Clinical utility of the prostate cancer gene 3 (PCA3) urine assay in Japanese men undergoing prostate biopsy

Authors


Abstract

What's known on the subject? and What does the study add?

  • It is known that a prostate cancer gene 3 (PCA3) urine assay is superior to serum PSA level or PSA-related indices for predicting a positive biopsy result in European and US men.
  • This is the first report on PCA3 in a large cohort of Japanese men. The diagnostic value of the PCA3 score in Japanese men was similar to those reported in European and US men. The study concludes that a combination of PSA density and PCA3 score may be useful for selecting patients who could avoid an unnecessary biopsy.

Objective

  • To examine the diagnostic performance of the prostate cancer gene 3 (PCA3) score for prostate cancer in Japanese men undergoing prostate biopsy.

Patients and Methods

  • This Japanese, multicentre study included 647 Asian men who underwent extended prostate biopsy with elevated prostate-specific antigen (PSA) and/or abnormal digital rectal examination (DRE).
  • Urine samples were collected after DRE.
  • The PCA3 score was determined using a PROGENSA PCA3 assay and correlated with biopsy outcome. Its diagnostic accuracy was compared with that of serum PSA level, prostate volume (PV), PSA density (PSAD), and free/total PSA ratio (f/t PSA).

Results

  • A total of 633 urine samples were successfully analysed (the informative rate was 98%). Median PSA was 7.6 ng/mL.
  • Biopsy revealed cancer in 264 men (41.7%). The PCA3 score for men with prostate cancer was significantly higher than that for men with negative biopsies (median PCA3 score: 49 vs. 18; P < 0.001). The rate of positive biopsy was 16.0% in men with a PCA3 score of <20 and 60.6% in those with a PCA3 score of ≥50.
  • Using a PCA3 score threshold of 35, sensitivity and specificity were 66.5 and 71.6%, respectively.
  • The area under the curve of the PCA3 score was significantly higher than that of the f/t PSA in men with PSA 4–10 ng/mL (0.742 vs 0.647; P < 0.05).
  • In men with PSAD < 0.15 and PCA3 < 20, only three (4.2%) out of 72 men had prostate cancer.

Conclusions

  • The PCA3 score was significantly superior to f/t PSA in predicting a positive biopsy result for prostate cancer in Japanese men with PSA 4–10 ng/mL.
  • The combination of PSAD and PCA3 score may be useful for selecting patients who could avoid an unnecessary biopsy.
Abbreviations
PCA3

prostate cancer gene 3

f/t PSA

free/total PSA ratio

mRNA

messenger RNA

PV

prostate volume

PSAD

PSA density

AUC

area under the receiver–operator curve

Introduction

Serum PSA level has been widely used to detect prostate cancer [1]. It is organ-specific, but not cancer-specific. Several conditions, including BPH and prostatitis, may be associated with an elevated PSA level. An elevated PSA level is likely to be associated with prostate cancer, but the low specificity of PSA limits its use as a screening test and results in a large number of unnecessary biopsies [2]. Several modifications of PSA-related indices such as PSA isoforms and volume-referenced PSA have been investigated to improve the specificity of PSA in detecting prostate cancer [3-5]. Free/total PSA ratio (f/t PSA) is widely used in clinical practice to differentiate prostate cancer from BPH in men with grey zone PSA levels, but this does not have sufficient specificity to reduce unnecessary biopsies. Thus, more accurate and reliable assessments are needed to select candidates for prostate biopsy.

Bussemakers et al. [6]reported that prostate cancer gene 3 (PCA3) produces an untranslated, prostate-specific messenger RNA (mRNA) that is highly overexpressed in prostate cancer tissue compared with its level in normal or benign tissue. Several studies have shown that PCA3 urine assay is superior to serum PSA level or various PSA isoforms for predicting prostate cancer in European and US men, and it could be used as a diagnostic tool to select biopsy candidates [7-10]. We have previously demonstrated the high specificity of PCA3 urine assay in detecting prostate cancer in a limited number of Japanese men at a single institution [11]. In the present multicentre study, we investigated the diagnostic performance of this assay in a large cohort of Japanese men.

Methods

The study protocol was approved by the institutional review boards and all men provided written informed consent before enrolment in the present study. A total of 647 men with elevated serum PSA levels and/or an abnormal DRE were enrolled. They underwent systematic extended prostate biopsy (≥8 cores) at one of four Japanese institutions (Kyoto Prefectural University of Medicine, Japanese Red Cross Medical Centre, University of Tsukuba and Kinki University) from 2009 to May 2011. The ethnic origin of all patients was Asian. Among the 647 cases, 158 had a previous negative biopsy. The exclusion criteria were as follows: previous history of prostate cancer, taking medication known to affect serum PSA levels, UTIs, and history of invasive treatment for BPH. The first voided urine sample was collected after a DRE, and the urine specimen was examined using a PROGENSA PCA3 assay according to a previously described method [11]. The PCA3 score was determined using PCA3 mRNA copy divided by PSA mRNA copy. The f/t PSA was measured in men with PSA 4–10 ng/mL. Prostate volume (PV) was measured using ultrasonography and PSA density (PSAD) was calculated by dividing PSA by PV. Clinical and pathological outcomes such as clinical stage, Gleason score and percentage of positive cores (% positive cores) in men diagnosed with prostate cancer were correlated with PCA3 score. The % positive cores was calculated as the number of positive cores divided by the number of cores taken, and patients were divided into two groups according to % positive cores (≤33% or >33%). Indolent cancer was defined, according to the Epstein criteria, as clinical stage T1c, PSAD < 0.15, Gleason score ≤ 6, and <3 positive cores on a six-core biopsy, which was replaced by % positive cores ≤ 33 with biopsy sampling of more than six cores [12]. The Mann–Whitney test was used to compare continuous variables among the groups. The chi-squared test was used to assess nominal variables. Bivariate analysis (Pearson's correlation coefficient ‘r’) was used to test the linearity of relationships among the variables. Areas under the receiver–operator curves (AUCs) were compared. Multiple stepwise logistic regression analysis was used to determine the significant predictors of positive biopsy among variables such as repeated biopsy or not, PSA, PV, PSAD and PCA3 score. These statistical analyses were performed using commercially available software (SPSS version 12.0, Chicago, IL, USA). A P value of <0.05 was considered to indicate statistical significance.

Results

Among 647 urine samples, 633 were successfully analysed (the informative rate was 98%). The median (range) age, PSA level and number of biopsy cores taken were 69 (42–89) years, 7.6 (1.4–1908) ng/mL and 12 (6–59), respectively. Two patients had a six-core biopsy. There was no relationship between the PCA3 score and serum PSA level (r = 0.049, nonsignificant). Prostate biopsy was positive for prostate cancer in 264 men (41.7%). The characteristics of the men with positive and negative biopsies are shown in Table 1. The median PCA3 score in men with prostate cancer was significantly higher than that in those without prostate cancer (49 vs 18, P < 0.001). The positive rate of biopsy is shown in Table 2. We excluded all men who had a PSA level > 50 ng/mL and men who had initial biopsy with PSA of 20–50 ng/mL from further analyses because of the high yield of positive biopsy results; thus, the remaining 578 men were entered for analysis. The percentage of men with positive biopsy increased with increasing PCA3 score (Fig. 1). In men with a PCA3 score < 20, 16.0% (38/237) had a positive biopsy. When the PCA3 score was ≥50, the percentage of patients with a positive biopsy was 60.6% (106/175). Sensitivity, specificity, positive and negative predictive values of PCA3 scores at different PSA thresholds are shown in Table 3. Using a PCA3 threshold of 35, the sensitivity, specificity and diagnostic accuracy were 66.5, 71.6 and 69.7%, respectively. The AUCs of PSA, PV, PSAD and PCA3 score in 561 available men were 0.583, 0.706, 0.712 and 0.748, respectively. There was a significant difference in AUC between PSA and PCA3 score (P < 0.001), but not between PSAD and PCA3 score. Thirty-eight out of 408 men with PSA of 4–10 ng/mL had missing values of either PV or f/t PSA. In 370 available men with PSA of 4–10 ng/mL, the AUCs of PSA, f/t PSA, PV, PSAD and PCA3 score were 0.557, 0.647, 0.686, 0.692 and 0.742, respectively. There was a significant difference in AUC between f/t PSA and PCA3 score (P < 0.05), but not between PSAD and PCA3 score (Fig. 2). On univariate regression analysis, all variables had a significant association with biopsy outcome. Multivariate logistic regression analysis showed that PCA3 score (P < 0.001), PSAD (P < 0.001), PV (P < 0.01) and repeated biopsy (P < 0.01) were independent factors predicting biopsy outcome (Table 4). Totals of 21.1% (32/152), 22.9% (27/118) and 51.5% (150/291) of patients had a positive biopsy in patients with PSAD < 0.15, 0.15–0.2 and ≥0.2, respectively. The percentage of men with positive biopsy according to the combination of PSAD and PCA3 score is shown in Fig. 3. The percentage of patients with a positive biopsy increased with higher PCA3 scores in subgroups based on PSAD. Only three (4.1%) out of 72 cases with PSAD < 0.15 and PCA3 score < 20 had prostate cancer. A total of 43% had a positive biopsy in men with PSAD < 0.15 and PCA3 score ≥50. In 264 men diagnosed with prostate cancer, a total of 72, 103 and 89 cases had Gleason scores ≤ 6, 7, and ≥8, respectively. Median (range) PCA3 scores in men with Gleason scores ≤ 6, 7 and ≥8 were 45 (4–280), 51 (4–288), and 45 (1–250), respectively. There was no significant difference in PCA3 score among these three groups. A total of 88, 133, 35 and eight patients had clinical stage T1c, T2, T3, and T4, respectively. Median (range) PCA3 scores in men with clinical stage T1c, T2, T3, and T4 were 44 (6–242), 55 (4–280), 49 (1–288), and 51 (15–123), respectively. There was no significant difference in PCA3 score among the four categorical groups of clinical stage. There was a significant association between PCA3 score and % positive cores (r = 0.166, P < 0.01). Data on % positive cores was not available in two patients. A total of 164 and 98 cases had % positive cores ≤ 33, and >33, respectively. A total of 12 and 248 cases had indolent cancer and significant cancer, respectively. There was no significant difference in PCA3 score between % positive cores ≤ 33 and >33 (median PCA3 score 47 vs 58), or between indolent cancer and significant cancer (median PCA3 score 39 vs 49).

Figure 1.

Percentage of men with positive biopsy by PCA3 score range (N = 578).

Figure 2.

A, Receiver – operator curve (ROC) analysis to predict positive biopsy results (n = 561), PSA: 0.583, PV: 0.706, PSAD: 0.712, PCA3 score: 0.748. PSA vs. PCA3 score: P < 0.001, PV, PSAD vs. PCA3 score N.S. B, ROC analysis to predict positive biopsy result in patients with PSA between 4 and 10 ng/mL (n = 370), PSA: 0.557, f/t PSA: 0.647, PV: 0.686, PSAD: 0.692, PCA3 score: 0.742, PSA vs PCA3 score P < 0.001, f/t PSA vs. PCA3 score: P < 0.05, PV, PSAD vs. PCA3 score N.S.

Figure 3.

Percentage of men with positive biopsy by combination of PCA3 score range and PSAD range.

Table 1. Characteristics of patients with negative and positive biopsy results.
CharacteristicNegative biopsy, n = 369Positive biopsy, n = 264P
Median (range)Median (range)
  1. a22 cases not available. N.S., nonsignificant.
Age, years67 (42–89)71 (49–88)<0.001
PSA, ng/mL7.0 (1.4–42.6)9.0 (2.2–1908)<0.001
PVa, mL38 (9.4–130)29.1 (8.2–109)<0.001
PSAD0.18 (0.03–1.38)0.36 (0.07–80.84)<0.001
No. of cores12 (6–59)12 (6–40)N.S.
PCA3 score18 (0–381)49 (1–288)<0.001
Table 2. Positive rate of prostate cancer by serum PSA range.
PSA,ng/mLTotalInitial biopsyRepeat biopsy
nProstate cancer (%)nProstate cancer (%)nProstate cancer (%)
≤4225 (22.7)215 (23.8)10 (0)
4–10408140 (34.3)316120 (38.0)9220 (21.7)
10–2013164 (48.8)8544 (51.8)4620 (43.5)
20–504629 (63.0)2923 (79.3)176 (37.9)
>502626 (100)2525 (100)11 (100)
Total633264 (41.7)476217 (45.6)15747 (29.9)
Table 3. Sensitivity, specificity, positive and negative predictive values at different PCA3 score thresholds.
PCA3 scoreSensitivitySpecificityPositive predictive valueNegative predictive value
1094.929.244.390.6
2082.354.851.984.0
3566.571.658.178.3
5049.381.060.673.0
10018.694.867.866.3
Table 4. Univariable and multivariable logistic regression analysis for positive biopsy.
VariableUnivariableMultivariable
OR95% CIPOR95% CIP
Repeat biopsy0.6260.422–0.930<0.050.5210.319–0.849<0.01
PSA1.0581.022–1.096<0.01  N.S.
PV0.9550.942–0.968<0.0010.9780.963–0.992<0.01
PSAD60.88518.671–198.536<0.00118.8834.805–74.208<0.001.
PCA3 score1.0171.013–1.022<0.0011.0151.010–1.020<0.001

Discussion

In the present study, we investigated the ability of a PCA3 urine assay to predict the prostate biopsy outcome in a large cohort of patients from four major institutions in Japan. We found the highly informative specimen rate of 98% using a PROGENSA PCA3 assay, which verified the results of multiple studies [7-11]. We observed an increasing incidence of prostate cancer in men with a higher PCA3 score. The diagnostic performance of PCA3 urine assay for prostate cancer was excellent, with an AUC of 0.748 in Japan, compared with other reports in North America and Europe ranging from 0.66 to 0.69 [8-10, 13]. Adam et al. [14] reported that the AUC of PCA3 was 0.7054 in a South African setting consisting of 68% black men. At a PCA3 threshold of 35, which is considered to be a better balanced value between sensitivity and specificity, the specificity of the PCA3 score was 71.6% in Japanese men, compared with 72–76% in North American and European men and 50% in South African men. These results showed that the diagnostic performance of PCA3 assay in Japan was similar to that reported in different regions and ethnic groups.

Free/total PSA ratio is widely used to stratify the risk of prostate cancer in men with PSA 4–10 ng/mL, and a lower f/t PSA is more likely to be found in association with prostate cancer [4]. In the present study, we found that the diagnostic performance of PCA3 score surpassed that of f/t PSA in men with PSA 4–10 ng/mL. The AUC of the PCA3 score was highest among the variables analysed and it was significantly higher than that of f/t PSA (0.742 vs 0.647; P < 0.05). In the placebo arm of the Reduction by Dutasteride of Prostate Cancer Events (REDUCE) trial, 1140 men who had a negative baseline biopsy received a PCA3 score before repeat biopsy at 2 years [15]. In its largest repeat biopsy study to date, the AUCs of PCA3 score, PSA, and f/t PSA were 0.693, 0.612 and 0.637, respectively. A significant difference was found between PCA3 score and PSA, but the difference between PCA3 score and f/t PSA did not reach statistical significance (P = 0.065). In 463 European men with repeat biopsy, the AUC of PCA3 score was higher than that of f/t PSA (0.658 vs 0.578); however, the difference in AUCs between PCA3 score and f/t PSA did not reach statistical significance either [10]. We confirmed that PCA3 score was significantly superior to f/t PSA and PSA in predicting biopsy outcome in Japanese men with grey zone PSA levels.

Multivariable logistic regression showed that PCA3 score (P < 0.001), PSAD (P < 0.001), PV (P < 0.01) and repeated biopsy (P < 0.01) were independent predictors of positive biopsy in the present study. Several studies showed that PCA3 score was significantly cooperating with PSA and PV for predicting prostate cancer [9, 10, 13]. PSA correlates with PV in men without prostate cancer; thus, PSAD (PSA divided by PV) was used for more accurate assessment to improve the specificity in diagnosis of prostate cancer; however, the commonly used PSAD threshold of 0.15 could not detect > 40% of cancers, resulting in limited usefulness in a routine clinical setting [5]. In the present study, 21.1% of men with PSAD < 0.15 had a positive biopsy. Thus, when a PSAD threshold value of 0.15 was applied in our population, a considerable number of prostate cancer cases were missed. When the PCA3 score was combined with PSAD, we observed that the rate of positive biopsy increased, even in the subgroup of men with PSAD < 0.15 (Fig. 3). When the PCA3 score was <20 in men with PSAD < 0.15, only three (4.1%) out of 72 men had prostate cancer. By contrast, 43% of men with PCA3 score ≥50 and PSAD < 0.15 had a positive biopsy. The combination of PSAD and PCA3 score stratifies the risk of prostate cancer and predicts a low risk of prostate cancer, suggesting that these men could avoid unnecessary biopsy. It is notable that these three patients with cancer (PSAD < 0.15 and PCA3 score <20) had significant cancer with % positive cores of ≤33 and Gleason score > 6 as biopsy pathological features.

Several studies have shown the significant relationship between PCA3 score and tumour volume in prostatectomy specimens and the ability of PCA3 score to discriminate low-volume/low-grade cancer with the aim of selecting patients who are candidates for active surveillance [16, 17]. Ploussard et al. [17]reported that a high PCA3 score of >25 was an important predictor of large tumour volume with an odds ratio of 5.4 (P = 0.1) and significant cancer with an odds ratio of 12.7 (P = 0.003). In the present study, we found a significant relationship between PCA3 score and % positive cores; however, there was no difference in PCA3 score between % positive ≤ 33 and >33, or among subgroups by Gleason score and clinical stage. Further investigation of the correlation between PCA3 score and pathological outcomes on prostatectomy specimens will be needed in Japan.

In the present cohort, men with a wide range of PSA levels were enrolled. A PCA3 urine assay would be irrelevant in men with a high PSA level (the positive rate of prostate cancer was 100% in men with PSA > 50 ng/mL, and 79.3% in men with PSA 20–50 ng/mL at initial biopsy); thus, these were excluded from the analysis for prediction of biopsy outcome. Furthermore, not all men received prostate volume measurement because this was a multi-institutional study. These features might have influenced the results.

In conclusion, this Japanese multicentre study shows that the PCA3 urine assay could improve the prediction of prostate cancer and may help in selecting men who might benefit from prostate biopsy. The percentage of patients with positive biopsy increased with higher PCA3 scores. PCA3 score was superior to f/t PSA for predicting prostate cancer in patients with PSA 4–10 ng/mL. A combination of PSAD and PCA3 score may be useful for selecting patients who could avoid an unnecessary biopsy.

Acknowledgements

We wish to thank Fujirebio Inc. for their technical assistance.

Conflict of Interest

None declared.

Ancillary