Development and internal validation of preoperative transition zone prostate cancer nomogram

ADULT UROLOGY

DEVELOPMENT AND INTERNAL VALIDATION OF PREOPERATIVE TRANSITION ZONE PROSTATE CANCER NOMOGRAM THOMAS STEUBER, FELIX K.-H. CHUN, ANDREAS ERBERSDOBLER, ALBERTO BRIGANTI, ALEXANDER HAESE, MARKUS GRAEFEN, THORSTEN SCHLOMM, LUC VALIQUETTE, HARTWIG HULAND, AND PIERRE I. KARAKIEWICZ

ABSTRACT Objectives. Up to 20% of men may harbor a transition zone (TZ) prostate cancer (PCa) at radical prostatectomy (RP). TZ PCa may be associated with more favorable RP pathologic findings than peripheral zone (PZ) PCa. To identify these men, we developed a model capable of predicting the probability of TZ PCa at RP. Methods. The study cohort consisted of 945 consecutive men treated with RP, with clinical stage, prostatespecific antigen (PSA) level, and detailed biopsy and RP pathology data available. The preoperative variables were used as predictors in the multivariate logistic regression models to predict the rate of TZ PCa at RP. PCa was defined as a TZ tumor when more than 50% of the planimetrically measured tumor volume was situated within the TZ. Regression coefficients were used to develop nomograms, which were subjected to 200 bootstrap resamples to reduce overfit bias. Results. TZ PCa at the final pathologic examination was recorded in 110 patients (11.6%). After 200 bootstraps, the most parsimonious and most accurate nomogram was 77.3% accurate in predicting the probability of TZ PCa. Conclusions. This nomogram is ideally suited to identify patients with markedly elevated, nearly metastatic serum PSA levels who harbor a TZ tumor that is highly treatable by RP. UROLOGY 68: 1295–1300, 2006. © 2006 Elsevier Inc.

T

he anatomic model that divides the prostate into a peripheral zone (PZ), central zone, and transitional zone (TZ) was introduced by McNeal et al.1 These distinct anatomic zones may harbor prostate cancer (PCa). Up to 25% of prostate tumors are situated within the TZ at radical prostatectomy (RP).2– 4 Importantly, TZ PCa differs from PZ PCa. Specifically, TZ tumors are associated with high preoperative prostate-specific antigen (PSA)

P. I. Karakiewicz was partially supported by the Fonds de la Recherche en Santé du Québec, University of Montreal Health Center Foundation, Department of Surgery, and Les Urologues Associés du University of Montreal Health Center. From the Departments of Urology and Pathology, and Martini Clinic–Prostate Cancer Center, University of Hamburg, Hamburg, Germany; and Cancer Prognostics and Health Outcomes Unit, University of Montreal, Montreal, Quebec, Canada Reprint requests: Pierre I. Karakiewicz, M.D., Cancer Prognostics and Health Outcomes Unit, University of Montreal Health Center, 1058, rue St.-Denis, Montréal, QC H2X 3J4 Canada. E-mail: [email protected] Submitted: January 30, 2006, accepted (with revisions): August 14, 2006 © 2006 ELSEVIER INC. ALL RIGHTS RESERVED

levels and greater tumor volumes relative to PZ PCa. Conversely, TZ PCa exhibits lower biopsy Gleason sums. Despite these seemingly adverse PSA characteristics, TZ PCa is associated with a greater rate of organ-confined disease and a lower rate of biochemical recurrence relative to PZ PCa.5–12 These features make patients with TZ PCa candidates for RP. However, to date, no validated tools are available that can discriminate between TZ and PZ tumors in patients with adverse preoperative features. Therefore, definitive treatment could be denied to some men with TZ PCa. We addressed this void13,14 and developed a tool capable of accurately predicting the probability of TZ PCa at RP. MATERIAL AND METHODS PATIENT POPULATION From February 1997 to December 2003, the clinical and pathologic data were prospectively gathered for 1154 consecutive patients treated at the University of Hamburg, Germany by RP for biopsy proven, clinically localized PCa. Of these, 209 0090-4295/06/$32.00 doi:10.1016/j.urology.2006.08.1066 1295

patients were excluded because of missing data. Of these 209 patients, 33 were missing RP TZ versus PZ classification, 5 were missing the biopsy Gleason sum, 64 were missing the pathologic Gleason sum, 62 were missing the prostate volume, and 45 were missing biopsy location information. The analyses targeted the 945 remaining patients.

CLINICAL AND PATHOLOGIC EVALUATION The clinical stage was assigned by the attending urologist according to the 1992-2002 TNM system. Between six and eight needle cores were obtained under transrectal ultrasound guidance. The pretreatment PSA level was measured before digital rectal examination and transrectal ultrasonography. Each biopsy core was individually interpreted, and the distribution of cores was recorded for each patient. The percentage and length of PCa were quantified in each core and graded according to the Gleason system. To calculate the cumulative percentage of cancer in all biopsy cores, we divided the total length of biopsy tissue involved with cancer by the total length of biopsy tissue examined. All prostatectomy specimens were step sectioned at 3 mm according to the Stanford protocol and graded according to the Gleason system. Tumor foci were identified. Computer-assisted planimetry was then used to define the cumulative tumor volume. Cases with more than 50% of the cumulative tumor volume within the TZ were classified as TZ PCa.4,11 No patient received neoadjuvant androgen deprivation therapy.

STATISTICAL ANALYSIS All statistical tests were performed using the Statistical Package for Social Systems, version 10 (SPSS, Chicago, Ill) and S-PLUS Professional, version 1 (MathSoft, Seattle, Wash). Two-sided tests and a significance level of 0.05 were used in all statistics. We compared the distribution of PSA level, clinical stage, prostate volume, biopsy Gleason sum, detailed biopsy information on tumor location and extent, and Partin pathologic stage according to PCa zone origin (TZ versus PZ) using either the chi-square or independent sample t test. The PSA level, clinical stage, biopsy Gleason sum, total prostate volume, and detailed biopsy information were used as predictors in the univariate and multivariate logistic regression models predicting the presence of TZ PCa at RP. Regression coefficients were used to develop a nomogram predicting the probability of TZ PCa. Then, 200 bootstrap resamples were used for internal validation of the accuracy estimates and to reduce overfit bias. The bootstrap-corrected area under the curve represents an average of 200 internal validations. All variables were forced into the multivariate model. Fast backward variable selection was used to define the most parsimonious and accurate model. No external validation (split sample) was performed.

RESULTS The descriptive characteristics of the study cohort are shown in Table I. Of the 945 patients, 110 (11.6%) had TZ versus 835 (88.4%) with PZ tumors. The pretreatment mean and median PSA value was 14.0 and 10.3 ng/mL for TZ tumors versus 8.2 and 6.5 ng/mL for PZ tumors (P ⬍0.001). Clinical Stage T1c and T2a was recorded in 102 (92.7%) TZ and 724 (86.7%) PZ tumors, respectively (P ⫽ 0.03). Of all TZ tumors, 94 (85.4%) had a biopsy Gleason sum of 6 or 7 versus 793 (95%) of their PZ PCa counterparts (P ⬍0.001). The mean and median total prostate volume were not signif1296

icantly different statistically (49.1 and 45.5 versus 48.1 and 44.0 cm3, respectively, P ⫽ 0.6) between TZ and PZ PCa. Most TZ tumors (n ⫽ 83, 75.5%) had significantly fewer positive biopsy cores (either one or two) compared with their PZ counterparts (n ⫽ 494, 59.2%, P ⫽ 0.02). According to biopsy location, TZ PCa showed a trend toward more positive biopsy cores at the apex (P ⫽ 0.08) and significantly fewer positive biopsy cores in the mid-prostate (P ⬍0.001) or at the base (P ⬍0.001). Exclusively positive apical biopsy cores were significantly more frequent statistically in TZ tumors (P ⬍0.001). The mean and median percentage of tumor in all biopsy cores was significantly greater statistically in PZ tumors (7.8 and 5.0 versus 14.2 and 9.0 cm3, respectively, P ⬍0.001). Table II shows the univariate and multivariate logistic regression models for the entire cohort. On the univariate analyses, all variables were highly statistically significant (all P ⱕ0.01) predictors of the presence of TZ tumor at the final pathologic examination, except for clinical stage (P ⫽ 0.1), total prostate volume (P ⫽ 0.6), and the presence of positive biopsy cores exclusively at the midprostate (P ⫽ 0.5) and base (P ⫽ 0.3). Of all predictors, the number of positive biopsy cores at the base (67.5%) was most informative, followed by PSA level (66.1%), cumulative percentage of cancer in all biopsy cores (63.4%), and overall number of positive biopsy cores (60.4%), for which 50% accuracy represents a flip of a coin and 100% represents perfect predictions. In the multivariate full model that included all variables, PSA level, biopsy Gleason sum, and the cumulative percentage of cancer in all biopsy cores were highly significant statistically (P ⱕ0.009), none of the remaining variables reached statistical significance (all P ⬎0.05). The multivariate predictive accuracy of the full model was 76.4% and exceeded that of any individual predictor. After applying the fast backward variable selection method, the nonsignificant and uninformative variables were eliminated. This resulted in the reduced model, which included PSA level, biopsy Gleason sum, presence of positive biopsy cores exclusively at the mid-prostate, number of positive cores at the base, and cumulative percentage of cancer in all biopsy cores (Table II). Its accuracy was 77.3% and exceeded that of any individual predictor as well as did the full model. The assessment of the nomogram axes (Fig. 1A) revealed that biopsy Gleason sum was inversely related to the probability of TZ PCa at RP and that the probability of TZ PCa at RP increased if the mid and/or base cores were not involved by PCa. Figure 1B depicts the calibration plot of the model, which shows the relation between the nomogram probability and UROLOGY 68 (6), 2006

TABLE I.

Descriptive characteristics for 945 men treated with radical prostatectomy

Variable Patients (n) Age (yr) Mean Median Range PSA (ng/mL) Mean Median Range Clinical stage (TNM 2002) T1c T2a T2b T2c T3 Prostate volume (cm3) Mean Median Range Biopsy Gleason sum ⬍6 6 7 (3 ⫹ 4) 7 (4 ⫹ 3) ⬎8 Overall No. of positive cores 1 2 3 4 5 6 Positive cores at apex (n) 0 1 2 Presence of positive cores exclusively at apex Positive cores at mid-prostate (n) 0 1 2 Presence of positive cores exclusively at mid-prostate Positive cores at base (n) 0 1 2 Presence of positive cores exclusively at base Cumulative percentage of cancer in all biopsy cores Mean Median Range Organ confinement Extracapsular extension Positive surgical margin status

Entire Cohort

TZ Tumors

PZ Tumors

P Value

945 (100)

110 (11.6)

835 (88.4)

—

62 62 39–74

62 62 50–73

62 62 39–74

8.9 6.8 0.1–74.6

14.0 10.3 2.1–74.6

8.2 6.5 0.1–73.2

661 (69.9) 165 (17.5) 91 (9.6) 21 (2.2) 7 (0.7)

91 (82.7) 11 (10.0) 7 (6.4)

570 (68.3) 154 (18.4) 84 (10.1) 21 (2.5) 6 (0.7)

48.2 44.0 13.3–140.0

1 (0.9) 49.1 45.5 13.3–104

48.1 44.0 15–140

0.9

⬍0.0001

0.03

0.6

32 (3.4) 595 (63.0) 223 (23.6) 69 (7.3) 26 (2.8)

15 (13.6) 67 (60.9) 22 (20.0) 5 (4.5) 1 (0.9)

17 (2.0) 528 (63.2) 201 (24.1) 64 (7.7) 25 (3.0)

⬍0.0001

318 (33.7) 259 (27.4) 210 (22.2) 81 (8.6) 46 (4.9) 31 (3.3)

50 (45.5) 33 (30.0) 17 (15.5) 7 (6.4) 2 (1.8) 1 (0.9)

268 (32.1) 226 (27.1) 193 (23.1) 74 (8.9) 44 (5.3) 30 (3.6)

0.02

404 (42.8) 427 (45.2) 114 (12.1) 108 (11.4)

32 (29.1) 63 (57.3) 15 (13.6) 33 (30.0)

372 (44.6) 364 (43.6) 99 (11.9) 75 (9.0)

0.08

317 (33.5) 509 (53.9) 119 (12.6) 76 (8.0)

55 (50.0) 47 (42.7) 8 (7.3) 7 (6.4)

262 (31.4) 462 (55.3) 111 (13.3) 69 (8.3)

⬍0.0001

320 (33.9) 483 (51.1) 142 (15.0) 155 (16.4)

68 (61.8) 37 (33.6) 5 (4.5) 14 (12.7)

252 (30.2) 446 (53.4) 137 (16.4) 141 (16.9)

⬍0.0001

13.5 8.0 1.0–83.0 664 (70.3) 276 (29.2) 210 (22.2)

7.8 5.0 1.0–34.0 90 (81.8) 19 (17.3) 21 (19.1)

14.2 9.0 1.0–83.0 574 (68.7) 257 (30.8) 189 (22.6)

⬍0.0001

⬍0.0001

0.5

0.3

0.005 0.003 0.4

KEY: TZ ⫽ transition zone; PZ ⫽ peripheral zone; PSA ⫽ prostate-specific antigen. Data in parentheses are percentages.

UROLOGY 68 (6), 2006

1297

TABLE II.

Univariate and multivariate analyses predicting transition zone prostate cancer at radical prostatectomy in 945 men Univariate Model

Predictor PSA Biopsy Gleason sum 6 vs. ⬍6 7(3 ⫹ 4) vs. ⬍6 7(4 ⫹ 3) vs. ⬍6 ⱖ8 vs. ⬍6 Clinical stage T2a vs. T1c T2b vs. T1c T2c vs. T1c T3 vs. T1c Prostate volume Overall No. of positive cores Positive cores at apex Presence of positive cores exclusively at apex Positive cores at mid-prostate Presence of positive cores exclusively at mid-prostate Positive cores at base Presence of positive cores exclusively at base Cumulative percentage of cancer in all biopsy cores Predictive accuracy (%) (95% CI)

Multivariate Models

Predictive Accuracy (%)

OR

P Value

1.07 — 0.14 0.12 0.09 0.05 — 0.45 0.52 0.001 1.04 1.00 0.73 1.44 4.34

⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 0.004 0.10 0.02 0.11 0.74 0.97 0.62 0.001 0.01 ⬍0.001

66.1 56.7

0.54 0.75

Full Model

Reduced Model

OR

P Value

OR

P Value

⬍0.001 0.001 ⬍0.001 0.001 0.002 0.01 0.78 0.21 0.78 0.78 0.77 0.19 0.66 0.14 0.96

1.1 — 0.17 0.2 0.12 0.06

⬍0.001 ⬍0.001 ⬍0.001 0.001 0.002 0.01

50.4 60.4 57.1 60.2

1.11 — 0.18 0.20 0.12 0.05 — 0.63 0.87 ⬍0.001 0.71 0.99 0.85 1.86 1.03

⬍0.001 0.49

60.4 50.3

1.06 0.42

0.89 0.12

0.25

0.002

0.33 0.72

⬍0.001 0.27

67.5 51.3

0.53 1.32

0.12 0.60

0.37

⬍0.001

0.95

⬍0.001

63.4

0.96

0.009

0.95

0.001

57.2

76.4 (76.2–76.8)

77.3 (77.0–77.7)

KEY: OR ⫽ odds ratio; PSA ⫽ prostate-specific antigen; CI ⫽ confidence interval.

the observed rate of TZ PCa. The nomogram predictions were very close to ideal predictions, as indicated by the 45° line. COMMENT In general, high pretreatment total PSA levels and large tumor volumes are indicative of aggressive, more advanced, disease with a resultant poor prognosis.15 Despite high PSA values and large tumor volumes, TZ PCa may show highly favorable pathologic features.3,4,6,7 Furthermore, TZ PCa is associated with more favorable biologic features, emphasizing the distinctness of TZ PCa versus PZ PCa.6 – 8 Except for tumor biology, we have confirmed these features as shown in Table I. Noguchi and colleagues4 also reported that despite high PSA values, TZ PCa cases will have lower pathologic tumor stages and lower rates of biochemical recurrence (P ⫽ 0.0002) when matched with PZ PCa by tumor volume and percentage of high-grade disease. Noguchi et al.16 reconfirmed these findings in a subcohort of men with organconfined disease. However, the independent predictor status of TZ PCa could not always be confirmed.3 In addition to the clinical and pathologic 1298

features, specific biopsy patterns were identified in men with TZ PCa. Augustin et al.14 reported lower positive biopsy rates in the mid-prostate and at the base (P ⫽ 0.02 and P ⬍0.001) and a greater positive biopsy rate at the apex (P ⫽ 0.002), relative to PZ PCa. We corroborated these findings (Table I). Taken together, TZ PCa demonstrated specific PSA levels, biopsy findings, RP pathologic features, and biochemical recurrence patterns relative to their PZ counterparts. We relied on this established distinctiveness of the clinical and biopsy characteristics to develop a model predicting TZ PCa at RP, as previously suggested by Augustin et al.14 and Steuber et al.13 Our results have confirmed that specific variables such as PSA level and biopsy features represent statistically significant predictors of TZ PCa. Moreover, these variables are highly informative predictors of TZ PCa, and their combined contribution yielded a model that is 77.3% accurate in predicting the probability of TZ PCa at RP. These variables included PSA level, biopsy Gleason sum, presence of positive biopsy cores exclusively at the mid-prostate, number of positive cores at the base, and cumulative percentage of cancer in all biopsy cores. UROLOGY 68 (6), 2006

FIGURE 1. (A) Nomogram predicting presence of TZ PCa at final pathologic examination and (B) corresponding calibration plot. BxGleasonSum ⫽ biopsy Gleason sum; ⫹BxCore@Midonly ⫽ presence of positive biopsy cores exclusively at mid-prostate; #⫹BxCores@Base ⫽ number of positive cores at base; Total%CaVolume ⫽ cumulative percentage of cancer in all biopsy cores; probability of TZ PCa ⫽ probability of TZ PCa at final pathologic examination. Instructions for nomogram: to obtain nomogram predicted probability of TZ PCa, locate patient 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 to assess the individual probability of TZ PCa on the “probability of TZ PCa” line. Instructions for calibration plot: perfect predictions correspond to 45° line. Points estimated below 45° line correspond to nomogram over prediction, and points situated above 45° line correspond to nomogram under prediction. Nonparametric, smoothed curve indicates relationship between predicted probability and observed proportion of TZ PCa on final pathologic examination.

The assessment of the nomogram axes showed that high PSA levels are predictive of TZ PCa. Moreover, TZ PCa is more likely to exhibit a lower biopsy Gleason sum. Finally, TZ PCa rarely shows positive cores at the mid or base of the gland. These findings are consistent with anterior large volume tumors. These tumors are less extensively sampled by biopsy schemes that focus on the PZ. Thus, the UROLOGY 68 (6), 2006

percentage of cores involved by PCa is usually lower in TZ PCa than it is in PZ PCa. The combination of these variables reliably discriminates between TZ and PZ PCa in 77.3% of cases. Several applications of our findings should be considered. Our model is particularly useful in determining who may be at risk of having a TZ PCa versus a locally advanced or even metastatic PZ PCa. Thus, our nomogram may help in deciding who should be offered definitive treatment versus systemic therapy, because not all men with elevated PSA levels have metastatic disease. Instead, a patient may have TZ PCa and may be most effectively treated with definitive therapy. For example, a man with a high PSA value, who has a high nomogram-predicted probability of a TZ PCa, could be offered definitive therapy, and a man with a high PSA level and a low-predicted TZ PCa probability might be a less suitable candidate for definitive treatment. Our nomogram represents the first multivariate, internally validated attempt at the prediction of TZ PCa at RP. We are unaware of other models that address this outcome. Moreover, we have not only shown a statistically significant model, but have also clearly demonstrated that it is capable of predicting the probability of TZ PCa at RP highly accurately. Graefen et al.17 has recently shown that, despite statistical significance, variables may offer no ability to improve overall model predictive accuracy. Our method circumvented this difficulty and confirmed the accuracy of our model. Moreover, our modeling technique allows the user to assess the relative importance of the individual predictors, as evidenced by the effect of these variables on the risk axes of the nomogram (Fig. 1A). Additionally, our predictions are presented as a percentage. Such a format is easier to interpret than the odds ratios provided in standard logistic regression models. Furthermore, our nomogram is accompanied by its performance characteristics, which allows the user to assess the relation between the nomogram-predicted and observed rate of TZ PCa at RP. The calibration plot shows that the predicted probabilities are virtually the same as the observed rate of TZ PCa, indicating excellent performance. Finally, our model is superior to other modeling techniques, because it does not require computational facilities or custom-designed software. It can be provided in a user-friendly paper-based format, perfectly suited for use in a busy clinical practice. Our study had distinct limitations. We relied on systematic sextant and eight-core biopsy schemes for diagnosis. The recent, more extensive, biopsy regimens may provide more detailed pathologic information and possibly could more accurately predict for TZ pathologic features. However, this is speculative, and the prediction of TZ PCa should 1299

ideally be validated in a prospectively collected data set. An additional limitation of this study resulted from its single-institution design, which might undermine the generalizability of its findings. Moreover, the overall accuracy of our model was not perfect (77.3%). Therefore, as many as 22.7% of predictions could be incorrect. This limitation is shared by other predictive models in which the accuracy has rarely exceeded 80%.18 Additionally, we have not confirmed the accuracy of the nomogram in an external data set. Instead, we proved its accuracy using a statistically accepted surrogate of external validation, namely with bootstrapping, which simulates the use of the nomogram under novel testing conditions. Two hundred bootstraps imitate the application of the nomogram in 200 novel cohorts and represent a statistical surrogate of 200 external validations. Finally, the accuracy of our model could potentially be improved by integrating additional novel predictor variables. For example, in general, biomarkers such as transforming growth factor-beta and interleukin-6 or human glandular kallikrein 2 were shown to improve the predictive accuracy of the preoperative biochemical recurrence nomogram.19,20 CONCLUSIONS TZ PCa at the final pathologic examination represents an important consideration in treatment decision-making, even in contemporary patients. Nearly one third of patients with PCa will exhibit TZ PCa at the final pathologic examination, and our nomogram can accurately identify those men. REFERENCES 1. McNeal JE, Redwine EA, Freiha FS, et al: Zonal distribution of prostatic adenocarcinoma: correlation with histologic pattern and direction of spread. Am J Surg Pathol 12: 897–906, 1988. 2. Reissigl A, Pointner J, Strasser H, et al: Frequency and clinical significance of transition zone cancer in prostate cancer screening. Prostate 30: 130 –135, 1997. 3. Augustin H, Hammerer PG, Blonski J, et al: Zonal location of prostate cancer: significance for disease-free survival after radical prostatectomy? Urology 62: 79 – 85, 2003. 4. Noguchi M, Stamey TA, Neal JE, et al: An analysis of 148 consecutive transition zone cancers: clinical and histological characteristics. J Urol 163: 1751–1755, 2000. 5. Augustin H, Erbersdobler A, Hammerer PG, et al: Prostate cancers in the transition zone: part 2— clinical aspects. BJU Int 94: 1226 –1229, 2004.

1300

6. Grignon DJ, and 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 Suppl 19: 267–269, 1994. 7. Greene DR, Wheeler TM, Egawa S, et al: A comparison of the morphological features of cancer arising in the transition zone and in the peripheral zone of the prostate. J Urol 146: 1069 –1076, 1991. 8. McNeal JE: Cancer volume and site of origin of adenocarcinoma in the prostate: relationship to local and distant spread. Hum Pathol 23: 258 –266, 1992. 9. Stamey TA, Yemoto CM, McNeal JE, et al: Prostate cancer is highly predictable: a prognostic equation based on all morphological variables in radical prostatectomy specimens. J Urol 163: 1155–1160, 2000. 10. Stamey TA, and Yemoto CE: The clinical importance of separating transition zone (TZ) from peripheral zone (PZ) cancers (abstract). J Urol 159(suppl): 221, 1998. 11. Stamey TA, Sozen TS, Yemoto CM, et al: 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 159: 2009 – 2012, 1998. 12. Shannon BA, McNeal JE, and Cohen RJ: Transition zone carcinoma of the prostate gland: a common indolent tumour type that occasionally manifests aggressive behaviour. Pathology 35: 467– 471, 2003. 13. Steuber T, Karakiewicz PI, Augustin H, et al: Transition zone cancers undermine the predictive accuracy of Partin table stage predictions. J Urol 173: 737–741, 2005. 14. Augustin H, Erbersdobler A, Graefen M, et al: Differences in biopsy features between prostate cancers located in the transition and peripheral zone. BJU Int 91: 477– 481, 2003. 15. Stamey TA, McNeal JE, Yemoto CM, et al: Biological determinants of cancer progression in men with prostate cancer. JAMA 281: 1395–1400, 1999. 16. Noguchi M, Stamey TA, McNeal JE, et al: Preoperative serum prostate specific antigen does not reflect biochemical failure rates after radical prostatectomy in men with large volume cancers. J Urol 164: 1596 –1600, 2000. 17. Graefen M, Ohori M, Karakiewicz PI, et al: Assessment of the enhancement in predictive accuracy provided by systematic biopsy in predicting outcome for clinically localized prostate cancer. J Urol 171: 200 –203, 2004. 18. Augustin H, Eggert T, Wenske S, et al: Comparison of accuracy between the Partin tables of 1997 and 2001 to predict final pathological stage in clinically localized prostate cancer. J Urol 171: 177–181, 2004. 19. 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 21: 3573–3579, 2003. 20. Steuber T, Vickers AJ, Haese A, et al: Risk assessment for biochemical recurrence prior to radical prostatectomy: significant enhancement contributed by human glandular kallikrein 2 (hK2) and free prostate specific antigen (PSA) in men with moderate PSA-elevation in serum. Int J Cancer 118: 1234 –1240, 2006.

UROLOGY 68 (6), 2006