Journal Article Page
Jump to
European Urology
Volume 58, issue 5, pages e51-e52, November 2010Prostate Cancer
External Validation of Urinary PCA3-Based Nomograms to Individually Predict Prostate Biopsy Outcome
00597-X/assets/eulogo1.jpg)
Accepted 22 June 2010, Published online 2 July 2010, pages 727 - 732
Abstract Full Text Full-Text PDF (317 KB) Create Platinum Slide Series
- Abstract
- Introduction
- Materials and methods
- Results
- Discussion
- Conclusions
- Contributions
- References
- Authors
- Data
3. Results
Patient characteristics are shown in Table 1. Overall, PCa was detected in 41.1% (n = 255) of patients. As shown in Fig. 1A, median PCA3 score was 29 (range: 1–301). Median PCA3 score was significantly higher in patients with PCa versus patients without PCa (48 vs 20, p < 0.001). Overall, PSA levels ranged from 1.0 to 82.7 ng/ml (mean and median: 7.3 and 6.1 ng/ml, respectively) and DRE was suspicious in 16.9%. Increasing age and PSA level, suspicious DRE, and decreasing prostate volume were associated with PCa at biopsy with statistically high significance (p < 0.001).
Table 1 Risk factors comparing biopsy-negative and biopsy-positive men
| Variables | Validation cohort (n = 621) | |||
|---|---|---|---|---|
| Total | PCa negative | PCa positive | p value | |
| Patients, No. (%) | 621 (100) | 366 (58.9) | 255 (41.1) | – |
| Age, yr | <0.001 | |||
| Mean | 63.0 | 62 | 64.5 | |
| Median | 63 | 62 | 65 | |
| Range | 35–90 | 38–83 | 35–90 | |
| PSA (ng/ml) | <0.001 | |||
| Mean | 7.3 | 6.5 | 8.4 | |
| Median | 6.1 | 5.8 | 6.6 | |
| Range | 1–82.7 | 1.0–22.4 | 2.4–82.7 | |
| DRE, No. (%) | <0.001 | |||
| Suspicious | 105 (16.9) | 40 (10.9) | 65 (25.5) | |
| Unsuspicious | 516 (83.1) | 326 (89.1) | 190 (74.5) | |
| Total prostate volume (ml) | <0.001 | |||
| Mean | 48.0 | 52.6 | 41.3 | |
| Median | 44 | 47.2 | 38 | |
| Range | 10–148 | 12.4–148 | 10–132 | |
| Previous biopsy session, No. (%) | 0.60 | |||
| No | 467 (75.2) | 278 (76.0) | 189 (74.1) | |
| Yes | 154 (24.8) | 88 (24) | 66 (25.9) | |
| Gleason score | – | – | – | |
| ≤3 + 3 | 146 (57.3) | |||
| 3 + 4 | 67 (26.3) | |||
| 4 + 3 | 21 (8.2) | |||
| ≥4 + 4 | 21 (8.2) | |||
| PCA3 assay score (%) | <0.001 | |||
| Mean | 45.8 | 32.6 | 64.7 | |
| Median | 29 | 20 | 48 | |
| Range | 1–301 | 1–215 | 4–301 | |
| ≤ 17 | 195 (31.4) | 163 (44.5) | 32 (12.5) | <0.001 |
| >17 | 426 (68.6) | 203 (55.5) | 223 (87.5) | |
| ≤24 | 264 (42.5) | 208 (56.8) | 56 (22.0) | <0.001 |
| >24 | 357 (57.5) | 158 (43.2) | 199 (78.0) | |
| ≤35 | 347 (55.9) | 256 (69.9) | 91 (35.7) | |
| >35 | 274 (44.1) | 110 (30.1) | 164 (64.3) | <0.001 |
| – | – | – | ||
PCa = prostate cancer; PSA = prostate-specific antigen; DRE = digital rectal examination; PCA3 = prostate cancer antigen 3.
References in context
-
Patient characteristics are shown in Table 1.
Go to context
Fig. 1 Box plots of median PCA3 assay scores with comparison of biopsy negative and biopsy positive men in total (A), initial biopsy (B) and repeat biopsy (C) cohort.PCA3 = Prostate cancer gene 3; PCa = prostate cancer.
00597-X/assets/gr1/gr1_584x468.jpg)
References in context
-
As shown in Fig. 1A, median PCA3 score was 29 (range: 1–301).
Go to context -
Median PCA3 score was significantly higher in patients with PCa compared with patients without PCa (47 vs 17, p<0.001) (Fig. 1B).
Go to context -
Median PCA3 score was 45 (range: 1–289) and was significantly higher in patients with PCa versus those without (53 vs 37; p=0.002) (Fig. 1C).
Go to context
Of all men, 75.2% (n = 467) compared with 24.8% (n = 154) underwent an initial versus repeat biopsy. Initial biopsy revealed PCa in 40.5% (n = 189) of patients. Median PCA3 score on initial biopsy was 25 (range: 1–301). Median PCA3 score was significantly higher in patients with PCa compared with patients without PCa (47 vs 17, p < 0.001) (Fig. 1B). Median PSA level was 5.6 ng/ml (range: 1–57.2 ng/ml). Higher age, increasing PSA level, suspicious DRE, and decreasing prostate volume were statistically significantly associated with PCa at initial biopsy (p < 0.001).
At repeat biopsy PCa was diagnosed in 42.9% (n = 66) men. Median PCA3 score was 45 (range: 1–289) and was significantly higher in patients with PCa versus those without (53 vs 37; p = 0.002) (Fig. 1C). Serum PSA levels in men at repeat biopsy ranged from 1.8 to 82.7 ng/ml (mean and median: 7.3 and 6.1 ng/ml, respectively). Except for age and suspicious DRE, higher PSA level, lower prostate volume, and higher PCA3 score were significantly associated with PCa at repeat biopsy (p < 0.01).
Table 2 displays accuracy estimates in the external validation cohort (n = 621) stratified according to four different PCA3 codings. The AUCs were 0.73 versus 0.75 versus 0.74 versus 0.74 when continuously coded versus cut-off 17 versus cut-off 24 versus cut-off 35 for PCA3 were used.
Table 2 External validation accuracy estimates
| Variables | Accuracy |
|---|---|
| Age, PSA, DRE, PV, prevBx + PCA3 (cc) | 0.73 |
| Age, PSA, DRE, PV, prevBx + PCA3-17 | 0.75 |
| Age, PSA, DRE, PV, prevBx + PCA3-24 | 0.74 |
| Age, PSA, DRE, PV, prevBx + PCA3-35 | 0.74 |
PSA = prostate-specific antigen; DRE = digital rectal examination; PV = prostate volume; prevBx = history of previous biopsy (yes vs. no); PCA3 score = prostate cancer antigen 3 score; PCA3-17 = PCA3 assay score threshold 17; PCA3-24 = PCA3 assay score threshold 24; PCA3-35 = PCA3 assay score threshold 35.
References in context
-
Table 2 displays accuracy estimates in the external validation cohort (n=621) stratified according to four different PCA3 codings.
Go to context -
Our external validation findings demonstrate that accuracy depending on PCA3 coding ranged from 0.73–0.75 (Table 2).
Go to context
Fig. 2A–D displays nomogram calibration plots of the previously developed models stratified according to PCA3 coding, as applied to the external validation data set. On each calibration plot, the predicted probability of the previously reported nomogram is represented on the x-axis and the actual proportion of biopsy-proven PCa is represented on the y-axis. The 45° line indicates perfect agreement between predicted probability and observed proportion of PCa. Within the external validation cohort, equally excellent calibration of all investigated PCA3 codings is demonstrated, virtually overlapping the 45° line of perfect prediction.
Fig. 2 Nomogram calibration plots within external validation cohort stratified to various prostate cancer gene 3 (PCA3) coding.AUC = area under the curve (= predictive accuracy); PCA3(cc) = PCA3 assay score continuously coded; PCA3-17 = PCA3 assay score threshold 17; PCA3-24 = PCA3 assay score threshold 24; PCA3-35 = PCA3 assay score threshold 35.
00597-X/assets/gr2/gr2_584x384.jpg)
References in context
-
Fig. 2A–D displays nomogram calibration plots of the previously developed models stratified according to PCA3 coding, as applied to the external validation data set.
Go to context -
Moreover, external calibration virtually overlaps the 45° line of perfect prediction, indicating an almost perfect agreement between predicted and observed probability of PCa at biopsy (Fig. 2A–D).
Go to context
Article information
PII: S0302-2838(10)00597-X
DOI: 10.1016/j.eururo.2010.06.038
© 2010 European Association of Urology, Published by Elsevier B.V.
Recommend this article
Currently this article has a rating of 0. Please log in to recommend it.
Copyright © 