of Radical Prostatectomy in Thai Men with Prostate Cancer

Research Article External Validation of an Artificial Neural Network and Two Nomograms for Prostate Cancer Detection

Use of early PSA velocity to predict eventual abnormal PSA values in men at risk for prostate cancer {

CONTEMPORARY UPDATE OF PROSTATE CANCER STAGING NOMOGRAMS (PARTIN TABLES) FOR THE NEW MILLENNIUM

Development and Internal Validation of a Prostate Health Index Based Nomogram for Predicting Prostate Cancer at Extended Biopsy

european urology 54 (2008)

Prostate Cancer: Is There Standard Treatment? Who has prostate cancer? In this article:

Post Radical Prostatectomy Radiation in Intermediate and High Risk Group Prostate Cancer Patients - A Historical Series

Supplemental Information

NIH Public Access Author Manuscript World J Urol. Author manuscript; available in PMC 2012 February 1.

Best Papers. F. Fusco

Saturation Biopsy for Diagnosis and Staging and Management of Prostate Cancer

Prostate-specific antigen density as a parameter for the prediction of positive lymph nodes at radical prostatectomy

Percentage of Gleason Pattern 4 and 5 Predicts Survival After Radical Prostatectomy

EUROPEAN UROLOGY 61 (2012)

of Nebraska - Lincoln

Extended 12-Core Prostate Biopsy Increases Both the Detection of Prostate Cancer and the Accuracy of Gleason Score

Predicting Prostate Biopsy Outcome Using a PCA3-based Nomogram in a Polish Cohort

Controversies in Prostate Cancer Screening

Prognostic Value of Surgical Margin Status for Biochemical Recurrence Following Radical Prostatectomy

Saturation Biopsy for Diagnosis and Staging and Management of Prostate Cancer

, De La Taille Alexandre * INSERM : U955, Universit é Paris XII Val de Marne, IFR10, FR. , Universit é Paris XII Val de Marne, Créteil,FR

When PSA fails. Urology Grand Rounds Alexandra Perks. Rising PSA after Radical Prostatectomy

Although the test that measures total prostate-specific antigen (PSA) has been

Computer simulated additional deep apical biopsy enhances cancer detection in palpably benign prostate gland

european urology 55 (2009)

Clinical Evaluation of the PCA3 Assay in Guiding Initial Biopsy Decisions

Saturation Biopsy for Diagnosis, Staging, and Management of Prostate Cancer

10/2/2018 OBJECTIVES PROSTATE HEALTH BACKGROUND THE PROSTATE HEALTH INDEX PHI*: BETTER PROSTATE CANCER DETECTION

External validation of the Briganti nomogram to estimate the probability of specimen-confined disease in patients with high-risk prostate cancer

ONCOLOGY LETTERS 8: , 2014

Prospective validation of a risk calculator which calculates the probability of a positive prostate biopsy in a contemporary clinical cohort

Case Discussions: Prostate Cancer

J Clin Oncol 25: by American Society of Clinical Oncology INTRODUCTION

A NEURAL NETWORK PREDICTS PROGRESSION FOR MEN WITH GLEASON SCORE 3 4 VERSUS 4 3 TUMORS AFTER RADICAL PROSTATECTOMY

A schematic of the rectal probe in contact with the prostate is show in this diagram.

or more transrectal ultrasonography (TRUS)-guided ng/ml and 39% if it was 20.0 ng/ml. of >10 ng/ml have prostate cancer [3], many other

Distribution of prostate specific antigen (PSA) and percentage free PSA in a contemporary screening cohort with no evidence of prostate cancer

Nomogram to Predict Insignificant Prostate Cancer at Radical Prostatectomy in Korean Men: A Multi-Center Study

Oncology: Prostate/Testis/Penis/Urethra

Original Paper. Urol Int 2013;90: DOI: /

Prostate Case Scenario 1

Although current American Cancer Society guidelines

Pathological and clinical characteristics of large prostate cancers predominantly located in the transition zone

Number: Policy *Please see amendment for Pennsylvania Medicaid at the end of this CPB.

Prostate Cancer Detection and High Grade PIN

Prostate Biopsy in 2017

Outcomes Following Negative Prostate Biopsy for Patients with Persistent Disease after Radiotherapy for Prostate Cancer

Q&A. Overview. Collecting Cancer Data: Prostate. Collecting Cancer Data: Prostate 5/5/2011. NAACCR Webinar Series 1

Comparative Analysis Research of Robotic Assisted Laparoscopic Prostatectomy

Early outcomes of active surveillance for localized prostate cancer

Original Article - Urological Oncology.

Performance of transperineal template-guided mapping biopsy in detecting prostate cancer in the initial and repeat biopsy setting

Radical prostatectomy as radical cure of prostate cancer in a high risk group: A single-institution experience

Developing a new score system for patients with PSA ranging from 4 to 20 ng/ ml to improve the accuracy of PCa detection

Prostate Biopsy. Prostate Biopsy. We canʼt go backwards: Screening has helped!

J Clin Oncol 28: by American Society of Clinical Oncology INTRODUCTION

MEDICAL POLICY Genetic and Protein Biomarkers for Diagnosis and Risk Assessment of

ND AT THE TIME OF HOLMIUM LASER ENUCLEATION OF THE PROSTATE FOR BENIGN PROSTATIC HYPERPLASIA: PREDICTING ITS PRESENCE AND GRADE IN ANALYZ

Pathologic Results of Radical Prostatectomies in Patients with Simultaneous Atypical Small Acinar Proliferation and Prostate Cancer

Table 1. Descriptive characteristics, total prostate-specific antigen, and percentage of free/total prostate-specific antigen distribution Age Groups

Oncologic Outcome and Patterns of Recurrence after Salvage Radical Prostatectomy

Significance of Atypical Small Acinar Proliferation and High-Grade Prostatic Intraepithelial Neoplasia in Prostate Biopsy

PROSTATE MRI. Dr. Margaret Gallegos Radiologist Santa Fe Imaging

The Clinical Potential of Pretreatment Serum Testosterone Level to Improve the Efficiency of Prostate Cancer Screening

Transcription:

european urology 50 (2006) 498 505 available at www.sciencedirect.com journal homepage: www.europeanurology.com Prostate Cancer High Incidence of Prostate Cancer Detected by Saturation Biopsy after Previous Negative Biopsy Series Jochen Walz a, *, Markus Graefen c, Felix K.-H. Chun a, Andreas Erbersdobler b, Alexander Haese a, Thomas Steuber a, Thorsten Schlomm a, Hartwig Huland a, Pierre I. Karakiewicz d a Department of Urology, University Medical Centre Eppendorf, Hamburg, Germany b Department of Pathology, University Medical Centre Eppendorf, Hamburg, Germany c Martini-Clinic, Prostate Cancer Centre, Hamburg d Cancer Prognostics and Health Outcomes Unit, University of Montreal Health Center, Montreal, Quebec, Canada Article info Article history: Accepted March 13, 2006 Published online ahead of print on March 29, 2006 Keywords: Prostate Biopsy Cancer Saturation Diagnosis Abstract Objectives: We explored the yield of saturation biopsy and developed a nomogram predicting the probability of prostate cancer (PCa) on the basis of saturation biopsy. Materials and methods: Between 2001 and 2004, saturation biopsies (average of 24 cores) were performed in 161 men with persistently elevated prostate specific antigen (PSA) level (median, 9 ng/ml). All had at least two previously negative, eight-core biopsy sessions. PCa predictors on saturation biopsy were integrated within multivariate nomograms. Results: PCa detection was 41% (n = 66 of 161). PSA density and transition zone volume were the most significant predictors of PCa on saturation biopsy. The accuracy of the nomogram with the best performance characteristics was 72%. Conclusions: Saturation biopsy may be indicated in men with a persistent suspicion of PCa. High-risk individuals can be identified accurately with our nomogram. # 2006 European Association of Urology. Published by Elsevier B.V. All rights reserved. * Corresponding author. Department of Urology, University Medical Centre Eppendorf, Martnistr 52, 20246 Hamburg, Germany. Tel. +49 40 42803 3443; Fax: +49 40 42803 6837. E-mail address: jwalz@uke.uni-hamburg.de (J. Walz). 1. Introduction Elevated serum prostate specific antigen (PSA) represents an indication for prostate biopsy. However, there is controversy regarding the nature and the extent of diagnostic workup in men with a persistently elevated serum PSA after several negative prostate biopsies. The yield of serial prostate biopsies decreases substantially after the first two sessions. For example, Keetch [1] showed that detection rates drop from 34% to 19%, to 8% and finally to 7%, when up to four consecutive biopsies sessions are performed. Djavan [2] showed detection rates that were 22%, 10%, 5% and 4%, respectively 0302-2838/$ see back matter # 2006 European Association of Urology. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.eururo.2006.03.026

european urology 50 (2006) 498 505 499 from biopsy session one to four. However, these series relied on six- or eight-core biopsy schemes and may not reflect detection rates associated with more detailed gland sampling, such as multicore or saturation biopsy. The saturation biopsy scheme was introduced initially by Borboroglu et al. [3] and consists of at least 20 biopsy cores. In most saturation biopsy series, between 20% and 34% of men have prostate cancer (PCa) despite two previously negative biopsies [3 8]. However, a substantially lower detection rate also has been reported, and overdetection of clinically insignificant PCa is a concern [8,9]. We report our experience with saturation biopsy in 161 men with at least two previously negative biopsies and persistently elevated serum PSA. We describe detection rates of PCa, as well as the pathologic characteristics of the tumours that were treated with radical prostatectomy (RP). Finally, we assess predictors of PCa on saturation biopsy and provide a nomogram to assist with selection of patients for saturation biopsy. 2. Material and methods From November 2001 to September 2004, 161 consecutive men (mean age, 63.7 years; range: 43 79 years) underwent saturation biopsy at our institution. Saturation biopsy was performed on an outpatient basis with antibiotic prophylaxis and under intravenous sedation. Our indications for saturation biopsy are given in Table 1; biopsies were performed when the main criterion and one of the secondary criteria were fulfilled. Previous biopsies were performed predominantly at outside institutions and consisted of at least eight cores. The majority of included patients represented referrals. Except for total gland and transition zone (TZ) volumes, as well as the number of previously negative biopsies, we did not have information pertaining to detailed PSA history or previous biopsy schemes. Consequently, this information could not be considered in risk factor analyses. At initial or repeat biopsy, total gland and TZ volumes were measured routinely and were provided on the referral sheet. Saturation biopsies were performed by using a modified Stewart [4] scheme and consisted of at least 18 cores. The number of cores taken at saturation biopsy was adjusted according to the preference of the attending urologist and/or according to prostate size. Biopsy specimens were stored individually in 5% formalin solution. Patients were discharged the same day. A subset of patients (n = 32 of 66) with PCa underwent RP at our institution. Histologic analysis of the RP specimens was performed by using the 3-mm Stanford step-section technique [10]. The 2002 TNM classification and the Gleason system were used [11]. Pathologic tumour volume was calculated as described previously [19]. The location of the tumour was determined according to its predominant location, within either the peripheral zone (PZ) or the TZ. If >70% of the tumour was found within the peripheral zone, the tumour was considered a PZ cancer. Of 161 patients, 115 (71.4%) records were fully evaluable and were included in univariate and multivariate logistic regression models predicting the presence of PCa on saturation biopsy. Of 46 exclusions, three had high-grade prostatic intraepithelial neoplasia (HGPIN) and 10 had atypical small acinar proliferation (ASAP) on saturation biopsy; neither condition represented frank PCa or benign pathology. The remaining 33 exclusions were due to missing variables: percent free PSA (%fpsa, n = 27), PSA density (n = 4), TZ PSA density (n = 6), prostate volume (n = 4) and TZ volume (n = 6). Statistics were performed with S-Plus Professional, version 1 (MathSoft Inc, Seattle, WA, USA). Two-sided tests and a significance level of 0.05 were used in all statistics. Univariate and multivariate logistic regression models were built to predict the probability of cancer on saturation biopsy. Reduced model selection was performed by using a backward step-down selection process, which used the Akaike s information criterion (AIC) as the stopping rule [13,14]. Regression coefficients were used to generate nomograms. The predictive accuracy of each nomogram was quantified with the area under the receiver operating curve (AUC). Two hundred bootstrap resamples were used to reduce over fit bias of the AUC estimates. Performance of individual and combined predictors was explored graphically with calibration plots. 3. Results Saturation biopsy detected 66 (41%) new cancers. Ten (6.2%) patients harbored ASAP, and 3 (1.8%) had HGPIN. In a subgroup of 13 patients with ASAP on previous biopsy, 10 had PCa on saturation biopsy, and 4 PCas were diagnosed in 12 patients with previous HGPIN. Table 2 shows the descriptive statistics for the entire cohort (n = 161), as well as for the subset of Table 1 Indication for saturation biopsy and number of patients with this indication Indication No. of patients 1. Criteria (fulfilled) At least two sets of previous negative multicore (8 cores) biopsy series n =161 2. Criteria (one fulfilled) High PSA level (10 ng/ml) n =81 Significant rising PSA level (0.75 ng/ml) n =55 ASAP in previous biopsy n =13 HGPIN in previous biopsy n =12 ASAP = atypical small acinar proliferation; HGPIN = high-grade prostatic intraepithelial neoplasia; PSA = prostate specific antigen.

500 european urology 50 (2006) 498 505 Table 2 Patient characteristics Variables Patients included in nomogram No. of patients (%) All patients No. of patients (%) Preoperative PSA (ng/ml) Mean (median) 13.2 (9.4) 13.5 (10) Range 3.3 125.7 3.3 125.7 %fpsa Mean (median) 15.8 (14.9) 15.7 (14.7) Range 3.0 44.9 3.0 44.9 PSA density (ng/ml/cc) Mean (median) 0.2 (0.2) 0.2 (0.2) Range 0.0 1.0 0 1.4 Transition zone density (ng/ml/cc) Mean (median) 0.4 (0.3) 0.4 (0.3) Range 0.1 2.0 0.1 2.8 Age (yr) Mean (median) 63.4 (64.0) 63.7 (64.0) Range 43.0 77.0 43.0 84.0 No. of previous biopsies series (%) 2 72 (62.6) 97 (60.2) 3 30 (26.1) 35 (21.7) 4 8 (7) 12 (7.5) 5 4 (3.5) 4 (2.5) 6 1 (0.6) 7 10 (6.2) Prostate volume (cc) Mean (median) 66.4 (58.0) 66.6 (56) Range 25.0 218.0 15.0 218.0 Transition zone volume (cc) Mean (median) 39.3 (33.0) 40.2 (32) Range 9.0 125.0 9.0 140.0 No. of cores Mean (median) 24.5 (24.0) 24.2 (24) Range 20.0 32.0 18.0 32.0 Cancer (%) Absence of cancer 64 (55.7) 82 (50.9) Presence of cancer 51 (44.3) 66 (41.0) 2 (HGPIN) (%) 3 (1.9) 2 (ASAP) (%) 10 (6.2) Total 115 161 ASAP = atypical small acinar proliferation; fpsa = free prostate specific antigen; HGPIN = high-grade prostatic intraepithelial neoplasia; PSA = prostate specific antigen. evaluable patients with complete records, who were included in regression models (n = 115). Positive cores were mainly in the far lateral zone (79%), the medio-lateral zone (36%) and the TZ (18%). Of all, 37% of patients PCa were found within a single biopsy sector. In 21%, two sectors harbored PCa. In 17%, three sectors were involved, and, finally, PCa was present in four or more sectors in 25%. Three patients (4.5%) had exclusive TZ pathology, without coexistent PZ pathology. Thus it can be Table 3 Characteristics of patients treated with radical prostatectomy or radiation therapy Variables Radical prostatectomy Radiation therapy n = 33 15 PSA (ng/ml) (median) 9.1 10.2 %fpsa (median) 9.9 13.2 PSA density (ng/ml/cc) 0.21 0.27 Biopsy-Gleason 6 (n [%]) 25 (75%) 11 (73%) Biopsy-Gleason 7 (n [%]) 11 (25%) 4 (27%) No. of previous 2.4 2.4 biopsies (average) No. of positive 2.4 3 cores (average) %PCa per core (average) 14 17 fpsa = free prostate specific antigen; PCa = prostate cancer; PSA = prostate specific antigen. postulated that the omission of TZ biopsies would have resulted in the lack of detection of three of 66 (4.5%) PCas and would have decreased the overall detection rate from 41 to 39%. Thus, TZ cores may not always be necessary. Of 66 patients with PCa, 32 (48%) underwent RP at our institution. Nine patients underwent high-dose brachytherapy, six had external beam radiotherapy and six were treated with hormonal therapy. The remaining 13 patients were undecided or were lost to follow-up. The characteristics of the patients treated with RP and radiotherapy are shown in Table 3. There were no significant clinical differences between these two groups. RP pathologic stages were as follows: pt2a in 16% (n = 5), pt2b in 3% (n =1),pT2c in 56% (n = 18), pt3a in 12.5% (n = 4), pt3b in 9% (n =3)andpT4in3%(n = 1). Twenty-four patients (75%) had organ-confined PCa. Gleason pattern 4 or 5 was identified in 14 (43.7%). Mean tumour volume was 3.0 cc (range, 0.002 14.3). Clinically insignificant PCa 13 was defined as absence of high-grade components, tumour volume less than 0.5 cc and pathologic organ confinement. It was identified in five of 32 (15.6%) men. The predominant tumour location was within the PZ in 43.3%, within the TZ in 36.6% and in both locations in 20%. The overall saturation biopsy complication rate was 2.5% and consisted of two acute urinary retentions, one acute prostatitis and one reactive syncope. Two hospital admissions were necessary for intravenous antibiotics. Table 4 shows univariate and multivariate logistic regression models predicting PCa on saturation biopsy. In univariate analyses, %fpsa, TZ density and TZ volume emerged as the most significant variables, followed by PSA density and prostate volume. %fpsa, PSA density and TZ volume were

european urology 50 (2006) 498 505 501 Table 4 Univariate and multivariate logistic regression model for prediction of prostate cancer on saturation biopsy Variables Univariate model Multivariate OR; p value Predictive accuracy (%) Full model OR; p value Reduced model OR; p value Age 1.030; 0.362 52.5 1.054; 0.174 1.062; 0.100 Preoperative PSA 1.009; 0.514 52.8 1.025; 0.555 %fpsa 0.909; 0.001 74.3 0.938; 0.053 0.924; 0.006 PSA density 20.155; 0.025 69.2 0.001; 0.215 Transition zone density 9.702; 0.001 73.1 36.775; 0.182 Prostate volume 0.977; 0.003 72.3 1.013; 0.694 1.042; 0.063 Transition zone volume 0.965; 0.001 72.8 0.961; 0.322 0.924; 0.010 No. of previous biopsies 0.876; 0.586 51.4 0.835; 0.535 No. of cores 0.936; 0.318 73.1 0.957; 0.575 Predictive accuracy (%) 72.2 75.2 fpsa = free prostate specific antigen; OR = odds ratio; PSA = prostate specific antigen. Fig. 1 Model calibration plots. (A) Calibration plot for prediction of cancer on saturation biopsy according to percent of free PSA level, where x-axis represents predicted probability of prostate cancer and y-axis represents observed fraction with evidence of prostate cancer on saturation biopsy according to percent of free PSA. Calibration plot for prediction of prostate cancer on saturation biopsy using the full (B) and reduced (C) prediction model, where x-axis represents the full (B) or reduced (C) model, respectively, and predicted probability of prostate cancer on saturation biopsy; y-axis represents observed fraction with evidence of prostate cancer on biopsy. Perfect prediction would correspond to the 45-degree line. Points estimated below the 45-degree line correspond to nomogram overprediction, whereas points situated above the 45-degree line correspond to nomogram underprediction.

502 european urology 50 (2006) 498 505 correlated negatively with the risk of PCa. Conversely, TZ density and prostate volume were correlated positively. Examination of nomogram axes reveals that the risk of PCa increases in men with high TZ densities and large total gland volume. This implies that high-risk individuals may have large PZ volumes. Of all risk factors, %fpsa demonstrated the highest predictive accuracy (74.3%), followed by TZ density (73.1%) and TZ volume (72.8%). All variables subsequently were included in a multivariate model, which was 72.2% accurate in predicting PCa on saturation biopsy. This full multivariate model was between 0.6% and 2.1% less accurate than one of the three most informative univariate predictors. Assessment of the performance characteristics of univariate predictors revealed important departures from ideal predictions (Fig. 1A). For example the most informative univariate predictor, %fpsa, substantially overestimated the probability of cancer for predicted values between 20% and 40%. Similarly important departures from ideal predictions were noted for all other univariate predictors (data not shown). Assessment of the full-model calibration plot demonstrated minimal departures from ideal predictions (Fig. 1B). In an attempt to reconcile predictive accuracy and model performance, we applied fast backwards variable elimination to the full model, with the intent of identifying the most parsimonious and the most accurate model [13,14]. Removal of PSA, PSA density, TZ density, number of previous biopsy session and number of cores resulted in the most accurate model. This reduced model included patient age, %fpsa, prostate volume and TZ volume. Its 200 bootstrap-corrected predictive accuracy (75.2%) exceeded the 200 bootstrap-corrected accuracy of the full model (72.2%). Moreover, its accuracy exceeded the 200 bootstrap-corrected accuracy of each individual predictor (range, 51.4 74.3%). The calibration plot of this final model (Fig. 1C) revealed departures from ideal predictions, which exceeded those noted for the full model. For example, for predicted values between 0% and 30%, the model underestimated the observed values. Conversely, predicted values between 30% and 60% overestimated the observed values. The departures from ideal predictions were less pronounced than those noted for the best individual predictor (i.e., %fpsa). 4. Discussion At saturation biopsy (average, 24 cores), 41% of the men were diagnosed with PCa. In the most previous saturation biopsy series (Table 5), the yield of saturation biopsy ranged from 20% to 34% and suggests that saturation biopsy represents an effective method for detecting previously missed PCas [3 8]. Our cancer detection rate exceeds those reported in all previous series, in which the highest (34%) rate was reported by Stewart et al. [4] with 23 biopsy cores. Except for one series [9], previous series reported median PSA values from 8.6 to 9.4 ng/ml, which are comparable to the 9.4 ng/ml in the current series (Table 4). Unfortunately, the referral pattern of our cohort makes detailed assessment of other patient characteristics impossible. Detection of insignificant PCa is a risk of any initial or repeat biopsy. It may be as high as 33.5% with 12 cores versus 22.7% with sextant biopsy [20]. We report the presence of insignificant PCa, defined as absence of high-grade components, organ confinement and tumour volume less than 0.5 cc [12] in only 5 (15.6%) of 32 men treated with RP. Unfortunately, the assessment of clinical significance may be performed only for those treated with RP and cannot be assessed for the remaining 34 patients. However, the clinical parameters did not differ significantly between those treated with RP and those treated with radiation (Table 3). In addition to reporting the yield of saturation biopsy, our goal was to develop a model capable of accurately predicting the probability of PCa on saturation biopsy. Our model is 72% accurate and relies on variables that can be readily obtained prior to saturation biopsy. Table 5 Comparison of saturation biopsy studies Author Year Sample size PCa detection rate (%) No. of cores No. of previous biopsies Average PSA level (ng/ml) Borboroglu et al. 2000 57 30 22.5 2.1 8.6 (mean) Stewart et al. 2001 224 34 23 1.8 8.7 (median) Fleshner et al. 2002 37 13.5 32 38 4.2 22.4 (median) Rabets et al. 2004 116 29 22.8 1.7 9.2 (mean) Patel et al. 2004 100 25 20 24 1.7 9.4 (mean) Current series 2004 161 41 24.2 2.5 9.4 (median) PCa = prostate cancer; PSA = protein specific antigen.

european urology 50 (2006) 498 505 503 Several interesting observations may be derived from close examination of the axes defining the nomogram risk variables. PSA and %fpsa represent the most commonly used predictors of biopsy outcome. PSA is of only limited value in our cohort. Most commonly encountered PSA values are located on the low side of the risk point scale and contribute to few risk points. Thus, in most patients, PSA exerted a minimal effect on the probability of a positive saturation biopsy. %fpsa contributes more importantly to risk points than PSA. Assessment of other nomogram axes shows that other variables, Fig. 2 Nomograms. To obtain nomogram predicted probability of saturation biopsy (for full and reduced model), locate the patient s variable values at each axis. Draw a vertical line to the Point axis to determine how many points are attributed for each variable value. Sum the points for all variables. Locate the sum on the Total Points line. Draw a vertical line towards the Prob. of CA axis to determine the patient cancer probability. PSA = prostate specific antigen; %fpsa = percent free PSA; PSAD = PSA density; TZ-D = transition zone density; Tot.Vol = total prostatic volume; TZVol = transition zone volume; Prev. bx = number of previous biopsies; # cores = planned number of cores; Prob. of PCa = probability of prostate cancer.

504 european urology 50 (2006) 498 505 such as total gland and TZ volume as well as their PSA-to-volume ratios, represent highly informative predictors of saturation biopsy outcome. PSA density and TZ volume represent the most important and informative predictors in this population. The capacity of these variables to predict the risk may be explained by the relative inability of previous biopsy schemes to detect cancer in large glands. Our results indicate that, when saturation biopsy is contemplated after one or several negative biopsies, the risk of finding cancer is highest in men with large prostates and small TZ volumes. These men have large PZ volumes. Besides important clinical observations, our regression models provided important insight about the combined effect of saturation biopsy risk factors. We have shown that, when variables are used alone, they might be associated with high predictive accuracy. For example, the accuracy rates of %fpsa, TZ PSA density and TZ volume in predicting cancer on saturation biopsy were74.3%, 73.1% and 72.8%, respectively. However, their predictive accuracy was associated with suboptimal performance characteristics, manifested by important differences between observed and predicted probabilities of PCa at saturation biopsy. These differences rendered those seemingly accurate predictors undesirable, when risk variables were used in isolation. Our data demonstrate that the combined contribution of all predictors within the full model (Fig. 2A), substantially improved the performance characteristics, as evidenced by calibration plots (Fig. 1A C). Despite its better ability to predict the observed rate of PCa at saturation biopsy, the full model was less accurate (72.2%) than selected highly significant and informative factors, such as %fpsa (74.3%). Assessment of the calibration plot in Fig. 1 demonstrates that the curve depicting %fpsa is substantially farther away from the ideal 45-degree prediction line than the performance curve of the full model. Better predictive accuracy of %fpsa relative to the whole model can be explained by areas of underprediction (curve above ideal 45-degree prediction line) and areas of overprediction (curve below ideal 45-degree prediction line), which cancel out each other in a more symmetric fashion than seen for the whole model. Therefore, it appears that, in the case of saturation biopsy, a more complete model (Fig. 2A) is superior to a model with fewer variables (Fig. 2B), in which higher predictive accuracy is traded for worse performance characteristics. In consequence, we recommend the use of the full model. The compromise between discriminant and calibration properties of the full model still yielded highly accurate predictions, relative to accuracy reported for two previously validated repeat biopsy nomograms [15,16]. The first revealed predictive accuracy of 70%, and the second was 71% accurate. Therefore, 72.2% accuracy of the current full model is superior to those two previous studies and confirms that the outcome of repeat biopsy, either saturation or regular, may be predicted in a highly accurate fashion. The predictive accuracy of repeat biopsy models (range, 70 72%) is on average inferior to models predicting the outcome of initial biopsy, which range from 69% to 0.91% [15 17]. It appears therefore, that a repeat biopsy outcome is more difficult to predict than the outcome of an initial biopsy. By extrapolation, it might be postulated that prediction of saturation biopsy, which represents the third or fourth biopsy, is even more difficult. In this light, predictive accuracy of 72.2% appears acceptable. Moreover, recent reports suggest that nomograms are more accurate than clinical estimates [18]. Finally, a model with nonperfect accuracy may be better than no model at all. There are a number of limitations in our study. For example, we were unable to assess whether presence of HGPIN and/or ASAP represents a risk factor for PCa on saturation biopsy. Our sample was limited by a small number of patients with information on these two conditions prior to saturation biopsy because of the referral pattern of the population. Other nomogram limitations include its suboptimal accuracy, a flaw shared with virtually all prognostic models [13,14]. 5. Conclusion Saturation biopsy represents a necessary and a safe diagnostic procedure for men with persistent suspicion of PCa after several negative biopsies. The detection rate of PCa is high (41%) and the likelihood of clinically insignificant cancers is not increased (15.6%) relative to initial or repeat biopsy. Finally, our nomogram can accurately (72%) identify those at risk of having PCa at saturation biopsy. Conflicts of interest Pierre I. Karakiewicz is partially supported by the Fonds de la Recherche en Santé du Québec, the CHUM Foundation, the Department of Surgery and Les Urologues Associés du CHUM.

european urology 50 (2006) 498 505 505 References [1] Keetch DW, Catalona WJ, Smith DS. Serial prostatic biopsies in men with persistently elevated serum prostate specific antigen values. J Urol 1994;151:1571 4. [2] Djavan B, Ravery V, Zlotta A, et al. Prospective evaluation of prostate cancer detected on biopsies 1, 2, 3 and 4: when should we stop? J Urol 2001;166:1679 83. [3] Borboroglu PG, Comer SW, Riffenburgh RH, Amling CL. Extensive repeat transrectal ultrasound guided prostate biopsy in patients with previous benign sextant biopsies. J Urol 2000;163:158 62. [4] Stewart CS, Leibovich BC, Weaver AL, Lieber MM. Prostate cancer diagnosis using a saturation needle biopsy technique after previous negative sextant biopsies. J Urol 2001;166:86 91, discussion 91 2. [5] Chan TY, Chan DY, Stutzman KL, Epstein JI. Does increased needle biopsy sampling of the prostate detect a higher number of potentially insignificant tumors? J Urol 2001;166:2181 4. [6] Taylor JA, Gancarczyk KJ, Fant GV, McLeod DG. Increasing the number of core samples taken at prostate needle biopsy enhances the detection of clinically significant prostate cancer. Urology 2002;60:841 5. [7] Rabets JC, Jones JS, Patel A, Zippe CD. Prostate cancer detection with office based saturation biopsy in a repeat biopsy population. J Urol 2004;172:94 7. [8] Patel AR, Jones JS, Rabets J, DeOreo G, Zippe CD. Parasagittal biopsies add minimal information in repeat saturation prostate biopsy. Urology 2004;63:87 9. [9] Fleshner N, Klotz L. Role of saturation biopsy in the detection of prostate cancer among difficult diagnostic cases. Urology 2002;60:93 7. [10] McNeal JE, Villers AA, Redwine EA, Freiha FS, Stamey TA. Histologic differentiation, cancer volume, and pelvic lymph node metastasis in adenocarcinoma of the prostate. Cancer 1990;66:1225 33. [11] Gleason DF. Histologic grading and clinical staging of prostatic carcinoma. In: Tannenbaum M, editor. Urologic pathology: the prostate. Philadelphia: Lea & Ferbiger; 1977. p. 171 97, chp 9. [12] Dugan JA, Bostwick DG, Myers RP, et al. The definition and preoperative prediction of clinically insignificant prostate cancer. JAMA 1996;275:288. [13] Van Houwelingen JC, Le Cessie S. Predictive value of statistical models. Stat Med 1990;9:1303. [14] Atkinson AC. A note on the generalized information criterion for choice of a model. Biometrika 1980;67:413. [15] Lopez-Corona E, Ohori M, Scardino PT, Reuter VE, Gonen M, Kattan MW. A nomogram for predicting a positive repeat prostate biopsy in patients with a previous negative biopsy session. J Urol 2003;170:1184 8. [16] Yanke BV, Gonen M, Scardino PT, Kattan MW. Validation of a nomogram for predicting positive repeat biopsy for prostate cancer. J Urol 2005;173:421 4. [17] Karakiewicz PI, Benayoun S, Kattan MW, et al. Development and validation of a nomogram predicting the outcome of prostate biopsy based on patient age, digital rectal examination and serum prostate specific antigen. J Urol 2005;173:1930 4. [18] Specht MC, Kattan MW, Gonen M, Fey J, Van Zee KJ. Predicting nonsentinel node status after positive sentinel lymph biopsy for breast cancer: clinicians versus nomogram. Ann Surg Oncol 2005;12:654 9. [19] Henke RP, Kruger E, Ayhan N, Hubner D, Hammerer P, Huland H. Immunohistochemical detection of p53 protein in human prostatic cancer. J Urol 1994;152:1297 301. [20] Singh H, Canto EI, Shariat SF, et al. Improved detection of clinically significant, curable prostate cancer with systematic 12-core biopsy. J Urol 2004;171:1089 92.