european urology 49 (2006) 820–826

available at www.sciencedirect.com journal homepage: www.europeanurology.com

Prostate Cancer

Development and Internal Validation of a Nomogram Predicting the Probability of Prostate Cancer Gleason Sum Upgrading Between Biopsy and Radical Prostatectomy Pathology Felix K.-H. Chun a,d,1, Thomas Steuber a,1, Andreas Erbersdobler c, Eike Currlin a, Jochen Walz a, Thorsten Schlomm a, Alexander Haese a, Hans Heinzer a, Michael McCormack d, Hartwig Huland a,b, Markus Graefen b, Pierre I. Karakiewicz d,* a

Department of Urology, University of Hamburg, Hamburg, Germany Martini Clinic– Prostate Cancer Center, University of Hamburg, Hamburg, Germany c Department of Pathology, University of Hamburg, Hamburg, Germany d Cancer Prognostics and Health Outcomes Unit, University of Montreal, Montreal, Quebec, Canada b

Article info

Abstract

Article history: Accepted November 10, 2005 Published online ahead of print on December 22, 2005

Objective: Previous reports indicate that as many as 43% of men with low grade PCa at biopsy will be diagnosed with high-grade PCa at RP. We explored the rate of upgrading from biopsy to RP specimen in our contemporary cohort, and developed a model capable of predicting the probability of biopsy Gleason sum upgrading. Materials and Methods: The study cohort consisted of 2982 men treated with RP, with available clinical stage, serum prostate specific antigen and biopsy Gleason scores. These clinical data were used as predictors in multivariate logistic regression models (LRM) addressing the rate of Gleason sum upgrading between biopsy and RP pathology. LRM regression coefficients were used to develop a nomogram predicting the probability of Gleason sum upgrading and was subjected to 200 bootstrap resamples for internal validation and to reduce overfit bias. Results: Overall, 875 patients were upgraded (29.3%). In multivariate LRMs, all predictors were highly significant (all p values 10–20 ng/ml; >20 ng/ml) clinical stage (1992 AJCC T1c, T2a, T2b, T2c) and ultrasound determined prostate gland volume (30 cm3; >30–75 cm3; >75 cm3). Subsequently, we determined the predicted rate of Gleason sum upgrading and the predictive accuracy of the D’Amico model in our patient population. All statistical tests were performed using S-PLUS Professional, version 1 (MathSoft Inc., Seattle, Washington). Moreover, all tests were two-sided with a significance level at 0.05.

3.

Results

Patient characteristics are shown in Table 1. Pretreatment PSA levels ranged between 0–125 ng/ml. PSA values over 50 ng/ml were recorded in 207 patients (7.6%) and 2755 (92.4%) had PSA values below 20 ng/ml. Clinical stages T1c and T2a accounted jointly for 2444 patients (81.9%). Of all, 2880 (96.5%) had a biopsy Gleason sum  7 and were at highest risk of being upgraded to a more aggressive Gleason variant. Table 2 shows the concordance between biopsy and prostatectomy Gleason sum. Upgrading was recorded in 875 (29.3%), whereas 14.3% were downgraded. These data also indicate that 36.7% of patients were upgraded from biopsy Gleason 6 or less to pathologic Gleason sum 7 or higher. Conversely, 13.5% were downgraded from biopsy Gleason sum of 7 or higher to a pathologic Gleason sum of 6 or less. The same Gleason sum at biopsy and pathology was found in 1681 (56.4%). In men subjected to sextant biopsy, the rate of upgrading was 24.2% vs. 30.5% in those who underwent a 10core biopsy (odds ratio 1.2, p = 0.07). The overall rate

822

european urology 49 (2006) 820–826

Table 1 – Descriptive characteristics for 2982 evaluable prostate specimen sides Variables Pre-operative PSA Mean (median) Range

Number of patients 9.6 (7.0) 0–125

Clinical Stage (%) T1c T2a T2b T2c T3

1951 (65.4) 493 (16.5) 349 (11.7) 108 (3.6) 81 (2.7)

Biopsy Gleason primary (%) 1 2 3 4 5

5 (0.2) 140 (4.7) 2522 (84.6) 310 (10.4) 5 (0.2)

Biopsy Gleason secondary (%) 1 2 3 4 5

3 (0.1) 176 (5.9) 2030 (68.1) 742 (24.9) 31 (1.0)

Biopsy Gleason sum (%) 2 3 4 5 6 7 8 9

2 (0.1) 4 (0.1) 25 (0.8) 258 (8.7) 1704 (57.1) 887 (29.7) 73 (2.4) 29 (1.0)

Pathological Gleason primary (%) 2 3 4 5

32 (1.1) 2576 (86.4) 368 (12.3) 6 (0.2)

Pathological Gleason secondary (%) 2 3 4 5

326 (10.9) 1393 (46.7) 1223 (41.0) 40 (1.3)

Pathological Gleason sum (%) 4 5 6 7 8 9

5 (0.2) 347 (11.6) 1045 (35.0) 1527 (51.2) 19 (0.6) 39 (1.3)

Upgrading Gleason sum (%)

875 (29.3)

of upgrading from biopsy Gleason sum 6 or less to pathologic Gleason 7 or higher was 36.7%. Of men subjected to sextant biopsy, 38.0% were upgraded from 6 or less to 7 or more vs. 32.0% of those who underwent a 10-core biopsy (odds ratio 1.3, p = 0.03). Table 3 shows the temporal changes in the rate of Gleason sum upgrading and indicates that the rate

of upgrading decreased over time from 52.2 to 27.8% ( p = 0.003). Table 4 shows the univariate and multivariate logistic regression models for PSA, clinical stage, and primary and secondary biopsy Gleason score with corresponding univariate and multivariate predictive accuracy estimates. In univariate analyses, primary and secondary biopsy Gleason scores were highly significant predictors of final Gleason sum upgrading ( p < 0.001). Of all predictors, secondary biopsy Gleason score (0.697) represented the most informative predictor and was followed by primary biopsy Gleason score (0.604), PSA (0.524), and clinical stage (0.504). In multivariate analyses all variables were highly significant ( p < 0.001). The multivariate 200 bootstrap-corrected predictive accuracy was 80.4% and exceeded the most informative univariate predictor, namely secondary biopsy Gleason score (69.7%). Fig. 1 shows the regression coefficient-based nomogram. High PSA values, an intermediate clinical stage (T2b/c), as well as a low primary and/or secondary biopsy Gleason scores are risk factors for Gleason sum upgrading at final pathology. The calibration plots are shown in Fig. 2. Its X-axis represents the nomogram predictions, the Y-axis represents the observed rate of Gleason upgrading.

Fig. 1 – Nomogram predicting Gleason sum upgrading between biopsy and radical prostatectomy. Legends: PSA: Prostate-specific antigen (ng/ml); BX Gleason Pri: Primary biopsy Gleason score; BX Gleason Sec: Secondary biopsy Gleason score; P(Upgrade): probability of biopsy Gleason sum upgrading. To obtain nomogram predicted probability of biopsy upgrading, 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 be able to assess the individual probability of biopsy Gleason sum upgrading on the ‘‘P(Upgrade)’’ line.

823

european urology 49 (2006) 820–826

Table 2 – Concordance between biopsy and prostatectomy Gleason sum Pathologic Gleason sum

Biopsy Gleason sum 2

3

4

5

6

7

8

9

Total

4 5 6 7 8 9

0 1 1 0 0 0

0 1 0 3 0 0

0 5 12 8 0 0

0 87 91 80 0 0

3 219 841 634 3 4

2 34 95 735 8 13

0 0 1 54 7 11

0 0 4 13 1 11

5 347 1045 1527 19 39

Total

2

4

25

258

1704

887

73

29

2982

Grade agreement between biopsy and final pathology = 1681 (56.4%). Upgraded at RP = 875 (29.3%). Upgraded by 1 grade = 749 (25.1%). Upgraded by 2+ grades = 126 (4.2%). Downgraded at RP = 426 (14.3%). Downgraded by 1 grade = 369 (12.4%). Downgraded by 2+ grades = 57 (1.9%).

Table 3 – Annual upgrading rate Year of biopsy 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

Total number of cases

Rate of upgrade (%)

46 78 66 106 110 171 199 258 300 403 393 445 399

52.2 30.8 31.8 37.7 27.3 26.3 24.6 29.1 25.3 29.5 26.2 35.1 27.8

Chi-square p value = 0.003.

The 45-degree line represents ideal predictions. The nomogram calibration plot demonstrated virtually ideal predictions. The rate of predicted Gleason upgrading closely paralleled the observed rate of Gleason upgrading and nearly corresponded to the 45-degree line (Fig. 2A). The D’Amico model predictions are represented by open circles in Fig. 2B, whereas the curve represents the logistic calibration of the model. The model curve and individual data points show important departures from ideal prediction. The overall accuracy of the D’Amico model was 0.523, where 0.5 equals a toss of a coin and 1.0 represents perfect prediction.

Table 4 – Univariate and multivariate logistic regression models predicting Gleason sum upgrading between biopsy and final pathology Predictors

Univariate predictive accuracy

Pre-operative PSA Clinical Stage T1c T2a T2b T2c T3 Biopsy Gleason primary 1–2 3 4–5 Biopsy Gleason secondary 1–2 3 4–5 Predictive Accuracy OR – odds ratio; PSA – Prostate-specific antigen.

0.524 0.504

0.604

0.697

Univariate model OR; p value 1.007; 0.112 –; 0.783 1.000; – 1.049; 0.661 1.057; 0.662 0.975; 0.909 0.742; 0.265 –;