Risk of repeat biopsy and prostate cancer detection after an initial extended negative biopsy: longitudinal follow-up from a prospective trial

Authors


Abstract

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

  • Even after a negative set of prostate biopsies, the risk of undetected prostate cancer remains clinically significant. Predictive markers of such a risk are undefined.
  • In addition to PSA and PSAD, low prostate volume and %fPSA are interesting time-varying risk factors and are relevant in biopsy decision-making.

Objective

  • To assess prospectively the time-varying risk of rebiopsy and of prostate cancer (PCa) detection after an initial negative biopsy protocol.

Patients and Methods

  • Over a period of 10 years, 1995 consecutive patients with initially negative biopsies were followed.
  • Rebiopsies were performed in patients who had a persistent suspicion of PCa.
  • Predictive factors for rebiopsy and for PCa detection were tested using univariate, multivariate and time-dependent models.

Results

  • A total of 617 men (31%) underwent at least one rebiopsy after a mean follow-up of 19 months.
  • PCa detection rates during second, third, and fourth sets of biopsies were 16.7, 16.9 and 12.5%, respectively. The overall rate of detected PCa was 7.0%.
  • The 5-year rebiopsy-free and PCa-free survival rates were 65.9 and 92.5%, respectively.
  • Indications for rebiopsy were more frequently reported in patients having a high prostate-specific antigen (PSA) level (P = 0.006) or a high PSA density (PSAD; P < 0.001) and in younger patients (P = 0.008). The risk of PCa on rebiopsies was not correlated with age, but significantly increased more than twofold in cases of PSA >6 ng/mL, PSAD >0.15 ng/mL/g, free-to-total PSA ratio (%fPSA) <15, and/or prostate volume <50 mL. Time-dependent analyses were in line with these findings.
  • The main study limitation was the lack of control of the absence of PCa and PSA kinetics in men not rebiopsied.

Conclusions

  • The overall risk of detected PCa after an initial negative biopsy was low.
  • In addition to PSA and PSAD, which are well-used in rebiopsy indications, low prostate volume and %fPSA are interesting time-varying risk factors for PCa on rebiopsy and could be relevant in biopsy decision-making.
Abbreviations
PCa

prostate cancer

PSAD

PSA density

%fPSA

free-to-total PSA ratio

HGPIN

high grade prostatic intra-epithelial neoplasia

ASAP

atypical small acinar proliferation

OR

odds ratio

PSAV

PSA velocity

Introduction

Because of the widespread use of PSA level as an early detection marker and the lack of generalized targeted biopsies on imaging lesions, prostate cancer (PCa) detection is based on an indirect tissue sampling leading to an unquantifiable false-negative rate, which means a substantial number of men will undergo repeat biopsies in their lifetime for an overall PCa detection rate on rebiopsy of 15–30% [1-4]. Several studies have investigated the risk factors associated with detection of PCa on subsequent biopsies [1, 5-7]. Prognostic models and nomograms that would be useful for counselling patients in the decision to undergo repeat biopsy have also been developed [5, 8]. The aim is to identify risk factors that might reduce the number of unnecessary biopsies. Unfortunately, few series have used longitudinal data and rarely in a prospective manner [3, 4]. Logistic regression, the most frequently used model in previous studies, ignores time variations in cancer detection. Time-dependent models provide an interesting approach to longitudinal analysis of rebiopsy and PCa detection risk.

By testing clinicopathological variables as potential risk factors for rebiopsy outcomes, we aimed to estimate the probability of rebiopsy and of PCa detection for populations whose risk factor patterns vary in different ways as time elapses after a negative biopsy. Analyses were performed in our prospective single-centre cohort of consecutive patients who underwent a standardized 21-core biopsy protocol as first and subsequent biopsy schemes.

Materials and Methods

The study population included patients whose first 21-core biopsy was negative. Between December 2001 and April 2012, 1995 consecutive patients suspicious for PCa prospectively underwent an extended 21-core biopsy protocol and were found to have no PCa on this first set of biopsies. Indications for first prostate biopsies were abnormal DRE and/or a PSA level >4 ng/mL (or 3 ng/mL in patients <60 years old) and/or a free-to-total PSA ratio (%fPSA) <10%. Patients were followed by urologists from our institution. Medical visits and PSA tests were scheduled within a 6-month interval the first year after the negative biopsies, and then every year. The time and the results of rebiopsies were recorded in a prospective database. Indications for rebiopsies were: PSA level persistently elevated, increase in PSA level during follow-up, and persistent nodule on DRE. Patients having a diagnosis of high grade prostatic intra-epithelial neoplasia (HGPIN)/atypical small acinar proliferation (ASAP) on first biopsies were excluded as immediate rebiopsy was recommended for these patients because of their higher risk of PCa [2].

All patients had undergone a standardized 21-core biopsy protocol as previously described [9]. Three dedicated surgeons performed all biopsies. The 21-sample biopsy protocol included prostate ultrasonography to evaluate the prostate volume using the prolate ellipsoid formula. All patients received local anaesthesia. The biopsies were performed in the following order: (i) six sextant biopsies (standard 45° angle); (ii) three biopsies in each peripheral zone (80° angle); (iii) three biopsies in each transition zone; and (iv) three biopsies in the midline peripheral zone. Thus, the 6-core scheme included the sextant biopsies. The 12-core scheme added the six additional lateral peripheral biopsies. The 21-core scheme added the three midline cores and the six transition zone cores. Each prostate core was given a specific number according to the biopsy protocol, and thus, mapped for location. Each core was placed in separate pots and analysed separately. Two senior uropathologists reviewed the cores.

Prebiopsy clinical and biological data were abstracted from a computerized prospective database. Findings from pathological assessment included the length of cores, the number of positive cores and their location, the percent of cancer involvement in any positive core, and the biopsy Gleason score. Insignificant PCa was defined according to the Epstein criteria: PSA density (PSAD) ≤0.15 ng/mL/g; Gleason score ≤6; <3 positive cores; and <50% of cancer involvement in any core [10]. PSA, PSAD, age, %fPSA and prostate volume were assessed as continuous variables and as ordinal categories in statistical models.

The main endpoint was to study the risk of rebiopsy and of PCa detection during the follow-up. PCa detection rates for each biopsy round were assessed. Predictive factors of the risk (i) of rebiopsy; and (ii) of PCa were studied in a univariate, multivariate and time-dependent analysis. For independent variables, a Student t-test was used for quantitative variables and a chi-squared test (or a Fisher's exact test as appropriate) was used for qualitative variables. Time-dependent analyses included Kaplan–Meier analysis and the log-rank test to compare rebiopsy-free and PCa-free survivals; and a Cox regression model to assess the independent values of studied variables, and to calculate hazard ratios (HRs) with their 95% CIs. Statistical analyses were performed using SPSS software (SPSS Inc, Chicago, IL, USA). A P value <0.05 was considered to indicate statistical significance.

Results

Characteristics of the overall cohort are listed in Table 1. During the study period, 1995 patients were followed after an initial negative extended 21-core scheme (Fig. 1). The majority of patients (n = 1378) did not undergo a subsequent set of biopsies, and were followed during a mean period of 58.8 months. The overall rate of detected PCa was 7.0% (139 PCa cases).

Figure 1.

Flow diagram of the study.

Table 1. Clinicobiological characteristics of the study population (N = 1995).
Characteristic 
PSA, ng/mL 
Mean7.5
Median (range)6.0 (0.15–55.0)
PSA >10 ng/mL, %18.2
PSA >6 ng/mL, %47.4
PSAD, ng/mL/g 
Mean0.173
Median (range)0.139 (0.006–1.608)
PSAD >0.20 ng/mL/g, %21.7
PSAD >0.15 ng/mL/g, %36.5
%fPSA: 
Mean17.7
Median (range)16.0 (1.0–68.0)
%fPSA >15, %59.4
Age, years 
Mean63.2
Median (range)63.0 (33.1–85.8)
Age >60 years, %65.2
Clinical stage T2, %8.2
Prostate volume, mL 
Mean50.0
Median (range)40.0 (10.0–270.0)
Prostate volume >50 mL, %29
PCa detected 
 n (%)139 (7.0)

Thirty-one percent of men underwent at least one rebiopsy after a mean follow-up of 19 months. After excluding detected PCa cases, 32.3% of these rebiopsied patients underwent a third set of biopsies (8.3% of the overall cohort), 34.8% a fourth set (2.4% of the overall cohort), and then, 33.3% a fifth set (0.7% of the overall cohort). PCa detection rates during second, third, fourth and fifth sets of biopsies were 16.7, 16.9, 12.5 and 14.3%, respectively.

Overall, 139 PCas were detected. The overall detection rate was 22.5% in the rebiopsy subpopulation. Pathological features of these cancers are shown in Table 2. Insignificant PCa was seen in 17.3%.

Table 2. Pathological features of the PCa cases detected on repeat biopsy.
 PCa detected on repeat biopsy, n = 139
  1. aUpdated Epstein criteria: PSA 15 ng/mL/g, Gleason 6, <3 positives cores, <50% cancer involvement per core.
Biopsy Gleason score, n (%) 
6107 (77.0)
722 (15.8)
87 (5.0)
93 (2.2)
Biopsy Gleason score >6, n (%)32 (23.0)
Primary Gleason pattern 4, n (%)20 (14.4)
No. of positive cores 
Mean2.39
Median (range)2.0 (1–14)
Core involvement, % 
Mean20.6
Median (range)14.0 (1–95)
Insignificant PCaa, %24 (17.3)
Core length, mm 
Mean11.96
Median (range)11.95 (7.0–19.0)

The 2- and 5-year rebiopsy-free survival rates were 75.3 and 65.9%, respectively. In the overall cohort, the 2- and 5-year PCa-free survival rates were 95.1 and 92.5%, respectively. Among the 617 men who underwent at least one rebiopsy, the 2- and 5-year PCa-free survival rates were 84.9% and 78.1%, respectively.

Predictive factors for the need for rebiopsy were studied in Table 3. Indications of rebiopsy were more frequently reported in patients with a higher PSA level (P = 0.006) and a higher PSAD (P < 0.001) and in younger patients (P = 0.008). After an initial negative set of biopsies, the risk of undergoing at least a second set of biopsies was increased 1.4-fold or 1.6-fold in patients with PSA >6 ng/mL (P < 0.001) or a PSAD >0.20 ng/mL/g (P < 0.001), respectively. An initial biopsy performed before the age of 60 years increased this risk 1.33-fold (odds ratio [OR] 0.75; P = 0.004). The rebiopsy-free survival rates were significantly different after stratification by PSA level, PSAD and patient age. Survival curves are shown in Fig. 3. No single factor was independently predictive of rebiopsy-free survival in a Cox multivariate regression model.

Table 3. Predictive factors for the risk of rebiopsy: univariate, multivariate (logistic regression), and time-dependent analyses (log-rank test, Cox regression).
 RebiopsyUnivariate analysisLog-rank testMultivariate Cox regression
No (n = 1378)Yes (n = 617)PORPP
Mean PSA level7.27.90.006   
PSA >10 ng/mL, %17.120.60.064 0.038 
PSA >6 ng/mL, %44.753.3<0.0011.41<0.0010.633
Mean PSAD ng/mL/g0.1290.166<0.001   
PSAD >0.20 ng/mL/g, %19.227.2<0.0011.58<0.0010.097
PSAD >0.15 ng/mL/g, %32.346.0<0.0011.79<0.001 
Mean %PSA18.017.10.223   
%PSA >15, %60.557.40.418 0.5620.985
Mean age, years63.362.40.008   
Age >60 years, %67.260.60.0040.750.0010.217
Clinical stage T2, %7.99.00.383 0.5420.153
Mean prostate volume, mL41.641.70.982   
Prostate volume >50 mL,%29.527.40.335 0.7170.713

Predictive factors for PCa detection were studied in Table 4. PCa was more frequently detected in patients with a higher PSA level (P < 0.001), higher PSAD (P < 0.001), a smaller prostate volume (P < 0.001) and a lower %fPSA (P = 0.014). Thus, at the time of the initial set of biopsies, the risk of PCa on rebiopsies was increased more than twofold in patients with PSA >6 ng/mL, PSAD >0.15 ng/mL/g, a %fPSA <15, and/or a prostate volume <50 mL. All these statistical differences remained significant in time-dependent univariate analyses (log-rank tests, Fig. 2). HR estimates from Cox regression models are listed in Table 5. PSA and PSAD were positively and significantly associated with risk of PCa. %fPSA and prostate volume were inversely and significantly correlated with PCa risk. Age did not reach statistical significance. In Cox multivariate regression analysis, taking into account these factors, PSAD >20 ng/mL/g was the most predictive independent factor for PCa detection on rebiopsies (HR 2.34; P = 0.012; Table 4, Fig. 3). The 5-year PCa-free survival rate was 97.1% in patients with PSAD <0.10 ng/mL/g compared with 86.1% in patients with PSAD >0.20 ng/mL/g (P < 0.001).

Figure 2.

Left column: rebiopsy-free survival curves stratified by prostate volume (log-rank: P = 0.717), %fPSA (log-rank: P = 0.562), and age (log-rank: P = 0.001). Right column: PCa-free survival curves stratified by prostate volume (log-rank: P < 0.001), %fPSA (log-rank: P = 0.004), and age (log-rank: P = 0.809). PSAD and PSA were significantly associated with rebiopsy-free and PCa-free survival rates.

Figure 3.

PCa-free survival curves stratified by PSAD using four different thresholds: <0.10, 0.10–0.15, 0.15–0.20 and >0.20 ng/mL/g. The 5-year PCa-free survival rate was 97.1% in patients with PSAD <0.10 ng/mL/g compared with 86.1% in patients with PSAD >0.20 ng/mL/g (P < 0.001).

Table 4. Predictive factors for the risk of PCa detection: univariate, multivariate (logistic regression) and time-dependent analyses (log-rank test, Cox regression).
 PCaUnivariate analysisLog-rank testMultivariate Cox regression
No (n = 1856)Yes (n = 139)PORPPHR
Mean PSA level, ng/mL,7.39.4<0.001    
PSA >10 ng/mL, %17.7%25.2%0.0271.570.0330.705 
PSA >6 ng/mL, %46.1%64.0%<0.0012.08<0.001  
Mean PSAD ng/mL/g0.1350.216<0.001    
PSAD >0.20 ng/mL/g, %20.340.3<0.0012.66<0.001  
PSAD >0.15 ng/mL/g, %34.958.3<0.0012.60<0.0010.0122.34
Mean %PSA17.915.20.014    
%PSA >15, %61.142.20.0030.470.0040.063 
Mean PSAV, mL0.1010.2870.813    
>0.75 mL, %34.333.10.797 0.4690.701 
Mean age63.064.00.150    
Age >60 years, %65.066.90.655 0.8090.844 
Clinical T2:8.28.70.813 0.9200.867 
Mean prostate volume, mL42.234.5<0.001    
Prostate volume >50 mL30.013.3<0.0010.36<0.0010.082 
Table 5. Hazard ratio estimates and 95% CIs from a Cox regression model: endpoint of time to detection of PCa.
 PHR95% CI
PSA, ng/mL   
<4 1 
4–60.0722.080.94–4.63
6–100.0023.461.58–7.61
>100.0023.561.58–8.02
PSAD, ng/mL/g   
<0.10 1 
0.10–0.150.0092.521.25–5.06
0.15–0.200.0033.051.48–6.28
>0.20<0.0015.632.95–10.75
%fPSA, %   
<15% 1 
15–25%0.0090.490.29–0.84
>25%0.1120.500.21–1.18
Age   
<60 years 1 
60–65 years0.7090.9160.58–1.45
65–70 years0.3211.250.81–1.93
>70 years0.9630.990.62–1.59
Prostate volume   
>70 mL 1 
50–70 mL0.1751.850.76–4.50
30–50 mL<0.0014.482.07–9.71
<30 mL0.0053.641.47–9.02

To evaluate possible selection bias resulting from urologists’ indications for rebiopsy, an additional model that included only participants who underwent at least one rebiopsy was constructed. This model gave similar results. PSA velocity (PSAV) was also tested and was not associated with the risk of PCa (0.29 vs 0.10 ng/mL/year; P = 0.813; Table 4).

Table 6 shows comparisons between pathological features of PCa detected on the first set of rebiopsies and those detected on the subsequent ones. There were trends towards more favourable features for PCa detected on the subsequent sets of biopsies; however, statistical differences were not significant.

Table 6. Comparisons of pathological features between PCa detected on first rebiopsy or on subsequent rebiopsies (third round or more).
PCa casesOn first rebiopsyOn subsequent rebiopsiesP
n = 103n = 36
Median PSA, ng/mL8.58.10.380
Median age, years66.965.60.419
Median %PSA13.015.00.633
Median PSAD, ng/mL/g:0.2330.2480.625
Biopsy Gleason score >6, %21.926.50.583
Primary Gleason pattern 4, %14.314.70.952
Median no. of positive cores2.01.00.144
Median % of core involvement14.210.50.449

Discussion

When a patient has an initial negative biopsy and there are persistent clinical suspicions of cancer from DRE or PSA testing, or suspicious pathological findings in initial biopsy specimens, at least one set of repeat biopsies can be warranted [11]. Since it is a fact that in any technique of initial and repeat biopsy there is a significant sampling error, we experience a ‘repeat biopsy dilemma’ in a large number of patients presenting as clinically suspicious for cancer even after multiple sessions of negative biopsies. The identification of PCa detection risk factors after an initial negative set of biopsies appears clinically relevant as it might reduce the number of unnecessary biopsies, and thus, avoid cost and psychological distress [12, 13]. Several studies have investigated the risk factors associated with detection of PCa on subsequent biopsies [1, 5-7]. The clinical variables generally studied are age, a suspicious DRE, PSA, %fPSA, PCAV and PSAD, but few studies have assessed survival analyses that enabled consideration of time-varying risk profiles [3, 4]. Logistic regression, which has been the most frequently used model in previous studies, ignores time variations in cancer detection. Recently, Gann et al. [3] highlighted this limitation and raised interest in longitudinal analysis of PCa risk after an initial negative biopsy.

Since 2001, the present authors have performed a 21-core biopsy protocol on every patient referred to us for PCa suspicion [2, 9]. Characteristics of each patient and follow-up data were recorded to allow prospective survival analyses. Interestingly, we found that predictive factors for rebiopsy were not systematically similar to predictive factors for PCa detection. Urologists tended more frequently to rebiopsy young patients who were not at higher risk of PCa. In addition to PSA and PSAD, which are commonly used in rebiopsy indications, our findings highlighted that low prostate volume and %fPSA were better indicators of PCa on rebiopsy and might be more relevant in biopsy decision-making. The risk of PCa on rebiopsy was significantly higher (more than twofold higher) in patients with PSA >6 ng/mL, PSAD >0.15 ng/mL/g, %fPSA <15, and/or prostate volume <50 mL. These factors can identify subgroups with unusually low risk of PCa after a negative set of biopsies and can help physicians by avoiding unnecessary biopsies. To illustrate, a patient with a gland >50 mL and a PSAD <0.10 ng/L/g has a predicted probability of PCa detection of ∼ 2% 5 years after initial biopsies compared with a sevenfold increased risk (14%) in a patient with a PSAD >0.20 ng/mL/g and a gland <50 mL. PSAV was not a significant predictor for PCa in our analysis. The predictive value of PSAV for PCa detection on repeat biopsy remains debatable in the literature [3, 14-17]. Auprich et al. [16] also found that %fPSA performed best at second and subsequent repeat biopsies and that the diagnostic potential of PSAV might apply only for patients at the third or more repeat biopsy. In their prospective longitudinal analysis, Gann et al. [3] showed that PSAV was not associated with PCa risk unless PSA was removed from the model, in line with the present findings.

The ideal goal of PCa detection should not be a simple increase in the positive biopsy rate, but the detection of all clinically significant cancers without the over-detection of clinically insignificant ones. Even if the subsequent rounds of biopsies increase the number of detected PCa cases, it is important to know if such a detection improvement is clinically relevant. The counterpart is that the repeat biopsy strategies have been hypothesized to increase the potential risk of over-treating patients whose tumours induce very low risk to life [18, 19]. In the present study, only 17% of insignificant PCa as defined by the Epstein criteria were reported and ∼25% of detected PCa had a Gleason score 7–9.

The main limitation of the present study was lack of absence of PCa and of PSAV in non-rebiopsied men as a control. This introduced a verification bias as we cannot be sure that all patients who did not undergo a rebiopsy and all patients with subsequently negative rebiopsies were cancer-free, but this limitation could not be overcome. PSAV was not recorded in non-rebiopsied men so we could not study this variable as a potential risk factor of performing repeat biopsy. Vickers et al. [17], however, found no evidence to support the recommendation that men with high PSAV should be biopsied in the absence of other indications. We were unable to include some potentially interesting factors such as family history or race. We also chose to exclude patients having a diagnosis of HGPIN/ASAP on first biopsies as most studies have shown that HGPIN/ASAP might be significantly associated with a higher rate of PCa on rebiopsy [2, 4, 20-22]. Currently, although a finding of single-focus HGPIN does not warrant repeat biopsy, multifocal HGPIN indicates a twofold risk of positive biopsy on delayed interval biopsy within 3 years, and ASAP in previous biopsy pathology is an almost certain clinical indication for repeat biopsy within 3–6 months because ∼40% of patients with ASAP are found to have cancer that was not identified during initial biopsy.

Another limitation is that we used a 21-core biopsy protocol from the first set of biopsies. Conclusions from our patient population may not be generalized to a population typically encountered by urologists in practice and subject to a standard 12-core biopsy scheme. In a previous study comparing the results of rebiopsy between men who underwent a first standard 10–12-core scheme and men who underwent a 21-core protocol, we reported that an initial saturation sampling has a stronger negative predictive value than standard procedures [23]. Moreover, we found that the main risk factors of cancer on repeat procedure were equivalent between the two groups. Thus, we hope that our present findings might be generalizable to all patients who are candidates for initial biopsies, regardless of the number of cores sampled.

In conclusion, determining the patient risk for PCa after an initial negative biopsy has gained clinical relevance with regard to reducing the number of unnecessary biopsies. Our time-dependent analyses could estimate the probability of rebiopsy and of cancer detection for subgroups of patients whose risk factors vary as time elapses. After an initial negative extended 21-core biopsy protocol, the risk of rebiopsy increases over time to reach 25 and 35%, 2 and 5 years after the first procedure. The PCa detection rate remains stable during the two first consecutive rebiopsies with a rate of ∼17%. Urologists tend to rebiopsy young patients, who are not at higher risk of PCa, more frequently. In addition to PSA and PSAD, which are commonly used in rebiopsy indications, low prostate volume and %fPSA are better indicators of PCa on rebiopsy and might be more relevant in biopsy decision-making.

Conflict of Interest

No financial disclosure. No funding sources.

Ancillary