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

european urology 51 (2007) 949 955 available at www.sciencedirect.com journal homepage: www.europeanurology.com Prostate Cancer Zonal Origin of Localized Prostate Cancer Does not Affect the Rate of Biochemical Recurrence after Radical Prostatectomy Felix K.-H. Chun a,b,1, Alberto Briganti b,1, Claudio Jeldres b, Andreas Erbersdobler c, Thorsten Schlomm a, Thomas Steuber a, Andrea Gallina b, Jochen Walz a, Paul Perrotte b, Hartwig Huland a,d, Markus Graefen d, Pierre I. Karakiewicz b, * a Department of Urology, University of Hamburg, Hamburg, Germany b Cancer Prognostics and Health Outcomes Unit, University of Montreal, Montreal, Quebec, Canada c Department of Pathology, University of Hamburg, Hamburg, Germany d Martini Clinic, Prostate Cancer Center, University of Hamburg, Hamburg, Germany Article info Article history: Accepted July 11, 2006 Published online ahead of print on July 28, 2006 Keywords: Transition zone prostate cancer Radical prostatectomy Biochemical recurrence Abstract Objective: To investigate whether transition zone (TZ) prostate cancers demonstrate different rates of biochemical recurrence after radical prostatectomy compared to peripheral zone (PZ) cancers. Methods: In 1262 consecutive patients treated with radical prostatectomy, computerized planimetry defined tumour origin as either TZ tumours (>70% TZ location) or PZ. Kaplan-Meier and multivariate Cox regression models tested the association between zonal origin and the rate of biochemical recurrence (prostate-specific antigen >0.1 ng/ml and rising). We used the Harrell s concordance index to quantify the accuracy of various Cox regression models. Results: TZ prostate cancers were diagnosed in 115 patients (9.1%). Biochemical recurrence was recorded in 16 TZ and in 201 PZ prostate cancers patients. In multivariate Cox models, the rate of biochemical recurrence was not significantly different between TZ and PZ prostate cancers ( p = 0.4). Combined multivariate predictive accuracy of biochemical recurrence predictions was 81.2% accurate when zonal origin was included versus 81.0% when zonal origin was omitted. Conclusions: The zonal origin of prostate cancers does not affect the rate of biochemical recurrence after radical prostatectomy. # 2006 European Association of Urology. Published by Elsevier B.V. All rights reserved. 1 These authors contributed equally to the manuscript. * Corresponding author. Cancer Prognostics and Health Outcomes Unit, University of Montreal Health Center, 1058, Rue St-Denis, Montréal, Québec H2X 3J4, Canada. Tel. +1 514 890 8000 35336; Fax: +1 514 412 7363. E-mail address: pierre.karakiewicz@umontreal.ca (P.I. Karakiewicz). 0302-2838/$ see back matter # 2006 European Association of Urology. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.eururo.2006.07.008

950 european urology 51 (2007) 949 955 1. Introduction Currently, the Kattan postoperative nomogram represents the most widely used tool to predict the rate of biochemical recurrence (BCR) after radical prostatectomy [1]. It relies on the combination of preoperative serum prostate-specific antigen (PSA) and on pathologic stage variables. However, the predictive accuracy of this nomogram, as well as of virtually all other predictive tools, is not perfect, and efforts continue to improve the accuracy of these tools. Some of these efforts have met with limited success, whereas others that rely on novel markers, such as TGF-b1 and IL-6, have resulted in improved predictive accuracy [2]. Here, we examine the effect of zonal tumour origin, which distinguishes peripheral zone (PZ) cancers from their transition zone (TZ) counterparts, as several authors suggested that origin may represent an important variable [3 5]. Approximately 20% of prostate cancers (PCa) originate from the TZ [3] and were suggested to have distinct biological characteristics relative to PZ cancers [4 8]. Specifically, TZ tumours tend to display higher PSA and higher tumour volumes. Conversely, they have lower biopsy Gleason sums. Moreover, they are frequently associated with more favourable pathologic disease characteristics and with better prognosis [9 11]. For example, Augustin et al. found a 70% 5-yr BCR-free rate for PZ PCa versus 80% for TZ PCa [12]. Stamey et al. reported an even more striking difference, where TZ PCa showed an 81% BCR-free rate versus 53% for PZ PCa [13]. Noguchi et al. confirmed these data and reported 72% BCR-free rate in TZ PCa versus 49% in TZ PCa [14]. Taken together, these findings suggest that PCa zonal origin represents an important variable in prediction of BCR. Based on these reports, we hypothesized that zonal origin may successfully discriminate between PZ and TZ cancers, with the latter having a lower rate of BCR and a more favourable prognosis. We postulated that zonal tumour origin represents an independent multivariate predictor of BCR. Moreover, we hypothesized that zonal tumour origin may increase the combined predictive accuracy of established predictors of BCR. 2. Methods 2.1. Patient population Between January 1996 and January 2004, 1262 consecutive patients treated with radical prostatectomy were subjected to meticulous pathologic assessment, which focused on detailed planimetric, computer-assisted assessment of zonal tumour origin. The goal of this modified, computer-assisted planimetric pathology protocol was to study the potential gains in predictive accuracy associated with routine quantification of the variable defining zonal tumour origin. Pretreatment PSA, presence of extracapsular extension (ECE), seminal vesicle invasion (SVI), lymph node invasion (LNI), the pathologic Gleason sum, and the rate of PSA recurrence after radical prostatectomy were also available for all patients. 2.2. Clinical and pathologic evaluation Pretreatment PSA was measured before digital rectal examination. All prostatectomy specimens were processed according to the Stanford protocol [3] and were graded according to the Gleason system [15]. Tumour areas were marked on whole-mount slides with a water-resistant pen. By identifying the TZ boundary, the location of the largest tumour area was determined. Zonal tumour origin measurements were performed using a computerized planimetric method [16]. Tumours were classified as TZ cancers when more than 70% of tumour volume was located within the TZ. In cases of extensive tumour multifocality, TZ cancer was assessed when the largest tumour focus (i.e., index tumour) was confined to the TZ or when the cancer volume of multiple tumour foci located in the TZ boundary was greater than 50% of total PCa volume. In all patients PSA values were measured quarterly in the first year, followed by biannual measurements in the second and annual measurements in the third year after radical prostatectomy. BCR was defined as a postoperative PSA of 0.1 ng/ml and rising after an initial undetectable PSA. The first PSA value above or equal to 0.1 ng/ml was used to define the time to BCR. Patients without evidence of BCR were censored at the last follow-up. 2.3. Statistical analyses BCR-free probability after radical prostatectomy was calculated using the Kaplan-Meier method. Cox regression models using PSA and radical prostatectomy pathology variables addressed BCR rates after prostatectomy. Pretreatment PSA, ECE, SVI, and LNI were used as base predictor variables. Zonal tumour origin complemented the multivariate model, and its independent multivariate predictor status was tested. Cox regression coefficients were then used to calculate the accuracy of BCR predictions, with and without zonal tumour origin. This method was used to quantify the increment in predictive accuracy associated with the addition of the zonal tumour origin to all base predictor variables. Two hundred bootstrap resamples were used for internal validation of all accuracy estimates and to reduce overfit bias. All statistical tests were performed using S-PLUS Professional, version 1 (MathSoft Inc., Seattle, Washington, USA). All tests were twosided with a significance level set at p < 0.05. 3. Results Descriptive characteristics of the 1262 patients are shown in Table 1. TZ PCa was present in 9.1% (115)

european urology 51 (2007) 949 955 951 Table 1 Overall patient characteristics and comparison of characteristics according to zonal origin of cancer Variables Number of patients TZ PZ p PSA (ng/ml) Mean (median) 11.2 (8) 17.7 (13.2) 10.5 (7.7) 0.001 Range 0.5 125.0 1.1 82.6 0.5 125.0 RP Gleason sum 0.001 2 5 146 (11.5%) 56 (48.7%) 90 (7.8%) 6 366 (29.0%) 10 (8.7%) 356 (31.0%) 7 10 750 (59.5%) 49 (42.6%) 701 (61.1%) Presence of ECE 502 (39.8%) 37 (32.2%) 465 (40.5%) 0.08 Presence of SVI 202 (16.0%) 11 (9.6%) 191 (16.7%) 0.05 Presence of LNI 74 (5.9%) 4 (3.5%) 70 (6.1%) 0.3 Zonal origin TZ 115 (9.1%) PZ 1147 (90.9%) TZ: transition zone cancer; PZ: peripheral zone cancer; PSA: preoperative serum prostate-specific antigen; ECE: extracapsular extension; SVI: seminal vesicle invasion; LNI: lymph node invasion; RP: radical prostatectomy. versus 90.9% (1147) with PZ PCa. The follow-up ranged from 0 to 96 months (mean: 45.1; median: 45.6). Pretreatment mean and median PSA values were 17.7 and 13.2 ng/ml for TZ tumours versus 10.5 and 7.7 ng/ml for PZ tumours ( p < 0.001). BCR occurred in 17.2% (217) of all patients. BCR occurred in 13.9% (16) of the TZ PCa patients and in 17.5% (201) of the PZ PCa patients ( p = 0.3). The rate of ECE, SVI, and LNI were consistently lower in men with TZ PCa. Gleason sums were more favourable in patients with TZ PCa than their PZ counterparts ( p < 0.001). Fig. 1 shows Kaplan-Meier plots of BCR-free survival. Fig. 1A illustrates the overall BCR rate; Fig. 1B displays BCR rates according to zonal tumour origin. The log rank test comparing BCR-free survival according to zonal tumour origin was not statistically significant ( p = 0.4). Table 2 shows univariate and multivariate Cox regression models. In univariate Cox analyses, all base predictor variables (i.e., pretreatment PSA, radical prostatectomy Gleason sum, ECE, SVI, and LNI) were highly statistically significant ( p < 0.001), except for zonal origin ( p = 0.4). The most informative univariate predictors of BCR consisted of ECE (72.6%), pretreatment PSA (70.6%), and Gleason sum (68.7%). Zonal tumour origin demonstrated 50.8% accuracy, where 50% is equivalent to a flip of a coin and 100% equals perfect prediction. In multivariate Cox analyses, we fitted two models. Both controlled for PSA and all radical prostatectomy pathologic variables. One multivariate model (i.e., the base model) did not include the zonal tumour origin variable, and the other model contained all base model variables plus the zonal origin. In the second multivariate model, zonal tumour origin did not reach independent multivariate predictor status ( p = 0.4). Moreover, the multivariate predictive accuracy of the model that included the zonal tumour origin was 81.2% versus 81.0% for the base model. 4. Discussion Approximately 20% of localized PCa patients who undergo a radical prostatectomy fail biochemically [3]. Pretreatment PSA and histopathologic radical prostatectomy features, such as Gleason pattern, ECE, SVI, and LNI, represent established and informative predictors of disease progression after radical prostatectomy [17 19]. Previous studies suggested that the zonal tumour origin represents one of the most important discriminants of BCR. In Stamey et al. s series of 791 radical prostatectomies, 25% of all T1c PCas were in the transition zone [13]. Moreover, TZ tumours demonstrated an 81% 5-yr cure rate, relative to 53% seen in PZ tumours. Noguchi et al. examined 148 TZ PCa and 79 matched PZ tumours [14]. In this series, TZ tumours demonstrated a BCR-free rate of 71.5% versus 49.2% for PZ PCa. These striking observations suggest that the zonal origin of PCa warrants close examination. However, existing models of BCR after radical prostatectomy do not include this key variable. Based on the potential value of zonal tumour origin, we hypothesized that the distinction between TZ and PZ PCa may add to our ability to accurately predict BCR rates after radical prostatectomy. To test our hypothesis, we used most stringent methodological analyses, as suggested by Kattan [20]; in addition to demonstrating its multivariate independent predictor status, a novel marker should enhance the overall model predictive accuracy. We added this methodology to the

952 european urology 51 (2007) 949 955 Fig. 1 Rate of biochemical recurrence after radical prostatectomy of the entire cohort and according to zonal tumor origin. (A) Kaplan-Meier curve of overall rate of biochemical recurrence-free survival (mean recurrencefree survival: 6.8 yr; median recurrence-free survival: not reached; 1-yr recurrence-free probability: 93.5% [95% CI, 92.0 94.9]; 5-yr recurrence-free probability: 80.6% [95% CI, 78.1 82.9]). (B) Kaplan-Meier curve of rate of biochemical recurrence-free survival according to zonal tumor origin (log rank test: transition zone (TZ) vs. peripheral zone (PZ) prostate cancer: p = 0.4; TZ: mean recurrence-free survival: 6.7 yr, median recurrence-free survival: not reached, 1-yr recurrence-free probability: 96.4% [95% CI, 92.9 99.9], 5-yr recurrence-free probability: 81.5% [95% CI, 73.5 90.5]; PZ: mean recurrence-free survival: 7.0 yr, median recurrencefree survival: not reached, 1-yr recurrence-free probability: 93.3% [95% CI, 91.8 94.7]; 5-yr recurrence-free probability: 80.5% [95% CI, 78.0 83.1]). standard univariate and multivariate tests of the variable defining the zonal origin of PCa. Our univariate analyses (Kaplan-Meier and Cox regression) demonstrated that the rate of BCR is not significantly different between TZ and PZ PCas ( p = 0.4). Testing the predictive accuracy of the variable coding the zonal origin demonstrated that TZ versus PZ origin is only 50.8% accurate in predicting the rate of BCR, which is about equal to a toss of a coin. Conversely, established predictors of BCR, such as ECE (73.3%), PSA (70.6%), and radical prostatectomy Gleason sum (68.7%), demonstrated substantially higher accuracy. These findings indicate that the zonal origin alone cannot accurately identify those who are destined to fail biochemically. Our multivariate analyses confirmed our univariate findings. Zonal tumour origin failed to reach independent multivariate predictor status ( p = 0.4), after controlling for the effect of PSA, ECE, SVI, LNI, and radical prostatectomy Gleason sum. Finally, its addition to established multivariate predictors failed to improve predictive accuracy (81.2% with zonal origin vs. 81.0% without zonal origin). Taken together, our findings suggest that zonal tumour origin does not fulfil the characteristics of a novel and informative marker of BCR. Based on these findings, it does not appear warranted to determine the zonal origin in radical prostatectomy specimens on a routine basis, as this distinction does not improve our ability to predict PCa prognosis. Our findings refute Stamey et al. s [13] and Noguchi et al. s [14] contention that BCR rates differ according to PCa zonal origin. Moreover, our data refute Augustin et al. s univariate analyses of BCR rate according to zonal origin of the tumour, where TZ tumours (n = 63) exhibited 80% BCR-free rate versus 70% for PZ tumours (n = 244) [12]. Unfortunately, Augustin et al. failed to include all TZ and PZ tumours in their multivariate analysis. Instead, they opted for a matched analysis, where each TZ tumours assigned to a PZ counterpart on a 1:1 basis and similar clinical and pathologic characteristics were used for matching. Such matching of retrospective data is fraught with difficulties, as many unknown characteristics cannot be accounted for and numerous biases may be introduced. In the face of this potential methodological limitation, the actuarial comparison of BCR rates according to matched TZ and PZ zonal origin should be interpreted with caution. Hence, we did not qualify Augustin s findings as conclusive, especially in the light of Stamey et al. s and Noguchi et al. s reports [13,14], where very strong differences in BCR rates in favour of TZ cancers were shown. However, our methodologically detailed analysis, which included univariate, multivariate, and predictive accuracy tests, failed to confirm Stamey et al. s [13] hypothesis and corroborated the matched analysis findings of Augustin et al. [21]. After using the most stringent methodology, as suggested by Kattan [20], our findings demonstrate that zonal tumour origin did not represent an

european urology 51 (2007) 949 955 953 Table 2 Univariate and multivariate Cox regression models predicting biochemical recurrence after radical prostatectomy Type of analyses Cox regression models predicting time to PSA-recurrence Univariate analyses Multivariate analyses Variables Relative risk; p-value Predictive accuracy (%) Base model relative risk; p-value Base model + zonal origin relative risk; p-value PSA 1.03; 0.001 70.6 1.02; 0.001 1.02; 0.001 Presence of ECE 6.4; 0.001 73.3 2.5; 0.001 2.5; 0.001 Presence of SVI 5.2; 0.001 65.6 1.6; 0.003 1.6; 0.004 Presence of LNI 5.7; 0.001 57.7 2.1; 0.001 2.1; 0.001 RP Gleason sum ; 0.001 68.7 ; 0.001 ; 0.001 6 vs. 2 5 0.4; 0.07 0.5; 0.1 0.5; 0.08 7 10 vs. 2 5 4.9; 0.001 2.4; 0.008 2.3; 0.02 Zonal origin TZ vs. PZ 0.8; 0.4 50.8 0.8; 0.4 Predictive accuracy a 81.0 81.2 PSA: preoperative serum prostate specific antigen; ECE: extracapsular extension; SVI: seminal vesicle invasion; LNI: lymph node invasion; RP: radical prostatectomy. a Predictive accuracy denotes the ability of individual variables and multivariate models to discriminate between patients who recur from those who do not recur. The base model contains all variables except for the zonal origin of the cancer. independent or an informative predictor of BCR after radical prostatectomy (Table 3). As noted by Kattan, statistical significance is not synonymous with predictive ability. Instead, Kattan recommended that a novel marker should not only be judged according to its multivariate statistical significance, but should increase the combined predictive accuracy of base predictors, in addition to confirming the independent, multivariate predictor status of this marker. We followed this recommendation and tested the accuracy of the base multivariate model in which zonal origin was omitted, with 200 bootstraps resamples to reduce overfit bias. We found that the base model was 81.0% accurate in predicting the probability of BCR. Inclusion of zonal origin was associated with a negligible increment in predictive accuracy (+0.2%). Thus, this final analytic step also failed to demonstrate any benefit in consideration of PCa zonal origin. The low rate of BCR observed in the TZ cohort (13.9%) represents a limitation of our study [22]. Longer follow-up and associated higher rate of BCR may have translated into a more significant difference in BCR rates. Other differences between our data and those of Stamey et al. [13] and Noguchi et al. [14] may relate to pathologic characteristics of included tumours; for instance, our series may have included patients with overall lower tumour volumes. Smaller PZ tumours may obliterate the difference in BCR rates. Moreover, our pathologic analysis differed from that of the previous studies, which only considered the index tumour, defined as the dominant tumour [13,14]. Conversely, we considered all areas consistent with invasive PCa in our tumour definition. This may have diluted the effect of the distinction between TZ and PZ PCa. We believe that our method of defining PCa is superior, however, because it accounts for the multifocal Table 3 Comparison of studies investigating tumour location (transition vs. peripheral zone) as a risk factor of biochemical recurrence after radical prostatectomy Reference Number of patients Definition of BCR (ng/ml) Univariate RR; p-value Type of analyses Multivariate RR; p-value Predictive accuracy Stamey et al. 366 0.07 NA; <0.001 NA NA Stamey et al. 372 0.07 NA; <0.001 NA NA Noguchi et al. 158 0.07 NA; <0.001 NA NA Noguchi et al. 191 0.07 1.9; 0.01 1.9; 0.01 NA 52 (OC only) 12.4; 0.002 13.5; 0.002 Augustin et al. 307 0.1 0.99; 0.03 1.0; 0.6 NA Present series 1262 0.1 0.8; 0.4 0.8; 0.4 unaffected BCR: biochemical recurrence; RR: rate ratio; NA: not available; OC: organ confinement.

954 european urology 51 (2007) 949 955 nature of PCa. Multifocality represents a key feature of contemporary PCas and cannot be ignored in the pathologic evaluation of the tumour. Although, we disagree with the findings of Stamey et al. [13] and Noguchi et al. [14] regarding the differences in BCR-free rates after radical prostatectomy, our findings need to be interpreted with care. We conclude that pathologic distinction between TZ and PZ cancers does not translate into a prognostic benefit. However, before the pathologic findings are known, the distinction between TZ and PZ pathology is crucial. As Stamey et al. [13] noted, key characteristics distinguish men with TZ cancers from those with PZ cancers [8]. Despite high PSA values, those in the former group harbour cancers that have more favourable Gleason sums. These findings were confirmed in our descriptive analyses (Table 1). Similarly, fewer TZ PCa had ECE, SVI, and LNI. Therefore, despite high PSA, men with TZ cancers have more favourable pathologic characteristics. Therefore, when TZ PCa is suspected, definitive therapy should be strongly considered. After definitive therapy is delivered and when pathologic stage is known, the distinction between TZ and PZ origin blurs, especially when pathologic stage variables are held constant. 5. Conclusions IfPSA,Gleasonsum,ECE,SVI,andLNIareallequal, patients with TZ PCa at radical prostatectomy may expect the same rate of BCR as their counterparts with PZ cancers. The addition of zonal origin to these established pathologic predictors and PSA did not enhance the ability to prognosticate BCR after radical prostatectomy. Nonetheless, the clinical distinction between TZ and PZ PCa remains important, as the former clearly demonstrates more favourable pathologic features, despite higher PSA. References [1] Stephenson AJ, Scardino PT, Eastham JA, et al. Postoperative nomogram predicting the 10-year probability of prostate cancer recurrence after radical prostatectomy. J Clin Oncol 2005;23:7005 12. [2] Kattan MW, Shariat SF, Andrews B, et al. The addition of interleukin-6 soluble receptor and transforming growth factor beta1 improves a preoperative nomogram for predicting biochemical progression in patients with clinically localized prostate cancer. J Clin Oncol 2003; 21:3573 9. [3] McNeal JE, Redwine EA, Freiha FS, Stamey TA. Zonal distribution of prostatic adenocarcinoma: correlation with histologic pattern and direction of spread. Am J Surg Pathol 1988;12:897 906. [4] Shannon BA, McNeal JE, Cohen RJ. Transition zone carcinoma of the prostate gland: a common indolent tumour type that occasionally manifests aggressive behaviour. Pathology 2003;35:467 71. [5] Steuber T, Karakiewicz PI, Augustin H, et al. Transition zone cancers undermine the predictive accuracy of Partin table stage predictions. J Urol 2005;173:737 41. [6] Greene DR, Wheeler TM, Egawa S, Dunn JK, Scardino PT. A comparison of the morphological features of cancer arising in the transition zone and in the peripheral zone of the prostate. J Urol 1991;146:1069 76. [7] Aus G, Abbou CC, Bolla M, et al. EAU guidelines on prostate cancer. Eur Urol 2005;48:546 51. [8] Erbersdobler A, Fritz H, Schnoger S, et al. Tumour grade, proliferation, apoptosis, microvessel density, p53, and bcl-2 in prostate cancers: differences between tumours located in the transition zone and in the peripheral zone. Eur Urol 2002;41:40 6. [9] Stamey TA, Yemoto CM. The clinical importance of separating transition zone (TZ) from peripheral zone (PZ) cancers. J Urol 1998;159:221. [10] Sakai I, Harada K, Hara I, Eto H, Miyake H. A comparison of the biological features between prostate cancers arising in the transition and peripheral zones. BJU Int 2005;96: 528 32. [11] Grignon DJ, Sakr WA. Zonal origin of prostatic adenocarcinoma: Are there biologic differences between transition zone and peripheral zone adenocarcinomas of the prostate gland? J Cell Biochem 1994;19:267 9. [12] Augustin H, Erbersdobler A, Graefen M, et al. Biochemical recurrence following radical prostatectomy: a comparison between prostate cancers located in different anatomical zones. Prostate 2003;55:48 54. [13] Stamey TA, Sozen TS, Yemoto CM, McNeal JE. Classification of localized untreated prostate cancer based on 791 men treated only with radical prostatectomy: common ground for therapeutic trials and TNM subgroups. J Urol 1998;159:2009 12. [14] Noguchi M, Stamey TA, Neal JE, Yemoto CE. An analysis of 148 consecutive transition zone cancers: clinical and histological characteristics. J Urol 2000;163:1751 5. [15] Gleason DF. Histologic grading and clinical staging of prostate carcinoma. In: Tannenbaum M, editor. Urologic pathology: the prostate. Philadelphia, PA: Lea and Febiger; 1977. p. 171 97. [16] 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. [17] Kattan MW, Eastham JA, Stapleton AM, Wheeler TM, Scardino PT. A preoperative nomogram for disease recurrence following radical prostatectomy for prostate cancer. J Natl Cancer Inst 1998;90:766 71. [18] Stamey TA, McNeal JE, Yemoto CM, Sigal BM, Johnstone IM. Biological determinants of cancer progression in men with prostate cancer. JAMA 1999;281:1395 400.

european urology 51 (2007) 949 955 955 [19] D Amico AV, Whittington R, Malkowicz SB, et al. A multivariate analysis of clinical and pathological factors that predict for prostate specific antigen failure after radical prostatectomy for prostate cancer. J Urol 1995; 154:131 8. [20] Katz EM, Kattan MW. How to judge a tumor marker. Nat Clin Pract Oncol 2005;2:482 3. [21] Augustin H, Hammerer PG, Blonski J, et al. Zonal location of prostate cancer: Significance for disease-free survival after radical prostatectomy? Urology 2003;62:79 85. [22] Pelzer AE, Bektic J, Berger AP, et al. Are transition zone biopsies still necessary to improve prostate cancer detection?: Results from the Tyrol screening project. Eur Urol 2005;48:916 21. Editorial Comment Michael Kattan kattanm@ccf.org Chun et al. have done an elegant job of looking at whether zonal origin is useful for predicting biochemical failure following radical prostatectomy. Rather than simply looking at the multivariable analysis p-value and hazard ratio, they have examined the incremental predictive accuracy zonal origin, and found it to be very limited. In short, knowledge of zonal origin does not enhance our ability to predict recurrence after adjusting for standard prognostic factors. Chun et al. carefully compared a model which contains zonal origin (plus the other prognostic factors) to a model that lacked zonal origin (i.e., had only the standard prognostic factors). These models were nearly identical in their ability to predict recurrence. It is important to compare these two models since a change in predictive accuracy can be caused by (1) the addition of a valuable biomarker, (2) an increase in sample size, (3) an improvement in the collection/standardization of data, and (4) modeling. Chun et al. held constant causes (2) through (4) to isolate the effect of (1). It would seem adding zonal origin to a postoperative nomogram would provide negligible benefit.