Analysis of outcomes after radical prostatectomy in patients eligible for active surveillance (PRIAS)


Correspondence: Albert El Hajj, Department of Urology, CHU Mondor, 51 avenue du Maréchal de Lattre de Tassigny, 94000 Créteil, France.



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

  • Several criteria have been described to select patients with prostate cancer in active surveillance (AS) protocols; however, the risk of missing unfavourable disease remains.
  • We report the risk of misclassification using the Prostate Cancer Research International: Active Surveillance (PRIAS) study in an analysis of pathological results after radical prostatectomy. We also define predictors of favourable disease that can be used to better select patients eligible for AS, as well as risk factors associated with disease progression.


  • To identify the risk of failure of active surveillance (AS) in men who had the Prostate Cancer Research International: Active Surveillance (PRIAS) criteria and had undergone radical prostatectomy (RP), by studying as primary endpoints the risk of unfavourable disease in RP specimens (stage >T2 and/or Gleason score >6) and of biochemical progression after RP.

Patients and Methods

  • We assessed 626 patients who had the PRIAS criteria for AS defined as T1c/T2, PSA level of ≤10 ng/mL, PSA density (PSAD) of <0.2 ng/mL per mL, Gleason score of <7, and one or two positive biopsies. All patients underwent immediate RP at our department between January 1991 and December 2010.
  • Multivariate logistic regression was used to test factors correlated with the risk of unfavourable prostate cancer.
  • The risk of progression was tested using multivariate Cox regression models.
  • Biochemical recurrence-free survival (BFS) was established using the Kaplan–Meier method


  • Pathological study of RP specimens showed upstaging (>T2) in 129 patients (20.6%), upgrading (Gleason score >6) in 281 (44.9%) and unfavourable disease in 312 patients (50%).
  • There was a statistically non-significant trend for BFS at P = 0.06.
  • Predictors of favourable tumours were age <65 years (P = 0.005), one vs two positive biopsies (P = 0.01) and a biopsy core number >12 (P = 0.005).
  • Preoperative factors predicting disease progression were a PSAD of >0.15 ng/mL2 (P = 0.008) and biopsy core number of ≤12 (P = 0.017).


  • Even with stringent AS criteria, the rate of unfavourable disease remains high.
  • Predictive factors of unfavourable disease and biochemical progression should be considered when including patients in AS protocols.

active surveillance


biochemical recurrence-free survival


extracapsular extension


European Randomized Study of Screening for Prostate Cancer


hazard ratio


interquartile range


Prostate cancer Research International: Active Surveillance


PSA density


positive surgical margin


radical prostatectomy


seminal vesicle invasion


Active surveillance (AS) is gaining popularity as a treatment option for selected men with low-risk localised prostate cancer, thus reducing the risk of overtreatment of patients with insignificant disease [1, 2]. Close follow-up with serial PSA measurements and repeated biopsies reduces the risk of undersampling and allows identification of progression to higher risk disease, which could allow treatment in the ‘window of opportunity’ for cure [2].

Several criteria for patients' inclusion in AS protocols have been described; however, the risk of missing unfavourable prostate cancer remains [2, 3]. The Rotterdam section of the European Randomized Study of Screening for Prostate Cancer (ERSPC) and the Department of Urology of the Erasmus Medical Centre in Rotterdam have initiated the prospective, observational Prostate Cancer Research International: Active Surveillance (PRIAS) study to validate the management of prostate cancer with AS [4]. The latest outcomes from that study showed that repeat biopsies at a median of 1 year, led to re-classification of prostate cancer in 21.5% of patients [5]. The number of initial positive cores and higher PSA density (PSAD) were found as the main predictors of re-classification [5].

We sought to identify the risk of failure of AS protocols in men who had the PRIAS criteria [4] and have undergone radical prostatectomy (RP) at our department, by studying as primary endpoints the risk of unfavourable disease in RP specimens and of biochemical progression after RP.

Patients and Methods

From January 1991 to December 2010, we identified 2700 patients who had undergone RP for clinically localised prostate cancer in our department. After obtaining Institutional Review Board approval, patients were included based on the PRIAS study criteria: clinical stage T1c or T2 disease, PSA level of ≤10 ng/mL, Gleason score of ≤6, PSAD of <0.2 ng/mL2, and fewer than three positive biopsy cores [4]. All patients underwent clinical evaluation including DRE, serum PSA level measurement and TRUS. All biopsies and RPs were performed in the Department of Urology, and specimens were evaluated by senior uropathologists. Data from clinical evaluation, biopsy, RP specimens, and follow-up were recorded in a prospective database.

Of the patients in the database, we identified 626 (23%) patients who were eligible for AS using the PRIAS criteria. We studied the pathological findings from RP specimens, i.e. Gleason score, extracapsular extension (ECE), seminal vesicle invasion (SVI), positive surgical margins (PSMs) and PSA outcome. Unfavourable disease was defined as non-organ confined disease (pathological stage > pT2) and/or upgraded disease (Gleason score >6) in the RP specimens. We sought to identify preoperative factors that could predict favourable disease. We also studied biochemical recurrence-free survival (BFS) for favourable and unfavourable disease, and analysed the preoperative and postoperative data separately to identify predictors for disease progression in these patients.

The qualitative data were tested using the chi-square test or Fisher's exact test as appropriate, and the continuous data were tested using Student's t-test. The Mann–Whitney test was used in cases that were not normally distributed. Multivariate logistic regression was used to test factors correlated with the risk of unfavourable disease. The risk of progression was tested in multivariate Cox regression models. BFS was established using the Kaplan–Meier method. Curves were tested with the log-rank test. The limit of statistical significance was defined as P < 0.05.


In all, 626 patients with a median age of 63 years had the PRIAS study criteria for AS; 513 patients (82%) were T1c on diagnosis and 443 (70.8%) had one positive biopsy core (Table 1). The median number of total biopsy cores was 15. Pathological study of RP specimens showed: 118 pT3a (18.8%) and 11 pT3b (1.8%) (TNM 2009). Gleason score was 6 in 342 patients (54.6%), 7 in 271 (43.3%) and ≥8 in 10 patients (1.6%; Table 1).

Table 1. Clinical and pathological outcomes.
Median (IQR): 
Age, years63 (58–68)
PSA level, ng/mL5.67 (4.5–7.2)
Prostate volume, mL54 (42–66)
PSAD, ng/mL20.11 (0.08–0.14)
Clinical stage, n (%): 
T1c513 (82)
T2113 (18)
Biopsy features: 
Positive cores, n (%): 
1443 (70.8)
2183 (29.2)
Biopsy Gleason score 6, %100
Median (IQR) total number of biopsy cores15 (9–21)
N (%): 
pTNM classification: 
pT2a118 (18.8)
pT2b29 (4.6)
pT2c341 (54.4)
pT3a118 (18.8)
pT3b11 (1.8)
Prostatectomy Gleason score 
6342 (54.6)
7271 (43.3)
≥810 (1.6)
Upstaging (>T2)129 (20.6)
Upgrading (Gleason >6)281 (44.9)
Significant disease312 (50)
PSM113 (18)

In all, 113 patients had PSMs (18%), 38 of those were T3a (33.6%), two were T3b (1.7%), and 83 had unfavourable disease (73.4%).

Upstaging (>T2) was found in 129 patients (20.6%), upgrading (Gleason score >6) in 281 (44.9%) and in total 312 patients (50%) had unfavourable disease. The median follow-up was 24 months and BFS at 5 and 10 years was 93.3% and 85.6% for favourable tumours, and 87.6% and 82.8% for unfavourable tumours, respectively (P = 0.06; Fig. 1)

Figure 1.

BFS for favourable vs unfavourable prostate cancer.

Predictors of favourable disease on final pathology were patient age of <65 years (P = 0.005), one vs two positive biopsies (P = 0.01) and a biopsy core number of >12 (P = 0.005; Table 2).

Table 2. Predictors of favourable disease: multivariate logistic regression analysis.
VariablePOdds ratio (95% CI)
Age <65years0.0051.61 (1.15–2.24)
Clinical stage0.7390.93 (0.61–1.42)
PSAD0.1981.29 (0.87–1.91)
Total biopsy cores (>12)0.0051.59 (1.15–2.22)
Positive cores (one vs two)0.011.60 (1.12–2.28)

Preoperative factors predicting disease progression were a PSAD of >0.15 ng/mL2 (P = 0.008) and biopsy core number of ≤12 (P = 0.017; Table 3). Statistical analysis of postoperative pathological findings showed on univariate analysis a strong correlation of BFS with Gleason score (P = 0.01), pathological stage (P < 0.001) and PSMs (P < 0.001).

Table 3. Multivariate cox regression models analysis of preoperative and pathological predictors of biochemical recurrence.
VariablePHR (95% CI)
Positive cores (two vs onw)0.5810.83 (0.44–1.59)
Age ≥65 years0.0570.56 (0.30–1.02)
Clinical stage >T20.5710.81 (0.39–1.68)
PSAD > 0.15 ng/mL20.0082.35 (1.25–4.34)
Total biopsy cores ≤120.0172.16 (1.15–4.05)
Pathological Gleason score >60.360.85 (0.6–1.2)
Pathological stage >T20.0062.37 (1.27–4.44)
PSM<0.0013.38 (1.86–6.21)

Only PSMs (P < 0.001) and pathological stage (P = 0.006) were significant predictors of disease progression on multivariate analysis (Table 3 and Fig. 2).

Figure 2.

BFS according to tumour significance and margin status.


Several criteria have been described to detect patients with low-risk or insignificant prostate cancer who might be eligible for AS protocols [2, 6]. Epstein et al. [7] and Ohori et al. [8] defined ‘insignificant’ prostate cancer as no pattern 4 disease, no ECE or lymph node invasion, and a tumour volume of <0.5 mL. In 2004, an update of the preoperative Epstein criteria was reported in a contemporary analysis of T1c prostate cancer [9]. The updated criteria consist of a PSAD of ≤0.15 ng/mL2, Gleason score of ≤6, fewer than three positive cores and <50% of cancer involvement in any core. To date, these are the most widely used preoperative criteria for predicting insignificant prostate cancer, and more recent attempts to predict insignificant prostate cancer after positive biopsies and before surgery are based on this updated definition [10, 11]

The Epstein criteria are not perfect and misclassify ≈30% of patients who would have unfavourable pathological features in their RP specimen. This risk ranges from 16% to 42% in selected series [12]. In the study by Lee et al. [11] only 37% of men who fulfilled the preoperative Epstein biopsy criteria had insignificant prostate cancer as defined by the current definition of insignificant prostate cancer in RP specimens. Nevertheless, the authors emphasised that the preoperative contemporary Epstein criteria were highly predictive for favourable disease (58% of organ-confined and Gleason 6 prostate cancer) and cure by RP. In a recent update of the John Hopkins' AS cohort, Tosoian et al. [13] reported a 30.6% biopsy re-classification at a median follow-up of 2.7 years, the rate of biopsy re-classification was 8.9 per 100 person-years (all causes) and 4 per 100 person-years based on Gleason upgrading.

In AS protocols, studies comparing entry criteria have emphasised the risk of misclassification and the limitations of currently available AS criteria. Recent studies of repeat biopsies in men under AS have also emphasised the risk of encountering upgraded and/or upstaged disease [14]. Long-term follow-up of AS series might modify the definition of insignificant prostate cancer with the identification of prognostic factors of disease progression.

Most published AS series have used less stringent criteria than the current definition of insignificant prostate cancer. The most current AS criteria are a Gleason score of ≤6, PSA level of ≤10 ng/mL, and clinical stage T1–T2a disease and additional biological or biopsy criteria depends on institutional preferences. Other characteristics may include multiple pathological biopsy parameters, e.g. <33% of biopsy cores, fewer than two positive cores, and/or <50% of cancer involvement. This consideration argues for a more liberal definition of insignificant prostate cancer.

In the present study, we adopted the criteria of the prospective, observational PRIAS study originating from the ERSPC. Entry criteria included: T1c and T2 disease, PSA level of ≤10 ng/mL, PSAD of <0.2 ng/mL2, Gleason score of <7 and one or two positive biopsies [4, 5, 15]. The percentage of cancer invasion in a biopsy core was not investigated in the present study, because it was considered that pathological evaluation of each core was not a standard procedure in all centres. This would have been a limiting factor for participation in their protocol [4, 5]. In the preliminary PRIAS results, 82 of the 500 men enrolled in the study switched from AS to treatment at 2-years follow-up; 83% were based on protocol and 17% on patient's anxiety or choice. The pathological results were available for 24 patients who underwent RP. Four patients (16%) had ECE, 12 (50%) had a Gleason score upgrading and nine patients had PSMs (38%) [15, 16]. It is difficult to compare the present results with these findings, as only a few of the treated patients underwent RP, and this was performed after a median of 2 years AS during which time the prostate cancer might have progressed. Furthermore, the decision to operate was based on upgrading on second biopsy or PSA doubling-time. The latest published results from the PRIAS study, showed that repeat biopsy at a median of 1.03 years follow-up in 757 patients, led to a 21.5% rate of re-classification [5].

The present results showed upgrading of 44.9% and upstaging of 20.6%. There was ECE in 118 patients (18.8%) and 11 (1.8%) had SVI. Unfavourable disease was found in half the patients who had the PRIAS criteria for AS. In an analysis of the same criteria by Suardi et al. [17] there was upstaging in 7.1% and upgrading in 28.2%. These results are comparable with those for other published AS protocols [18, 19]. In a study evaluating pathological data of patients eligible for current AS protocols, upgrading varied from 23% to 35%, ECE from 7% to 19% and SVI from 2% to 9% [6].

In the present study, patients aged <65 years (P = 0.005), with a single positive biopsy (P = 0.005) and a >12-core biopsy scheme (P = 0.01) were more likely to harbour favourable disease. In the latest PRIAS results, the number of positive biopsies (two vs one) predicted the risk of re-classification as well as PSAD. However, age and total number of biopsy cores were not significant [5]. Suardi et al. [20] tested the PRIAS criteria according to age, they found that after the age of 70 years, the rate of unfavourable prostate cancer characteristics was 41% compared with 23.2% and 24.1% in patients in the previous age tertiles (ages 63 years and 63.1–69 years, respectively). A threshold of 65 years was significant in the present study (odds ratio 1.61, 95% CI 1.15–2.24, P = 0.005). A recent study by Ploussard et al. [21] reported that men eligible for AS based on a 21-core strategy have cancers with a lower extent of disease on biopsies and a lower risk of unfavourable disease on RP specimens regardless of how AS criteria are defined, compared with men eligible in a 12-core scheme. As more biopsies were taken in the patients in the present study: 15 (interquartile range [IQR] 9–21) than in PRIAS: 10 (IQR 8–12) [5], PRIAS would in theory show even higher upgrading numbers after RP.

Although half the patients in the present study had unfavourable prostatic disease this did not have an impact on BFS at 5 and 10 years. Preoperative predictors of treatment failure for patients eligible for AS were PSAD of >0.15 ng/mL2 and biopsy core number of ≤12. The John Hopkins AS protocol included in its criteria PSAD of <0.15 ng/mL2 [22], while PRIAS raised the limit to 0.2 ng/mL2. An interesting study by Kundu et al. [23], compared pathological outcomes in four PSAD groups, PSAD range 0.15–0.19 ng/mL2 compared with 0.1–0.14 ng/mL2 had increased risk of a tumour volume of >0.5 mL (33 vs 25%, P < 0.001); however, there was no significant difference in PSMs and Gleason score >7.

We also found that pathological predictors of failure after RP were upgrading, upstaging and PSMs (univariate analysis). However, on multivariate analysis Gleason score upgrade was not significant. This could be due to the fact that only 10 patients had a Gleason score of ≥8 and that only 26.4% of Gleason 7 upgrades had predominant Gleason 4 foci. However, upstaging was strongly correlated with disease progression and PSM was found to be the most important predictor of disease progression in these patients. In a recent study of biochemical recurrence in 1384 patients undergoing robot-assisted RP, Menon et al. [24] reported that on multivariate analysis, the strongest predictors of biochemical recurrence were pathological Gleason grade 8–10 (hazard ratio [HR] 5.37, 95% CI 2.99–9.65, P < 0.001) and pathological stage T3b/T4 (HR 2.71, 95% CI 1.67–4.40, P < 0.001). While another study by Barocas et al. [25] found ECE, PSM and pathological Gleason score ≥7 as independent predictors of biochemical recurrence on multivariate analysis.

The particularity of the present study is that these patients could have been candidates for AS; however, the pathological results predictive for recurrence are comparable with those of published series for candidates for radical treatment [24, 26, 27]. Nonetheless, the PRIAS criteria have selected patients for Gleason score, as although there was a 44.9% rate of upgrading, most (73.6%) were Gleason (3 + 4), which could explain our present finding that upgrading, was not significant for biochemical recurrence.

In light of these results, we consider that the studied AS criteria are not sufficient to select patients for favourable prostate cancer. A PSAD of >0.15 ng/mL2 was one of the strongest predictors of disease progression in the present study; we thus think that a PSAD of <0.15 ng/mL2 would be more selective than the 0.2 ng/mL2 value included in PRIAS. Moreover, further criteria are needed particularly percentage of positive core on biopsy [28, 29], which could provide better assessment of tumour volume than the number of positive cores alone. We recommend an extended 21-core biopsy, which confers a lower risk of unfavourable disease as previously described by Ploussard et al. [21]. Repeat biopsy before inclusion in AS has also been suggested and could reduce undersampling of prostate cancer [3]. We suggest early repeat biopsy in patients presenting predictive factors of failure of AS, as defined in the present study: two vs one positive biopsies, and diagnosis based on a standard 12-core biopsy. Although younger patients were more likely to harbour favourable disease, we recommend re-biopsy for these men who would benefit most from treatment in case aggressive disease was found. Furthermore progress in modern MRI may allow safer inclusion of patients in AS protocols [30].

We acknowledge that the present study has some limitations that warrant discussion. The short median follow-up limits the interpretation of long-term biochemical recurrence results. Selection bias is also a concern, as this is a retrospective study of patients who elected surgical treatment; it is then difficult to evaluate what the outcomes of AS would have been for these men. Another weakness of the present study is that it included patients operated on before the 2005 International Society of Urological Pathology modified Gleason score system, who might have more aggressive tumour characteristics in comparison with contemporary patients. The inclusion of the former in the study may have increased the observed risk of upstaging and/or upgrading. We have tested these variables before and after 2005, and found no difference in upstaging (18.5% before 2005 vs 22.4% after 2005, P = 0.235), but upgrading was significantly higher before 2005 (49.3% vs 40.1%, P = 0.024). This finding could also be explained by the higher median number of biopsy cores in the contemporary patients (16 vs 12, P < 0.001), which could have reduced the risk of undersampling.

In conclusion, the present study shows that even with stringent AS criteria half the present patients had unfavourable disease on final pathology. Therefore inclusion of patients in AS protocols should consider extended biopsy core sampling. Early repeat biopsy in patients with two positive cores or diagnosed based on a standard 12-core protocol is recommended to minimise the risk of misclassification. A PSAD of <0.15 ng/mL2 was a strong predictor of favourable outcome. Inclusion of these criteria could help reduce the risk of failure in patients on AS.

Conflict of Interest

None declared.