- Top of page
- SUBJECTS AND METHODS
- CONFLICT OF INTEREST
Prostate biopsy has a significant false-negative cancer-detection rate [1,2]. Men with an initial negative biopsy are often suspected to harbour ‘missed’ prostate cancer. The question of how to appropriately follow these men and when, and for whom, to recommend repeat biopsy, have not been totally resolved [2–4]. Decisions for repeat biopsy have been based on many factors, including the likelihood that a repeat biopsy will be positive, and the characteristics of the cancers that are discovered on repeat biopsy [2–4].
There are few reports from the USA describing the rates at which men undergo repeat biopsy or the factors that clinicians and patients use when selecting repeat biopsy, especially in a large, representative population. The Prostate, Lung, Colorectal, and Ovarian (PLCO) cancer screening trial is a multicentre, randomized trial designed to evaluate the effectiveness of screening in reducing cause-specific mortality . The cohort of men enrolled in the screening arm of the PLCO was previously examined for their initial biopsy after a positive screen . We now analyse the experience of repeat biopsy in men from this cohort whose initial biopsy was negative. This cohort had annual prostate screening through the PLCO, but the follow-up of positive screen results was at the discretion of the subject’s healthcare provider and was not carried out under the auspices or direction of PLCO. Thus, the experience of this cohort for repeat biopsy might be generally representative of practice patterns in the USA.
We describe the rates and patterns of repeat biopsy (second to fourth) and factors associated with the likelihood of repeat biopsy in men with initially a negative biopsy taken for a suspicious PSA level and/or a suspicious DRE, and present the positivity rates at the second to fourth biopsy.
SUBJECTS AND METHODS
- Top of page
- SUBJECTS AND METHODS
- CONFLICT OF INTEREST
The design and baseline results of the PLCO trial were reported previously [5–7]. Briefly, men and women aged 55–74 years were randomized between November 1993 and July 2001, with almost 155 000 persons randomized. Exclusion criteria for men in the PLCO included a history of prostate, lung, or colorectal cancer, surgical removal of the entire prostate, having taken finasteride in the previous 6 months and, starting in 1995, having had more than one PSA blood test in the previous 3 years. At the time of randomization, subjects completed a self-administered demographic and medical/screening history questionnaire. Men in the screening arm had their PSA level tested and a DRE at baseline (year 0) and then annually to the third year, and had a PSA test with no DRE at 4 and 5 years. A PSA level of >4 ng/mL or a finding of nodularity or induration on DRE was considered a positive screen. Men with previous positive screens were invited back for further screening as long as they did not have a diagnosis of prostate cancer.
Men were referred to their private physicians or health plans for the follow-up of positive screening tests; the PLCO trial protocol did not recommend a specific diagnostic algorithm. The PLCO obtained medical records related to the diagnostic follow-up of positive screens and trained medical-record abstractors recorded information on diagnostic procedures. Follow-up diagnostic PSA test results were recorded, but the results of follow-up diagnostic DREs were not generally recorded. Cancer diagnoses, and findings of high-grade prostatic intraepithelial neoplasia (PIN) on biopsy, were also recorded. Information on the number of biopsy cores was not collected.
The cohort used in this analysis consisted of all men with an initial negative biopsy after a positive PSA or DRE screen at study years 0–3 through to the end of 2002. Men were censored at 3 years after their first biopsy, the end of the screening phase of PLCO (6 years after study entry), death or a diagnosis of prostate cancer, or on 31 December 2003, whichever came first.
The Kaplan–Meier method was used to estimate the probability of having a second biopsy , and to estimate the probability of a third or fourth biopsy, conditional on having had a negative second or third biopsy. Rates of repeat biopsy were calculated as the total number of repeat biopsies (including third and fourth biopsies) divided by the total number of person-years at risk of repeat biopsy.
We analysed risk factors for repeat biopsy during two periods of follow-up, i.e. the year directly after the initial biopsy and before any repeat PSA or DRE, and the 3 years after the repeat PSA test or DRE. During the first period, the risk of repeat biopsy was analysed as a function of the PSA level and DRE result at the screening examination just before the initial biopsy, and baseline covariates. During the second period, the risk of repeat biopsy was analysed as a function of the first repeat DRE, the first repeat PSA level (whether diagnostic or screening), and the PSA velocity (calculated from the first repeat PSA test and the immediately previous PSA test), in addition to baseline covariates. Baseline covariates included: age, family history of prostate cancer, history of prostate problems, previous prostate biopsy (before PLCO), presence of high-grade PIN at initial biopsy, and calendar and study year of first biopsy. Separate models were run for men with elevated screening PSA levels and men with a positive DRE but normal PSA levels at the time of first biopsy
- Top of page
- SUBJECTS AND METHODS
- CONFLICT OF INTEREST
Table 1 describes the cohort used in this analysis; it was 90% non-Hispanic White, with a mean (range) age of 65 (55–79) years. In all, 63% of men were referred for initial biopsy due to a positive (i.e. elevated) PSA level, while the rest (37%) were referred due to a positive DRE (and normal PSA level).
Table 1. The cohort in the PLCO study at risk of a repeat biopsy
| White (non-Hispanic), %|| 90|
| Black, %|| 4|
|Mean (range) age at 1st biopsy, years|| 65 (55–79)|
|Previous prostate biopsy (before PLCO), %|| 12|
|Family history of prostate cancer, %|| 8|
|History of prostate problems, %|| 38|
|Study year of first biopsy, %|
| 0|| 36|
| 1–3|| 64|
|Calendar year of first biopsy, %|
| 1993–97|| 45|
| 1998–99|| 32|
| 2000–2002|| 23|
|Screen result (% with PSA level, ng/mL) at year of 1st biopsy|
| >4|| 63|
| ≤4, + DRE|| 37|
Table 2 shows the observed rates of repeat biopsy by reason for referral for the initial biopsy. For men with an (initial) positive PSA level, the probability of having a second biopsy within 3 years of the first biopsy was 43%. The probabilities of a third or fourth biopsy within 3 years, conditional on having had a previous negative biopsy, were similar, at 44% and 43%, respectively. The men with an (initial) positive DRE had significantly lower probabilities of second (13%) and third biopsies (22%) than the men with a positive PSA test. The rate of repeat biopsy (per 100 person-years) for the 3-year period after the initial biopsy was 20.3 for men with positive PSA test and 5.4 for men with a positive DRE (P < 0.001; Table 2). The median interval between the first and second biopsies was 1.1 years for PSA-positive men and 1.0 year for DRE-positive men; the median intervals were slightly shorter between the second and third biopsies. During the 3-year period after the initial biopsy, the ultimate outcome is categorized in Table 2 as follows: no further biopsy, second biopsy positive, second biopsy negative with no further biopsy, and second biopsy negative with a third biopsy. Among men with a second biopsy, biopsy positivity rates were 23% in men with a positive PSA test (724) and 17% in men with a positive DRE (130). Positivity rates on the third biopsy were 21% for PSA positive men (248) and 17% (24) for DRE-positive men, while PSA-positive men had a positivity rate of 25% (65) on fourth biopsy (only five DRE-positive men had a fourth biopsy, with one being positive).
Table 2. Probabilities and rates of repeat biopsy
|Variable||Reason for referral for initial biopsy|
|PSA >4||DRE + PSA ≤ 4|
|Mean percentage (sd) probability of:|
| 2nd within 3 years of 1st||43 (1.2)||13 (1.1)*|
| 3rd within 3 years of 2nd†||44 (2.2)||22 (4.1)*|
| 4th within 3 years of 3rd†||43 (4.7)||33 (12.8)|
|Mean rate (sd):|
| Repeat rate (per 100 person-years)|
| 3-year period after 1st‡||20.3 (0.7)|| 5.4 (0.4)*|
|Median (IQR) interval, years between:|
| 1st and 2nd¶|| 1.1 (0.6–1.9)|| 1.0 (0.3–1.9)|
| 2nd and 3rd¶|| 1.0 (0.6–1.6)|| 0.9 (0.4–1.2)|
|Ultimate outcome (3-year period after 1st), %|
| No further biopsies||58||87|
| 2nd positive||10|| 2.1|
| 2nd negative, no 3rd||22|| 8.8|
| 2nd negative, 3rd||10|| 1.8|
Men took various paths to the first repeat biopsy, as shown in Fig. 1 (positive PSA) and 1B (positive DRE). In each group, relatively few men (3–5%) proceeded directly to biopsy; most (61–76%) proceeded directly to the next screen, while 14–28% had repeat PSA or DRE tests outside the study (6–7% received no further diagnostic follow-up or screening). About a fifth (19%) of the men having a repeat PSA outside the study had a repeat biopsy before the next PLCO screen, compared with only 5% of men having a repeat DRE outside of the study. After the next PLCO screen, 38% of men with a positive PSA test and 9% of men with a positive DRE had a repeat biopsy within 2 years.
Figure 1. Paths after initial biopsy for men with a positive PSA test (A) and positive DRE (B) at the first biopsy. Men can proceed directly to repeat biopsy (top), to diagnostic PSA/DRE before the next screen (middle), or to the next PLCO screen (bottom). Men with a diagnostic PSA/DRE can proceed to biopsy or to next PLCO screen, and men with the next PLCO screen can proceed (within 2 years) to biopsy. Percentages are based on numbers in the previous boxes. Percentages from the first biopsy do not add to 100% as some men had no further diagnostic follow-up or screening.
Download figure to PowerPoint
Table 3 shows the results of the proportional-hazards modelling examining risk factors for a first repeat biopsy. In men with a positive PSA level, during the period before any subsequent PSA test, high-grade PIN on initial biopsy significantly increased the probability of a repeat biopsy (hazard ratio, HR, 4.8). After the first subsequent PSA test, both the PSA level (HR 0.3, 1.0, 1.5, and 1.8 for PSA < 4, 4–7, 7–10 and >10 ng/mL, respectively) and PSA velocity (HR 1.5 for >1.0 ng/mL/year) were significantly associated with a repeat biopsy. Men aged >65 years had a modest but statistically significantly lower probability of biopsy (HR 0.7). High-grade PIN was not significantly associated with repeat biopsy among these men.
Table 3. Risk factors for a first repeat biopsy, from the proportional hazards models
|HR (P) for factor||Period before any subsequent PSA/DRE||Period after subsequent PSA/DRE|
|PSA-positive men, PSA level, ng/mL|
|7–10*||1.2 (0.6)||1.5 (<0.001)|
|>10*||0.9 (0.9)||1.8 (<0.001)|
|PSA velocity >1 ng/mL/year*||–||1.5 (<0.001)|
|+ DRE*||0.8 (0.7)||1.2 (0.2)|
|High-grade PIN†||4.8 (<0.001)||1.3 (0.1)|
|Baseline study year†||1.1 (0.7)||1.1 (0.1)|
|Calendar year 93–97†||0.95 (0.9)||1.1 (0.5)|
|Age >65 years†||1.4 (0.2)||0.7 (<0.001)|
|+ DRE*||–||3.8 (<0.001)|
|PSA >2.5 ng/mL*||2.3 (0.04)||2.3 (<0.001)|
|PSA velocity >1 ng/mL/year*||–||2.3 (0.001)|
|High-grade PIN†||6.0 (<0.001)||2.4 (0.01)|
|Baseline study year†||0.9 (0.8)||2.1 (<0.001)|
|Calendar year 93–97†||0.3 (0.03)||1.4 (0.2)|
|Age >65 years†||2.0 (0.1)||1.1 (0.8)|
For men with a positive DRE, during the period before any subsequent PSA or DRE determination, high-grade PIN at initial biopsy (HR 6.0) and a screening PSA level of 2.5–4 ng/mL at the time of the first biopsy (HR 2.3) were significantly associated with a repeat biopsy. Men with an initial biopsy in 1993–97 were less likely to have a direct repeat biopsy than men with an initial biopsy in 1998–2002 (HR 0.3). After a subsequent PSA test or DRE, a PSA level of >2.5 ng/mL (HR 2.3), a PSA velocity of >1 (HR 2.3), a positive DRE (HR 3.8), high-grade PIN (HR 2.4) and baseline study year (HR 2.1) were all significantly associated with higher rates of repeat biopsy.
- Top of page
- SUBJECTS AND METHODS
- CONFLICT OF INTEREST
The optimum management of men with suspicious results from prostate cancer screening tests (PSA, DRE) and an initial negative biopsy has not been determined. We sought to use data from the PLCO trial to estimate the rates of repeat biopsy and to identify factors associated with the decision to take a repeat biopsy in men in a large, multicentre prostate cancer screening trial.
The repeat biopsy rates over a 3-year period after an initial negative biopsy were 43% in men with an elevated PSA level and 13% in men with an abnormal DRE. Rates of third or fourth biopsy, among those with a previous second or third negative biopsy, were comparable or higher. Most men in this cohort undergoing biopsy did so only after an additional round of screening. As the diagnostic follow-up in the PLCO was the responsibility of the subject’s personal healthcare providers (primary and urological) and was not under the control of PLCO, these results should be generally representative of the experience of men in the USA who are undergoing regular prostate screening. Although the PLCO cohort does not represent a true national sample, the men in this cohort were evaluated at hundreds of different medical practices in nine metropolitan areas and one rural area across the USA. Repeat biopsy rates showed relatively little variation across the 10 screening centres. For PSA-positive men, centre-specific 3-year repeat biopsy rates were 31–54%, with eight of 10 centres having rates of 36–50%. For DRE-positive men, nine of 10 centres had 3-year repeat biopsy rates of 10–16%; one small centre had a rate of 7% based on only 30 subjects.
The decision to recommend a repeat biopsy should depend, in part, on the likelihood that the biopsy will be positive (i.e. that the first biopsy missed cancer). Therefore, factors predictive of having a repeat biopsy should parallel those that are thought to predict a positive biopsy. Previous studies have found that PSA level, PSA velocity, free PSA, and/or PSA density, and the presence high-grade PIN at initial biopsy, were associated with cancer on repeat biopsy [9–11]. We found that PSA and PSA velocity, as determined after the initial biopsy, were independently associated with the probability of having a repeat biopsy in both PSA-positive and DRE-positive men. In DRE-positive men a repeat positive DRE at the next screen was also significantly associated with repeat biopsy. In both PSA- and DRE-positive men, high-grade PIN was associated with repeat biopsy.
Previous studies on this same cohort examined the screen positivity rate and the rate of first biopsy among screen-positive men [6,7]. For example, these studies reported that ≈8% of men had elevated PSA levels at baseline and that the 3-year probability of having an initial biopsy in these men was 64%, while ≈6% had (only) a positive DRE at baseline and 27% of these had a biopsy within 3 years. Combining the data from these analyses with those presented here, we calculate that among all men undergoing the baseline prostate screen, the rate of prostate biopsy per 100 person-years during the screening period of the trial (the first 6 years) was 3.5. Biopsy rates were 20.2 and 6.8 for men with a positive PSA and a positive DRE screen at baseline, respectively. This can be compared with the rates of repeat biopsy shown in Table 2 of 20.3 and 5.4 among men with a positive PSA and positive DRE, respectively, and initial negative biopsy. In all, 23% of all biopsies taken during the screening period were repeat biopsies.
There is relatively little reported on repeat biopsy rates in populations of men being screened for prostate cancer. Eggener et al. reported on the experience of repeat biopsy in men involved in a prostate cancer screening study in the area of St. Louis, MO, USA. A cohort of 1011 men with a PSA level of 2.6–4.0 ng/mL and a negative initial biopsy were followed for a mean of 6 years; 51% had a repeat biopsy, and 27% of those men were diagnosed with cancer. Nam et al. studied 1500 men in two Ontario clinics who were initially referred for biopsy because of a PSA level of >4 ng/mL or an abnormal DRE. Among men with a negative initial biopsy, they reported a repeat biopsy rate of 32.5% over a year. An initial PSA level of 10–20 ng/mL and >20 ng/mL were associated with significantly greater risks of repeat biopsy, with odds ratios of 3.6 and 4.5, respectively, compared to an initial PSA of <10 ng/mL. In addition, men with high-grade PIN or atypia on initial biopsy had a greater risk of repeat biopsy (odds ratio 3.2). Naya et al. found that 24% of 720 men with an initial negative biopsy had a repeat biopsy within a year; it was not stated whether these men, whose median initial PSA level was ≈6 ng/mL, were having regular prostate screening.
A limitation of the present study is that we did not collect data on the number of cores taken at each biopsy, so we cannot analyse this as a factor affecting the probability of a repeat biopsy. In addition, biopsies in the absence of a reported cancer diagnosis were generally tracked only during the 1-year period after a positive screen, so some (negative) biopsies outside this period might have been missed. Based on the number of positive biopsies reported outside this period and the observed rate of repeat biopsy positivity, we estimate that ≈10% of the repeat biopsies could have been missed, so that the rates reported here might underestimate the true rates by ≈10%.
In conclusion, factors associated with an increased likelihood of having a repeat prostate biopsy after a positive screen and initial negative biopsy in the PLCO cancer screening trial include an elevated total PSA level, a PSA velocity of >1 ng/mL/year, DRE positivity and the presence of high-grade PIN at initial biopsy. Because the follow-up was at the discretion of the patient’s physician and urologist, and was not determined by a trial algorithm, these data might be generally applicable to community practice in the USA for repeat prostate biopsy in men undergoing periodic PSA-based prostate cancer screening.